1 DEVELOPMENT OF COLLAGEN-HYD ROXYAPATITE NANOSTRUCTURED COMPOSITES VIA A CALCIUM PHOSPHATE PRECURSOR MECHANISM By SANG SOO JEE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008
2 2008 Sang Soo Jee
3 To my family and my lab members
4 ACKNOWLEDGMENTS Most of all, I really appreci ate the strong support from my ad visor, Dr. Gower. She guided my research and also provided a financial and emotional support for me and my family. In particular, I really appreciate her precious guidance in wri ting well-organized papers and proposals. I must mention the considerable help from my committee members. Dr. Douglas, he gave me a lot of scientific comments for my work and strong backgrounds of organic materials I studied in my Ph. D work. Dr. Mecholsky also provided the proper attitude toward the world of science and I really enjo yed his comments and corrections on my proposal. I took two classes of Dr. Sigmund which gave me an idea for interpreting the interfacial condition of precursor, and he also gave me a lot of comments on my zeta-poten tial measurement works which were great help to understand my suspicious precursor system Dr. Long also showed me how I can make a connection between materials en gineering and biology. I also appreciate great many advices from Dr. Cheol-young Kim in Inha university. Many thanks must be given to my family in Korea for their devoted emotional support. My parent kept calling me to encourage me and my younger brother and sister were always been on my side. For my wife, there may be no way to say thank you in written words. She is my best supporter giving me an enormous energy wh ich keeps me working and living. My daughter, Yoonjung, she is always my energy booster during my graduate school period. Lastly, I have to say great thanks to my co lleagues in the biomimetics lab; Rajendra (Dr. Kasinath) who provided great help and a lot of co mments for my work at the beginning of my work and still giving me various ideas, Matt (Dr. Olszta) who gave me many instructions for my project, Xingguo (Dr. Chen) who is my most favorite Chinese scie ntist, Sara w ho did a lot of works for this dissertation and gave me bright ideas and comments for my work, Mark who is my best American buddy even thou gh his attitude to politics is to tally different with mine, One
5 of my best friend in US, Fairland (Dr. Amos) who taught me a lot about biomineralization, Chiwei who gave me great comments for my wor k, Taili (Dr. Thula) who gave me a great impression on biomedical approach for biomaterials. I also have to say thanks to many Korean colleagues in UF. Seoung-Woo (Dr. Lee), he showed me a model of graduate student and family man and also helped me from the first day of my USA life. Sung-Huan (Dr. Lee), he also gave me great helps from the beginning of my graduate school period. Yi-Yeon (Dr. Kim), who was my senior in biomimetics lab, she taught me how to survive in the biomimetics lab with great humors. Besides them, I really thanks to all Korean friends who made my best year (20042005) in UF (Woo-Chul Kwak, Hyun-Sik Kim, Joo-Ro Kim, Jong-Dae Lee, Tae-Gon Kim, Y un-Jae Moon, Joo-Hee Mo on, Do-Won Jung). And, I hope every Korean student in UF, especially Young-mee Kim, Do-sung Son, Dong-Jo Oh and Sang-Hyun Eom, will do their best and achieve what they want.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........9 LIST OF FIGURES.......................................................................................................................10 ABSTRACT...................................................................................................................................15 CAHP TER 1 INTRODUCTION..................................................................................................................17 2 BACKGROUND ....................................................................................................................25 Bone: From a Stru ctural Standpoint ....................................................................................... 25 Components of Bone.......................................................................................................25 Collagen...................................................................................................................26 Hydroxyapatite.........................................................................................................27 Structure of Secondary Bone........................................................................................... 28 Mechanism of Bone Formation.............................................................................................. 30 Bone Formation: From a Biological Standpoint ............................................................. 31 Bone Formation: From a Materials Science Standpoint.................................................. 34 Bone Grafts.............................................................................................................................38 First Generation of Bone Grafts......................................................................................39 Second Generation of Bone Grafts..................................................................................40 Third Generation of Bone Grafts.....................................................................................41 Intrafibrillar Minera lization of Collagen Fibrils w ith the Polym er-Induced-LiquidPrecursor (PILP) Process.................................................................................................... 44 3 M INERALIZATION OF TYPE-I COLLAGEN SPONGES VIA AN ENZYMEAIDED PILP PROCESS ........................................................................................................63 Introduction................................................................................................................... ..........63 Materials and Methods...........................................................................................................65 Sample Preparation..........................................................................................................65 X-Ray Diffraction (XRD) analysis.................................................................................. 66 Scanning Electron Microscopy (SEM) Analysis.............................................................66 Transmission electron micr oscopy (T EM) analysis........................................................ 66 Confocal Microscopy Study............................................................................................67 Results and Discussion......................................................................................................... ..68 Analysis of Precipitates from the Mineralization Solution ............................................. 68 Analysis of Collagen Sponge Mineralized by Enzym e-Aided PILP Process................. 70 Study of Depth of Penetration of Po lym er Incorporating with Precursor....................... 73 Conclusions.............................................................................................................................74
7 4 INTRAFIBRILLAR MINERALIZATI ON OF BOVINE ACHI LLES TENDON VIA THE PILP PROCESS .............................................................................................................85 Introduction................................................................................................................... ..........85 Materials and Methods...........................................................................................................88 Sample Preparation..........................................................................................................88 X-Ray Diffraction (XRD) analysis.................................................................................. 89 Scanning Electron Microscopy (SEM) Analysis.............................................................91 Transmission electron micr oscopy (T EM) analysis........................................................ 91 Thermogravimetric analysis (TG)...................................................................................91 Results and Discussion......................................................................................................... ..92 Mineralization of Bovine Tendon with Low Molecular W eight Poly-aspartic Acid...... 92 Mineralization of turkey tendon as a prelim inary experiment................................. 92 Mineralization of bovine tendon..............................................................................93 Improvement of Mineralization of Bovine Tendon by EDTA Treatm ent....................103 Conclusion............................................................................................................................107 5 MINERALIZATION OF SYNTHETIC COLLAGEN SPONGE VIA THE PILP PROCESS .............................................................................................................................131 Introduction................................................................................................................... ........131 Materials and Methods.........................................................................................................133 Mineralization of Collagen Sponges.............................................................................133 X-Ray Diffraction (XRD) analysis................................................................................ 134 Scanning Electron Microscopy (SEM) Analysis...........................................................135 Transmission electron micr oscopy (T EM) analysis...................................................... 135 Thermogravimetric and Differentia l Therm al Analysis (TG/DTA).............................. 135 Results and Discussion......................................................................................................... 136 Mineralization of Collagen Sponges by the PILP Processes Induced by the Different Molecular W eight Poly-aspartic Acids...................................................... 136 Enhancement of Mineraliza tion by Buffer Treatm ent or Solution Replacement.......... 143 Tris-buffer solution treatment................................................................................ 143 Mineralization solution replacement...................................................................... 144 The Thermal Analysis of Mineralized Collagen Scaffolds........................................... 145 Conclusions...........................................................................................................................149 6 CHARACTERISTICS OF AMORPHOUS CALICUM PHOS PHATE PRECURSOR FORMED BY THE PILP PROCESS................................................................................... 169 Introduction................................................................................................................... ........169 Materials and Methods.........................................................................................................173 Thermogravimetry Analysis (TGA).............................................................................. 173 The size distribution analysis of poly-aspartic acid...................................................... 174 Zeta-potential measurement.......................................................................................... 174 Energy dispersive x-ray sp ectroscopy (EDS) Analysis ................................................. 175 Results and Discussion......................................................................................................... 175 Conclusions...........................................................................................................................183
8 7 CONCLUSIONS ...................................................................................................................195 LIST OF REFERENCES.............................................................................................................204 BIOGRAPHICAL SKETCH.......................................................................................................217
9 LIST OF TABLES Table page 2-1 Principle constituents of bone*.......................................................................................... 48 6-1 Sample preparation for zeta-potential analysis of functionalized polystyrene beads with poly-L -aspartic acid or precu rsors formed by the PILP process............................. 189 6-2 Zeta-potential (mV) of polystyren e beads functionalized with carboxyl group (COOH) or am ine group (NH2), which arenegatively (COO-) or positively (NH3 +) charged at pH 7.4.............................................................................................................190
10 LIST OF FIGURES Figure page 2-1 Tropocollagen molecule..................................................................................................... 49 2-2 Collagen fibrils assembled into a quarter-staggered arrangement.. ...................................50 2-3 Hierarchy of osteonal bone structure, as illustrate d by W einer and Wagner. ...................51 2-4 Based on tomographic TEM images, Landis et al. provide a schem atic diagram of mineralized turkey tendon as ossification progresses........................................................52 2-5 Selected area electron diffraction (SAED) patterns of equine bone, and a schem atic diagram of mineralized collagen fibrils showing the deck-of-cards structure................... 53 2-6 TEM micrographs of A) decalcified zebra fish b one, and B), C) untreated fish bone...... 54 2-7 Activated osteoclast at the resorptive bone site. ................................................................ 55 2-8 TEM micrographs of the initial stage of calcificati on in A) avian em bryonic bone and B) unstained collagen fibril which was mineralized by in vitro experiment.............. 56 2-9 History of bone graft development.................................................................................... 57 2-10 Stress-strain relationship of various bone grafts.. ..............................................................58 2-11 The TEM micrograph of colla gen fibril produced by Kikuchi et al. using the coprecipitation m ethod. ........................................................................................................ 59 2-12 Formation of calcium carbonate film via the PILP process............................................... 60 2-13 Intrafibrillar minerali zation of a collagen fibril via the PILP process. .............................. 61 2-14 The SEM micrographs of the m ineralized part of hen tendon........................................... 62 3-1 The pH of mineralization solutions containing various concentrations of polyaspartate points to a clear difference in the cr ystallization pathways. ........................76 3-2 The XRD patterns of precipitates coll ected from the mineralization solutions containing various concentration of polyasp artate (0, 25, 50 g/ml) after 1 day of reaction...............................................................................................................................77 3-3 The XRD patterns of aged precip itates collected from mineralization solutions containing various concentration of polyasp artate (0, 25, 50 g/ml) after 10 days of reaction...............................................................................................................................78
11 3-4 Scanning Electron Microscopy (SEM) micrographs of a collagen sponge (left) that was m ineralized for 4 days, and the magni fied image of the surface precipitate (right).................................................................................................................................79 3-5 The SEM micrographs of collagen sp onges before and after m ineralization.................... 80 3-6 The SEM micrographs of mineralized colla g en after acid treatment (left) and bleach treatment (right) to selectively remove the mineral and organic phases, respectively...... 81 3-7 Transmittance Electron Microscopy (TEM ) m icrographs of the collagen sponge before and after mineralization.......................................................................................... 82 3-8 Confocal fluorescence microscopy images of FITC-labeled polyaspartate transported into a turkey tendon scaf fold.............................................................................................. 83 3-9 Proposed mechanism of in trafibrillar m ineralization via the enzyme-aided PILP process................................................................................................................................84 4-1 Collagen fibrils and struct ural factors producing peaks in X-ray diffraction analysis. ...109 4-2 Schematic diagram illustrating the alignm ent of collagen fibers in bovine tendon relative to the x-ray beam path......................................................................................... 110 4-3 The XRD patterns of turkey tendon minera lized with the PIL P process for various times (black: 0 day, red: 3 days of minerali zation, blue: 4 days of mineralization)....... 111 4-4 The XRD patterns of bovine tendon minera lized by the PILP process for 6 days.. ........ 112 4-5 The XRD patterns (mode I) of bovine tendons reacted with various reaction solutions. ..........................................................................................................................113 4-6 The XRD patterns (mode I orientation) of bovine tendon m ineralized by the PILP process with various concentration of poly-( )-DL-aspartic acid (Mw: 5,500 Da) for 6 days..........................................................................................................................114 4-7 The XRD patterns (Mode II orientation) of bovine tendon m ineralized by the PILP process with various co ncentrations of poly-( )-DL-aspartic acid (Mw: 5,500 Da) for 6 days..........................................................................................................................115 4-8 The EDS analysis of bovine tendon minera lized via the P ILP process with 50 g/ml of poly-( )-DL-aspartic acid for 6 days....................................................................... 116 4-9 The SEM micrographs of the surface of bovine tendon m inera lized without polyaspartic acid.....................................................................................................................117 4-10 The SEM micrographs and EDS analysis of bovine tendons m ineralized with the PILP process containing various concentrations of poly-( )-DL-aspartic acid............ 118
12 4-11 The TEM micrographs of bovine ten don before (A, B) and after (C, D) m ineralization..................................................................................................................119 4-12 The XRD patterns (Mode I) of bovine te ndons m ineralized with the PILP process containing poly-L-aspartic acid (polyAsp) for 8 days..................................................... 120 4-13 The XRD patterns (Mode II) of bovine tendons m ineralized with different molecular weight polymers via the PILP process containing 100 g/ml of poly-L-aspartic acid (polyAsp) for 8 days........................................................................................................ 121 4-14 The XRD patterns of isotropic bovine te ndons m ineralized with the PILP process containing 100 g/ml of poly-L-aspa rtic acid (polyAsp) for 8 days............................... 122 4-15 The SEM micrographs and EDS analysis of bovine tendons m ineralized with the PILP process containing 100 g/ml of poly-Laspartic acid (polyAsp) for 8 days......... 123 4-16 The TEM analysis of bovine tendon minera lized with the PIL P process containing 100 g/ml of poly-L-aspartic acid (Mw ; 32,200 Da) for 8 days.................................... 124 4-17 High resolution TEM (HRTEM) micrographs of bovine tendon m ineralized with the PILP process containing 100 g/ml of poly-Laspartic acid (Mw: 32,200) for 8 days... 125 4-18 The SEM micrographs and EDS analysis of bovine tendons m ineralized with the PILP process containing 100 g/ml of pol y-L-aspartic acid (Mw: 32,200 Da) for 8 days..................................................................................................................................126 4-19 The XRD patterns of bovine tendon and EDTA-treated bovine tendon which were m ineralized by the PILP process containing 100 g/ml of poly-L-aspartic acid (Mw: 10,300 Da) for 8 days.......................................................................................................127 4-20 The TEM micrographs of EDTA-treated bovine tendon m ineralized with the PILP process containing 100 g/ml of poly-L-aspa rtic acid (Mw: 10,300 Da) for 8 days...... 128 4-21 The XRD patterns of bovine tendon and EDTA-treated bovine tendon which were m ineralized by the PILP process containing 100 g/ml of poly-( )-DL-aspartic acid (Mw: 5,500 Da) for 8 days....................................................................................... 129 4-22 Thermogravimetric analysis (TGA) of bovine tendons m ineralized by the PILP process containing 100 g/ml of poly-L-as partic acid (Mw:10,300 or 32,200 Da) for 8 days...............................................................................................................................130 5-1 The XRD patterns of Collagen Matrix s ponge m ineralized by the PILP process for various times, using 50 g/ml of poly-( )-DL-aspartic acid (Mw: 5,500)..................151 5-2 Optical microscope (Left) and TEM (Right) m icrographs of solution borne precipitate formed with low molecular weight polymer.................................................. 152
13 5-3 The SEM micrographs and EDS analysis of a collagen sponge m ineralized by the PILP process induced with 50 g/ml of poly-( )-DL-aspartic acid (Mw: 5,500) for various times.................................................................................................................. ..153 5-4 The XRD patterns of collagen sponges mi neralized by the PIL P process containing 50 g/ml of poly-L-aspartic acid (Mw: 10,300) for various times.................................. 154 5-5 The SEM micrographs and EDS analysis of collagen sponges m ineralized by the PILP process containing 50 g/ml of polyL-aspartic acid (Mw: 10,300) for various times.......................................................................................................................... .......155 5-6 The SEM micrographs of collagen sp onges m ineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 10,300) for various times................ 156 5-7 The XRD patterns of collagen sponges mi neralized by the PIL P process containing 50 g/ml of poly-L-aspartic acid (Mw: 32,200) for various times.................................. 157 5-8 The SEM micrographs and EDS analysis of collagen sponge m ineralized by the PILP process containing 50 g/ml of polyL-aspartic acid (Mw: 32,200) for various times.......................................................................................................................... .......158 5-9 The SEM micrographs and EDS analysis of the inner part of a collagen sponge m ineralized by the PILP process with 50 g/ml of poly-L-aspartic acid (Mw: 32,200 Da) for 16 days.................................................................................................................159 5-10 A) Bright field and B) dark-field TEM m icrographs of a collag en sponge mineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 32,200Da) for 16 days........................................................................................................................160 5-11 Thermogravimetry analysis of collag en sponges m ineralized by the PILP process induced with 50 g/ml of poly-L-as partic acid for 8 and 16 days................................... 161 5-12 The SEM micrographs of collagen sponges after tris-buffer treatm ent for various times.......................................................................................................................... .......162 5-13 The XRD patterns of collagen sponges pre-tr eated with tris-buffer solution for 5 days before the mineralization using the PILP process............................................................163 5-14 The SEM micrographs of a collagen sponge that was pre-treate d with tris-buffer solution for 5 days before the 2 days of m ineralization using the PILP process............. 164 5-15 Thermogravimetry analysis of collagen sponges m ineralized by the PILP process for 16 days.............................................................................................................................165 5-16 Thermo Gravimetric and Differential Therma l analysis (TG/DTA) of a pure collagen sponge..............................................................................................................................166
14 5-17 TG/DTA of collagen spo nges m ineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 32,200 Da) for 8 and 16 days........................................... 167 5-18 TG/DTA of collagen sponge m ineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 10,300 Da) for 8 and 6 days.................................................. 168 6-1 Electrical double layer formed on a charged solid surface.............................................. 185 6-2 Thermogravimetry (TG) analysis of co llagen sponges which were m ineralized by the PILP process at different he ights in the reaction vessel.................................................. 186 6-3 Particle size analysis of poly-L-aspartic acid in various solutions. The shape of polyL-aspartic acid is assumed to be sphe rical (a random coil) for analysis.......................... 187 6-4 Particle size analysis of poly-L-aspartic acid (Mw: 32,200 Da) in the m ineralization solution of the PILP process for 1 hour (A) and 3 days (B)............................................ 188 6-5 The EDS analysis of COOH functionaliz ed polystyrene beads reacted with the various solutions indicated at table 6-1. ...........................................................................191 6-6 Interfacial condition of polystyrene beads functionalized with carboxyl (-COOH) groups.. .............................................................................................................................192 6-7 The EDS analysis of PSNH2 functionalized polystyrene bead reacting with various solutions indicated in table 6-1. Spot EDS analysis was done for polystyrene beads during SEM examination.................................................................................................193 6-8 Interfacial condition of polystyrene b eads which were functionalized with am ine (NH2) groups....................................................................................................................194
15 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DEVELOPMENT OF COLLAGEN-HYD ROXYAPATITE NANOSTRUCTURED COMPOSITES VIA A CALCIUM PHOSPHATE PRECURSOR MECHANISM By Sang Soo Jee August 2008 Chair: Laurie B. Gower Major: Materials Science and Engineering Bone is an interpenetrating inorganic/organic composite th at consists of mineralized collagen fibrils, which is hierarch ically organized into various structures. The structure of mineralized collagen fibril, in which nano-crys tals of hydroxyapatite are embedded within the collagen fibrils, provides remarkable mechanical and bio-resorptive properties. Therefore, there have been many attempts to produce collagenhydroxyapatite composites having a bone-like structure. However, duplication of even the mo st fundamental level of bone structure has not been easily achieved by conventio nal nucleation and growth tec hniques, which are based on the most widely accepted hypothesis of bone mineralization. In nature, the collagen fibril is mineralized via intrafibrill ar mineralization, which produces preferentially oriented hydroxyapati te nano-crystals occupying the interstices in collagen fibrils. Our group has demonstrated that intrafibrillar mineralization can be achieved by using a new method based on the Polymer-Induced Liquid-Precurso r (PILP) mineralization process. In the PILP process, a poly-anionic additive can produce an amorphous calcium phosphate precursor which enables us to achieve intrafibrillar mineralization of collagen. It is thought that the precursor is pulled into the in terstices of the collagen fibr ils via capillary forces, and upon solidification and crystallization of the precursor produces an inte rpenetrating composite with the nanostructured architecture of bone.
16 In this dissertation, to de monstrate the effectiveness of the PILP process on the intrafibrillar mineralizati on of collagen fibril, various collagen scaffolds, such as turkey tendon, bovine tendon and synthetic collagen sponge, were mineralized by the PILP process. Various poly-aspartates with different mo lecular weight were also used for the optimization of the PILP process for the mineralization of the collagen sca ffolds. With the systematic researches, we discovered that the molecular weight of poly-aspa rtic acid affects the de gree of intrafibrillar mineralization of collagen scaffolds. High mol ecular weight poly-aspartic acid could produce a stable and dispersed amorphous precursor, l eading to a high degree of intrafibrillar mineralization. The mineral c ontent of the collagen sponge mineralized using high molecular weight poly-aspartic acid was e quivalent to the mineral conten t of bone. According to X-ray diffraction analysis of the mineralized collage n, the size and composition of the intrafibrillar hydroxyapatite produced by the PILP process were almost identical to carbonated hydroxyapatite in bone. The selective area elect ron diffraction patterns indicated that the  direction of hydroxyapatite is roughly aligned along the c-axis of collagen fibr il, leading to the formation 002 arcs. Using dark field imagi ng, it was possible to visualize the preferentially oriented hydroxyapatite in TEM. Thermal analysis of mine ralized collagen also s howed a reduction in the thermal stability of collagen, which is similar to that observed in the collagen in bone, due to the presence of intraf ibrillar hydroxyapatite. Now, we confidently suggest that the PILP process can provide a new way to develop synthetic bone-like composites whose nano-struct ure is very close to the nano-structure of natural bone. Moreover, we hope that our successf ul intrafibrillar minerali zation of collagen via the precursor mechanism revives discussion of hypothesis of bone mineralization via the amorphous calcium phosphate phase.
17 CHAPTER 1 INTRODUCTION As a society has m odernized a nd technologies of medical cares have evolved, the desire to live a longer and healthier life ha s increased. Therefore, there has been high demand for new synthetic organs for transplantation [1, 2]. In particular, bone substitutes have been gathering the interest of researchers in various areas because bone is one of the most frequently damaged organs in humans [3, 4]. According to stat istical research, about 6.3 million fractures were reported in the United States of America every year, and about 9 percen t of those patients suffered needed some kind of bone graft . The fracture of the hip is the most common fracture in the United States, and the surgery of hip replacement had been performed about 152,000 times in the year 2000, and the number of opera tions for hip replacement is expected to be 272,000 cases in 2030 . In the case of sales of bone grafts, the market of bone grafts was estimated to be about 1.6 billi on US dollar in 2002, and continuous increase of bone graft market is now expected [4, 7]. Therefore, comprehensive research about bone substitutes, in which researchers with various backgr ounds should be incorporated, has been required to satisfy the demands from patients and medical service providers. Bone is the major framework of vertebrate sp ecies which supports and protects the internal organs [8-10]. Besides its function of physical support and protection, bo ne plays the pivotal role of maintaining the concentration of inorganic ions, such as cal cium and phosphate, through the continuous resorption and r econstruction of bone [11, 12]. For the physical protection of organs or support of body, bone must have remark able mechanical properties, such as high tensile and compressive strength, ductility and high toughness [13, 14]. The density of bone, as a framework, is also required to be relatively low for high performance of the human body against gravity. As a reservoir of inorganic ions, bone should be resorptive by metabolism even though
18 bone contains the thermodynamically stable mineral phase of carbona ted hydroxyapatite, in which carbonate partially substitutes for the phosphate group in hydroxyapatite crystal . Those unique properties of bone, such as low density, high toughne ss, high mechanical strength and bioresorbability, are thought to be associated with the nanost ructured architecture of bone . The major components of bone can be categorized into three parts; i) inorganic mineral, ii) organic phases including extracel lular matrix (collagen), non-colla genous proteins and cells, and iii) water [13, 16-19]. Those three parts are pr ecisely engineered to achieve the required properties of bone. The basic building block of bone is a mineralized collagen fibril . Collagen fibrils are mineralized with carbonate d hydroxyapatite plates which are embedded in the collagen fibril and roughly  aligned with uniaxia l direction along the long axis of fibril [20-22]. Hydroxyapatite platelets are interconnected with coll agen molecules via chemical bonding through non-collagenous proteins [23, 24]. This basic building block of bone, a mineralized collagen fibril, is precisely assemble d into higher levels of structure, such as concentric lamellae or plywood-like lamellae, with hierarchy to achieve the unique mechanical and chemical properties of bone [1 1]. The hierarchy of bone coul d be explained by the way that sub-structures of bone are organized into th e final structure [10, 11, 25]. The mineralized collagen fibrils are assembled into collagen fibe r bundles with uniaxial orientation, and collagen fibers are arranged into the layer by layer struct ure in parallel orient ation. The lamellae of mineralized collagen fibers are al so organized into a cylindrical structure called an osteon in which a nerve system is penetrating. The osteon al bone with concentric layer-by-layer structure is arranged into two types of bone: compact bone with densely packed osteonal bone and trabecullar bone with porous structure.
19 From a materials science sta ndpoint, the basic building bloc k of bone could be simply considered as a ceramic-reinforced polymer com posite or polymer-reinforced ceramic composite . Ceramic substances in a polymer matrix can increase the modulus and mechanical strength because ceramic materials normally are stiffer and have higher mechanical strength than that of the polymer matrix, and the numerous nanocry stals help to preven t crack propagation by deflecting the path of crack grow th . In the case of polymer matrix, polymer can provide ductility and resilience to a cer amic-reinforced polymer composite for high toughness [23, 28]. By using a composite, one may also reduce the de nsity of materials due to the combination of light and heavy substances, in comp arison to the material consisti ng of only ceramic components. In the macroscopic aspect of bone structure, the layer-by-layer structure of bone also can increase the mechanical strength . The lamellae consist of bundl es of parallel collagen fibers in discrete layers in which the orientation of collagen fiber in each layer is gradually changed, presumably due to the cholesteri c nature of collagen . Th e structure of bone lamellae is analogous to the plywood structure, which s hows a remarkable mechanical strength in comparison to the monolithic wood plate . The size of the carbonated hydroxyapatite nanocrystals embedded in the narrow space within collagen fibrils is also an important factor for the reso rptive property of bone. The nanoscaled hydroxyapatite increases the surface area of the mineral phase. The increased surface area of nano-sized crystal, as well as lattice impuritie s and imperfections, could enhance the chemical reactivity of crystals and make bone easier to be resorbed by acids or enzymes, even though hydroxyapatite is the most stable mineral phas e in the calcium phosphate mineral systems [15, 18, 32].
20 As briefly explained in the previous para graphs, bone is an organ which is damaged frequently and often requires grafts to aid in the healing process. It is also a multifunctional organ in which various organic substances and in organic minerals are intr icately interlocked and arranged to accomplish the desired functions. Therefore, the development of bone substitutes has become a difficult but attractive challenge for many researchers. When bone is considered as a nano-engineered composite material, however, the approach from the point of view of materials science could make it easier to develop bone substitutes. From a biological standpoint, the autograft, which is defined as bone tissue extracted from the patient, is considered as the gold standard of bone substitution . However, b ecause autografts have several disadvantages, such as limited amount of bone graft material wh ich can be extracted and additional surgery for the graft extraction, alternative methods for bone s ubstitution are required . On the other hand, from a materials science stand point, synthetic bone grafts co uld be potential candidates for replacing autografts because synthetic bone graf ts do not necessarily have size limitations, and they can reduce the risk of transferring inf ectious diseases from the donor [4, 7, 34, 35]. The development of synthetic bone grafts has been continuously investigated for decades [35, 36]. The evolution of synthe tic bone grafts can be divided into three generations depending on the development of materials . The first generation of bone grafts are composed of nontoxic metal or ceramic materials such as titanium, stainless steel and alumina. The bone grafts of the first generation provide mechanical support or bridge a void between fr actured bone parts as a form of pin, plate or bulk, but they seldom provide a reactive surface which can make direct bonding with new bone, with the exception of some pure or surface-modified titanium or titanium alloys [37, 38]. Therefore, the first generation bone grafts are often considered as a synthetic bioinert material. After 1970, many re searchers began to inve stigate surface active
21 materials, such as hydroxyapatite, collagen, and bioglass, which provide suitable surface conditions for osteoconduction, the prolifera tion of bone forming cells under physiological conditions [39-41]. Some bioresorbable materi als, which dissolve after implantation as new bone is growing, are also used as second generati on of bone grafts . However, because the mechanical and chemical properties of the fi rst and second generation bone grafts are not matched with those of bone, they can cause several severe problems, such as a bone resorption by the stress shielding effect, detachment of the implant, a nd an immunological response by being recognized as a foreign material [43, 44]. Therefore, new types of third generation bone grafts have recently been developed. Various composites with non-toxic polymers and ceramics or metals are good examples of the third genera tion bone grafts [4, 45], in which they provide various bioactive and bioinert materials together to mimic bo ne properties . However, composites of combinations of polymers, ceramic s or metals, such as hydroxyapatite/titanium, hydroxyapatite/polycaprolactone and hydroxyapatite/chitosan, stil l have similar problems to those of the first and second gene ration bone grafts such as the stress shielding effect causing detachment of implant and uncontrolled-dissolu tion of implant [43, 46-48] Therefore, there have been continuing attempts to make com posites having a bone-like structure with collagen and hydroxyapatite, which are major co mponents of natural bone [49-56]. Bone, as described previously, is a hierarchical composite with nanostructured architecture [10, 57]. Although bone could be simply considered as a polym er/ceramic composite from a view of materials scienc e, it is still difficult to duplicate bone s structure or fabricate synthetic bone by the simple fabrication methods that have frequently been used in materials science area because bone formation is precisely controlled by the living cells, such as osteoclast and osteoblast. Therefore, various techniques have been investigated to duplicate the bone structure,
22 which is considered to provide the remarkab le physical and chemical properties of bone. Unfortunately, various trials with the conven tional approaches for th e fabrication of nanocomposites have not been successful at mimick ing the fundamental structure of bone: collagen fibrils with intrafibrillar mineral [ 15, 45]. The conventional approaches to hydroxyapatite/collagen composites are to initiate the nucleat ion of hydroxyapatite on the collagen substrate placed in a mineralization so lution, such as a simulated body fluid (SBF) or modified SBF [58-60]. Although the conventional methods are us eful to coat the collagen substrate with hydroxyapatite, intrafibrillar mine ralization, in which hydro xyapatite plates are embedded in the collagen fibril, could not be ach ieved. Therefore, biomimetic approaches have been highlighted as an al ternative method for the intr afibrillar mineralization. Biomimetics is the study of mimicking, adap ting or specializing th e design, mechanism and function of biological substances to develop novel technologies . Biomineralization is one of the branches of biomimetics in which va rious biominerals, such as calcium carbonate and silicon oxide in marine species or calcium phos phate in bone or tooth, are investigated to determine how their unique architectures or morphol ogies are originated or achieved [62, 63]. In biomineralization research, it is worthwhile to study the function of the organic components of natural biominerals, which are in soluble organic matrix and water-s oluble acidic proteins [64-68]. The insoluble organic component is thought to te mplate or to provide compartments for the growth of biominerals, and water-soluble protei ns are thought to control the crystal growth by inhibition of crystal growth or alteration of the rate of crystal growth in specific crystallographic directions . Here, we concentrate on th e function of acidic solu ble proteins for the fabrication of synthetic bone graft.
23 Gower discovered that the addition of acidi c polymers, such as polyacrylic acid or polyaspartic acid, to a mineraliz ation solution, can induce or st abilize the amorphous precursor of calcium carbonate, and the proc ess of adding acidic polymers as a process-directing agent was named the PILP (Polymer-Induced-L iquid-Precursor) process [70, 71]. With the PILP process, it is possible to form a fluidic amorphous precursor for calcium carbonate, and it is observed that fluidic precursors are agglomerated into various mo rphologies, such as a film, fiber or core-shell, depending on the substrates or organic matrix, whic h then crystallize into crystalline phases such as calcite or aragonite [72-74]. In the case of the calcium phosphate system, it has previously been shown by our group the possibility of mimicking bone structure with the PILP process [15, 26]. In this dissertation, I will show how the amorphous precursor phase generated via the PILP process causes intrafibrillar mineralization of collagen, how the process can be optimized for the fabrication of bone grafts with various collagen scaffolds. Chapter tw o contains the background information of bone structure, bone formation, and the biomimetic approach called the PILP process to achieve intraf ibrillar mineralization. In chapter th ree, an enzyme-aided PILP process is introduced for the intrafibrillar mineraliza tion of a collagen sponge, and the advantages and disadvantages of enzyme-aided PILP are discusse d. Chapter four describes the mineralization of tendon with densely packed collagen fibers havi ng a parallel arrangemen t similar to lamellar bone. The effect of molecular weights of the polymer additi ves on the optimization of the intrafibrillar mineralization is the main topic of th e chapter four. In chap ter five, intrafibrillar mineralization of synthetic collagen sponges is disc ussed, and optimization of such intrafibrillar mineralization is investigated by controlling the molecular weight of polymer additive or tris-
24 buffer treatment of collagen sponge. Chapter si x shows the electrosta tic character of the precursor, which may play an important role in the in trafibrillar mineralization of collagen fibril.
25 CHAPTER 2 BACKGROUNDS Bone: From a Structural Standpoint As previously discussed in the Chapter 1, bone has a nanostructural architecture which is thought to be associated with the high perform a nce of bone, such as its remarkable mechanical strength, high toughness and biores orptive properties [11, 15, 75]. Therefore, the structure of bone and structural factors, such as collagen fibrils and their arrangement with hydroxyapatite crystals and their alignment, must be unde rstood to develop new bone grafts having the fundamental nanostructure of bone. Bone can be categorized into two types, according to the stages of bone formation . Primary bone is simply considered as an early stage of bone which is formed in soft cartilage tissue where bo ne had not previously existed . The rate of primary bone formation is relatively fast and the structure of bone is very loosely organized in the form of a woven structure, and the mineralization of hydroxyapati te is not considered to be governed by the collagen [15, 76, 77]. Primary bone is gradually replaced by secondary bone via a remodeling process done by osteoclast and osteoblast cells . During the remodeling process, the structure of bone is reorganized with the ordered st ructure we are interested in. Therefore, the structure and the structural factors of secondary bone are now discussed. Components of Bone Bone is composed of m ajor th ree components; inorganic phase, various organic phases and water [10, 79, 80]. The weight contents of each component are shown in Table 2-1 . Among those components, it is worthwhile to discu ss collagen and hydroxyapatite, which occupy about 80 wt % of the bio-composite, to in vestigate the structural characteristics of secondary bone.
26 Collagen Collagen is one of the most abundant proteins within vertebrates, and it is the m ain extracellular matrix prov iding the basic organic framework of bone . Up to now, more than 27 types of collagens are identified, but only a few of them form fiber conf ormations. More than 90% of fibrous proteins found in nature are type I collagen, which is also the type of collagen that is the main organic portion of bone . As shown in Figure 2-1, collagen molecules, called tropocollagen, are composed of two 1 and one 2 polypeptide chains with a (Gly-x-y)n sequence . These three polypeptide chains are self-assembled into tropocollagen with triplehelical structure via hydrophobic, electrostatic, and hydrogen bonding inte ractions [83-86]. During fibrillogenesis, Cand Npropeptides at the end of the individual molecules are cleaved by enzymes, and telopeptides, which are short nontriple-helix amino acid chains at the end of tropocollagen, are exposed . Then, as shown in Figure 2-2, fi ve tropocollagen molecules are self-assembled into a quarte r-staggered arrangement, whic h was proposed by Hodge and Petruska [87, 88],, creating 67 nm of axial peri odicity [85, 86, 89]. The collagen molecules are then cross-linked by enzymatically oxidized lysyl residues in the telopeptides. This very unique periodic arrangement of collage n molecules leads to the formation of gap zones, which is considered an important structural factor of the collagen fibril in the intrafibrillar mineralization via the PILP process . The repetitive nature of the gap zones generates the periodic banding patterns seen in transmittance electron microscope (TEM) micrographs of collagen fibrils . Collagen fibers are formed via the assembly of co llagen fibrils, and vary in diameter from 1 to 20 m . The macro-mechanical property of collage n fibers ranges from 1 to 8 GPa (Youngs modulus), and collagen provides elasticity and resilience leadi ng to high fracture toughness . Recently, a new micromechanical test me thod with atomic force microscopy has been
27 introduced to measure the Youngs m odulus of single collagen fibril s, and the reported value was 1.4 0.3 GPa . Although there are some devi ations between macroand micromechanical properties, the function of collagen fibers with respect to the mechanical properties of bone is considered important for providing the ductility and toughness of the co mposite. Besides the function of collagen fibrils as a structural organic framework, collagen polypeptides with specific sequence of amino acids provide a specific site for osteoblast cell adhesion . It is thought that the amino acid seque nce Arg-Gly-Asp (RGD; arginine -glycine-aspartic acid) in collagen can induce the osteoblast cell attachment through binding with integrin, which is a cell surface receptor [96-98]. Hydroxyapatite Hydroxyapatite is one of the calcium phosphate minerals, and its chemical formula is Ca10(PO4)6(OH)2. The crystal system of hydroxyapati te is hexagonal and its space group is P63/m [99-101]. In the case of hydroxyapatite in bone, hydroxyapatite has structural defects such as calcium deficiency, and sodium, carbonate and magnesium impurities. Therefore, hydroxyapatite in bone is considered a non-stoi chiometric hydroxyapatite whose chemical formula can more accurately be written as Ca10-x(PO4)6-y(HPO4)y(CO3)(OH)2-x (x = amount of Mg, Na or vacancy, y = around 0.7) . Weiner et al. [11, 21, 103] observed mineralized colla gen fibrils from bone with TEM, and suggested that plate-like hydroxyapa tite crystals are well-aligned and stacking in the collagen fibrils with deck-of-cards struct ure. More recent direct observa tions of bone hydroxyapatite with AFM also confirmed that hydroxyapatite extrac ted from young bovine bone has a plate-like crystal habit (in the older literature, the mor phology was often described as needle-like). The size of the non-stoichiometric hydroxyapatite reported has varied, with ranges from 10 -50 nm in the length, 6-30 nm in the width, and 0.7 10 nm in thickness, depending partially on the sample
28 preparation and mostly on types of bone, maturation and species [104-106] This nano-scaled hydroxyapatite is chemically active due to the large surface area and various defects, even though hydroxyapatite is thermodynamically stable [ 15, 18]. Therefore, n on-stoichiometric bone hydroxyapatite can be dissolved by metabolic pathwa ys to maintain the appropriate ion balance in the body. Structure of Secondary Bone The m acroscopic structure of bone is varied depending on the type, age and species from which the bone is extracted . Even though it is an ultimate goal to fabricate bone grafts whose nano, micro and macro-structures are almost identical to the struct ure of real bone, the engineering of the higher levels of the bone hierarchy is beyond the objective of this dissertation. Therefore, it is worthwhile to understand the nano structure of bone (the nano-structure), before attempting to fabricate a synthetic bone graft, because this fundamental building block of bone is basically identical in various types of bone [10, 11]. Weiner et al.  divided the hierarchy of bone into 7 levels, as shown in Figure 2-3. The whole bone (level 7) is composed of compact bone and spongy bone (level 6). Osteonal bone is composed of compact bone and spongy bone arranged as a hollow structural element. In compact bone, osteonal bone is very densely packed with a desired arrangement to endure tensile and compressive stress. On the other hand, s pongy bone is loosely packed osteonal bone, and it has optimal structure to tolerate the compressi ve stress while reduci ng the weight of bone. Osteonal bone is composed of mineralized coll agens which are precisel y aligned into desired structure, such as parallel, woven-like (random), radial and plywood-like array, to construct the concentric lamellae of osteonal bone (level 4). As described previously, the basic building block, consisting of a mineralized collagen fibril (level 2), is the same in the structure of all secondary bone. Mineralized collagen fibrils, in which hydroxyapatite (level 1) plates are embedded
29 between tropocollagens within the collagen fibril (level 1), is formed by the intrafibrillar mineralization. As shown in Figure 2-4, during intrafibrillar minerali zation, hydroxyapatite was observed to form first within the gap zone fi rst, and it grows or fills the space between tropocollagen molecules and grooves formed by the quarter-stagger arrangement of tropocollagens as the collagen beco mes fully mineralized [90, 108]. Intrafibrillar hydroxyapatit e in bone has an interesting st ructural feature in that the crystallographic orientation of hydr oxyapatite is aligned along the caxis of the collagen fibril. The  direction of hydroxyapatite in bone is roughly aligned along the caxis of collagen fibrils leading to the ( 002) and (004) set of arcs in selected area el ectron diffraction (SAED) patterns of bone (Figure 2-5 A) [15, 104]. Traub et al.  also investigated the uniaxial crystallographic orientation of hydroxyapatite in bone and suggested the d eck-of-card structure of a mineralized collagen fibril (Figure 2-5 B). However, our gr oup has argued that for the case of a deck-of-card arrangement of hydroxyapatite, the (112), (211), (310) and (300) planes can not all diffract the incident electron beam from th e same direction at the same time because the beam is diffracted by the planes which are almost parallel to the incide nt beam direction [15, 109]. For example, when the orientation of the incident beam is [1 10], (112) and (002) planes can diffract the incident beam because the plan es are parallel to th e incident beam ([1 10] (002) = 0, direction of plane is perpendicular to plane surface). By comparison, the dot product between [1 10] and (211), (310) and (300) is not zero, indicating these planes are not parallel to the incident beam direction when the incident be am is diffracted by (002) and (112) at the same time. Therefore, the caxis of hydroxyapatite plates should be oriented along the long axis of the collagen fibril, as shown in Figure 2-5 B, with some tilting disorder, as seen by the arcing of the (002) spots. But, hydroxyapatite plate also has some rota tional disorder along a or b direction,
30 which is not coincident with th e illustration of the arrangement of hydroxyapatite plates in figure 2-5 B. Besides the intrafibrillar hydroxya patite in bone, there is anothe r type of hydroxyapatite in bone. Ge et al.  investigated the skeleton of zeb rafish with TEM and AFM, and they discovered that the mineralized collagen fibril s were thicker than co llagen fibrils whose hydroxyapatite was removed by formic acid due to the interfibrillar hyd roxyapatite crystals which occupied the space between collagen fibrils. As shown in Figure 2-6 A, the diameter of the decalcified collagen fibril is around 83 nm, and space that was thought to have been occupied by the interfibrillar hydroxya patite crystals was also observed between collagen fibrils. In the case of untreated zebrafish bone (Figure 2-6 B), the average diameter of mineralized collagen fibrils was 151 13.5 nm, which is larger than th e diameter of decalcified collagen fibril, and this observation is consistent with the result from high resolution TEM (Figure 2-6 C), in which the diameter of mineralized coll agen was around 145 nm, as illustrate d in the schematic diagram. The increase in the diameter of mineralized collagen is because th e surface hydroxyapatite (interfibrillar hydroxyapatite), which fills the space (white arrow) between collagen fibrils, makes mineralized collagen look thic ker. Moreover, interfibrillar hydroxyapatite seems to act as an adhesive between the mineralized collagen fibr ils to form fully mineralized collagen fibers. The formation of interfibrillar hydroxyapatite is also thought to be cl osely associated with collagen fibrils or non-collagenous proteins because the morphology of hydroxyapatite is different from hydroxyapatite clusters precipitat ed from supersaturated calcium and phosphate solution without organic phases. This will be discussed in the later section of Chapter 2. Mechanism of Bone Formation Although th e mechanism of bone formation ha s been studied for many decades, various mechanisms of bone formation are still being de bated. Because bone is formed by the metabolic
31 reaction of various cells, and in situ observati on of bone formation at th e nano-scale is almost impossible, even with highly developed observa tion techniques, the ex act mechanism of bone formation is unclear. Therefore, various approaches from in vitr o model systems might give us a better chance to clarify the mechanism of bone formation ( i.e. the materials chemistry aspects). Bone Formation: From a Biological Standpoint There are many types of bone, such as flat bone (skull and scapula) and long bone (tibia, fe mur), and they are identified by the developm ent mechanisms of bone: intramembranous and endochondral bone development . In th e case of intramembranous ossification, mesenchymal cells are condensed and differentiate d into osteoblast direct ly, and the cells form the woven bone which is later remodeled into lamellar bone . On the other hand, in endochondral bone development, mesenchymal cells are differentiated into chondroblasts which form cartilaginous matrix which subsequently becomes ossified [111, 112]. The common feature of these two types of bone development is that th e final structure of each type of bone is formed and maintained by a remodeling process . For instance, in the case of endochondral ossification for long bone, the layer of cartilage called epiphyseal cartilage grows in the longitudinal direction and is gra dually calcified as a form of primary bone. Then, primary bone is replaced with the optimal structure by a re modeling process in which various proteins (collagen and and non-collagenous pr oteins)and cells, such as osteoc yte, osteoclast, osteoblast, are involved [15, 113]. In the case of primary bone, hydroxyapatite is formed in a proteoglycan matrix as a spheric cluster, a nd crystals are deposited around the collagen fibrils throughout the ground substance, which is the amorphous organi c substance in connective tissues containing water, proteoglycans, and glycosaminoglycans [76, 114, 115]. In contrast to secondary bone, collagen in cartilage, whose diameter is somewhat smaller than that of the collagen fibrils found in bone, does not appear to direct the mineralization process .
32 In secondary bone formation, primary bone is replaced with secondary bone having a more optimal structure by the remodeling process. Three cellular reactions, such as activation, resorption and formation, are involv ed in remodeling and the activitie s of cells, such as osteocyte osteoclast and osteoblast, which are responsible for the each reac tion, are correlated without time intervals between reactions . During activation process, pre-osteoclasts, which are mononuclear cells, are activated by the signals fr om osteocytes, bone marrow cells, possibly in response to deformation of bone . Pre-oste oclasts attach to the bone surface via integrins and are fused into an osteoclast. Osteoclast is multinucleated cell containing about 4 to 20 nuclei, and are responsible for the bone re sorption process [119, 120] The bone resorption process begins with the attachment of an oste oclast on calcified bone surface via transmembrane adhesion receptors called integrin which can re cognize specific RGD amino acid sequences at the surface of the bone matrix [ 120]. After attachment of the osteoclasts, bone encapsulated by the sealing zone is dissolved by th e acidification and prot eolysis of the bone. In osteoclasts, type II carbonic anhydrase produces hydrogen ions from the reaction between carbon dioxide and water by enzymatic reaction, and hydrogen ions are delivered into a resorp tive pit via the ruffled border membrane by vacuolar H+-ATPase (Figure 2-7) [119, 121, 122]. Hydrogen ions localized at the resorptive p it decreases the pH, and dissolves hydroxya patite crystals. In addition to hydrogen ions for acidification, enzymes such as cat hepsin K or collagenase are also released from osteoclast to digest collagen matrix  In the case of the formation process, the osteoblast is responsible for th e production of collagen and gr ound substance . Osteoblasts are differentiated from pre-osteoblasts which orig inate from mesenchymal stem cells . At the early stage of bone formation in the remodeli ng process, osteoblast rapidly produces collagen which forms a thick osteoid seam consisting of type-I collagen and ground substances, such as
33 chondroitin sulfate and osteocalcin [111, 125-127]. In the case of the deposition of mineral phase, the mechanism of hydroxyapatite formation is still not clarified. However, the ionic concentration of calcium and phosphate must be increased locally to tr igger the formation of hydroxyapatite. Various potential candidates that are thought to accumulate the ionic components have been suggested. For examples, mitochondria , matrix vesicles  and ground substance [130, 131] containing various types of proteins have been studied as potential candidates. Among them, matrix vesicles, which originate from cytoplasmic processes of the chondrocyte or osteoblast, are t hought to be responsible for the deposition of hydroxyapatite in primary bone because hydroxyapatite was observed in the vesicles at the early stage of primary bone formation [132, 133]. It is known that matrix vesicles can accumulate calcium ions when isolated vesicles are incubated in synthetic cartilage lymph, a nd that matrix vesicles which contain alkaline phosphatase, stim ulate the release of inorgani c phosphate ions from organic phosphate to increase the concentration of phospha te ions . Moreover, various lipids or proteins in matrix vesicles play an important role in the accumulation of ionic components . Therefore, the calcium and phosphate ions locally concentrated by vesicles or proteins probably initiate the formation of the calcium phosphate phas e in the vesicle. However, although there are several reports showing th e presence of matrix vesicles in cal cified turkey tendon, it is somewhat difficult to consider the matrix vesicle mech anism as a strong hypothesis of hydroxyapatite formation in secondary bone because turkey tendon is not mineralized via the remodeling process, and matrix vesicles ar e seldom found in secondary bone formation . Therefore, it is worthwhile to consider other substances wh ich can increase the local ionic concentration and trigger hydroxyapatite formation. Our group has focused on non-collagenous proteins, which are found at the mineralization front as a potential candidate th at can induce intrafibrillar
34 mineralization of collagen fibrils . Non-colla genous proteins are nega tively charged due to a large number of glutamic and aspartic acid re sidues, which can attract calcium ions under physiological conditions . In addition, it is known that the non-collagenous proteins play a pivotal role of the mineraliz ation of bone [76, 79, 135]. The hypothesis of bone formation by non-collagenous proteins will be discusse d more in a later section. Bone Formation: From a Ma terials Science Standpoint Although the m acroscopic structure of collagen fibers, in which collagen fibrils are aligned in a parallel orientation, is hard to fabricate, but the reconstitution of collagen fibrils with native banding structure is now possible in the laboratory setti ng [136, 137]. Therefore, the investigation of the bone formati on from a material science sta nd point has concentrated on the development of the intrafibrillar mineralization process which can be achieved in the laboratory setting and the characterization of bone tissues via various techniques. To the materials scientist, how hydroxyapatite is nucleated and how hydroxyapatite nuclei grow into the narrow space betw een tropocollagen molecules are cen tral to the question, what is the mechanism of bone formation?. Therefore, there has been a lot of research with various techniques, such as X-ray diffraction, in fra-red spectroscopy and transmission electron microscopy, to characterize bone mineral and to analyze how bone mineral is formed from natural bone samples. X-ray diffraction is a powerful method to inves tigate the crystalline phases. With the observation of bone samples with X-ray diffraction, amorphous calcium phosphate had been considered as a precursor material of bone hydroxyapatite because the intensity of bone hydroxyapatite is always less than that of sy nthetic hydroxyapatite due to the contribution of amorphous calcium phosphate which produces very broad peak . Posner et al.  investigated rat bones of different ages and calculated the contents of amorphous and crystalline calcium phosphate. With results from rat bone, they obs erved the content of
35 amorphous calcium phosphate was reduced and the ra tio of Ca/P was increased as the maturation of bone progressed. Those results were thought to be reasonable because the ratio of Ca/P of amorphous calcium phosphate (1.5), which is incorporated with high content of HPO4 2and water, is far less than that of hydroxyapatite (1.67) and the transformation of metastable amorphous calcium phosphate to stable hydroxyapatite can be eas ily done by the dissolution of amorphous calcium phosphate leading to nucleation of hydroxyapatite [140-142]. On the other hand, Glimcher et al.  investigated embryonic chick bone using radial distribution function (RDF) analysis, and found no evidence of amorphous calcium phosphate in the early stage of bone formation. In their repo rt , they calculated the root mean square (RMS) of RDF over the region 2.5 7.5 wh ere the contribution of amorphous calcium phosphate would be strong, and over the region 10 -25 where the contribution of amorphous calcium phosphate would be nil, and they argued that the ratio of the RMS over 10 -25 to the RMS over 2.5 7.5 would be gradually increased as the maturation of bone increased if the amorphous calcium phosphate was involved in the early stage of bone formation. However, the value of this ratio varied randomly within the range of 0.38 to 0.42 (0.40 for embryo bone, 0.38 for 5-weeks chick bone and 0.40 for 1-year chick bone), implying the probability of having amorphous calcium phosphate was very low. Mo reover, in the case of low integrated peak intensity of bone hydroxyapatite which was considered as an other evidence of the amorphous calcium phosphate, the decrease in the integrated intensity of the x-ray peak was also observed from hydroxyapatite incorporating carbonate, which substitutes the phosphate group in bone hydroxyapatite . From a conventional nucleation mechanism standpoint, their argument seems to be reasonable because the bone mineralization incor porates with collagen matrix. Because calcium
36 and phosphate ions are already s upersaturated in body plasma, heterogeneous nucleation can be induced if the appropriate active site for nucleation is provided. An active site on collagen has been suggested . As shown in Figure 2-8, in the initial stage of cal cification of both avian embryonic bone and collagen fibril, th e crystals are formed within collagen fibrils with regular spacing . This observation implies that co llagen fibrils provide a specific region for heterogeneous nucleation of hydroxyapatite at th e initial stage of calcification . Katz et al. [147, 148] suggested that the molecular circumstances around the gap zone in collagen fibrils may provide the appropriate space for mineral deposition, and Landis et al. [90, 149] also confirmed that hydroxyapatite crys tals nucleate in the gap zone using 3D reconstruction of TEM images. In recent reports, carboxyl groups from am ino acid residues in collagen fibrils, such as glutamic and aspartic acid, were introduced as a key factor of heterogeneous nucleation of hydroxyapatite [45, 50]. In these reports, they argued that the carboxyl groups on the collagen fibrils could attract calcium and phosphate io ns locally and initia te the nucleation of hydroxyapatite. The formation of oriented hyd roxyapatite on a synthetic self-assembly monolayer of arachidic acid (CH3(CH2)18COOH) also supports th e hypothesis that carboxyl groups nucleate hydroxyapatite. Sato [100, 101] reported that hydroxyapatite that nucleated on a self-assembled monolayer grew with  pref erential orientation like bone hydroxyapatite [101, 150]. In that paper, he argued that calcium a nd phosphate ions, which are adjacent to each other at the (001) plane during nucle ation process, are combined with carboxyl groups in the monolayer for hydroxyapatite nucleation, and orde red carboxyl groups in the monolayer may control the crystallographic prope rties of hydroxyapatite. Beside s the carboxyl groups in amino acid residues, carbonyl groups in the collagen b ack bone may also provide nucleation sites for hydroxyapatite on collagen fibril surface . When the mineralized collagen was observed by
37 Fourier transform infrared spectroscopy (FT-IR), the amide I peak at 1675 cm-1, which corresponds to C=O stretch, was shifted to lower wave number due to th e adsorption of calcium ions on carbonyl groups which could attract calcium ions, and also co uld act as an active site for the heterogeneous nucleation of hydroxyapatite on collagen fibrils . Although there was evidence that collagen fibrils can initiate the nucl eation of hydroxyapatite, which supports the conventional nucleation mechanism for the bone mi neralization, efforts to induce intrafibrillar mineralization of collagen by the conventional nu cleation mechanism in the laboratory level were unsuccessful. Therefore, some researchers, including our group, have returned to the old theory of amorphous calcium phosphate serving as a transient mineral precursor phases for bone hydroxyapatite [152, 153]. In marine systems, various transient mine ral phases have been reported from different invertebrate species. The hydrated iron oxide, wh ich is very poorly ordered, is formed as a precursor phase to crystalline magnetite in chit ons . Amorphous calcium carbonate is one of the most famous transient precursors, it form s in many marine species such as mollusk shell  and sea urchin spine . It is t hought that the pr ecipitation of am orphous calcium carbonate may help the formation of non-equi librium morphologies of calcium carbonate crystals because amorphous calcium carbonate can be molded into various shapes [157-160]. With the observations of amorphous calcium car bonate precursor in invertebrate systems, therefore, the speculation of amorphous calcium phosphate as a transient precursor to bone hydroxyapatite seems plausible. Crane et al.  recently reported that the formation of octacalcium phosphate minerals is observed in intramembranous mineralization of murine calvarial tissue, and it is thought to be indirect evidence that b one hydroxyapatite is formed via a transient mineral, and maybe even amorphous calci um precursor . However, to bring this
38 previously rejected theory back into debate, another key factor which may play an important role in the formation of amorphous calci um phosphate, must be introduced. Our group is researching the mimetic peptides emulating the non-collag enous proteins that are found at the bone mineraliza tion front [17, 82]. Non-collage nous proteins contain large amounts of acidic amino acids, such as aspartic acid and phosphoserine, which can have polyanion characteristics by deprotonation of th e acidic functional groups [67, 162-164]. Even though the function of non-collagenous proteins is still elusive, they are thought to play important roles in biomineralization, such as in inhibition or enhancem ent of the nucleation of hydroxyapatite, when they are dissolved in the so lution or they are immobilized on the substrate [165-167]. Termine et al.  also argued that when non-co llagenous proteins were extracted by demineralization processes, the demineralized matrix could not be remineralized without noncollagenous proteins. Recently, our group has su ccessfully produced a mineralized collagen substrate with intrafibrillar mineralization using acidic polymer as a process-directing agent for inducing amorphous calcium phosphate , and based on such observations, we hypothesize that the non-collagenous proteins enriched with acidic residues may he lp the formation of transient precursors to bone hydr oxyapatite. The intrafibrillar mineralization of collagen fibril with acidic polymer will be discussed more in a last section of this chapter. Bone Grafts As discussed previously, the developm ent of bone grafts has b een a serious research thrust for many decades. Therefore, a brief considerat ion of the advantages and disadvantages of various types of bone grafts cu rrently in use will be helpful. Bone grafts have several requirements for replacing damaged bone. First, bone grafts must pr ovide mechanical and structural support for the missing bone part. Bone fillers for the cavity of defective gaps or various metal plates or screws for bridging the fracture are good examples of bone grafts for
39 mechanical purposes. Appropriate chemical prope rties of bone grafts are also required. To enhance the bone healing process, bone graft should provide appropriate surface conditions to decrease immunological response an d increased proliferation of bone forming cells . From a methodological standpoint, autografting is the most preferred way to replace the damaged bone . Because the defective bone is replaced by th e bone graft which is extracted from the patient having replacement surgery, autografting causes lo w immunological response and fast healing. However, limited quantity and the morbidity afte r extraction make it highl y desirable to develop a synthetic bone graft [4, 45]. As shown Figure 2-9, synthetic bone grafts can be categorized into several generations depending of the development of synthetic materials, and each generation has its own characteristics . First Generation of Bone Grafts Before 1950, m any materials were used for biom edical purposes without consideration for the immunological response of the body to the materials . Therefore, many materials implanted in the human body were either toxi c or pathogenic leading to the failure of implantation. In the 1950s and 1960s, the first generation of biomaterials was introduced for clinical use [40, 170]. Ceramic ma terials, such as alumina (Al2O3) and zirconia (ZrO2), and metals, such as stainless steel, Co-Cr alloy and titanium, are exam ples of the first generation of biomaterials used for bone grafts. The goal of the first generation bone graft was to provide physical support for the fracture gap during th e bone healing process or replace the whole damaged bone with minimal toxic response. Although the bio-inert materials do not induce a toxic response with the host, they have an i nnate drawback caused by th eir natural mechanical properties. As shown in Figure 2-10, the mechan ical properties of major ceramic and metal bioinert materials are much higher than those of bone, which leads to the stress shielding effect [4, 44]. According to Wolffs law , when a bone graft that is stiffer than natural bone is
40 implanted, the applied stress on the bone tissue around the bone graft decr eases, and bone tissue with lower stress is gradually resorbed by the re modeling process, leading to the detachment of the bone graft. Moreover, because these 1st generation materials cannot bond to living tissue, an additional fixation of the implant is needed with screws or bone cement, and detachment can occur due to encapsulation of the implant by fibrous tissue . Some polymer materials, such as polyethylene, poly-tetrafluoroe thylene or poly-methyl-metacrylate, were also used for the first generation of bone graft due to the similarity of their mechanical properties to those of bone. While the first generation of polymeric implan ts provides desirable mechanical properties for bone graft, it still causes some problems, incl uding inflammation or foreign body reaction due to the debris produced by imp lant abrasion . Second Generation of Bone Grafts After Hench introduced bioglass, which s hows the direct bondi ng to bone tissue, biom aterials evolved into the second generation [40, 174, 175]. In contrast to the bio-inert materials, the second generation of bone gr afts showed bioactive and biodegradable characteristics . Bioglass is a typical bi oactive bone graft, and it produces hydroxyapatite on the bioglass surface, leading to the direct bonding with re-grown bone [39, 40]. AW glassceramic, in which apatite and wolla stonite crystals are embedded in glass matrix, is also known to form hydroxyapatite and produce direct bondi ng to bone . However, because the ceramic bioactive materials have low toughness and brittleness, the uses of ceramic bioactive materials are limited to low-weight bearing si tes such as cranial bone or maxillofacial reconstruction . In the case of biodegradabl e materials, such as tricalcium phosphate, poly-Llactic acid and poly-lactide-co-glycolide, they can be hydrolytically di ssolved in the human body by a metabolic reaction, leading to gradual repl acement with newly-formed bone. However, because the dissolution rate is somewhat faster th at the rate of bone grow th, greater control of
41 dissolution rate is still required for practical uses in large defectiv e sites , and the variability within individuals and bone types could pose pr oblems with matching rates of bone formation with implant degradation. Surface treatment of bio-inert materials was also studied to combine the advantages of the first and second generation biomaterials [ 38]. In the case of titanium, although it has osteoconductive characteristics, the bonding between bone and the implant is not always guaranteed. Therefore, the titanium su rface is often chemically modified to induce a stronger bond between the titanium surface and ne wly formed bone. For example, titanium is frequently coated with bioactive materials such as bioglass, hydroxyapatite or collagen [178-181]. In these cases, surface modified titanium can be used at the load-beari ng site and it can also produce a more robust interface between implant and bone. Third Generation of Bone Grafts Although first and second generation biom ateria ls have been used as bone grafts for decades, researchers began to consider composite made of the components of natural bone as alternative bone graft materials in the 1990s because the primary goa l of engineering bone grafts is the development of synthetic bone graft analogous to autograft. Bone can be considered at the fundamental level as a nanostructured collagen/hydroxyapatite composite. Therefore, the fabrication of bone grafts with both collagen a nd hydroxyapatite is now being investigated to take advantage of the desirable attr ibutes found in autograft tissues. The simplest method to fabricate a collage n/hydroxyapatite composite is the direct blending of collagen and hydroxyapatite. The co mbination of these two biomaterials was expected to provide an advantage, such as osteoinductivity, over each material and produce a bone-like composite [136, 137, 182]. However, cont rol over the uniformity and homogeneity of the hydroxyapatite/collagen mixture is still a diffi cult problem . Moreove r, in the case of the direct blending method, because hydroxyapatite particles are randomly mixed with fibrous
42 collagen without chemical bonding, there is no structural resemblance to real bone . This structural and compositional difference of such blended composites may hinder the interaction between cells and the composite. The most common method for fabricating collagen/hydroxyapatite composites is to mineralize reconstituted collagen scaffolds us ing the conventional nucleation and growth method for hydroxyapatite [59, 183, 184]. Si mulated body fluid (SBF), whose ionic concentration and composition is similar to body plasma, modified SBF, saturated calcium and phosphate solution are most frequently used as a mineralization solution for this process [176, 185]. As previously discussed, the collage n surface has carboxyl groups and carbonyl groups, which are thought to act as nucleation sites for hydroxyapatite on the surface. The ability of those functional groups to induce the nucleation of hydroxyapatite was investigated with the selfassembly monolayer with the negatively charge d functional groups and the surfactant for the self-assembly monolayer modified with hydrophilic polar groups such as phosphate, carboxyl and hydroxyl groups showed the rapid hydroxyapati te formation [186, 187]. Therefore, to enhance the ability of a collage n surface to induce hydroxyapatite formation, the collagen surface was modified by Goes to be more negatively charged with the sele ctive hydrolysis of carboxyamide side chains of asparagine and glutamine residue, and using this modified collagen, a thick and continuous hydroxyapatite coating wa s produced . However, the formation of hydroxyapatite was limited to the surface of co llagen, and the conventional nucleation and growth strategy has not been able to achieve the intrafibrillar mineralization. As previously mentioned, the structural and co mpositional similarity of bone grafts to real bone may enable some of the remarkable advant ages of autografts, su ch as osteoconductivity, osteoinductivity and low immunologi cal response. Therefore, si nce 1995, various techniques has
43 been introduced to fabricate bone-like composites [50-52, 137, 189]. These techniques are based on the co-precipitation method. In the co-precipitation method, fibril logenesis of collagen fibrils and precipitation of hydroxyapat ite by an increase in pH or the addition of mixing agent are induced at the same time [45, 53]. In this met hod, because hydroxyapatite particles are nucleated during the self-assembly of tropocollagen, hydroxya patite particles can be embedded inside of the collagen fibrils. As shown at Figure 211, the mineralized collage n showed an electron diffraction pattern similar to that of bone, characterized by (002) arcs, which implies that the hydroxyapatite crystals are aligned within the coll agen fibrils with uniax ial orientation . Because collagen/hydroxyapatite composites made by the co-precipitation method resemble mineralized collagen in natural bone, both comp ositionally and structurally, it also showed resorptive characteristics dur ing remodeling processes . However, although the coprecipitation method can pr oduce bone-like composites whose nanos tructure is analogous to that of natural bone, the long range or der of collagen fibril was not obs erved and it is hard to make bulk bone graft without additional treatments from the such as cross-linking processes or forming processes [51, 52, 190]. Recently, biomimetic approaches to fabr icate bone-like collagen/hydroxyapatite composites have been highlighted and are expect ed to prompt the development of the fourth generation of biomaterials for bone grafts [45, 170]. Those approaches include tissue engineering, molecular tailoring fo r cell interaction, and uses of modified proteins for specific uses [4, 170]. For example, to produce biores orbable polymer scaffolds by specific cellular reactions, the polymer molecules can be tailored to induce cellul ar interactions via specific molecular signals, or the immobilization of protei n for the proliferation of typical cell which can produce an enzyme to induce dissolution of polymer. In the case of bone grafts, the
44 immobilization of protein or gr owth factors on the collagen/hydroxyapatite composite may be an example of the new direction of th e fourth generati on of bone grafts. Intrafibrillar Mineralization of Collagen Fib rils with the Polymer-Induced-LiquidPrecursor (PILP) Process Calcium carbonate is a major component of the sh ells of invertebrate marine species . In these invertebrates, calcium carbonate crysta ls have various non-equilibrium morphologies such as fibrous, plate-like or film-like, and molded morphologies . How calcium carbonate skeletons or shells that have non-equilibrium morphologies form has undergone a paradigm shift in the biomineralization area. Polyanionic macromolecules found occluded in biominerals extracted from invertebrate species had been considered morphology-directing agents that selectively adsorb on the specific crystallographic f aces to control the rate of crystal growth [66, 67]. In light of more recen t findings, where an amorphous precursor has been found to be involved in their formation, the concept of stereo selective adsorption has fallen to the wayside. The Gower group has proposed a different function of the poly-anionic macromolecules associated with biominerals, which is based on in vitro observations of a process called the polymer-induced-liquid-precursor (PILP) process. The PILP process was initially discovered in calcium carbonate system [70, 192]. When anionic polymer, such as polyacr ylic acid or polyaspa rtic acid, is added into a mineralization solution containing calcium ions, the negatively charged polymer sequesters calcium ions and carbonate ions that are continuous ly introduced by gas diffusion into the mineralizing solution. When a critical ion concentrati on is reached, a highly hydrated pr ecursor phase is formed, which is recognizable microscopic droplets with th e optical microscope, as the solution undergoes liquid-liquid phase separation (Figure 2-12) . The liquid precursor droplets, initially nanoscopic in size, agglomerate into larger micron-sized droplets, which are then accumulated
45 on the bottom of a reaction vessel via gravity. The fluidic droplets coalesce to make a continuous amorphous film, which solidifies and cr ystallizes into the more thermodynamically stable crystalline phase as water and polymer are excluded. The most important characteristic of the highly hydrated precursor formed by the PILP pr ocess is the ability of the fluidic precursor to be deposited or molded into various shapes, such as tablets, films, fi bers, and that crystals derived from the PILP phase retain the shape delineated by the phase bo undary of the precursor phase [15, 158]. We thought that this process, in which the amorphous fluidic precursor can be molded within a porous matrix, might lead to intrafibrillar mineralization of collagen fibrils, which can be considered as a polymer matrix with interconnected nano-pores . The hypothesis of intrafibrillar mineralizati on of collagen is that the liquid precursor, formed by the combination of anionic polymer and ionic components of hydroxyapatite, can be drawn into the grooves and channels in collagen fi brils through the gap zone s via capillary action. The precursors would then be expected to solidi fy and crystallize into hydroxyapatite as the waters of hydration are excluded from the metast able precursor (Figure 2-13) . It was considered that the combination of interfacial energy between water, collagen fibril channels and precursors separated from mineralizing solutio n by phase boundary woul d draw the precursor into the fibrils by capillary forces. During crysta llization of the precursor, it is thought that the unstable amorphous precursor is thermodynamically tr ansformed into the stable crystalline phase. In comparison to other hypotheses, one needs to consider the case of uniaxial orientation of hydroxyapatite, where it has long been thought th at epitaxial growth of hydroxyapatite along collagen fibrils is responsible. However, in co ntrast to the epitaxial growth mechanism, we speculate that the alignment of hydroxyapatite crystals is the re sult of compartmentalization of crystal growth caused by the collagen fibrils We believe that crystallization occurs
46 spontaneously and continuously in the narrow sp ace generated by collagen molecules, and the growth of hydroxyapatite could be constrained such that a rapid gr owth direction takes dominates, leading to the naturally favored [ 001] direction along the ch annel between collagen molecules. In recent research, randomly or iented hydroxyapatite crysta ls were observed in naturally mineralized hen tendon . As s hown in Figure 2-14, hydroxyapatite clusters without uniaxial orientation were observed in the cavities even though the hydroxyapatite was formed on the collagen fibril, which is expected to induce the epitaxial gr owth of hydroxyapatite. It seems apparent that crystal nu cleation on collagen, at least not on the surface, does not lead to preferred orientation. The formation of bone via amorphous calcium phosphate was abandoned for 20 years after Glimcher  argued that there was no distinct evidence of the presence of amorphous calcium phosphate in embryonic chick bone. This was conc luded when the RDF peaks of chick bone did not show rapid fall-off after the long-range order region. We have presented an argument against this evidence in our resent paper , based on the fact that this is primary bone formation, or only small regions of secondary bone will have am orphous mineral at any given time. Moreover, there has been limited success at in vitro attempts to mimic the intrafibrillar mineralization of collagen fibrils via the conven tional nucleation and growth me thod. On the other hand, the intrafibrillar mineralization of collagen fiber can be achieved by the PILP process induced by acidic polymer. Moreover, there are many kind s of non-collagenous proteins, such as bone morphogenetic proteins, bone sialopr oteins, osteocalcin and osteopontin, that are enriched with acidic amino acids present at the bone mineraliza tion front. We suspect th at these proteins may act as process-directing agents that sequester ionic component s and water to form amorphous calcium phosphate precursors, which may act lik e the fluidic precursors in the PILP process
47 during the bone mineralization process. Among those, bone sialoprotein and osteopontin are possible candidates that could c onceivably act as a process-dir ecting agent used in the PILP process. Even though the proteins are not co mposed of only amino acids which are negatively charged (like poly-aspartic acid in the PILP process), bone si aloprotein or osteopontin has a strong negative charge character, which could be as strong as poly-as partic acid, due to phosphorylated functional groups. Therefore, in a real bone formation, bone sialoprotein or osteopontin may function the same way as of polyasp artic acid does in the PILP process, forming an amorphous calcium phosphate pr ecursor. According to Toroian , while proteins with a molecular weight smaller than 6,000 Da were able to be diffused into t ype-I collagen, proteins with a molecular weight is larger than 40,000 Da are excluded from porous type-I collagen. Therefore, the size of osteopontin (about 44,000 Da) and bone sialoprotein (about 70,000-80,000 Da) may prohibit the infiltration of these proteins into the collagen fibril, leading to incomplete intrafibrillar mineraliza tion of collagen. According to the NMR study of bone sialoprotein and osteopontin done by Fisher , however, those pr oteins are unstructured and very flexible. Those physical characters of the proteins may enable the effective in filtration of the proteins with high molecular weight into the narrow spaces between tropocollagen molecules to achieve intrafibrillar mineralization of the collagen fibr ils. However, demonstration of intrafibrillar mineralization via such as process in vitro ha s not been achieved. Therefore, with recent advances reported in the literature and the achieve ments we have made with the PILP process, we propose revisiting the amorphous precursor theory. Finally, it may be worthwhile to mention that the Glimcher group recently tried to remineralize demineralized bone with the addition of negatively charged poly-L-glutamic acid, which leaded to the enhancement of intrafibrillar mineralization of demineralized bone .
48 Table 2-1. Principle constituents of bone* *The contents of each component are varied depending on the species, age or type. Components Wt% Hydroxyapatite ~60 Carbonate ~4 Minor inorganic phases (Na,Mg,Cl,F,K,Zn) ~2 Collagen ~20 Non-collagenous proteins 3 Water ~9 Others (bone cells, polysaccharides, lipids, cytokines)
49 Figure 2-1. Tropocollagen molecule.
50 Figure 2-2. Collagen fibrils assembled into a quarter-staggered arrangement, and TEM micrograph of collagen fibrils with banding patterns that arise from the corresponding hole and overlap zones .
51 Figure 2-3. Hierarchy of osteona l bone structure, as illustra ted by Weiner and Wagner .
52 Figure 2-4. Based on tomogr aphic TEM images, Landis et al. provide a schematic diagram of mineralized turkey tendon as ossificati on progresses. Tropocollagen molecules (cylinders) are arranged with the quarter-staggered arrangement, marked by a 67 nm repeat resulting from the gap zones. Hydroxyapatite is first observed in the gap zone, and then grows to fill the space between tropocollagen molecules .
53 Figure 2-5. Selected area el ectron diffraction (SAED) patterns of equine bone , and a schematic diagram of mineralized collagen fi brils showing the deck-of-cards structure . The direction of colla gen fibril in A) is indicated by the white arrow. A) B)
54 Figure 2-6. TEM micrographs of A) decalcified zebra fish bone, and B), C) untreated fish bone. For decalcification of bone, formic acid (5 % v/v) was used. Black arrows in B) indicate the orientation of mineralized collagen fibrils, and the average diameter of mineralized fibrils is 151.2 nm. At C), SAED patterns for domain A is inserted, and a schematic diagram of the mineralized colla gen with surface hydroxyapatite was also illustrated. A) B) C)
55 Figure 2-7. Activated osteoclast at the resorptive bone site .
56 Figure 2-8. TEM micrographs of the initial stage of calcification in A) avian embryonic bone and B) unstained collagen fibril which was mi neralized by in vitro experiment . A) B)
57 Figure 2-9. History of bone graft development .
58 Figure 2-10. Stress-strain relations hip of various bone grafts .
59 Figure 2-11. The TEM micrograph of collagen fibril produced by Kikuchi et al. using the coprecipitation method . The SAED pattern of the minerali zed collagen is inset.
60 Figure 2-12. Formation of calcium carbona te film via the PILP process . Liquid-Liquid Phase Separation Deposition of Precursor Film Solidification and Crystallization Birefringent CaCO3 Film
61 Figure 2-13. Intrafibrillar mineralization of a collagen fibril via the PILP process . (a) The precursor droplets formed by anionic polymer (black arrow) adsorb to collagen fibrils and are drawn in via capillary action. (b) The fluidic precursor is distributed throughout the interstitial space of the a ssembled tropocollagen helices. (C) The amorphous precursor then solidif ies and crystallizes into hydroxyapatite with uniaxial crystallographic orientation.
62 Figure 2-14. The SEM micrographs of the minera lized part of hen tendon . Arrows in A indicate the cavities in which hydroxyapatite cluster wa s formed without preferential orientation. Bars indicates 30 (A), 30 (B ), 3 (C), 15 (D) and 3 m (E).
63 CHAPTER 3 MINERALIZATION OF TYPE-I COLLAGEN SPO NGES VIA AN ENZYME-AIDED PILP PROCESS Introduction Various inorganic-organic com posites in biolog ical systems, such as bone, teeth, shell and sea urchin spine, have been investigated to clarify how t hose structures are formed under ambient conditions, what the contribution of thos e structures is on their performance, and how such structures and properties can be mimicked in the laboratory or industry [63, 79]. Among those composites, bone has been widely researched in various areas, incl uding materials science, because bone has a nanostructured architecture providing outstanding mechanical strength and bioresorptive properties [4, 10, 11, 34, 45]. In addition, the demand for a new synthetic bone graft for the replacement of damaged bone is dr amatically increasing. It is thought that the remarkable physical and chemical properties of bone originate from certain structural characteristics of bone [15, 26]. Therefore, th ere have been various attempts to mimic bone structure and develop a bone graft with the structural advantage of bone. Bone has a nano-engineered structure which has been categorized into 7 levels of hierarchy . The foundation of this hier archy is the mineralized collagen fibril, which can be described as an intrafibrillarly mineralized collagen fibril consisting of preferentially oriented hydroxyapatite platelets embedded a nd aligned within the interstices of the co llagen fibrils [104, 197]. The hydroxyapatite crystals preferenti ally nucleate between tropocollagen molecules (triple helices of collagen peptides), which are in turn self-assembled with a quarter-staggered arrangement, generating a periodic array of gap zones . The nanostructure of the mineralized collagen fibril is considered to lie at the f oundation of bones outstanding mechanical properties (strength and toughness) . In addition, the meta stability of the extremely small nano-crystals of hydroxyapatite allows them to be resorbed by osteoclast during the natural remodeling
64 processes of bone . Therefor e, many researchers have tried va rious methods to duplicate this fundamental nanostructure of na tive bone. Many attempts have been introduced to prepare collagen/hydroxyapatite composites via the conv entional nucleation and growth techniques, which initiate nucleation of hydr oxyapatite crystals on the collag en fibers [59, 183]. However, with the conventional nuclea tion and growth method, a hydrox yapatite coat with random orientation is formed on the collag en surface . It has been suggested that this coating might prevent further mineralization of the collagen fibr il beneath the surface, inhibiting the formation of a native nanostructure of bone, which arises from intrafibrillar minera lization [45, 188]. We believe this is not actually the problem, but ra ther the mineralization mechanism is responsible for the lack of intrafibrillar mineral via the conventional crysta llization route. Our lab has successfully achieved the intrafibri llar mineralization of collagen using a novel method, called the Polymer-Induced-Liquid-Precurso r (PILP) process [15, 198]. In the PILP process, a fluidic calcium phosphate precursor is induced by process-directing agents, such as polyacrylic acid or polyas partate, which can then be drawn into gaps and grooves of the collagen fibrils by capillary actio n [15, 198]. Once the mineral precur sor has infiltrated the fibrils, it solidifies and crystallizes into nano-sized hydroxyapatite crys tals with the crystallographic orientation roughly parallel to the fibril axis, generating the (002) ar cs that are typically seen for bone. In the work presented in this chapter, intr afibrillar mineralization of collagen was achieved by an enzyme-aided PILP process, which more closely mimics the chemical environment found during the natural process of bone formation [1 99, 200]. Alkaline phosphatase, which is always present during bone formation, such as in hypert rophic cartilage before ossification, is an enzyme that catalyzes the release of phos phate groups from organic molecules via
65 dephosphorylation [188, 201, 202]. It is used here to provide c ontrolled release of inorganic phosphate groups (PO4 3-) from a commercial organophosphate es ter, rac-glycerol 1-phosphate, to induce amorphous calcium phosphate via the PILP process. Materials and Methods Sample Preparation A Type I collagen suspension (2.4 m l, OAI A 9909 NeuVisc, Neucol l) was transferred into a 2-cm-diameter-well stainless steel tray and washed with de-ion ized (DI) water 5 times using centrifuge. The collagen suspension was fr ozen with liquid nitrogen, and lyophilized for 24 hours under vacuum, to form a loosely pack ed collagen sponge for these mineralization studies. Separate solutions of CaCl2H2O and rac-glycerol 1-phosph ate (Sigma-Aldrich) were prepared in D.I. water to a concentration of 6mM. The mineralization solution was prepared by mixing equal amounts of the CaCl2H2O (6 mM) and the rac-glycerol 1-phosphate (6 mM) solutions , and the PILP process-directing agent, poly-( )-DL-aspartic acid (PolyAsp), was added into the solution at a concentration of 50 g/ml. The lyophilized collagen sponge was placed into the mineralization solution, and air bubbles trapped within the sponge were removed under vacuum for 30 minutes. Then, alkaline phosphatase (0.03 units/ml) was added to the mineralization solution containing 3 mM of calci um chloride and rac-glycerol 1-phosphate and poly-( )-DL-aspartic acid. The pH of the mi neralization solution was around 8.3, which is high enough to maintain the ac tivity of alkaline phosphatase. The reaction was run in a 37oC oven to emulate physiological c onditions, and the mineralization solution was replaced by a new solution every 3 days to provide enough mineral to fully mineralize the collagen scaffold. For etching studies, to examine the interp enetrating nature of the composite, the mineralized sponge was bleached with a 2 % of NaOCl solution for 1 hour, followed by rinsing
66 with de-ionized water several times. To remove the extra hydroxyapatite coat, the mineralized collagen was soaked in an HCl solution (which wa s adjusted to pH 2) for 20 seconds and washed with de-ionized water. X-Ray Diffraction (XRD) analysis X-Ray analysis was used to determ ine the crys tal structure of the precipitates from the surrounding mineralization solution. These prec ipitates were centrifuged at 5000 rpm for 15 minutes and rinsed with de-ionized water followed by ethanol for rapid drying. The powder obtained after drying was transferred to a glass slide and scanned with Cu-K X-ray radiation from a Philips XRD 3720 at 40 KV and 20 mA, using a step size of 0.01 with a time of 1.25 sec/step, over a 2 range of 4 50 Scanning Electron Microscopy (SEM) Analysis The m ineralized collagen sponge was washed with de-ionized water 3 times and lyophilized with a freeze dryer for 24 hours. Dried samples were mounted on an aluminum stub covered in double-sided copper ta pe, and then sputter coated w ith either Au/Pd or amorphous carbon. The morphologies of the surface of the mine ralized collagen were then analyzed using a 6400 JEOL SEM at 15 kV. Transmission electron microscopy (TEM) analysis Nano-structural analysis was perform ed on a 200cx JEOL TEM (200 kV) with bright field (BF), dark field (DF) and selected area elec tron diffraction (SAED) modes to determine the crystallographic orientation of the embedde d hydroxyapatite. For transmission electron microscopy (TEM) analysis, mineralized collage n was pulverized into a powder. The powder dispersed in ethanol was applied as a drop onto a copper TEM grid with lacy carbon. In the case of un-mineralized collagen, pulverized collag en powder on TEM grid was soaked in a 1 %
67 solution of phosphotungstic acid (PTA) for 30 second to stain the collagen fibrils. To prevent electron charging on the sample, it was slig htly coated with amorphous carbon. Confocal Microscopy Study Confocal m icroscopy was used to study the infiltration depth of fluorescently labeled precursor in a turkey tendon scaffold, which ha s densely-packed collagen fibers. This was compared to the same scaffold soaked in a solution containing polyaspartic acid and calcium ions only (without phosphate, to avoid PILP formation). The poly-( ., )-DL-aspartic acid (20 mg) was di ssolved in 2 ml of 0.1 M sodium carbonate buffer, which was prepar ed by mixing 8 ml of 0.2 M of Na2CO3 and 17 ml of 0.2 M NaHCO3. Fluorescein isothiocyanate (FITC, 5 mg) was dissolved in 0.5 ml of dimethyl sulphoxide, and then 200 l of FITC solution wa s gently added to the polymer solution. The solution mixture was sealed from light and incu bated at 4C in the refrigerator for 24 hours. After incubation, the polymer solu tion combined with dye was cen trifuged in a Centricon spin column with molecular weight cutoff of 3 kDa to remove any un-reacted dye. The final concentration of polymer was assumed not to ch ange because its average molecular weight is above the cutoff, although the polymer is not mo nodisperse, so there was likely some loss of lower molecular weight chains below the 3 kDa Centricon cutoff. The polymer labeled with FITC was used as a process-directin g agent for the PILP process. An Olympus Fluoview 500 confocal scanni ng unit mounted on an IX-81 inverted fluorescence microscope was used with an Ar la ser light source (488 nm excitation) and 505-525 nm bandpass filter. Scanning depths along the Z-direction were chosen to be 500 m with a 2.5 m step size, where the Z-directi on corresponds to a depth profile into the cross-section of the
68 tendon. A photomultiplier voltage (PMT) of 400V was used to optimize the fluorescence intensity, which is very bright near the surface. Results and Discussion Analysis of Precipitates from the Mineraliz ation Solution A preliminary study was done to test the conditio ns needed for enzymatic release, where it was found that alkaline phosphatase stimulated the release of phosphate ions from rac-glycerol 1-phosphate when the solutions pH was above 8.0. As shown in Figure 3-1, the initial pH of the mineralization solution was around 8.2 8.3, dependi ng on the concentration of poly-aspartic acid added to the solution. It is expected that alkaline phosphatase stimulates the release of phosphate ions at this pH. Th e pH of the mineralization solu tion gradually decreased as the reaction progressed. In the case of the control experiment, whic h did not contai n poly-aspartic acid, the pH of the solution rapidly dropped to around 6.8 within 10 hours of reaction, and white precipitates, which were birefr ingent under polarized beam, were formed after only 6 hours of reaction. The decrease in pH could be cau sed by the rapid formation of hydroxyapatite (Ca10(PO4)3(OH)2) leading to the consumption of OHions . In contrast to the control experiment, the pH of the PILP mineralizati on solution containing 25 or 50 g/ml of polyaspartic acid slowly decreased and stabilized after 10 hours of reaction. A gel-like precursor phase was formed in the solution surrounding th e collagen scaffold after 10 hours of reaction, and when it was examined by polarized optical microscopy, it was found to be amorphous phase (non-birefringent). It is thought that the addition of polymer inhibits the di rect crystallization of hydroxyapatite at the initial stag e and induces the formation of this gel-like calcium phosphate precursor. When the precipitate collected from the control was examined by x-ray diffraction, hydroxyapatite was observed after 24 hours of reacti on (Figure 3-2). On the other hand, the gellike precursor formed from the enzyme-aided PILP process (with 25 g/ml of poly-aspartic acid)
69 had crystallized into a mixture hydroxyapatite and brushite (CaHPO4H2O) after 24 hours of reaction. In the case of the enzy me-aided PILP process with 50 g/ml of polymer, brushite was the major crystalline phase after 24 hours of reaction. This sugge sts that the gel-like precursor initially transforms into brushite, which is a metastable calcium phosphate phase, leading to the pH plateau at around 7.6. When th e precipitates from the PILP pro cess (induced with either the 25 or 50 g/ml of poly-aspartic acid) were aged in the mineralization solution for 10 days, the brushite fully transformed into hydroxyapatite Judging by the intensity and broadness (half width) of peak at 32, the hydr oxyapatite produced by the 25 g/m l of poly-aspartic acid seems to have smaller crystal size and less defects th an hydroxyapatite formed from the 50 g/ml of poly-aspartic acid (Figure 3-3). Brushite and octacalcium phosphate are both metastable calcium phosphate phases but are more stable than the first formed amorphous calcium phosphate phase [143, 152, 161]. Therefore, it is thought that the poly-aspartic acid inhibited the rapid precipitation of hydroxyapatite, and the crystallization process followed the kinetically-driven route, first forming an amorphous precursor to hydroxyapatite, then passing through the formation of the metastable polymorph of brushite. It is also worthwhile to note that the three hydroxyapatite peaks around 32 ((211), (112) an d (300)) are overlapped and make one broad peak (Figure 3-2), presumably because the hydroxyapa tite crystals are either very small, or have defects, such as calcium deficiency and ion substitution [38, 203]. As shown in Figure 3-4, the morphology of the hydroxyapatite formed by the control experiment was typical of hydroxyapatite cluste rs that normally form by the conventional nucleation and growth route from calcium a nd phosphate supersaturated solutions. Although alkaline phosphatase pr oduces glycerol [HOCH2CH(OH)CH2OH] as a byproduct of dephosphorylation of rac-glycerol 1-phosphate, glycerol has not been found to change the
70 morphology of hydroxyapatite, indicating that glycerol is not involved in th e reaction. Because glycerol does not have a negatively charged hydr oxyl group due to the high pKa value of the primary alcohol (pKa=14.15), it is not expected to have much interaction with hydroxyapatite crystals or ions [204, 205]. Analysis of Collagen Sponge Mineralized by Enzyme-Aided PILP Process In the case of the collagen sponge m inerali zed with the enzyme-aided PILP process containing 50 g/ml of poly-aspartic acid, the co llagen fibers and surface were coated with mineral after 4 days of reaction (Figure 3-5 B), which could be readily identified because of the rough appearance of the fibrils, which is distinctly different from the non-mineralized scaffold, for this collagen sponge, showing a smooth fibrils prior to minera lization (Figure 3-5 A). In contrast to the hydroxyapatite clusters from th e control experiment (F igure 3-4), the initial hydroxyapatite coating was not in a cluster form but was a continuous and rough film which was apparently formed by the coalescence of the fluidic calcium phosphate precursor. As the mineralization progressed, the rough film-like coa ting turned into a smooth hydroxyapatite layer after 11 days of reaction, because an additional adsorption of fluidic precursors on the initial rough coat layer caused a smoother coating (Figure 3-5 C). In the case of the long mineralization time, because the formation of the precursor occurr ed at the initial stage of the reaction (within 24 hours), the mineralization solution was replac ed by a new solution every 3 days to keep providing fluidic precursors. Ne w precursors formed by the soluti on replacements are thought to cover the rough coat continuously to form this smooth overcoat of hydroxyapatite which covers the entire surface of the collagen sc affold, to the point that the fibrous texture of the collagen is masked by the thick overcoat. Hydroxyapatite is soluble under ac idic conditions. Therefore, to remove the excessive hydroxyapatite layer formed after 11 days of reaction, the mineralized collagen sponge was
71 soaked in an HCl solution (pH 2) for 20 seconds. As shown in Figure 3-6 (left), some regions still had a thick hydroxyapatite layer (top regi on); however, in the regions where excessive hydroxyapatite was removed (bottom region), mine ralized collagen fibers with a fibrous structure could be seen (see one fiber (arrow) protruding from th e thick hydroxyapatite layer). This demonstrates that the individual collag en fibrils on the collagen sponge had been completely covered by the hydroxyapatite layer form ed by the adsorption of fluidic precursors as the reaction progressed. A 2 % of NaOCl solution was used to remove collagen, and it was expected that any unmineralized part of collagen sponge, which was not protected by the crystalline layer, would be dissolved. When the collagen spo nge mineralized by the enzymeaided PILP process for 11 days was treated with NaOCl solution, most of the sponges surface was intact due to the continuous hydroxyapatite layer protecting the underlying collagen. As shown in Figure 3-6 (right), there was no morphological change in the collagen surface protected by the smooth hydroxyapatite coa ting (right bottom region of right image). Moreover, at the crack on the excessive hydroxyapati te layer (center region), indivi dual collagen fibers that were exposed to NaOCl solution remained intact. This implies that individual collagen fibers were also protected by a very thin hydroxyapatite coat or there was insuffici ent exposure time since the crack had formed. We believe the native structure of collagen is important for achieving intrafibrillar mineralization of collagen. The fluidic precursor formed by the PILP process is thought to be drawn into the fibrils through the gap zones that are generated by the quarter-stagger arrangement of the collagen molecules. As show n in Figure 3-7 A, the collagen fibrils used for the experiment have a native banding pattern wi th 64 nm of periodicity, indicating they should be suitable for intrafibrillar mineralization. When the collagen sponge was mineralized by the
72 enzyme-aided PILP process for 11 days (Figure 37 B), the collagen fibrils showed dark contrast in the TEM, which came from the hydroxyapatite cr ystals (the collagen was not stained). The diameter of this collagen fibril was over 400 nm which is likely the result of two or three collagen fibrils overlapping. In the case of br ight-field TEM image of mineralized collagen (Figure 3-7 B), individual crystals of hydroxyapatite cannot be observed. However, in the selective area electron diffracti on pattern (Figure 3-7 C), (002) arcs were observed, and those arcs are strong evidence of the uni axial orientation of hydroxyapatite within the collagen fibril. If hydroxyapatite crystals were perfectly aligned along the fiber direction, diffraction from the (002) planes should make a set of 2 spots, inst ead of the arcs, which i ndicate a degree of misorientation. Thus, the hydroxyapa tite crystals within the collagen fibrils are roughly aligned along the fiber direction with about 30 ( 15) of misori entation. To visualize the orientation of hydroxyapatite, one of (002) arcs was chosen by the objective aperture and a dark-field TEM image was constructed by the electrons from onl y the (002) diffraction. As shown in Figure 3-7 D, the needle-like bright spot s are well aligned along the fiber c -axis. In most of the recent literature, the shape of the hydroxyapa tite crystals in bone is consid ered as a platelet. However, in the dark-field image, needle-like spots were observed because the side-view of thin hydroxyapatite plates is projected on a 2D film. These TEM imag es provide strong evidence of intrafibrillar mineralization of collagen vi a the enzyme-aided PILP process. However, this evidence of the intrafibrill ar hydroxyapatite (TEM image) were not observed from all collagen fibers on the TEM grid observed, indicating the degree of intrafibrillar mineraliza tion was not very high. As shown in Figures 3-5 and 3-6, individual collagen fibers and whole surfaces were covere d with hydroxyapatite, even after 4 days of reaction. This surface hydroxyapatite is thought to prevent the further penetration of ions or
73 precursors into the inside of the collagen sponge. Because the formation of the fluidic precursor and the transformation of the precursor occurs wi thin 24 hours of reaction, a replacement of the mineralization solution by new solution at shorter time periods, such as 16 hours or 24 hours, could reduce the formation of surface hydroxyapatite and increase the degree of intrafibrillar mineralization of the collagen fibrils. Study of Depth of Penetration of Po lymer Incorporating w ith Precursor The infiltration depth of the fluidic precursor was examined by tracing the FITC-labeled poly-aspartic acid with a confocal microscope. In this experiment, turkey tendon was used as a substrate instead of the collagen sponge becau se turkey tendon has densely-packed collagen fibrils which can prevent the migration of FITC -labeled polymer through pore space, requiring that the transport of the polymer be through the interstices of the fibrils. Therefore, the difference between infiltration of polymer by diffusive transport (as a nucleator protein) versus capillary forces (PILP mechanism) will be empha sized. The FITC-labeled polymer was added in two types of solution: the mineralization solutio n for the enzyme-aided PILP process and 3 mM of calcium solution. In the form er case, capillary forces are ex pected to provide rapid and long distance transport of a fluidic precursor (which presumably contains the FITC-polyaspartate), while in the latter case, it was expected th at diffusion of a macromolecule through a porous matrix would be much slower and lead to lowe r penetration depth of this same polymer. In the case of polyaspartate with calcium ions, the penetration depth due to diffusive transport was around 100 m. On the other hand, the polyaspartate entrap ped within the fluidic precursor showed over 500 m of penetration depth. Polyaspartate may be in a somewhat extended conformation in water because the carboxylate functional groups will mostly be deprotonated and negatively charged at high pH. When cations such as calcium or sodium are added into the polyaspartate containing solution, the polyaspartate chains may partially collapse
74 because of the charge screening by cations, and even intramolecular bridges with the divalent cation. On the other hand, the size of polyaspar tate with calcium and phosphate ions does not change much (as shown in Chapter 6 for poly-aspart ic acid in various solu tions). Therefore, if the infiltration of polymer th rough the turkey tendon was done by diffusion, the polyaspartate with calcium ions alone (with collapsed chains) would be expected to migrate further. However, the result of the confocal micros copy experiment was the opposite. We believe that the confocal microscopy observations strongly su pport our hypothesis that the flui dic precursor is drawn into the interstices and gap zones of the collagen, not by diffusion, but by capillary forces. Moreover, the prevailing hypothesis about the function of the acidic non-collagenous proteins in bone, in which it has generally been assume d that the protein migrates into the gap zone and stimulates the nucleation of hydroxyapatite should be re-considered In Figure 3-9, a schematic diagram represen ting of the collagen sponge mineralization by the enzyme-aided PILP process is illustrated. At the initial stage of r eaction (less than 12 hours of reaction), the fluidic precursor for calcium phosphate is formed by the poly-aspartic acid process-directing agent. The fluidic precursor ad sorbs to the collagen fibrils and is drawn into the gap zones of fibrils by capillary force. The precursor phase with in the collagen fibrils solidifies and crystallizes into, possibly through a brushite intermediate phase, which then transforms into hydroxyapatite quickly as the reaction progresses. Further addition of precursor adsorbs onto the collagen surface to form a thick hydroxyapatite coating. Conclusions A collagen-hydroxyapatite com posite was fa bricated by the PILP process using an enzymatic process for phosphate release. SAED of the mineralized collagen shows the typical pattern of native bone, which is usually represented by a series of arcs ((002) and (004)) indicative of the lattice planes with roughly uniaxial  orientation. Confocal microscopy
75 demonstrated that the precursor phase can ach ieve a good depth of penetration into dense collagen scaffolds, and supports our hypothesis that the polymer en ters the collagen via capillary action, and not diffusive transpor t. With further optimization, the PILP process may be a promising method of fabricating bon e-like composites. This work shows that the PILP process can be induced using an enzymatic method for phos phate release, bringing it a step closer to emulating the physiological conditions present in bone formation.
76 0510152025 6.4 6.6 6.8 7.0 7.2 7.4 7.6 7.8 8.0 8.2 8.4 0 Pasp 25 Pasp 50 Pasp pHTime (hour) Figure 3-1. The pH of mineralization solutio ns containing various concentrations of polyaspartate points to a cl ear difference in the crystallization pathways. The concentration of calcium ion in the mine ralization solution was fixed at 3mM, and alkaline phosphatase was added (0.003 un it/ml) to provide a gradual release of phosphate ion from the organic phosphate-est er whose concentration was fixed to 3 mM. Reaction temperature was 37oC.
77 Figure 3-2. The XRD patterns of precipitates collected from the mineralization solutions containing various concentration of polyasp artate (0, 25, 50 g/ml) after 1 day of reaction. Metastable brushite was found in the polyaspartate directed reactions. Brushite may have also have occurred in the control reaction, but being short-lived, was only observed with the stabilizing influence of the polymer) 01020304050 B B B H HH H H H HH : HydroxyapatiteB : BrushiteH B B B B B B B50 pasp 0 pasp 25 pasp Intensity2(degree)
78 Figure 3-3. The XRD patterns of aged precipita tes collected from mi neralization solutions containing various concentration of polyasp artate (0, 25, 50 g/ml) after 10 days of reaction. All peaks can be assigned to hydroxyapatite at this time point, but the higher polymer concentration s hows more peak broadening. 01020304050 H H HH : Hydroxyapatite50 pasp 25 pasp 0 pasp Intensity2(degree)
79 Figure 3-4. Scanning Electron Micr oscopy (SEM) micrographs of a collagen sponge (left) that was mineralized for 4 days, and the magnified image of the surface precipitate (right). In this case, no polyaspartate was added to the mineralization solution, while the other conditions were the same as before with concentrations of calcium ion and organic phosphate-ester fixed to 3mM, and alkaline phosphatase added at 0.003 unit/ml. 10m 1m
80 Figure 3-5. The SEM micrographs of collagen sp onges before and after mineralization. A) A non-mineralized collagen sponge, which wa s rinsed with de-ionized water and lyophilized for 1 day. B) A collagen sponge mineralized by the enzyme-aided PILP process for 4 days and rinsed with de-ion ized water several times. C) A collagen sponge mineralized for 11 days. Mineraliza tion solutions contained 3 mM of calcium chloride, 3 mM of organic phosphate-ester, 0.003 unit/ml of alkaline phosphatase and 50 g/ml of polyaspartate. To provide a more continuous supply of calcium phosphate liquid precursor, the collagen sponge was transferred to a new mineralization solution every 3 days. A) B) C) 1m 5m 5m
81 Figure 3-6. The SEM micrographs of mineralized collagen after aci d treatment (left) and bleach treatment (right) to selectively remove the mineral and organic phases, respectively. The collagen sponge was mineralized by the enzyme-aided PILP process for 11 days. The mineralized collagen sponge was soaked in HCl solution (pH 2) for 20 seconds to remove an excessive hydroxyapatite coati ng, or 2 % of NaOCl solution for 1 hour to remove organic collagen that was not prot ected by the hydroxyapatite coating. Scale bars = 5 m (left) and 1 m (right).
82 Figure 3-7. Transmittance Electron Microscopy (TEM) micrographs of the collagen sponge before and after mineralization. A) TEM of a non-mineralized collagen fibril, which was stained by 1 % of phosphotungstic acid fo r 30 seconds, shows the native periodic banding patterns. B) Bright-field TEM micrograph and C) selective area diffraction patterns (SAED) of a collagen sponge that was mineralized for 11 days with the condition mentioned in Figur e 3-5. D) Dark-field TEM micrograph showing the oriented hydroxyapatite crysta ls that were illuminated by using the 002 arc selected by the objective aperture. This appears to be two overlapping fibrils that are slightly offset, giving a double spot/arc in th e SAED pattern. Bar in A) = 100 nm A) B) C) D)
83 Figure 3-8. Confocal fluorescence mi croscopy images of FITC-label ed polyaspartate transported into a turkey tendon scaffold. These show th e penetration depth of polyaspartate used in the enzyme-aided PILP process (3 days of mineralization) into the turkey tendon scaffold. FITC-labeled polyaspartate in corporated only with calcium ion (left) showed penetration ability less than 100 m. But, FITC-labeled polyaspartate formed in the calcium phosphate precursor (right) in filtrated the turkey tendon more than 500 m. Each bar on left scale represents 100m.
84 Figure 3-9. Proposed mechanism of intrafibrillar mi neralization via the enzyme-aided PILP process. Amorphous calcium phosphate precursor is formed in the mineralization solution when the polyaspartate is added to the enzyme-aided PILP process. The droplets adsorb to the collagen fibrils and are drawn into the gap zones by capillary forces. The precursor phase then solidifies and crystallizes, leading to the formation of intrafibrillar HA nano-crystals. If th e mineralization is extensive, a mineral coating will also form on the surface of the fibrils, forming a continuous hydroxyapatite layer that masks the fibr ous texture of the underlying fibrils. Oriented hydroxyapati Continuous hydroxyapatite overcoat Collagen Fiber Fluidic precursor Ga p Zone Alkaline phosphatase PO4 3Phosphate ester Ca2+ PO4 3PO4 3-Ca2+Collagen fibril Ca2+
85 CHAPTER 4 INTRAFIBRILLAR MINERALIZA TION OF BOVINE ACHILLES TENDON VIA THE PILP PROCESS Introduction Bone is m ainly composed of inorganic crystals called hydroxyapatite, collagen fibrils as an extracellular matrix, non -collagenous proteins and water [10, 79, 80]. Those structural motifs, such as hydroxyapatite and collagen fibrils, ar e systematically connected leading to the hierarchically nanostructured ar chitecture of bone. Weiner et al [10, 11] categorized bone into 7 levels of hierarchies: from the main component s of collagen and mineral, to whole bone. The structural foundation of bone is composed of mineralized collagen fibrils, in which hydroxyapatite plates are embedded with uniaxia l crystallographic orientation. The collagen fibril is composed of tropocollagen molecules which are self-assembled with a quarter-staggered arrangement leading to the periodic gap zones in collagen fibrils [81, 83, 84, 86]. Hydroxyapatite is initia lly found at the gap zone of colla gen fibrils, and spreads along the grooves between tropocollagens as the intersti tial space becomes mineralized [80, 147, 148]. These mineralized collagen fibrils with interpen etrated hydroxyapatite nano-crystals are the basic building blocks of bone, in which the arrangement of mineralized collagen fibrils is organized into higher level structures, leading to its rema rkable mechanical strengt h [15, 26]. Besides the mechanical strength of bone, th e poor crystallinity of hydroxyapatite caused by defects, such as substituted carbonate and calcium vacancies, and large surface area of th e nano-sized crystals, help the bio-absorptivity of hydroxyapatite crys tals during the remodeling process of bone [15, 18, 26]. Although the formation of bone has been invest igated for decades, th e mechanism of bone mineralization is still being debated. It is thought that organic substances, such as matrix vesicles, non-collagenous proteins and collagen, play an important role in the mineralization of
86 collagen fibrils [67, 76, 132, 145, 162]. Since the in fluence of those organic substances on the mineralization process is not yet clarified, many competing proposed mechanisms involving those organic substances have been introdu ced. Among those organic substances, we pay attention to the non-collagenous proteins, which are found at the minerali zation front [17, 82]. Non-collagenous proteins, such as bone sialoproteins, osteopontin, and biglycan, are highly acidic due to the carboxylate functionality of the side chai ns of aspartic and glutamic amino acids, and it could provide a binding site for calcium ions, along with sequestering phosphate ions simultaneously [ 163]. We speculate that locally concentrated calcium and phosphate ions by acidic proteins might form a fluidic calcium phosphate precursor, which could be delivered into collagen fibr ils by capillary action, subseque ntly being transformed into hydroxyapatite [15, 198]. Gower et al. [70, 71] discovered that when acidic polymer is added into a crystallizing solution as a process-directing agent, the poly mer can induce fluidic precursor phases of inorganic crystals via liquid-phase separation in the cal cium carbonate system, and that process was named the PILP (polymer-induced-li quid-precursor) process. With poly-aspartic acid mimicking the electrostatic environment of acidic non-collagenous pr oteins, we successfully mineralized collagen fibrils with the nano-st ructure of bone via the amorphous precursor mechanism . We hypothesize that the fluidic calcium phos phate precursor formed by the PILP process could infiltrate into the collagen fibrils via the gap zone and spread throughout the fibrils by capillary action. Then, the fluidic precursor could be solidified and crystallized into hydroxyapatite by the thermodynamic driving force. Previously, we have had promising results mineralizing reconstituted collagen sponges, which are purified and loosely packed, via the PILP process . However, we wished to mineralize a natural extracellular matrix with de nsely packed and oriented collagen fibrils in
87 order to determine the degree of infiltration that can be achieved in a dense packed scaffold, and to examine the x-ray diffraction of a bulk sample consisting of well oriented scaffold (since electron diffraction only measures isolated fibrils). Turkey te ndon is one of the biological substrates that has been used as an appropriate model to inve stigate the mineralization of bone because its collagen fibers are parallel and dens ely packed like collagen in bone [104, 206]. In addition, because tendon is naturally mineralized with intr afibrillar minerals as the turkey ages, it is considered an accessible model of secondary bone formation ( i.e., it is not based on a remodeling of primary endochondral bone, where it is difficult to isolat e the different bone structures). We have mineralized turkey tendon via the PILP proces s; however it is hard to argue that the intrafibrillar mineralization is accomp lished only by the PILP process because turkey tendon might contain other key factors, such as matrix vesicle or non-collagenous proteins, for the intrafibrillar minera lization. Therefore, in this study, bovine achilles tendon was chosen as a substrate to determine if intr afibrillar mineralizati on of a normally non-mineralizing tissue could be accomplished via the PILP process. Bovine ac hilles tendon is composed of densely packed collagen fiber bundles which are surrounded by a white sheath called epitendineum, and achilles tendon is not naturally mineraliz ed. Therefore, bovine tendo n can be considered as an appropriate extracellular matrix that can emphasi ze the effectiveness of the PILP process on the intrafibrillar minera lization . In this chapter, there are three main topics related with the intrafibrillar mineralization of bovine tendon. Firstly, the intera ction between bovine tendon and a fluidic precursor formed by the low molecular weight polymer. Secondly, the effect of molecular weight of polymer on the intrafibrillar mineralizati on will be also investigated. Las tly, the mineralization of bovine tendon
88 with and without EDTA-treatment for removal of calcification inhibitors in bovine tendon will be discussed. Materials and Methods Sample Preparation Bovine achilles tendon was kindly donated from the Meat Processing Center in University of Florida. Extracted tendon was ke pt in the 50 % of ethanol at 4oC. Because tendon has oriented collagen fibers, when tendon is scanned by an X-ray diffractometer, the peaks from the collagen fibrils can be diminished or pronounced depending on the orientation of the collagen fibrils with respect to the stag e of x-ray diffractometer (and th erefore the incident beam). Therefore, tendons were cut perpen dicular to the collagen fiber dir ection to investigate the effect of fiber orientation on the x-ray diffraction analysis, and disc-lik e tendon with 1 cm diameter and 1-2 mm of thickness were obtained. For some experiments, bovine tendon was minced to remove the influence of collagen fiber or ientation on x-ray diffraction analysis. The mineralization solution was prepared by mixing equal amounts of CaCl2H2O (9 mM) and K2HPO4 (4.2 mM). To maintain the pH of th e mineralization solution at 7.4 during the reaction, each stock solution was made by dissolvi ng chemicals in tris-buffer solution containing 0.9 wt% of NaCl. For the PILP process, poly-aspa rtates with different molecular weight were used as a process-directing agent. Poly-( )-DL-aspartic acid sodium salt (Mw : 5500 Da) or poly-L-aspartic acid sodium salt (M w : 10,300 or 32,200 Da) was added to CaCl2H2O (9 mM) stock solution at various concen trations. Tendons were soaked in the mineralization solution, and any air bubbles in tendon were removed under vacuum for 30 minutes. The reaction was run in a 37oC oven to emulate physiological conditions. After the mineralization, tendon was washed with de-ionized water several times and lyophilized for 24 hours to reduce the structural change in collagen fibrils during the drying process.
89 X-Ray Diffraction (XRD) analysis The ordered structure of collagen fibrils pr odu ces three distinct broad peaks in X-ray diffraction analysis . As shown in Fi gure 4-1, the crystallogra phic repeat units in polypeptides (may be distance between pitch and pitch in helix chain) produce a peak at around 31 (d = 0.29 nm), the average distance between helix chains produces a very broad peak at around 20 (d = 0.45 nm) and the la teral space between tropocollagen molecules produces a peak at around 8 (d = 1.08 nm). On the other hand, carbonated hydroxyapatite has major peaks at around 26 ((002) plane) and 32 ((211) plane) [4, 208, 209]. Therefore, there may be overlaps of peaks between hydroxyapatite peaks and broad collagen peaks. Moreover, because the collagen fibrils in bovine tendon is ali gned with the uniaxial di rection, the intensities of collagen peaks can be diminished or pronounced depending on the direction of collagen fibrils against the incident X-ray beam and the typical peaks of hydroxyapatite can overlap with the exaggerated collagen peak . For the x-ray diffraction analysis of mineralized bovine tendon, therefore, two types of sample arrangements, shown in Figure 4-2, were used to avoid peakoverlap and to distinguish the typical peaks of carbonated hydroxyapatite from collagen peaks. The sample setup for the oriented collagen fiber experiment is borrowed from the x-ray diffraction setup with a transmittance mode which was used in the turkey tendon experiment (shown in Figure 4-3). In the transmittance mode of x-ray diffraction, a beam diffracted by the sample should have a cone shape due to randomly distributed diffracting units. However, as shown in Figure 4-3 (below), when the prefer entially oriented collagen fiber is aligned perpendicular to an incident beam and the path of detector movement is assumed to be an imaginary Ewalds sphere on which the incident beam passes through, the only diffracted beam whose beam path is on the imaginary Ewald sphere can be detected by a detector. Therefore, in the transmittance mode with preferentially oriented sample shown at Figure 4-3, the
90 crystallographic plane almost parallel to the inci dent beam can diffract the beam more due to the many domains of the ordered unit, leading to an exaggeration of peak intensity of the parallel plane. In the case of preferen tially oriented hydroxyapatite, the crystallographic planes parallel to the incident, such as (002) and (004), c ould make exaggerated peaks shown at Figure 4-3 (upper). In the case of the coll agen fibers, the repeat units (F igure 4-2 a) aligned along the c -axis of oriented collagen fiber could make a peak around 13.5 (transmittance mode, Figure 4-3) corresponding to a peak at 31 (d = 0.29 nm) in normal x-ray analysis using Cu-K x-ray. Therefore, in this experiment, we assumed that, even though x-ray diffractometer in the University of Florida has reflectance mode, th e geometric condition between reflectance and transmittance modes might be analogous, leading to the exaggeration of peak intensity, when the oriented collagen fiber is used (this assumption was verified by the experiment which will be discussed in this chapter). Ther efore, as shown in Figure 4-2, when the collagen fiber is aligned as illustrated in mode I, the peak of collagen at 31, which is produced by the crystallographic repeat units along the fiber direction, is exaggerated and the peak at 20 is diminished. Therefore, the hydroxyapatite peak at 32 can overlap with th e collagen peak (31) but, the (002) peak of hydroxyapatite (26) is supposed to be distinguished from the collagen peak. In contrast to mode I, when mode II (Figure 4-2) is used, the collag en peak at 31 is subdue d and the hydroxyapatite peak at 32 is disti nguished from the collagen peak at 31. Mineralized bovine tendon with mode I or mode II arrangement was scanned with Cu-K X-ray radiation from a Philips XRD 3720 at 40 KV and 20 mA, using a step size of 0.01 with a time of 1.25 sec/step, over a 2 range of 4 40 (or 50).
91 Scanning Electron Microscopy (SEM) Analysis Mineralized tendon was washed with de-ionized water seve ral tim es and lyophilized by freeze drying for 24 hours. Dried samples were mounted on an aluminum stub covered in double-sided copper tape, and then sputter coated with amorphous carbon. The morphologies of the surface of mineralized collagen were then analyzed using a 6400 JEOL SEM at 15 kV. For elemental analysis of mineralized tendon, energy dispersive x-ray spectroscopy (EDS) analysis was also done during SEM analysis. Transmission electron microscopy (TEM) analysis Micro-structural analysis wa s perform ed on a 200cx JEOL TEM (200 kV) with bright field (BF), dark field (DF) and selected area elec tron diffraction (SAED) modes to determine the crystallographic orientation of the embedde d hydroxyapatite. For transmission electron microscopy (TEM) analysis, mineralized tendon was pulverized into powder and then the powder dispersed in ethanol was applied as a drop onto a copper TEM grid. In the case of unmineralized tendon, pulverized collagen powder on TEM grid was soaked in 1 %phosphotungstic acid (PTA) for 30 second to stain the collagen fibrils. To prevent electron charging, the samples on TEM grids were slight ly coated with amorphous carbon. For the compositional analysis, EDS attached to the TEM was done during TEM analysis. Thermogravimetric analysis (TG) Tendon was pulverized using liquid nitrog en and air-dried for 24 hours for the therm ogravimetric analysis. Tendon powder (10 20 mg) was transferred into Pt pan and fired under air condition. The heating rate was 5C/ minute and analysis was done between 30 and 1200C. To compare the degree of mineralizatio n, the weight loss value at 600C was used because organic substances in tendon were totally burned off after 600C.
92 Results and Discussion Mineralization of Bovine Tendon with Lo w Molecular Weight Poly-aspartic Acid Mineralization of turkey tendon as a preliminary experiment When turkey tendon, which has oriented collagen fibrils, was exam ined by the powerful synchrotron X-ray source ( = 0.652549 ) with a specific samp le arrangement shown in Figure 4-3, the collagen peak generated by the organized repeat units in the direction of the fiber axis was exaggerated. As shown in Figure 4-3, pure turkey tendon with a vertical arrangement (similar to mode I in Figure 4-2) to the incident beam has a broad peak at d = 0.29 nm due to the repeat units along the c -axis of collagen fiber. As the in trafibrillar minera lization progressed by the PILP process, the (002) and (004) peaks were exaggerated, relative to what one would expect for a randomly oriented sample, which normally show s a lower relative intensity than that of the (211) peak at d = 0.278 nm. This is because the ge ometry of the detector had a preference for the diffracted beam from the planes which are para llel to the incident beam. With vertical arrangement of turkey tendon, the alignment of (002) planes of intrafibrillar hydroxyapatite are parallel to the incident beam, leading to the exaggeration of (002) peak. After 4 days of mineralization, the (211) peak started to be developed from extraneous interfibrillar hydroxyapatite that was randomly deposited on the collagen surface without preferential orientation. Comparison of this time series in the other mode, when turkey tendon was horizontally mounted on the stage with the fiber orie ntation parallel to the incident beam (similar to Mode II in Figure 4-2), the development of (002) and (004) peak s by the preferential orientation of hydroxyapatite was not observe d, and the (211) peak of hydroxyapatite was developed from 3 days of reaction (Data is not shown here). Furt her mineralization studies were also performed at UF, where samples were examined by the x-ray diffractometer in MAIC, which uses the reflectance mode in contrast to the synchrotron x-ra y. Since bovine tendon has
93 oriented collagen fibers, the fibe r orientation against the incident beam should be considered to help distinguish the peaks of hydroxyapati te from exaggerated collagen peaks. Mineralization of bovine tendon When bovine tendons w ere mineralized by the PILP process, hydroxyapatite formation was not observed by the x-ray diffraction analys is. As shown in Figure 4-4, pure bovine tendon has a broad peak at 31 due to the organized repeat units of collagen fibrils (d = 0.29 nm) when tendon was scanned by x-ray in mode I. In the case of the control sample (tendon mineralized without an addition of polymer in the mineraliza tion solution), the (002) peak was developed at 26 and the (211) peak, which is supposed to deve lop at 32, seemed to be overlapped by an exaggerated collagen peak at 31. On the ot her hand, when bovine tendons were mineralized by the PILP process containing vari ous concentrations of poly-( )-DL aspartic acid (Mw: 5,500 Da), no hydroxyapatite peak was observed. However, it is worthwhile to note that increases in the intensity of the collagen peak (31) were observed from bovine tendons that were mineralized by the PILP process containing 25, 50, and 100 g/ml of poly-( )-DL aspartic acid. It could be argued that, in contrast to pure bovine tendon, collagen fibe rs might have swollen and rearranged in the solution during the mineralization process (anal ogous to annealing). This rearrangement of collagen fibrils might lead to the enhancement of peak intensity due to the increase in the size of ordered domains. Howe ver, since pure tendon wa s soaked in de-ionized water for several hours prior to lyophilization for the x-ray anal ysis, the increase in the peak intensity is presumed to not be a consequence of the swelling or re-arr angement of collagen fibrils, but to some other factors. We specu late that the chelation of calcium ion, or incorporation of this higher mass element betw een repeat units might enhance the scattering factor and lead to the increase of peak intensity.
94 When bovine tendons reacted with various soluti ons, some showed the increase in the peak intensity but some didnt. As shown Figure 4-5 A, when bovine tendon was reacted in a solution containing calcium (b), and when bovine tendon wa s mineralized by the PI LP process (e), the intensity of the collagen peak (31) was incr eased. However, when bovine tendon was reacted with the solution containing calcium ions and 25 g/ml of poly-aspartic acid, the increase in the peak intensity was not observed. In the case of bovine tendon reacti ng with the phosphotungstic acid (PTA) for 30 minutes, all typical peaks from collagen were gone. When the contribution of the pure collagen peak was subtracted from each of the x-ray patterns (Figure 4-5 B), the increase in the peak intensity was emphasized. As shown in Figure 4-5 B, the increases in the intensity of collagen peak at 31 were obs erved for bovine tendon mi neralized by the PILP process (e) and bovine tendon reacte d with calcium solution (b). On the other hand, the others showed no increase in the intensity (c) or an al teration of the x-ray pattern (d). According to Darwin [211, 212], the equation for peak intensity is given as Ihkl = Io(l3/ )(Vx L p A/V2)[Fhkl]2 (4.1) where Io = incident beam intensity, l = wavelength of radiation, = rotation veloci ty of crystal, Vx = volume of the crystal, L = Lorentz factor, p = polarization factor, A = absorption factor, V = volume of unit cell, [Fhkl] = structure factor. Of those variables, we are especially interested in [Fhkl] (structure facto r) to explain the increase of the collagen peak intensity, because th e rest of variables for the collagen peak at 31 are not expected to be changed dramatically by the reaction. According to Cullity , the structure factor [Fhkl] is given as [Fhkl] = fn exp(2i ( hun+ kvn+ lwn)) (4.2)
95 where fn = atomic scattering factor, ( h, k, l ) = miller index, ( un, vn, wn) = coordination of atom position. In the case of the [Fhkl] (structure factor) functi on, the function must have a value of zero or some value given by integer (value of exponential term) atomic scattering factor ( fn). In other words, if a structure of material can develop a peak at 2 which is determined by Bragg law depending on the d spacing between plane (d spacing depends on miller index), the structure factor for the peak at that 2 has a value, but if a material does not develop a peak at that 2 the structure factor is zero due to 0 value of exponential term of equation (4.2). Therefore, if the crystallographic repeat unit along the fiber direction is altered by external conditions, the peak will be developed at different 2 because of the change in the value of [Fhkl]. In the case of the alteration of x-ray patterns in th e bovine tendon reacting with PTA (d), the PTA incorporates at locations associated with specific functional gr oups of the polypeptide for the negative staining of the collagen fibril. The negative staining of the collagen fibrils i nduces a new structural repeat unit of collagen, leading to new ordered st ructure with electron-dens e PTA. It is thought that the [Fhkl] function for new repeat unit produces a broad peak around 28 and 50, and the high intensity from electron-dense PTA subdues the intensity of collagen pe ak, leading to a lack of an apparent peak at 31. In the case of bovine tendon reacting with calcium solution, there may be a chelation between negatively charged re peat units in the collagen and calcium ions without any change in the or der of repeat units. This can change the value of fn (atomic scattering factor). The atomic scattering factor can be expressed as the ratio of the amplitude of a wave scattered by one atom to the amplitude of a wave scattered by one electron. Therefore, the atomic scattering factor is normally increased as the atomic number of the element increases. This leads to the idea that the intensity of th e peak at 31 can be increased due to a calcium
96 chelation or incorporatio n with repeat units leading to a high atomic scattering factor from calcium ( fCa = around 14) The increase of peak intensity was most significant in the bovine tendon sample mineralized by the PILP process. In this case, if the amorphous precursor was drawn into the interstices between the tropocollagen molecules, th en the repeat units on these tropocollagen molecules might chel ate or incorporate calcium ( fCa = around 14) and phosphate ions ( fP = around 10), which could then provide a much higher atomic scattering factor value leading to the observed increase in peak intensity. However, in the case of the bovine tendon sample that reacted with a solu tion containing calcium ions and pol y-aspartic acid, it is thought that calcium ions may be more bound to the nega tively charged polymer such that chelation between collagen repeat units and calcium ions may not be significant enough to increase the peak intensity. When collagen fibrils contain the amorphous precursor or transient polymorphs of hydroxyapatite, the amorphous precu rsor within the fibrils will be transformed into hydroxyapatite with time. Therefore, if the increa se of collagen peak intensity is caused by the amorphous calcium phosphate precursor incorporating with repeat un its, hydroxyapatite peaks might develop from bovine tendon mineralized by the PILP process after aging in the tris-buffer solution. As shown in Figure 4-6, after aging the sample, there was no significant change in the X-ray patterns of bovine tendon mineralized by the PILP process with 50 g/ml of poly-aspartic acid. However, when bovine tendon, was minerali zed by the PILP proce ss with 100 g/ml of poly-aspartic acid for 6 days and then aged in th e tris-buffer solution, a new apex of X-ray peak was developed at 32.01, which corresponds to the (211) peak of hydroxyapati te. To distinguish the hydroxyapatite peak at 32.01 from an overlapping collagen p eak, the identical bovine tendon (Figure 4-6, 100 g/ml Pasp) was scanned by the X -ray diffractometer with the mode II setup in
97 order to subdue the collagen peak at 31o but exaggerate the collagen p eak at 22. As shown in Figure 4-7, the (211) peak at 32 was observe d, indicating the amorphous calcium phosphate precursor was transformed into hyd roxyapatite after the aging proc ess. However, no (211) peak was observed in bovine tendon mineralized by the PI LP process with the lo wer concentration of 50 g/ml of poly-aspartic acid, even though the incr ease of collagen peak intensity was observed. This might be because the rate of transformati on of precursor is different or the amount of precursor infiltrating into the collagen is different, depending on the concentration of polymer. It is worth noting that the collagen peak at around 8 developed after the mineralization process. It is thought that the peak intensity at 8 may increase due to the space between tropocollagen molecules being filled with amorphous precursor with high value of atom ic scattering factor. The presence of high mass ions such as calci um and phosphate in the amorphous precursor was also observed by EDS. As shown in Figure 4-8, co llagen fibrils showed a contrast that is not seen in non-stained collagen fibrils; in additi on, calcium and phosphate signals were detected by EDS. These evidences, such as the developmen t of the (211) peak of hydroxyapatite after aging, the contrast present in un-stained collagen fibril s, and the calcium and phosphate signals detected by EDS of the collagen fibrils, might support th e hypothesis that the amorphous precursor drawn into collagen fibrils could increase the intensity of the collagen peak. As shown in Figure 4-9, clusters of hydroxyapatite were observed on bovine tendon mineralized by the conventional nucleation and gr owth method (control experiment) for 6 days, and EDS and X-ray diffraction analysis (Figur e 4-4) also confirmed the precipitation of hydroxyapatite on bovine tendon. On the othe r hand, there was no precipitation of hydroxyapatite clusters when bovine tendons were mineralized by the PILP process containing various concentrations (25, 50 and 100 g/ml) of poly-( )-DL-aspartic acid (Mw: 5,500 Da).
98 As shown in Figure 4-10, the surface of bovine tendon was very smooth and hydroxyapatite clusters were not observed. In the case of ED S analysis, calcium and ph osphate were detected from bovine tendons mineralized by the PILP process containing 50 and 100 g/ml of polyaspartic acid, even though no hydroxyapatite coat formed by the agglomeration of precursor on the surface was observed. It may be because the amorphous calcium phosphate precursor is drawn into collagen fibrils instead of depositing on the tendon surface. When bovine tendons mineralized by the PI LP process containing 50 and 100 g/ml of poly-aspartic acid were obser ved by TEM (Figure 4-11 C and D), the banding patterns of collagen fibrils were observed without staini ng. Normally, the banding patterns of collagen fibrils are not observed without PTA staining because collagen fibrils are composed of polypeptides that cannot scatter electron beam strongly, as shown in Figure 4-11 A and B. Therefore, it is thought that the banding patt erns in bovine tendon mineralized by the PILP process might be generated by elect ron dense scatterers with high ma ss, such as crystalline phase or amorphous calcium phosphate precursor drawn into the collage n fibrils . As shown in Figure 4-11 C, the selective area electron diffraction patterns of co llagen fibers showed diffuse rings generated by amorphous phase, implying th at the banding patter ns are generated by amorphous electron dense material. Sometimes an amorphous calcium phos phate precursor that was not infiltrated into the collagen fibril wa s observed around collagen fibrils as shown in Figure 4-11 D. When bovine tendon was mineralized by the PILP process with low molecular weight polymer, the X-ray diffraction and TEM analysis showed that some amorphous precursor was infiltrated into collagen fibril s, even though intrafibrillar diffrac ting hydroxyapatite crystals were not observed. It is thought that lower molecu lar weight polymer can successfully induce the
99 formation of amorphous calcium phosphate precu rsor by inhibiting the rapid nucleation of hydroxyapatite crystals. However, most of calcium and phosphate ions were consumed as an agglomeration of a gel-like precursor in the solu tion, which prevented the effective infiltration of precursor into collagen fibrils. Therefor e, it is hard to argue that the poly-( )-DL-aspartic acid with low molecular weight is reliable for th e intrafibrillar mineralization of bovine tendon because it seems that there is va riability in the types of amorphous phase that might be induced. Therefore, to more effectively mineraliz e bovine tendon via the PILP process, more considerations regarding the characteristics of the process-directing agent were required. Mineralization of Bovine Tendon with Hig h Molecular Weight Poly-aspartic Acids Generally, the negative charge on the acidic polymer chains is increased as the molecular weight of the acidic polymers increases . In the case of the PILP process for calcium phosphate system, the acidic polymer sequester s calcium and phosphate ions to form an amorphous precursor with water molecules. Therefor e, when the ability of the polymer to attract ions is mainly considered, the ability of the pol ymer to form amorphous precursors is expected to be enhanced as the negative charge of the polymer increases. On the other hand, work in Gowers group with the calcium carbonate PILP sy stem found that higher molecular weight was not necessarily better for producing PILP (granular precipitates occur), and there was an optimal molecular weight range of 5 15 kDa. Therefore, a range of pol y-L-aspartic acid with different molecular weights (10,300 and 32,200 Da) was tested for the intrafibrilla r mineralization of bovine tendon. After mineralizing bovine tendon via the PILP pr ocess with the highest molecular weight available, poly-L-aspartic aci d (Mw: 32,200 Da), the formation of hydroxyapatite was observed by X-ray diffraction analysis. As shown in Figur e 4-12 c, the (002), (200) and (213) peaks of
100 hydroxyapatite developed after 8 days of mineralization. An increase in intensity of the collagen peak at 31 was also observed, which was seen as well in bovine tendon mineralized by the PILP process with low molecular weight poly-aspartic acid (Mw: 5,500 Da). However, in this case, the increase in the collagen peak intensity is thought to be caused by the overlap between collagen and hydroxyapatite peak at 32. When bovine tendon was mineralized by the PILP process containing poly-L-aspartic acid (Mw: 10,300 Da), an increase in collagen peak intensity (at 31) was observed (Figure 4-12 b). In this case, however, because ther e are no distinct peaks of hydroxyapatite crystals, so the increase in collagen peak intensity is thought to be caused by the chelation or incorporation of amorphous precursors between collagen fibrils. For the higher molecular weight poly-L-aspartic acid (Mw: 32,2 00 Da), when the identical samples that had been scanned with mode I (Figure 4-12) were sc anned with mode II (Fi gure 4-13), the (002) and (211) peaks of hydroxyapatite were observed. As shown in Figures 4-12 and 4-13, the orient ed collagen fibrils in bovine tendon act as an obstacle in the X-ray analysis of mineralized bovine tendon. Theref ore, bovine tendon was minced to remove the uniaxial orientation of colla gen fibrils. As shown in Figure 4-14 a, the minced bovine tendon has three p eaks at around 8, 20 and 31, which are readily observed now that the exaggerated collagen peaks caused by su ch strong orientational anisotropy were removed by mincing the bovine tendon. When the minced bovine tendon was mineralized with the PILP process containing poly-L-aspartic acid (Mw: 32 ,200 Da), hydroxyapatite peaks, identical to the X-ray patterns of hydroxyapatite in bone, were observed (Figure 4-14 c). In the case of hydroxyapatite in bone, because the size of the hydr oxyapatite crystals is extremely small, and crystals contain defects, the (211) (112) and (300) peaks all in th e vicinity of 32 are overlapped, leading to one broad peak at around 32. The X-ray patterns of mineralized bovine tendon also
101 showed a broad peak at 32, and other peaks su ch as (002), (310), (222) and (213) were also observed, while the collagen peaks were subdued by the high intensity of hydroxyapatite crystals. On the other hand, when bovine tendon was mine ralized with the PILP process containing intermediate molecular weight poly-L-aspartic ac id (Mw: 10,300 Da), only one broad peak at 31 developed (Figure 4-14 b). As disc ussed in the previous section, an increase in the intensity of the collagen peak at 31 was normally observed after mineralization due to the chelation or incorporation of amorphous precursors when the collagen peak was exaggerated by the oriented collagen fibers. In this case, however, because the effect of oriented collagen fibrils on the X-ray diffraction analysis was reduced by the isotropic sample preparati on, it is thought that the broad peak at 31 might be produced by amorphous cal cium phosphate having short-range order. When bovine tendon was mineralized by the PILP process containing the poly-L-aspartic acid (Mw: 10,300 Da), as shown in Figure 4-15 A, the precipitation of hydroxyapatite clusters, normally observed in the conventional nucleation and growth method, was not observed. Moreover, there was no morphological change in the collagen fibrils before and after mineralization. In the case of the EDS analysis (Figure 14-15 B), the content of calcium and phosphate ion was very low, which was expected from the X-ray diffraction analysis (Figure 1414 b). When bovine tendon was mineralized by th e PILP containing poly-L-aspartic acid (Mw: 32,200 Da), no surface hydroxyapatite, such as hydroxyapatite clusters or a hydroxyapatite coating formed by the agglomeration of precursor s, was observed, yet the EDS analysis showed a high content of calcium and phosphate ions. Like wise, according to the X-ray analysis (Figure 4-14 c), the bovine tendon was well mineralized with either nano-cr ystalline or poorly-crystalline hydroxyapatite. Therefore, the calcium and phospha te detected by the EDS analysis are thought to have originated from hydroxyapatite crystals that are embedded within collagen fibrils. This
102 intrafibrillar hydroxyapatite was al so confirmed by TEM. As shown in Figure 14-16 A, collagen fibrils showed banding patterns without the stai ning process, indicati ng the fibril contained electron dense material. Also needle-like crystals preferentially oriented in the direction of c axis of the collagen fiber, were also observed in the collagen fibrils. Th e selective area electron diffraction (SAED) pattern (Figur e 14-16 C) showed an almost id entical diffraction pattern to that of bone hydroxyapatite, which is characte rized by arcs ((002), (004)) produced by roughly uniaxially oriented hydroxya patite. When a dark field image of mineralized collagen fibrils was constructed with the electron beam diffracted by the (002) planes, as shown in Figure 14-16 B, needle-like white spots corres ponding to (002) plane of hydroxya patite were well-aligned along the c -axis of the collagen fibril. The shape of hydroxyapatite crystals in bone is usually described as a platelet. However, in the bri ghtand dark-field TEM images, because the sideview of hydroxyapatite platelets was projected on the 2-dimenti onal screen, needle-like white spot were generated from the dark field image. When examined by a high resolution (HR) TEM (Figure 4-17 B), an array of (002) planes with d-spacing of 0.34 nm was observed. This HRTEM micrograph also shows that the  direction of hydroxyapatite, which is perpendicular to the (002) planes, was well-aligned along the c -axis of collagen fibril. To summarize, when bovine tendon was mineralized by the PILP process using the higher molecular weight poly-L-asparti c acid (Mw: 32,200 Da), intrafibri llar mineralization of collagen fibrils was successfully achieved and confir med by TEM and HRTEM. In the case of SEM observation, a surface precipitation of hydroxyapati te crystals or hydroxyapatite coat was not observed even though X-ray diffraction analys is of bovine tendon showed well-developed hydroxyapatite peaks. Those results from SEM, TEM and X-ray diffrac tion may indicate that most of amorphous calcium phosphate precursors fo rmed by the PILP process are drawn into the
103 collagen fibrils contributing to th e intrafibrillar minera lization of collagen. On the other hand, when bovine tendon was mineralized by the PILP process induced by intermediate molecular weight poly-L-aspartic acid (Mw: 10,300 Da), although a broad peak (probably produced by the amorphous calcium phosphate) developed in the X-ray diffraction analysis, and only a small amount of calcium and phosphate was detected by EDS, the intrafibrillar mineralization was not observed by TEM. It is thought that only small portions of precursor formed by the poly-Laspartic acid (10,300 Da) could infilt rate into the collagen fibrils. It is not clear why most of the precursor formed by poly-L-aspartic acid (10,300 Da) is not drawn into the collagen fibrils. We speculate that calc ification inhibitors in bovi ne tendon may effectively pr event the infiltration of precursors, or the precursor formed by lowe r molecular weight polymer which may have different properties, such as size or surface charge, in comparison to the properties of the precursor formed by high molecular weight polymer, which might influence its adsorption to the collagen, or wetting and uptake via capillary action. Improvement of Mineralization of Bovine Tendon by EDTA Treatment As I m entioned in the introducti on, bovine tendon has rarely b een used to investigate bone formation because bovine tendon is not naturally mineralized, presumably due to calcification inhibitors preventing mineraliza tion of the collagen fibrils  In contrast to bovine tendon, turkey tendon has frequently been used as a model of secondary bone formation because turkey tendons are naturally mineralized wi th an intrafibrillar nano-stru cture. The study described in the previous section seemed to suggest that in hibitors may be present in the bovine tendon since the lower molecular polymer that had been succe ssful with other collagen scaffolds did not perform well with the bovi ne tendon. Therefore, to investigate the possibility of inhibitors being present that might inhibit the intrafibrillar minera lization of bovine tendon, we tried to see if we could extract inhibitors from the tendon using a process described by Glimcher group . The
104 followings results compare the mineralization be havior of bovine tendon with and without these so called calcification inhibitors. Besides the calcification inhi bitors, bovine tendon also has a physical barrier which prevents the mineralization of tendon via the PILP process. The collagen fiber bundles in bovine tendon are surrounded by epitendineum, which is a white fibrous sheath indicated by the white circle in Figure 4-18. This sheath may act as a physical barrier which prev ents infiltration of the amorphous precursor. For example, when EDS an alysis was done at the region (black circle) where the collagen fibers had been exposed to the mineralization solu tion, a high amount of calcium and phosphate was detected because the collagen fibrils were mineralized by the PILP process. However, calcium and phosphate peak s were not observed from sampling the region surrounded by the epitendineum, even though the ED S analysis for the neighboring region (black circle) showed calcium and phosphate. Therefor e, this physical barrier may be one factor preventing the mineralization of bovine tendon. The inhibitors in bovine tendon were removed by treatment with EDTA (ethylenediaminetetraacetic acid), which was used by Glimcher group in their study on chicken bone matrix to extract non-collagenous proteins . Bovine tendon was minced and soaked in 0.5 M of EDTA solution, and then stored at 4oC for 3 weeks. When EDTA-treated bovine tendon was mineralized by the PILP process cont aining poly-L-aspartic acid (Mw: 10,300 Da), the intrafibrillar minera lization of EDTA-treated bovine tend on was successfully achieved, even though the lower molecular weight poly-L-aspartic acid previ ously gave poor results of intrafibrillar mineraliza tion. As shown in Figure 4-19, bovine tendon was not mineralized very well by the PILP process and only a broad p eak, probably produced by a small amount of amorphous calcium phosphate, was observed when pol y-L-aspartic acid with a molecular weight
105 of 10,300 Da was used. However, when the EDTA -treated bovine tendon was mineralized with poly-L-aspartic acid (Mw: 10,300 Da ), the peaks of hydroxyapatite (identical to those of bone hydroxyapatite) were observed after 8 days of mineralization. In the case of TEM observation, the SAED pattern of the minera lized bovine tendon pre-treated by EDTA was almost identical to that of bone hydroxyapatite (Figur e 4-20 C). The dark field imag e constructed by the (002) arc showed that hydroxyapatite crysta ls were well-aligned along the fi ber direction (Figure 4-20 B). These results show that the remova l of inhibitors can enhance the intrafibrillar mineralization of bovine tendon via the PILP process. When an even lower molecular weight polymer (poly-( )-DL-aspartic acid (Mw: 5,500 Da) was used for the intrafibrillar mineraliza tion of pure bovine tendon, as shown in Figure 4-21 (upper pattern), there was no peak correspond ing to hydroxyapatite or amorphous calcium phosphate. However, with EDTA-treatment, a broa d peak at 31 was observed, and the intensity of the collagen peak at 8 was decreased. This trend in the X-ray analys is of mineralized bovine tendon was similar to that observed in the tendon mineralized with poly-L-aspartic acid (Mw: 10,300), as was shown in Figure 4-14. These re sults show that the amount of amorphous precursor that can infiltrate into the collagen fibrils can be enhanced once the calcification inhibitors are removed by the EDTA treatment. The amount of amorphous precursor drawn in to collagen fibrils can be estimated by thermogravimetric analysis (TGA). The orga nic substances, such as collagen and noncollagenous proteins, are entirely combuste d and decomposed between 150 and 600C. Therefore, the remaining material in the platinum pan at 600C is considered the mineral content (since the pure collagen leaves no ash at this temperature). As shown in Figure 4-22, the mineral content of bovine tendon minera lized with high molecular weight poly-L-aspartic acid (Mw:
106 32,200 Da) was 24 wt%. On the other hand, bov ine tendon mineralized with low molecular weight poly-L-aspartic acid (Mw: 10,300 Da) contained on ly 6 wt% of mineral at 600oC. These results were expected from the X-ray diffraction analyses (Fig ure 4-14), in which only bovine tendon mineralized with high molecular weight poly-L-aspartic acid (Mw: 32,200 Da) showed hydroxyapatite peaks. However, when EDTA-t reated bovine tendon was mineralized with low molecular weight poly-L-aspartic acid (Mw: 10,300 Da), the mineral content of bovine tendon doubled (6 wt% 12 wt%), again, as expected from X-ray diffraction analysis and TEM observation. Importantly, the minera l phase in TG analysis can be considered to have originated from intrafibrillar mineral, because the precipit ation of hydroxyapatite clusters or hydroxyapatite coatings was not observed from SEM analysis. In the case of bovine tendons (EDTA-treated and pure tendon) mineralized with low molecular we ight poly-L-aspartic acid (Mw: 10,300), the mineral content of EDTA-treated bovine tendon was about 12 wt%, twice that of pure bovine tendon (without EDTA-treatment). Moreover, becau se identical poly-L-aspartic acid was used to induce the PILP process, the pr operties of precursor should be same for the mineralization of both tendons (pure and EDTA-treated). Therefore, it seems that calcification inhibitors in bovine tendon can prevent the infiltra tion of amorphous precursors. Although it is not clear how inhibitors can prevent the infiltra tion of precursors, we speculate that inhibitors might alter the surface chemistry. For example, if they are enrich ed with acidic amino acids, the scaffold might repel the precursor cont aining acidic polymer, Ca2+ and PO4 3-, or they might cause steric hindrance around the gap zone to prevent the inf iltration of precursor. In the case of bovine tendon mineralized with high molecular weight poly-L-aspartic acid (Mw: 32,200 Da), the mineral content was about 24 % even though th e bovine tendon contained the calcification inhibitors. This implies that the precursor fo rmed by the high molecular weight poly-L-aspartic
107 acid can infiltrate into the collagen fibrils even in the presence of calcification inhibitors which prevents the infiltration of precursor formed by low molecular weight polymer. Moreover, it is worthwhile to note that the inhibitors seem to prohibit the rapid crys tallization of amorphous precursor. In the case of bovine tendon minerali zed with low molecular weight polymer (Mw: 10,300 Da), while bovine tendon contained 6 wt% of inorganic phase (TG analysis), X-ray diffraction did not show the development of hyd roxyapatite after 8 days of mineralization. However, when the calcification inhibito rs were removed by the EDTA treatment, hydroxyapatite was developed after 8 days of reacti on. It is not clear why the precursor formed by high molecular weight polymer can infiltrate into collagen fibrils containing the calcification inhibitors. We speculate that the precursor fo rmed by the high molecular weight poly-L-aspartic acid may have internal and inte rfacial conditions, such as sm all size, high fluidity, or a specifically charged interface, l eading to the effective intrafib rillar mineralization of bovine tendon, even with the calcifi cation inhibitors. Conclusion When bovine tendon m ineralized by the PILP process with low molecular weight polyaspartic acid (Mw: 5,500 Da) was scanned by X-ray, the intensity of collagen peak (31o) was increased after the mineralization process, which I believe is due to amorphous calcium phosphate precursor or ion chelation with the re peat units of the polypeptides. However, the intrafibrillar crystals were not observed. When bovine tendon was mineralized by the PILP process containing the high molecular weight poly-L-aspartic acid (M w: 32,200 Da), X-ray diffraction, SEM and TEM analysis showed that the bovine tendon was intrafibrillarly mineralized by the amorphous precursor formed by the higher molecular weight poly-aspartic acid. Even though bovine tendon c ontains calcification inhibitors, the tendon contained about 24 wt% of intrafibrillar hydr oxyapatite after the
108 mineralization process. When bovine tendon wa s mineralized by the PILP process containing low molecular weight poly-L-a spartic acids (Mw: 10,300 Da), ve ry small amounts of precursor infiltrated into the bov ine tendon due to the calcification in hibitors. X-ray diffraction showed only a broad peak of amorphous calcium phosphat e. However, no hydroxyapatite peak was observed. The removal of inhibitors by the ED TA treatment enhanced the intrafibrillar mineralization of bovine tendon with this lowe r molecular weight polymer (Mw: 10,300 Da), and there was some collagen peak enhancem ent and development of amorphous calcium phosphate peak with the lowest molecular we ight polymer (Mw: 5,500 Da). While bovine tendon mineralized with low molecular weight polymer (Mw: 10,300 Da) contained only 6 wt% of mineral, the EDTA-treated bov ine tendon contained 12 wt% of mine ral after mineralization.
109 Figure 4-1. Collagen fibrils and st ructural factors producing peaks in X-ray diffraction analysis. Collagen fibrils produce three distinct broa d peaks due to a) repetative amino acid units along the polypeptide chains (0.29 nm ), b) average distance between peptide chains within a tropocollagen helix(0 .45 nm), and c) lateral spacing of the tropocollagens (1.08 nm). In the case of X-ray using Cu(K ) radiation, a) produces a peak at around 31, b) produces a peak at around 20 and c) produces a peak at around 8. Image was borrowed from http://web.mit.edu/3.082/www/te a m1_f02/collagen.htm . c) b) a)
110 Figure 4-2. Schematic diagram illustrating the alignment of collagen fibers in bovine tendon relative to the x-ray beam path. Because th e collagen fibers in bovine tendon are well aligned, three distinguished peaks of collagen fibers fr om powder X-ray diffraction change depending on the alignment of the fi bers to the beam. The position of the three collagen peaks, with 2 around 8 (d=1.08nm), 20 (d=0.45 nm) and 31 (d=0.29 nm) degrees, could overlap with the t ypical peaks of hydroxyapatite, with 2 around 26 (002) and 32 (211) degrees. Therefore, two types of sample alignment are required to avoid the peak-overlap and to distinguish the typical peaks of hydroxyapatite. With mode I, the (002) peak of hydroxyapatite should be distinguishable, but the (211) peak will overlap with the collagen peak. With mode II, the hydroxyapatite (211) p eak is emphasized, but the ( 002) peak overlaps with the collagen peak. All samples for XRD anal ysis were lyophilized to minimize the collapse or shrinkage of collagen structure du ring the drying process. Mode I Mode II (002) plane (211) plane
111 Figure 4-3. The XRD patterns of turkey tendon mi neralized with the PILP process for various times (black: 0 day, red: 3 days of minerali zation, blue: 4 days of mineralization). Synchrotron X-ray source ( = 0.652549 ) was used for the characterization. The reaction concentrations of calcium chloride and potassium phosphate were 4.5 mM and 2.1 mM, respectively. Poly-aspartate (Mw: 6,200 Da) was added as a processdirecting agent for the PILP process (100 g/m l). The orientation of aligned collagen fibers in the turkey tendon was adjusted to be perpendicular to the direction of incident X-ray beam to examine the effect of preferential orientation of intrafibrillar hydroxyapatite, in which the (002) planes ar e aligned along the fiber orientation. Mineralized turkey tendon0.00E+00 2.00E-04 4.00E-04 6.00E-04 8.00E-04 1.00E-03 1.20E-03510152025 2Theta Intensity Path of detector movement Detector (004) (002) (112) d= 0.278nm Turkey tendon with oriented collagen fiber (white arrows) X-ray Diffracted beam
112 Figure 4-4. The XRD patterns of bovine tendon mi neralized by the PILP process for 6 days. The mineralization solution contained 4.5 mM of calcium chloride, 2.1 mM of potassium phosphate and various concentrations of poly-( )-DL-aspartic acid (Mw: 5,500 Da). Control sample didnt contain poly-aspartic acid. Because mode I was used, the intensity of the collagen peak around 31 is emphasized. Therefore, the formation of hydroxyapatite should be confir med by the (002) peak around 26 (black dot at the control samples). 0510152025303540 100g/ml Pasp 50g/ml Pasp 25g/ml Pasp Control Bovine tendon Intensity2(degree)
113 01 02 03 04 05 0 (e) (d) (c) (b) (a) Intensity2(degree) 01 02 03 04 05 0 (a) (b) (c) (e) (d) Intensity2(degree) Figure 4-5. The XRD patterns (mode I) of bovi ne tendons reacted w ith various reaction solutions. B) To investigate the increase in peak intensity of bovine tendon after each treatment, the contribution of untreated co llagen (a) was subtracted from each of the XRD patterns shown in A). (a) Bovine tendon washed with de-ionized water several times and lyophilized. (b) Bovine tendon r eacted with 4.5 mM of calcium chloride solution for 6 days. (c) Bovine tendon r eacted with 4.5 mM of calcium chloride solution containing 25 g/ml of poly-( )-DL-aspartic acid (Mw: 5,500 Da) for 6 days. (d) Bovine tendon stained with phosphotungstic acid for 30 minutes. (e) Bovine tendon mineralized by the PILP process with 50 g/ml of poly-( )-DLaspartic acid (Mw: 5,500 Da) for 6 days. The peak intensity at around 31 increased when the tendon was reacted with the calci um solution and the PILP mineralization solution. B) After tendon peak (a) subtracted A) Original
114 Figure 4-6. The XRD patterns (mode I orienta tion) of bovine tendon mineralized by the PILP process with various concentration of poly-( )-DL-aspartic acid (Mw: 5,500 Da) for 6 days. After XRD analysis, mineralized bovine tendons were re-soaked in a trisbuffer solution for 6 days to induce the cr ystallization of any amorphous precursor that may be present, without additional calcium and phosphate sources from solution. Red lines indicate 31.17 and 32.01, the peaks associated with collagen and (211) peak of hydroxyapatite, respectively. 510152025303540 50g/ml Pasp 6 days Rxn 100g/ml Pasp 6 days Rxn 100g/ml Pasp 6 days Rxn+6 days Tris 50g/ml Pasp 6 days Rxn+6 days Tris Intensity2(degree)
115 Figure 4-7. The XRD patterns (Mode II orientati on) of bovine tendon mineralized by the PILP process with various co ncentrations of poly-( )-DL-aspartic acid (Mw: 5,500 Da) for 6 days. After XRD analysis, mineralized bovine tendons were re-soaked in a Trisbuffer solution for 6 days to induce the crystallization of amorphous precursors without calcium and phosphate sources. Because Mode II was used, the formation of hydroxyapatite should be confirmed by the ( 211) peak at around 32 (black dot). 0510152025303540455055 100g/ml pasp, 6days of Rxn 100g/ml pasp,After tris treatment 50g/ml pasp,After tris treatment 50g/ml pasp, 6days of Rxn Bovine tendon Intensity2(degree)
116 Figure 4-8. The EDS analysis of bovine tendon mineralized via the PILP process with 50 g/ml of poly-( )-DL-aspartic acid for 6 days. Cu si gnal is generated by copper grid and reflected beam from chamber. EDS anal ysis was done at th e area showing high contrast (Point with Spectrum indica tes the position where EDS analysis was performed). Ca P
117 Figure 4-9. The SEM micrographs of the surf ace of bovine tendon mine ralized without polyaspartic acid. The hydroxyapatite crysta ls on the surface of the tendon which is exposed to the X-ray beam in Mode I (left) and (right), a magnified image of the platy hydroxyapatite crystals. EDS analysis ( bottom middle) was done to confirm the calcium phosphate crystals.
118 Figure 4-10. The SEM micrographs and EDS analysis of bovine tendons mineralized with the PILP process containing various concentrations of poly-( )-DL-aspartic acid. Bovine tendons were mineralized with 25 (A, B), 50 (C, D) and 100 (E, F) g/ml of poly-( )-DL-aspartic acid for 6 days. C) F) A) B) D) E)
119 Figure 4-11. The TEM micrographs of bovine tendon before (A, B) and after (C, D) mineralization. Tendon was mineralized by th e PILP process with C) 50 and D) 100 g/ml of poly-( )-DL-aspartic acid for 6 days. Inset in sample C) shows the selective area electron diffraction patter n, in which a diffuse ring indicates only amorphous material is present. A) B) C) D)
120 Figure 4-12. The XRD patterns (Mode I) of bovine tendons mineralized with the PILP process containing poly-L-aspartic acid (polyAsp) for 8 days, investiga ting the effect of polyAsp of different molecular weig hts (10,300 and 32,200 Da) on mineralization. The concentration of polymer was fixe d to 100g/ml. a: bovine tendon without mineralization, b: molecular weight of pol yAsp is 10,300 Da, c: molecular weight of polyAsp is 32,200 Da. (002) (200) (213) 5101520253035404550 c b a Intensity2(degree)
121 Figure 4-13. The XRD patterns (Mode II) of bovine tendons mineralized with different molecular weight polymers via the PILP process containing 100 g/ml of poly-Laspartic acid (polyAsp) for 8 days. a: bovine tendon without mineralization, b: molecular weight of polyAsp is 10,300 Da, c: molecular weight of polyAsp is 32,200 Da. (002) (211) 51 01 52 02 53 03 54 0 c b a Intensity2(degree)
122 Figure 4-14. The XRD patterns of isotropic bovine tendons minera lized with the PILP process containing 100 g/ml of poly-L-aspartic aci d (polyAsp) for 8 days. To minimize the orientation effect of ali gned collagen fibers in bo vine tendon, bovine tendon was minced and lyophilized before minera lization. a: bovine tendon without mineralization, b: molecular weight of pol yAsp is 10,300 Da, c: molecular weight of polyAsp is 32,200 Da. 5101520253035404550 c b a Intensity2(degree)39.81(310) 31.87(211) 26.03(002) 49.49(213) 46.63(222)
123 Figure 4-15. The SEM micrographs and EDS analysis of bovine tendons mineralized with the PILP process containing 100 g/ml of poly-L-as partic acid (polyAsp) for 8 days. The molecular weights of polyAsp added for the PILP process were 10,300 Da (A, B) and 32,200 Da (C, D). Bovine tendon was minced a nd lyophilized before mineralization. B) C) A) D)
124 Figure 4-16. The TEM analysis of bovine tendon mineralized with the PILP process containing 100 g/ml of poly-L-aspartic acid (Mw ; 32,200 Da) for 8 days. A) Bright-field TEM image of bovine tendon, B) Dark field TEM image constructed from the 002 arc of the diffraction pattern shown in C), s hows bright streaks corresponding to the numerous well-oriented nanocrystals of hydroxyapatite; and C) the corresponding selected area electron diffr action pattern of the minera lized bovine tendon shown in A). A) B) C)
125 Figure 4-17. High resolution TEM (HRTEM) micr ographs of bovine tendon mineralized with the PILP process containing 100 g/ml of poly-L-aspartic acid (Mw: 32,200) for 8 days. A) Bright-field image of bovi ne tendon, B) lattice image of aligned hydroxyapatite. The d-spacing of the (002) plane of hydroxyapati te (0.34 nm) is indicated on B (red arrow), and fiber dire ction is also indicated (white arrow). 0.34nm B) A)
126 Figure 4-18. The SEM micrographs and EDS analysis of bovine tendons mineralized with the PILP process containing 100 g/ml of pol y-L-aspartic acid (Mw: 32,200 Da) for 8 days. EDS analysis was done for a region co mposed of exposed collagen fibers (left EDS spectrum) and the region covered with epitendineum (right EDS spectrum), which shows no mineral content, indicating that the mineral precursor was unable to penetrate the sheath.
127 Figure 4-19. The XRD patterns of bovine tendo n and EDTA-treated bovine tendon which were mineralized by the PILP process containing 100 g/ml of poly-L-aspartic acid (Mw: 10,300 Da) for 8 days. For EDTA treatment, minced bovine tendon was reacted with EDTA at 4C for 3 weeks. This treatmen t dramatically enhanced the mineralization process, presumably by removing some type of inhibitory species. (002) (211) (213) (222) 10 20 30 40 50 NO EDTA treatment EDTA treatment Intensity2(degree)
128 Figure 4-20. The TEM micrographs of EDTA-tr eated bovine tendon mineralized with the PILP process containing 100 g/ml of poly-L-aspa rtic acid (Mw: 10,300 Da) for 8 days. A) Bright-field TEM image of mineraliz ed bovine tendon, and B) Dark-field TEM image constructed with the 002 arc, and C) selected area electron diffraction pattern of this mineralized bovine tendon. A) B) C)
129 Figure 4-21. The XRD patterns of bovine tendo n and EDTA-treated bovine tendon which were mineralized by the PILP process containing 100 g/ml of poly-( )-DL-aspartic acid (Mw: 5,500 Da) for 8 days. For ED TA treatment, minced bovine tendon was reacted with EDTA at 4C for 3 weeks. When EDTA-treated tendon was mineralized by the low molecular weight polymer, th e peak corresponding to amorphous calcium precursor (around 31) was developed. 10 20 30 40 50 NO EDTA treatment EDTA treatment Intensity2(degree)
130 Figure 4-22. Thermogravimetric analysis (T GA) of bovine tendons mineralized by the PILP process containing 100 g/ml of poly-L-as partic acid (Mw:10,300 or 32,200 Da) for 8 days. Heating rate of TGA was 5C/mi nute, and air flow was controlled to 100 cc/minute. The mineral content of th e bovine tendon composites at 600C are indicated. 20040060080010001200 0 20 40 60 80 100 Pure tendon tendon mineralized with Pasp (Mw:10,300) EDTA-treated tendon mineralized with Pasp (Mw:10,300) tendon mineralized with Pasp (Mw:32,200) 0% 6 % 12 % 24 % Weight Loss (%)Temperature (oC)
131 CHAPTER 5 MINERALIZATION OF SYNTHETIC COLL AGEN SPONGE VIA T HE PILP PROCESS Introduction Collagen is a natural organic polym er found in various types of connec tive tissues, such as bone, cartilage, ligament and skin . Among the 27 types of collagens identified, more than 90% of the fibrous ones are type I collagen presen t in bone . As mentioned in chapter 2, the type-I collagen fibril is compos ed of tropocollagens in which 3 distinct polypeptides having (Gly-X-Y)n sequences are self-assembl ed with a triple helix structure [83, 84]. Those tropocollagens then self-assemble into a collagen fibril with th e quarter-staggered arrangement, leading to gap zones and grooves in the collagen fibrils which ar e thought to be very important for the intrafibrillar mineralization of collagen via the PILP precursor mechanism [15, 85, 86]. Collagen itself is considered an osteoconductive and biodegr adable natural polymer, and much research has been done in the bone graft area to produce synthetic collagen scaffolds for the development of bone grafts with collagen itself or combined with other substances such as hydroxyapatite or growth factors [4, 217, 218]. To prepare a synthetic collagen scaffold (reconstituted collagen), first one must obtain the molecular protein by the following steps, including extraction, separation, purification, fibr illogenesis, and lyophilization . Normally, collagen is extracted from various natural collagen sources, such as porcine skin, bovine tendon or human placenta, and then by dissolving these s ources in HCl solution and filtering the solution [84, 219]. Since collagen sources normally contain insoluble a nd soluble forms of collagen, soluble collagen is usually sepa rated from the extract to produce the synthetic scaffold. The soluble collagen molecules are separated from the extract by the addition of NaCl, which precipitates soluble collagen mol ecules by decreasing their solubility . The precipitates are re-dissolved in acidic solution, and purified with dialysis against an aqueous phosphate buffer at
132 4oC. A reconstituted collagen fibril is then pr oduced by fibrillogenesis, which is induced by the control of pH, temperature, a nd a collagen scaffold (sponge) is formed by lyophilization . Natural collagen scaffolds can also be produced by a demineralization process. The inorganic substance (hydroxyapa tite) of bone can be remove d by EDTA or a diluted HCl solution, resulting in demineralized bone matrix (D BM) that has been used in bone grafts for decades . Demineralized bone matrix is know n as an osteoinductive bone graft because the undifferentiated stem cells surrounding the DBM can be differentiated into osteogenic cells and later transformed into osteoblast cells by growth factors in DBM. It is known that growth factors, which were not extracted by the demineralizat ion process, induce the differentiation and proliferation of stem cells in to osteogenic cells . However, although the DBM has a remarkable osteoinductive ability due to the intact gr owth factors such as bone morphogenetic proteins (BMPs), the uses of DBM are limite d because those intact proteins after the demineralization process can lead to problems, such as immunogenic response from host and species-to-species transmissible di seases [137, 221]. Therefore, th e uses of synthetic collagen scaffolds, which are purified with chemical and physical treatments, have been growing in interest for bone graft, tissue engineering or drug delivery systems. In regards to the development of synthetic bone grafts, the minera lization of synthetic collagen scaffolds has been a key issue because a mineralized collagen scaffold with the fundamental structure of bone is expected to replace the use of aut ograft tissue, which is considered the gold-standard of bone grafting. There have been many attempts to develop bonelike collagen/hydroxyapatite composites by usin g the conventional nucleation and growth method [59, 183, 184]. However, while those methods, in which a collagen substrate is used for inducing the nucleation of hydroxyapa tite, have achieved several successes in fabricating simple
133 collagen/hydroxyapatite composites, they are not able to produce a collagen/hydroxyapatite composite with the nanostructure of bone [4, 45]. Our group has introduced a novel technique called the polymer-induced liquid-precursor (PILP) process, from which we have developed a collagen/hydroxyapatite composite with the fu ndamental structure of bone, in which nanohydroxyapatite crystals are embedded in the collagen fibril [15, 26, 198]. However, systematic researches on the conditions that are optimal fo r mineralization of synthetic collagen sponges via the PILP process were needed. In particular, the collagen used in our groups initial studies was no longer available because the supply of th is collagen sponge (Cellagen) was no longer available through Sigma (via a Japanese based supplier; presumably ove r concerns about mad cow disease). This limited us to a less-than-compr ehensive understanding of the role of the PILP process in the calcium phospha te mineralization system. Since then, a more stable supply of a synt hetic collagen sponge with an appropriate structure for intrafibrillar mineralization ( i.e. fibrillar type I collagen with the native banding pattern) was found to be available through a dom estic company called Collagen Matrix Inc.. Therefore, systematic research into the mineraliz ation of synthetic collagen sponges via the PILP process has resumed, which includes testing va rious reaction parameters (such as polymer molecular weight), and characterization by XRD, SEM, TEM and TG/DTA. The results of this work lead us closer to the development of bone-like collagen/hydroxyapatite composites that may provide a high purity, synthetic bone graft to augment or replace the limited supply of autograft and allograft tissues. Materials and Methods Mineralization of Collagen Sponges As a collagen scaffold for a m ineralization, synthetic collagen sponge was purchased from Collagen Matrix, Inc. (New Jersey, USA). The dimensions of the collagen sponge disc are 13
134 mm in diameter and 3 mm in a thickness. The density of the sponge is 0.04 g/cm3 and the estimated pore size (provided by the company) varies from 75 to 350 m. The collagen sponge was pre-washed with de-ionized water to remove any salts or chemical agents introduced during fabrication. The collagen sponge disc was cut in half into a semicirc ular shape. Four semicircle collagen sponges were mineralized in 500 ml of mineralization solu tion. The ratio of number of sponges to volume of solution was mainta ined constant between every mineraliza tion experiment. The mineralization solution was prepared by mixing equal volumes of CaCl2H2O (9 mM) and K2HPO4 (4.2 mM). To maintain the pH of th e mineralization solution at 7.4 during the reaction, each stock solution was made by dissolv ing the calcium and potas sium salts in trisbuffer solution containing 0.9 wt% of NaCl. Poly -aspartates with different molecular weights were used as the PILP process-directing agent. Poly-( )-DL-aspartic acid sodium salt (Mw : 5500 Da) or poly-L-aspartic acid sodium salt (Mw : 10,300 or 32,200 Da) was added to the CaCl2H2O (9 mM) stock solution before mixing in the counterion stock solution, and the concentration of polymer was fixed at 50 g /ml. Collagen sponges were soaked in the mineralization solution under vacuum conditions for 30 minutes to remove any air bubbles trapped in the pores of the sponge The reaction was run in a 37oC oven to emulate physiological conditions. After certain reaction times, the mineralized collagen sponge was washed with deionized water several times and lyophilized for 24 hours to reduce the structural changes that occur in collagen fibrils during the drying process. X-Ray Diffraction (XRD) analysis Because the collag e sponge is composed of collagen fibers randomly distributed, the x-ray diffraction analysis settings (Mode s I and II) used for the tendon analysis, were not needed for the isotropic order of the synthetic sponge. Mi neralized collagen sponges were scanned with CuK X-ray radiation from a Philips XRD 3720 at 40 KV and 20 mA, using a step size of 0.01
135 with a time of 1.25 sec/step, over a 2 range of 24 40 to obse rve the major hydroxyapatite peaks developed at around 26 (002) and 32 (combination of (211) (112) and (300)). Scanning Electron Microscopy (SEM) Analysis The m ineralized collagen sponge was washed with de-ionized water several times and lyophilized by freeze drying for 24 hours. Dried samples were mounted on an aluminum stub covered in double-sided copper tape, and then sputter coat ed with amorphous carbon. The morphologies of the surface of mineralized collagen were then analyzed using a 6400 JEOL SEM at 15 kV. For elemental analysis of mineralized tendon, ener gy dispersive x-ray spectroscopy (EDS) analysis was done during SEM analysis. Transmission electron microscopy (TEM) analysis Micro-structural analysis wa s perform ed on a 200cx JEOL TE M (200 kV) with bright filed (BF), dark field (DF) and selected area elec tron diffraction (SAED) modes to determine the crystallographic orientation of the embedde d hydroxyapatite. For transmission electron microscopy (TEM) analysis, the mineralized coll agen sponge was pulverized into powder. The powder was then dispersed in ethanol and then applied as a drop onto a copper TEM grid. To prevent electron charging, the sample was slightly coated with amorphous carbon. Thermogravimetric and Differential Thermal Analysis (T G/DTA) The mineralized collagen sponge was pulverized using liquid nitrogen and dried at air for 24 hours in preparation for thermogravimetric an d differential thermal analysis (TG/DTA). Collagen sponge powder (10 20 mg) was transferre d into a platinum pan and fired under air conditions. Alumina powder was used for the st andard. Analysis was conducted from 30 to 800oC with a heating rate of 5oC/minute. To compare the degr ee of mineralization, the weight loss value at 600oC was interpreted as the mineral content because organic substances within the collagen sponge are totally combusted before 600oC in the oxygen-containing environment. In
136 some cases, a peak area under exothermic reaction peak was calculated by the analysis program in the instrument. Results and Discussion Mineralization of Collagen Sponges by the PILP Processes Induced by the Different Molecula r Weight Poly-aspartic Acids In the mineralization of bovine tendon, which c ontained calcification in hibitors, the degree of mineralization via the PILP process depended on the molecular weight of the poly-aspartate additives. This correlation of the polymer molecular weight dependence on the intrafibrillar mineralization of collagen fibrils needed to be confirmed with the sy nthetic collagen sponge system since it has no inhibitors. When the collagen sponge was minera lized with the PILP process containing the poly-( )-DL-aspartic acid (Mw: 5,500 Da), no hydroxyapatite peaks were observed, even after 16 days of mineralizat ion (Figure 5-1), which is generally long enough to find mineral in this system. Instead of a hydroxyapatite peak, a broad peak around 31o was observed after 3 days of reaction. The inte nsity, however, of this broad peak at 31o did not increase proportionately as the mineralization progressed It is thought that th is peak is produced by a small amount of amorphous calcium phosphate w ith short-range order, and the differences in surface roughness between samples may cause a non-proportional increase in intensity. The high intensity of amorphous calcium phosphate peak (between 31 a nd 32) from the sample after 3 days of mineralization (Figure 5-1) may be considered as a peak from crystalline phase due to the peak intensity as high as the intensity of (211) peak of hydroxyapat ite crystal (shown in Figure 5-4). However, because no (002) peak, which theoretically has a high relative intensity (54 out of 100), was observed, it was s till considered as an amorphous phase. It is worthwhile to note that there was an agglomeration of amorphous precursor in the solution as the mineralization progressed, when low molecular weight polyaspartate (Mw: 5,500
137 Da) was used for the PILP process. As show n in Figure 5-2 A, a spongy film formed by the agglomeration of amorphous precursor was observed on the glass slide on which the collagen sponges were resting. When collected for obser vation by TEM (Figure 5-2 B), SAED patterns of agglomerates showed a diffuse ring indicating that the precursor was amorphous even after 6 days of reaction. In contrast to hydroxyapatite clusters that are normally precipitated from solution and deposited on the slide in control ex periments (without the a ddition of poly-aspartic acid), the adhesion between precursor film and glass slide was so weak that the films just washed away by dipping the glass slide into water for ri nsing. When the collagen surface was observed by SEM, there was no morphological difference in the collagen sponge mineralized for different time periods. As shown in Figure 5-3 A, a sm all amount of precipita te was observed on the collagen sponge mineralized for 1 day. EDS anal ysis showed that the precipitate was sodium chloride and calcium chloride that had remained, even after washing proce ss. In the cases of collagen sponges mineralized for 8 days and 16 days (Figure 5-3 B and C), more precipitates were observed on the collagen sponge surface. ED S analysis confirmed that the precipitates were composed of sodium chloride, calcium chloride and amorphous calcium phosphate. The content of calcium and phosphate detected by the EDS did not linearly increase with the reaction time. It is thought that since the amorphous calci um phosphate is not chemically attached to the collagen, the amorphous calcium phosphate formed by the agglomeration of precursors may be washed out irregularly during sample preparatio n. Those results imply that the low molecular weight polyaspartate is not reliable for the intr afibrillar mineralization of collagen sponges. Even though there is some formation of amor phous calcium phosphate precursors by the low molecular weight poly-aspartate, most of the calcium phosphate precursor seems to be consumed by the agglomeration of precursor droplets in solution, and not by the infiltration of a liquid
138 precursor into the collagen fibrils. However, it is notable that both the nucleation and growth of hydroxyapatite clusters was inhibited by the additi on of polyaspartate, and the transformation of precursors to hydroxyapatite was also prohibited. When the collagen sponge was mineralized with the PILP process containing poly-Laspartic acid in the medium range of molecula r weight (Mw: 10,300 Da), a broad peak indicative of amorphous calcium phosphate was observed afte r 1 day of reaction. After 6 days of reaction, broad peaks of hydroxyapatite were observed at around 26o and 32o corresponding to the (002) and the combination of (211), (112), (300) peaks, respectively (Figure 5-4). This x-ray pattern of mineralized collagen is almost identical to th at of carbonated hydroxyapat ite in bone. The (211), (112), (300) peaks of carbonated hyd roxyapatite in bone are normally overlapped into one broad peak due to the peak broadening caused by the ex tremely small size of the crystals, and defects such as carbonate, impurities and calcium vacancies [208, 209, 222]. The morphology of the collagen surface also changed as the mineraliza tion process progressed. In the case of SEM observation at low magnification, the collagen su rface was smooth and film-like after 1 day of mineralization, with no calcium or phosphate detected by EDS (Figure 5-5 A). As the mineralization progressed, however, the previous ly smooth surface now appeared to have a fibrous structure after 3 days of mineralization (Figure 5-5 B). After 6 days of reaction, the collagen fibers on the surface appeared bumpy and rough (Figure 5-5 C). EDS analysis determined that the bumpy and rough collagen fi bers contained a high amount of calcium and phosphate, as compared to the carbon peak at ar ound 0 keV. According to the x-ray diffraction analysis, the bumpy and rough collagen fiber appears to be produced as hydroxyapatite is being formed. Given that the hydroxyapatite coat layer formed by the agglomeration of precursors was not observed for this polymer, the calcium a nd phosphate content in the collagen fiber as
139 observed in EDS and XRD suggest that the hydroxyapa tite crystals are embedded in the collagen fibers. The formation of bumpy collagen fibe rs versus the progressi on of mineralization was closely investigated by high magnification SEM. During the initial stage of mineralization (Figure 5-6 A and B), the fibrous surface develo ped from the smooth surface due to the nodules that were formed along the collagen fibrils. This nodular formation process is seen to continue into 6 days of reaction (Figure 5-6 C), at whic h time the peaks of hydroxyapatite were observed by the x-ray diffraction. As the mineralization proceeded, the nodules gradually grew and spread along the collagen fibers (Figure 5-6 D and E) until the whole surface became bumpy due to the formation and growth of nodules throughout the collagen fibrils (Figure 5-6 F). In the case of the mineralization with high molecular weight poly-L-aspartic acid (Mw: 32,200 Da), broad peaks of hydroxyapatite at 26o and 32o developed after only 3 days of reaction (Figure 5-7). After 3 days of mineralization, the nodules formed on th e collagen fibers had a large amount of calcium and phosphate as detect ed by EDS (Figure 5-8 B), similar to that observed in the collagen mineralized with low molecular weight poly-L-aspartic acid (Mw: 10,300 Da) at 6 days. The nodules grew and spread along the collagen fibers as they become cylindrical in shape. In contrast to the bum py surface on the collagen sponge mineralized with low molecular weight poly-L-a spartic acid (Mw: 10,300 Da) for 16 days (Figure 5-6 F)), these collagen fibrils, which had had multiple nodules, became thicker and uniform by the growth of the nodules until, after 16 days of mineraliza tion, the whole surface was composed of smooth and thick collagen fibers without nodules. Alth ough the surface collagen fibers were smooth and thick, indicating complete mineralization, fibrils fr om the interior of th e collagen sponge were not completely mineralized, even after 16 days of reaction. As shown in Fi gure 5-9 A, the inside of the collagen sponge was composed of the bumpy and rough surface due to the formation of
140 these nodules on the partially mineralized collage n fibers. The morphology was very similar to the morphology of collagen fibrils on the surface of the sponge mineralized for 8 days of reaction (Figure 5-9 B), indicating that the inside of collagen sponge was not yet completely mineralized. As judging from SEM observations, it appears that the mineralizati on was initiated at random positions on the collagen fibrils, leading to the formation of nodules. It is thought that the nodules on the collagen fibers are formed by the infiltration of precursor into the collagen, which is not occurring uniformly, where some portions of the fibrils are infiltrated by precursor before others during the initial stage of mineralization. Once th e fluidic amorphous precursor is drawn into the collagen fibrils, it solidifies and crystallizes in to hydroxyapatite crystals, which] replacing much of the water as it fills the in terstices of the fibrils, which makes the collagen structure appear swollen upon drying.. Therefore, while the portions of collagen fibrils filled with hydroxyapatite crystals maintain the volume of collagen fiber during the drying process, the regions of collagen fibrils filled with only wa ter and organic substances is shrunken by the drying process for SEM, leading to the formati on of swollen nodules. Then, as the collagen fibrils are completely mineralized, the nodule formation is no longer observed because whole fibrils are filled with intraf ibrillar hydroxyapatite crystals. It may be argued that the crystallization or solidification of precursors forming the nodules can be initiated at an active site for hydroxyapatite nucleation due to an appropriate combination of amino acid sequences that present specific functionality. However, the formation of nodules al ong the fibrils does not appear to be periodic or regular. Therefore, it is hard to argue that the regularly dispersed sites such as gap zones or ends of tropocollagens init iate the crystallization of the precursors. Although the EDS analysis showed very high content of calcium and phosphate in the collagen sponge mineralized with poly-L-aspa rtic acid (Mw: 10,300 or 32,200 Da), there was no
141 formation of a hydroxyapatite coating layer or clusters on the collagen sponge surface. Moreover, there was no agglomerati on of precursor forming an amorphous film in the solution or on the sponge (which was observed in the minera lization process with low molecular weight poly-( )-DL-aspartic acid (Mw: 5,500 Da)). The mineralization solution was clear after 16 days of reaction, indicating there was no waste of calcium and phosphate ions in the form of solution precipitates and agglomerat ion. Therefore, it is reasonab le to assume that the calcium and phosphate signals are origin ating from the intrafibrillar hydroxyapatite embedded in the collagen fibrils. As shown in Figure 5-10 A, th e mineralized collagen fibril showed very high contrast in TEM (without staining) due to the intrafibrillar hydroxyapatite with high electron density. According to the SAED pattern of the isolated collagen fibril (Figure 5-10 C), the fibril was mineralized with well orient ed hydroxyapatite nanocrystals cr eating a set of arcs, including the pronounced (002) arcs. The uniaxial orienta tion of hydroxyapatite was also confirmed by the dark field image constructed from the (002) arcs. As shown in Figure 5-10 B, the needle-like spots correspond to the uniaxial arra y of hydroxyapatite crystals that are aligned wi th their  direction parallel to the long axis of the collagen fibril. It is thought that the growth of hydroxyapatite is constrained by the narrow space between collagen molecules, keeping the dimensions of the crystals very small, as in bone In addition, if growth in the naturally favored  direction is rapid, it could lead to a preferred cr ystal orientation; but preferred orientation could also be caused by a specific nu cleating domain in the collagen. The mineral content of the collagen sponges mineralized by the PILP process were examined by TG analysis. As shown in Figure 5-11, the mineral content of the collagen sponge mineralized with poly-L-aspartic acid (Mw: 32,200 Da) increased from 58 wt% (8 days of mineralization) to 74 wt% after 16 days of mi neralization. Similarly, when collagen sponges
142 were mineralized with poly-L-aspartic acid (Mw: 10,300 Da), the mineral content of collagen sponge was 34 wt% and 60 wt% after 8 and 16 days of mineralization, respectively. In both cases, the mineral content increased as the mineralization progressed, although the higher molecular weight polymer leaded to a consider ably faster mineralization rate. The mineral contents after 16 days of mineralization with di fferent molecular weight poly-L-aspartic acids (32,200 and 10,300 DA) were both over 60 wt%, which is similar to the mineral content of bone, which has a composition of 60-70 wt% of minera l, 20-30 wt% of organic phases and 5-10 wt% of water [4, 10, 80]. There are tw o possible reasons why the differe nt molecular weights lead to differences in mineral content after 16 days. Fi rstly, the infiltration of the precursor into the collagen matrix may be affected by the differe nt molecular weight polymers due to some different characteristics of the precursor phase. For example, as the molecular weight of the polymer increases, the net surface charge of the polymer will be increased , which in turn can affect the characteristics of the precursor, such as size, interfacial environment between precursor and solution, and net surface charge of precursor droplets. This may influence the infiltration of the precursor thr ough the gap zones, which are known to have a special steric and electrostatic environment in comparison to the ov erlap zones of collagen fibrils. Secondly, the amount of precursor formed by the PILP process may be varied by the molecular weight of polyaspartic acid. However, if one polymer chain, wh ich is supposed to be dispersed evenly due to its negatively-charged functiona l groups, is assumed to form one precursor droplet, low molecular weight poly-aspartic acid should form more precursor droplets (of smaller volume) because the same weight of poly-L-aspartic ac id was added to the mineralization solution, but the overall weight/volume of precursor phase woul d likely be constant. Therefore, it is thought
143 that the difference in some characteristics of the precursor droplets formed by different molecular weight polymers may cause the variat ion in degree of intrafibrillar mineralization. Enhancement of Mineralization by Buffer Treatment or Solution Replacement Tris-buffer solution treatment As shown in the SEM an alysis of the mineralized collagen sponges, the surface morphology of the collagen sponge became fibrous as the mineralization r eaction progressed. It can be argued that the minerali zation of collagen fibril s increases the volume of collagen fibers by the formation and growth of nodules, causing the fibrous structure to be visually apparent. However, as shown in Figure 5-5 or 5-8, th e fibrous surface was developed before the appearance of nodule formation on th e collagen fibril. When the collagen sponge was soaked in tris-buffer solution without calci um and phosphate ions, the fibr ous structure appeared on the collagen sponge without mineralization. As shown in Figure 5-12 A, after 1 day of tris-buffer treatment, the surface initially seemed to be smooth with a thin film-like appearance. The surface became fibrous as the soaking time was in creased to four days. Normally, during the fibrillogenesis of collagen molecu les, the collagen precipitate contains two types of collagen: collagen fibrils and a non-fibrous amorphous co llagen gel. Although the company (Collagen Matrix, Inc.) did not inform us of the process fo r collagen sponge fabrica tion, it is thought that the non-fibrous collagen forms th e smooth film on this commercial collagen sponge. It may also be that, during the mineralization or tris-bu ffer treatment, the non-fibrous collagen can be dissolved into solution, such that collagen fi brils underneath the non-fibrous collagen film may be revealed. This non-fibrous collagen film seem s to act as a physical ba rrier that inhibits the infiltration of amorphous precursors in the initial stage of mineralization. When the collagen sponge, which had been soaked in tris-buffer so lution for 5 days to remove this non-fibrous collagen film, was mineralized by the PILP proc ess containing poly-L-aspa rtic acid (Mw: 32,200
144 Da), the peaks of hydroxyapatite developed afte r only 2 days of minera lization (Figure 5-13), while hydroxyapatite peaks were not observed until 3 days of mineralization when the collagen sponge was not pre-treated with tris -buffer solution (Figure 5-7). Ther efore, it is thought that the removal of non-fibrous collagen film allows bette r infiltration of the amorphous precursors into the collagen fibrils, without the physical hindrance of this non-fi brous film, and to crystallize into hydroxyapatite at a very ea rly stage of mineralization. The formation and growth of nodules on collagen fibers was also enhanced when the co llagen sponge had been treated with tris-buffer solution before the mineralization. As shown in Figure 5-14, the degree of nodule formation in tris-buffer treated collagen sponge mineralized for 2 days was very similar to that of the nontreated collagen sponge mineralized for 8 days (Figure 5-8 C), i ndicating that more portions of collagen fibers pre-treated by tris-buffer solution were infiltrated by the amorphous calcium phosphate precursor due to the removal of non-fi brous collagen film in comparison to the collagen sponge without tris-buffer treatment Although microscopic analysis and X-ray diffraction analysis showed the enhancement of mineralization, to clarify the effect of noncollagenous film on the degree of intrafibrillar mineraliz ation, the further qua ntitative analysis, such as TGA and direct weight measurem ent, should be done in future. Mineralization solution replacement When the collagen sponge (not pre-treated wi th tris-buffer) was m ineralized by the PILP process containing poly-L-aspartic acid (Mw: 32 ,200 Da), most of the pr ecursor absorption into the collagen fibrils was completed within 8 da ys of mineralization (F igure 5-11, 58 wt % of mineral after 8 days of mineralization). Theref ore, to increase the mineral content in the collagen sponge, 8 days of mineralization was pe rformed twice. As shown in Figure 5-15, the mineral content of collagen sponge mineralized tw ice is 81 wt% which is 7 wt% higher than the mineral content of collagen sponge mineralized for 16 days. Therefore, it is expected that a
145 continuous supply of amorphous calcium phosphate pr ecursor can increase the mineral content in mineralized collagen sponges. It is also worthwhile to note that there is about 3 4% weight loss between 600oC and 1200oC, even though the organic substa nce was completely decomposed under 600oC. The transformation of mineral phase from hydroxyapatite to tri-calcium phosphate (Ca3(PO4)2) was also observed after this high temper ature heat-treatment (Figure 5-15, x-ray diffraction pattern). As explained in chapte r 2, the mineral phase in bone is a carbonated hydroxyapatite in which some of the phosphate groups are substituted with carbonate groups. Therefore, the heat-treatment of carbonated hydroxyapatite at high temperature can cause the decomposition of carbonate and hydroxyl groups in hydroxyapatite, le aving tri-calcium phosphate crystals . Theref ore, the weight loss between 600oC and 1200oC is probably caused by the decomposition of carbonate and hydroxyl group from carbonated hydroxyapatite. This is also evidence that the hydroxyapatit e formed by the PILP process is carbonated hydroxyapatite, similar to the mineral phase in bone which has about 4 wt% of carbonate . In this in vitro system, the carbonate presumably comes from CO2 dissolved in the aqueous solution (which is not purged). The Thermal Analysis of Mi neralized Collagen Scaffolds The therm al analysis of collagen sponges mineralized by the PILP process containing poly-L-aspartic acid was done by TG/DTA. In th e case of pure collagen (Figure 5-16), three consecutive drastic weight losses within the ranges of 30-100C, 260 360C and 450-550C were observed due to the evaporation of wa ter (30-100C), and decomposition and combustion of organic collagen molecules (260 360C and 450-550C). The da ta showed an exothermic reaction around 517oC, which corresponds to the decomposition and combustion of thermally stable collagen molecules [224, 225]. During fi brillogenesis, collagen molecules are thought to form crosslinks and secondary bonds, such as hydrogen bonds, electros tatic bonds, and van der
146 waals bonds (the hydrophobic effect), that may pl ay an important role in the formation of thermally stable collagen  However, such bonds are no t generally stable at these temperatures, and therefore are more likely to result from some type of degradative thermal transformation into more stable bonds (perhaps -C=N-, -C Netc.) which could be verified by FT-IR spectroscope or Raman spectroscope. Interestingly, this t hermal stability ( i.e. high temperature DTA peak) is lost for the collagen in bone and dentin, and was suggested by Lozano  as being due to intrafibri llar hydroxyapatite crystal acting fractur e center. While we do not believe this explanation is accurate, since even cova lent crosslinks would not be able to persist to these temperatures, it is interesting that the pr esence of mineral somehow disrupts this thermal transformation peak (perhaps catalyzes the other low temperature degradation processes, such as straight combustion). However, to verify the m echanism of the decrease in thermal stability of collagen, further investigation with various techniques, such as FT-IR, Raman and DSC, must be done. In our system, when the collagen was mine ralized with the PILP process leading to the intrafibrillar mineralization of collagen fibril, th e high temperature, thermal stability peak of collagen also disappeared. As shown in Figure 5-17 A and B, as mineralization of the collagen sponge with poly-L-aspartic acid (Mw: 32,200D a) progressed, the exothermic reaction corresponding to the decomposition of thermally stable collagen at high temperature (450 550oC) was shifted to a lower temperature (440oC, Figure 5-17 A) after 8 days of mineralization. By 16 days of mineralization, this high temperat ure peak was shifted to lower temperature (485 oC ) and intensity was greatly decreased (Fi gure 5-17 B). Besides the small and broad exothermic peak from the thermally stable co llagen, the big and sharp exothermic peak was observed at around 337oC corresponding to the thermally unstable collagen molecule (Figure 517 B). This result shows that the PILP minerali zation of collagen changes the thermal stability
147 of collagen in a similar fashion to bone. Sakae et al.  showed that human dentin had a collagen matrix whose thermal stability was lower than that of pure collagen or collagen from demineralized dentin, and argued that the hydroxyapatite, which wa s embedded in the interstices between tropocollagens, might expand the collagen fibril structure and break the crosslinks between tropocollagen molecules, or act as a fr acture center leading to thermally weak collagen structure [224, 225]. Although we dont necessar ily agree with their suggested chemical mechanism, it is thought that the change in thermal stability of collagen after the PILP mineralization can be used as evid ence of the intrafibrillar minera lization of collagen fibrils, in which hydroxyapatite crystals ar e embedded in the collagen fibril, and not on the collagen surface, which does not alter the collagen ther mal characteristics. With calibration, this technique might even provide a useful means for quantifying the degree of in trafibrillar mineral, which cant be done with TGA because it cannot di stinguish between intraand interfibrillar mineral content. It is also worthwhile to note that an exot hermic reaction correspondi ng to the transition of amorphous calcium phosphate to hydroxyapatite is observed from the collagen sponge mineralized for 8 days. As shown in Figure 517 A, three distinct exothermic peaks were observed around 335oC, 440oC, and 540oC. The exothermic reactions around 335oC and 440oC are thought to be caused by decomposition and combustion of collagen molecules, whose thermal stability was decreased by th e intrafibrillar mineralization. However, it is hard to argue that the exothermic reaction at 540oC is due to the thermally stable collagen because, during the exothermic reaction, no significant weigh loss is observed. According to Kumar et al. , the amorphous calcium phosphate produced by a pl asma spraying process transformed into hydroxyapatite between 500oC and 650oC, depending on the composition of the amorphous
148 calcium phosphate. Therefor e, it is thought that the exothermic reaction around 540oC may correspond to the transition of amorphous calcium phosphate to hydroxyapatite. As mentioned earlier, the abso rption or degree of infiltra tion of amorphous precursors depended on the molecular weight of poly-aspartic acid used to i nduce the PILP process. Similar results were observed in the TG/DTA analysis. As shown in Figure 5-18 A and B, when the collagen was mineralized with poly-L-aspartic acid (Mw: 10,300 Da), the decrease in the thermal stability of collagen was also observed due to the intrafibrillar mineralization of collagen governed by the PILP process. However, when the area below the exothermic reaction peak between 380oC and 510oC in Figure 5-18 A was compared to the area below the exothermic reaction peak between 380oC and 510oC in Figure 5-17 A, the calcu lated area in Figure 5-18 A (10,300 Da poly-aspartic acid) was larger than that in Figure 5-17 A (32,200 Da poly-aspartic acid). Although the DTA is not proper method for the direct quantitative comparison between two areas, an indirect deduction of quantitative information is possible because the temperature difference between standard and sample also depe nds on amount of material s with different heat capacities (For examples, DTA results between sa mple 1 composed of 1g of copper and 2g of alumina sample 2 composed of 2g of copper a nd 1g of alumina will show different area of reaction peak at the same posit ion). Therefore, the result implies that the degree of the intrafibrillar mineralization is higher in the co llagen mineralized with high molecular weight poly-aspartic acid, because the area under the exothermic reacti on peak, indirectly corresponding to energy for the decomposition and combustion of thermally stable collagen, was smaller. The collagen sponge mineralized with high molecular weight poly-aspartic acid also contained more amorphous calcium phosphate solid than the coll agen sponge mineralized with low molecular weight polymer. The calculated areas of the e xothermic reactions of this phase transition
149 (around 530-560oC) were -35.12Vs/mg for the collagen mineralized with low molecular weight polymer and -53.44Vs/mg for the collagen minera lized with high molecular weight polymer. The calculated area in Figure 5-17 (32,200 Da poly-as partic acid) was almost twice as large as the area in Figure 5-18 (10,300 Da pol y-aspartic acid). Therefore, it is thought that the collagen sponge mineralized with high molecular weight poly-aspartic acid contains more amorphous calcium phosphate. It can be ar gued that the rate of transfor mation of the precursor may be slower when the amorphous precurs or is induced by the high molecu lar weight poly-aspartic acid, leaving a larger amount of amorphous calci um phosphate solid. However, when the development of the hydroxyapatite peaks observed by x-ray diffraction is cons idered, it is hard to argue that the rate of transformation of precursor formed by high molecular weight polymer is slower because the hydroxyapatite peaks are obs erved after only 3 days of reaction (The hydroxyapatite peaks were observed after 6 days of reaction from the collagen mineralized with 10,300 Da poly-aspartic acid). Conclusions The synthetic collagen sponge was successfully m ineralized with the PILP process leading to the intrafibrillar mineraliza tion of collagen fibrils. When th e collagen sponge was mineralized with low molecular weight poly-aspartic acid (Mw: 5,500 Da), although amorphous calcium phosphate precursor was formed, the precursor was wasted as a form of agglomeration of solution precipitates that prevented the intrafib rillar mineralization of collagen sponge. On the other hand, when the collagen sponge was mineraliz ed with high molecular weight poly-aspartic acid (Mw: 32,200 Da), the collagen fibrils were successfully mineralized with hydroxyapatite crystals that were embedded in the collage n fibrils with uniaxial orientation along the c -axis of the collagen fiber. Mineral contents matchi ng that of bone were obtained after 16 days of mineralization, and all of the minerals appeared to be intrafibrillar. The degree of mineralization
150 could be enhanced by the tris-buffer treatment for removing the non-fibrous collagen film, which apparently acts as a physical barrier against pr ecursor infiltration. Wh en the collagen sponge was mineralized with medium range of molecula r weight poly-aspartic acid (Mw: 10,300 Da), although some intrafibrillar mineralization of collagen could be achieved, the degree of mineralization was lower than that of collagen sponge mineralized with high molecular weight polymer. In the case of thermal analysis of mineralized collagen sponges, the th ermal stability of the collagen was gradually decrease d with increasing degree of intrafibrillar mineralization, corresponding well to the data de termined by other characteriza tion techniques for the polymer molecular weight trends, and matching the thermal characteristics of bone/dentin. An exothermic reaction for the transition of amorphous calcium phosphate solid was observed at the initial stage of mineralization, indicating that the in trafibrillar mineraliza tion was achieved via the precursor mechanism.
151 Figure 5-1. The XRD patterns of Collagen Matrix sponge mineralized by the PILP process for various times, using 50 g/ml of poly-( )-DL-aspartic acid (Mw: 5,500). The mineralization solutions for the PILP proce ss in this chapter a ll contained 4.5 mM of calcium chloride and 2.1 mM of potassium phosphate. 25 30 35 40 16 days 3 days 2 days 1 day collagen sponge2(degree)
152 Figure 5-2. Optical microscope (Left) and TEM (Right) micrographs of solution borne precipitate formed with low molecula r weight polymer, 50 g/ml of poly-( )-DLaspartic acid (Mw: 5,500), at 6 days of reac tion. Inset in B) contains a selective area electron diffraction pattern of the amorphous precipitate. A) B)
153 Figure 5-3. The SEM micrographs and EDS anal ysis of a collagen sponge mineralized by the PILP process induced with 50 g/ml of poly-( )-DL-aspartic acid (Mw: 5,500) for various times. Bars indicate 20 m. No proportional progress of mineralization was observed. A) 1 day B) 8 days C) 16 days
154 Figure 5-4. The XRD patterns of collagen sponge s mineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 10,300) for various times. The broad peak around 32 is composed of overlapping peaks of (211), (112) and (300), whose 2 for hydroxyapatite would correspond to 31.785, 32.192 and 32.924 degrees, respectively. (002) (211) (112) (300) 25 30 35 40 6 days 3 days 2 days 1 day collagen sponge2(degree)
155 Figure 5-5. The SEM micrographs and EDS anal ysis of collagen sponges mineralized by the PILP process containing 50 g/ml of polyL-aspartic acid (Mw: 10,300) for various times. Bars indicate 20 m. Note the rough and bumpy texture of the partially mineralized sponge at 6 days. A) 1 day B) 3 days C) 6 days
156 Figure 5-6. The SEM micrographs of collage n sponges mineralized by the PILP process containing 50 g/ml of poly-L-aspartic ac id (Mw: 10,300) for various times. Bars indicate 5 m except F) indicating 20 m As the mineralization progressed, the nodule grew along the collagen fibril, leading to thick fibril. B) 3 days A) 1 day D) 8 days E) 16 days C) 6 days F) 16 days
157 Figure 5-7. The XRD patterns of collagen sponge s mineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 32,200) for various times. Broad peak around 32 is composed of overlap of (211), (112) and (300) whose 2 are 31.785, 32.192 and 32.924, respectively. (002) (211) (112) (300) 25 30 35 40 3 days 2 days 1 day collagen sponge2(degree)
158 Figure 5-8. The SEM micrographs and EDS anal ysis of collagen sponge mineralized by the PILP process containing 50 g/ml of polyL-aspartic acid (Mw: 32,200) for various times. Bars indicate 5 m except D) indicating 20 m. As mineralization progressed, the surface became fibrous, and the formation of nodules was observed, which later grew along the collagen fibril, leading to the thick and uniform collagen fiber indicating the completion of mineralization. A) 1 day B) 3 days C) 8 days D) 16 days
159 Figure 5-9. The SEM micrographs and EDS analys is of the inner part of a collagen sponge mineralized by the PILP process with 50 g/ml of poly-L-aspartic acid (Mw: 32,200 Da) for 16 days. The inner part of colla gen sponge, whose th ickness is 1.5-2.0 mm, was exposed by cutting the collagen sponge into half with a razor blade. Bars indicate A) 20 and B) 5 m. A lower degree of mineralization is seen, in comparison to the surface of the sponge shown in Figure5-8 D. A) B)
160 Figure 5-10. A) Bright fiel d and B) dark-field TEM micrographs of a collagen sponge mineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 32,200Da) for 16 days. Dark field image is constructed using the 002 arc selected by the objective aperture, as shown in the SAED pattern in C). A) B) C)
161 Figure 5-11. Thermogravimetry analysis of co llagen sponges mineralized by the PILP process induced with 50 g/ml of poly-L-aspartic ac id for 8 and 16 days. The poly-L-aspartic acids with different molecular weights we re used to investigate the effect of molecular weight of polymer on the PILP mineralization. Heating rate of TG was 5oC/minute and air flow was controlled to 100 cc/minute. The mineral contents (wt %) at 600oC are indicated in Figure. 200 400 600 800 0 20 40 60 80 100 8 days, 10,300 Da 16 days, 10,300 Da 8 days, 32,200 Da 16 days, 32,200 Da 34 % 58 % 60 % 74 % Weight Loss (%)Temperature (oC)
162 Figure 5-12. The SEM micrographs of collagen sponges after tris-buffer treatment for various times. The collagen sponges were soaked in a tris-buffer solution and kept in 37oC oven for various times. Note= the amorphous film-like collagen seems to be replaced with a fibrillar texture with time. C) 4 days A) 1 day B) 3 days
163 Figure 5-13. The XRD patterns of collagen sponges pre-treated w ith tris-buffer solution for 5 days before the mineralization using the PI LP process. Poly-L-aspartic acid (Mw: 32,200 Da) was used for the PILP process and its concentration was 50 g/ml. Note: Compare this figure with Figure 5-7 show ing hydroxyapatite peak after 3 days of mineralization. (002) (211) (112) (311) 25 30 35 40 3 days 2 days 1 day collagen sponge2(degree)
164 Figure 5-14. The SEM micrographs of a collagen sponge that was pre-treated with tris-buffer solution for 5 days before the 2 days of mi neralization using the PILP process. PolyL-aspartic acid (Mw: 32,200 Da) was used fo r the PILP process and its concentration was 50 g/ml. Bar in the left image indicates 200 m and bar in magnified image (right) of area marked by white circle indicates 10 m. The degree of nodule formation is similar to 8 days of minerali zation sample which had not pre-treated with tris-buffer before the minera lization (Figure 5-8 C).
165 20040060080010001200 60 80 100 70 % 78 % 81 % 74 % 8+8 days of reaction 16 days of reaction Weight Loss (%)Temperature (oC) 20 25 30 35 40 : -TCP : -TCP Intensity2(degree) Figure 5-15. Thermogravimetry analysis of co llagen sponges mineralized by the PILP process for 16 days. One sponge was mineralized for 16 days, but, the other sponge was mineralized for 8 days twice. For the sec ond round of 8 days of mineralization, fresh mineralization solution was used. Heating rate of TG was 5oC/minute, and air flow was controlled to 100 cc/minute. Mineral contents at 600 and 1200oC are indicated. After TG analysis, the byproduct in the Pt pan was examined by XRD, showing it underwent a high temperature transfor mation to tri-calcium phosphate.
166 Figure 5-16. Thermo Gravimetric and Different ial Thermal analysis (TG/DTA) of a pure collagen sponge. Heating rate of TG/DTA was 5oC/minute and air flow was controlled to 100 cc/minute. Before analysis, the collagen sponge was soaked in deionized water for 1 day to remove solubl e salts, such as sodium chloride, and lyophilized. The temperatures of typical exothermic reactio ns seen in collagen are indicated. Notethe collagen undergoes a tw o-step degradation reaction, presumably transforming into some species with str onger bonds, which leads to the so called thermally stable collagen, with a high temperature peak at 517C. 517oC 359oC 296oC 200 400 600 800 0 20 40 60 80 100 DTA(mV) TG (%)Temperature-10 0 10 20 30 40 50 60 70
167 200 400 600 800 0 20 40 60 80 100 DTA(mV) TG (%)Temperature-10 0 10 20 30 40 50 60 70 200 400 600 800 0 20 40 60 80 100 DTA(mV) TG (%)Temperature-10 0 10 20 30 40 50 60 70 Figure 5-17. TG/DTA of collage n sponges mineralized by the PILP process containing 50 g/ml of poly-L-aspartic acid (Mw: 32,200 Da) for 8 and 16 days. Heating rate of TG/DTA was 5oC/minute and air flow was controlled to 100 cc/minute. The temperatures of typical exothermic reactions were indicated. Note the dramatic change in the DTA spectrum, in which the high temperature peak of the collagen is nearly completely lost. 335 oC 440 oC 540 oC 337 oC 485 oC A) 8 days B) 16 days
168 200 400 600 800 0 20 40 60 80 100 DTA(mV) TG (%)Temperature-10 0 10 20 30 40 50 60 70 200 400 600 800 0 20 40 60 80 100 DTA(mV) TG (%)Temperature-10 0 10 20 30 40 50 60 70 Figure 5-18. TG/DTA of colla gen sponge mineralized by the PI LP process containing 50 g/ml of poly-L-aspartic acid (Mw: 10,300 Da) for 8 and 6 days. Heating rate of TG/DTA was 5oC/minute and air flow was controlled to 100 cc/minute. The temperatures of typical exothermic reactions were indicat ed. NoteCompare area of exothermic reaction of ACP transition (at 550 oC) with the exothermic reaction in Figure 5-17 A. Area roughly corresponds to th e amount of ACP phase. A) 8 days B) 16 days 335 oC 550 oC 444 oC 335 oC
169 CHAPTER 6 CHARACTERISTICS OF AMORPHOUS CALI CUM PHOSPHAT E PRECURSOR FORMED BY THE PILP PROCESS Introduction As described in previous chapters, the PILP pro cess is considered as an effective pro cess to achieve intrafibrillar mineraliza tion of collagen fibrils. The PILP process forms an amorphous calcium phosphate precursor in which calcium an d phosphate ions are sequestered with a large amount of water molecules by anionic polymer, lead ing to the high fluidity of precursor . The fluidic character of the amorphous precursor a llows precursor droplets to be drawn into the interstices between tropocollagens, where the pr ecursor is solidified and crystallized into hydroxyapatite with preferred crystallographic orie ntation . As mentioned in chapter 2, the PILP (Polymer-Induced Liquid-Precursor) proc ess was initially developed by studying the calcium carbonate system . The addition of anionic polymer into th e mineralization solution induces hydrated amorphous calcium carbonate precursor, which reta rds the rapid crystallization of calcium carbonate. The fluidic and metastable precursor droplets are then agglomerated, as the reaction progresses, and depos ited on the glass slide as an amorphous film, which is later crystallized into crystalline film . The agglomeration of precursor droplets was also observed in the PILP process for the calcium phosphate system based on tris-buffer so lution containing 0.9 wt% of NaCl (as shown in chapter 4 and 5). When the lo w molecular weight polymer (5,500 Da) was used as a processdirecting agent, an amorphous f ilm formed by the agglomeration of precursors was observed at the bottom of mineralization vessel. However, in this case, the intrafibrillar mineralization of collagen could not be achieved we ll due to these solution precipitates. It is thought that amorphous precursor droplets formed by the low molecular weight poly-aspartic acid are agglomerated and precipitate before individual precursor droplets have a chance to adsorb to and
170 be infiltrated into the collagen scaffold. There c ould also be differences in the fluidity of the amorphous precursor, so that the amorphous calci um phosphate precursor (ACP) may be more similar to the ACP in the control reaction (withou t polymer), which is more like a gel and does not infiltrate the fibrils. Another possibility is that the agglomeration may cause less mobility and a larger volume of precursor may prohibit th e infiltration of precursor droplets into the collagen fibril via the narrow gap zone, leadi ng to a low degree of the intrafibrillar mineralization. In this case, it is assumed that small precursor droplets must enter from within the gap zone, since a large coating on the surface could presumably get pulled inside. Given that the samples that do mineralize well seem to do so from the inside out, it does appear that the droplets favor adsorbing strictly at the hole zones, and do not tend to make a mineral coating over the whole fibril surface. In contrast to the PILP process cont aining the low molecu lar weight polymer, agglomeration of amorphous precu rsor was not observed in the PILP process containing the high molecular weight poly-aspartic acids as a pr ocess-directing agent (M w: 10,300 and 32,200 Da). Moreover, a high degree of intrafibrillar mineraliza tion of collagen was achieved. Therefore, the stability of the amorphous precursor against a gglomeration is thought to be important for the intrafibrillar mineralization of collagen fibril. We thought that the behavior of this stable precursor, which is formed by the high molecular weight poly-aspartic acid, may be simi lar to that of charged particles in a stable colloidal suspension. When solid particles are di spersed in an aqueous solution, the solid surface develops an interfacial charge on the surface de pending on the pH of solution. Normally, at low pH, the surface is positively charged by protonati on, but at high pH, the surface is negatively charged by deprotonation or the ad sorption of hydroxyl ions . The interfacial charge on the
171 solid surface can induce the rearra ngement of local free ions in so lution and form a thin charged layer called a stern layer on the solid surface . As shown in Figure 6-1, when the solid surface is negatively charged, the negative charge can attract the free cations from the solution to form a compact layer (the stern layer) consisting of immobile c ounterions . After the stern layer, the density of counterion is gradually redu ced with the distance fr om the surface, forming the diffuse layer consisting of mobile counteri ons . Those two la yers are known as an electrical double layer . Ze ta-potential is the el ectrostatic potential at the shear plane, which is known to be close to the stern layer [ 227]. Zeta-potential is normally measured by the mobility of charged part icles under electric field . Wh en a charged particle with an electrical double layer migrates in response to electric field, the m obile ions in the diffuse layer, which are not tightly bound to the stern layer, un dergo dissipative movement. It creates a new interface called the shear plane between tightly bound ions and solution . The mobility of charged particles is proportional to the electrostatic potential at the shear plane. In the case of the PILP process, the negatively charged polymer with its sequestered calcium and phosphate ions can be assumed to be a charged core that behaves in a similar fashion to the charged particles in the colloid suspension. Therefore, this chapter will di scuss the electrostatic potential of PILP precursor particles obtained by the zeta-potential measurement. The zeta-potential of a charged particle is calcu lated from the electrok inetic phenomena of charged particles . Wh en an electric field is applied to the charged particle dispersed in the solution, the charged particle migrates with a certain velocity which is proportional to the strength of the electric field. The velocity of particle movement is normally measured by light scattering. The velocity per unit field strength, k nown as the electrophoretic mobility (), is used
172 for determining the zeta-potential ( ). When a particle with a char ge q is placed under an electric field (E), the particle is movi ng toward the oppositely-charged el ectrode with a force given by F = qE (6.1) In aqueous medium with a certain viscosity ( ), the Stokes law can be used for rewriting (6.1), and a new equation is given by F = 6 a (6.2) where a is the radius of the particle and is the velocity of the partic le. From the combination of equation (6.1) and (6.1), the electrophoretic mobility () can be written by = /E = q/(6 a) (6.3) In the case of a dilute electrolyte solution in which the thickness of electrical double layer ( -1) is sufficiently large enough to make a smaller than 0.1, th e zeta-potential ( ) is given by = q/(4 0a) (6.4) where and 0 are the dielectric constant of the liquid medium and the permittivity of vacuum, respectively. When the charge q in (6.4) is substituted by q in (6.3), zeta-potential is given by q = 6 a, and = 3 /(2 0) (6.5) However, in the case of PILP mineralization, because the ionic strength of the solution is relatively high (around 0.17M), the Smoluchowski approximation should be applied to equation (6.5) [230, 232-234]. After applying the Smoluc howski approximation, the equation for zetapotential under high ionic con centration can be obtained: = /( 0) (6.6) For the zeta-potential measurement for the amorphous calcium phosphate precursor, the Smoluchowski approximation was used.
173 Besides of the high ion concentration of th e mineralization solution, the size of the precursor droplets is another problem in the zeta-potential measurement. The size of the amorphous calcium phosphate precursor lies be tween 10 to 20 nm (the size distribution of precursor will be discussed in results and discussion section), making direct zeta-potential meter measurements difficult since the zeta-potential meter we have can hardly detect nano-scale particles. In this study, theref ore, functionalized polystyrene b eads were used for the substrate upon which the charged precursor was deposited, l eading to the variation of zeta-potential of polystyrene beads. Using this method, the char ged character of amorphous precursor can be verified indirectly. Materials and Methods Thermogravimetry Analysis (TGA) The collagen sponges were m ineralized at differe nt heights in a mineralization solution to investigate the agglomeration of precursors during the PILP process containing high molecular weight poly-aspartic acid. One collagen sponge was hung near the solu tion-air interface by a Teflon thread. A second collagen sponge was pl aced at the bottom of reaction vessel. The height difference between two collagen sponges was 7 cm. After 16 days of mineralization, both collagen sponges were washed with de-ionized water several times and pulverized after freezing with liquid nitrogen. After drying the pulveri zed collagen powder, each collagen sponge powder (10 20 mg) was individually transferred into a Pt pan and fired under an air condition using thermogravimetric analysis (TGA). The heating rate was 5oC/minute and analysis was done between 30oC and 1200oC. To compare the degree of mine ralization, the weight loss value at 600oC was used as a measure of mineral content b ecause organic substances within the collagen sponge powder are totally burned off by 600oC.
174 The size distribution analysis of poly-aspartic acid To m easure the size of polyaspartic acid which has seque stered calcium and phosphate ions, a size distribution analys is was done. Poly-L-aspartic acid (Mw: 32,200 Da) was added into four types of solutions to investigate the size of poly-aspartic acid under different conditions: tris-buffer solution, cal cium solution (4.5 mM of CaCl2), phosphate solution (2.1 mM of K2HPO4) and the mineralization solution (4.5 mM of CaCl2 and 2.1 mM of K2HPO4). The Nanotrac Particle Size Analyzer (Microtrac Inc., USA), which measures the size of particles by dynamic light scattering method, was used for the size distribution an alysis. For the analysis, the shape of polymer was assumed to be spherical and th e refractive index of proteins in the analyzer database was used for poly-aspartic acid. An op tical probe, which delivers the laser light to the suspension and scattered light back to the photodetector, was placed in solutions containing 100 g/ml of poly-aspartic acid and was kept stationary during measurement. Zeta-potential measurement As m entioned in the introduction section, be cause of a size problem of the precursor particles, polystyrene (PS) bead s (Polysicences Inc. USA, 0.1 m diameter) were used as a substrate for the indirect measur ement. The sample preparation condition is summarized in table 6-1. The final concentration of calcium and phosphate was adjusted to 4.5 mM and 2.1 mM, respectively, and the final volume was also adjust ed to 20 ml. Two types of polystyrene beads were used: polystyrene beads functionalized with amine groups, wh ich are positively charged at pH 7.4, and beads functionalized with carboxyl groups, which are negatively charged at that pH. The polystyrene beads were added to 20 ml of solution to achieve a concentration of 0.025 mg/ml. Before the measurement, the solution, de scribed in table 6-1, was incubated for 3 hours. The Brookhaven ZetaPlus (Brookhaven Instrume nts Corporation, USA) in PERC was used for the zeta-potential measurement. The polystyrene suspension was transferred into a colorless
175 cuvette with an equipped measuring kit and el ectrodes for applying the electric field. One hundred of measurements were performed to calcu late the average value of zeta-potential of the polystyrene beads with ions, poly-aspa rtic acid or amorphous precursors. Energy dispersive x-ray spect roscopy (EDS) Analysis To investigate ion com ponents on polystyrene be ad which is supposed to incorporate ions and precursor, EDS observations for polystyrene beads were performed. The polystyrene beads were collected by centrifugation (10,000 rpm, 30 mi nutes) and the collected beads were washed with de-ionized water and ethanol several times. The collected beads dried in air for 24 hours were mounted on an aluminum stub covered in d ouble-sided copper tape, and then sputter coated with amorphous carbon. The beads were examined using SEM to find an isolated bead on which EDS analysis could be conducted. The EDS an alysis was done by a 6400 JEOL SEM at 15 kV for the elemental analysis of a polystyrene bead incorporated with ions or precursors. Results and Discussion In the case of the PILP process in the calci um carbonate system, the size of the earliest formed droplets was measured by light scattering to be between 100 -250 nm, but they rapidly grew to micron sized dimensions over several minutes (in this particular vi al setup) . On the other hand, AFM analysis of CaCO3 films showed that the films can be composed of even smaller colloids. In this system, it is commonly observed that the amorphous calcium carbonate droplets settle by gravity down to the bottom of the reaction vessel due to the growth and agglomeration of precursor droplets. Therefore, when a silicon wafer with a hydrophilic pattern was vertically placed in the solution, the thickne ss of the calcium carbonate film on the pattern near the bottom became thicker as the particle size was larger . A similar experiment was done in the calcium phosphate system. As shown in Figure 6-2, one collagen sponge was hung near the air/solution inte rface, and the other was mineralized at the bottom (the solutions were
176 not stirred). If the precursor doe s in fact settle more rapidly by the growth and agglomeration of precursor droplets, the collagen s ponge at the bottom would be e xpected to show a higher degree of mineralization than th at of the collagen sponge mineralized at the top near the air/solution interface. However, there was no significant diff erence in the degree of mineralization between two collagen sponges (Figure 6-2). It is thought that, because the precursor droplets stay well dispersed without agglomeration in the mineraliz ation solution, both of the collagen sponges had an equal chance to adsorb an equal amount of precursor. Particle size analysis of the precursor was performed to investigate whether the precursor droplets are agglomerated or not. For the measur ement, the shape of polymer in the solutions was assumed to be spherical. As shown in Figure 6-3 A, the mean hydrodynamic volume diameter of poly-aspartic acid in tris-bu ffer solution (pH 7.4) was around 14 nm. The polyaspartic acid should be highly negatively charged at pH 7.4, leading to the fully stretched conformation of the chain . However, because tris-buffer solution contains about 0.9 wt% of sodium chloride, the polymer is thought to be s lightly collapsed due to the screening effect of sodium ions. When the poly-aspartic acid wa s dissolved in 4.5mM of calcium dichloride solution (Figure 6-3 B), the mean volume diameter of polymer decreased to around 8 nm. This is due to the strong screening e ffect of divalent calcium ions, which can make a bridge between charged functional groups (COO-) of poly-aspartic acid, leadi ng to shrinkage of the polymer coils. In the case of poly-aspartic acid in 2.1 mM of potassium phosphate solution (Figure 6-3 C), the polymer conformation seemed to be slightly expanded (around 14 15 nm) by the strong anions (PO4 3-) repelling the COOgroups in the polymer. When the poly-aspartic acid was added into the mineralization solution, whic h contained 4.5 mM of calcium and 2.1 mM of phosphate (Figure 6-3 D and also in Figure 6-4 A), the mean volume diameter of polymer, which
177 in this case is supposed to form the amorphous calcium phosphate precursor, was around 13 nm after 1 hour of reaction. Because the mean volume diameter of poly-aspartic acid in the mineralization solution is very similar to that of polymer in the tris-bu ffer solution, it could be argued that the formation of precursor was not completed yet after 1 hour of reaction. However, as shown in Figure 6-3 B, the in teraction between the poly-aspartic acid and calcium ion leading to the shrinkage of polymer is thought to be instantaneous. Ther efore, we thought that the polyaspartic acid can form the amorphous precu rsor immediately after addition. When the mineralization solution containing the poly-aspartic acid was kept for 3 days (Figure 6-4 B), the mean volume diameter of precursor slightly incr eased to 16 nm, and the distribution of precursor size was broadened. Moreover, neither the pr ecipitation of crystal line phase, nor the agglomeration of precursor was observed when the mineralization solution was kept for 1 month. Those results imply that the agglomeration of precursor is not significant when high molecular weight poly-aspartic acid is used for the PILP process. We thought that the precursor might be dispersed without agglomeration due to the in terfacial charge formed by the calcium and phosphate sequestered by the negatively charged pol y-aspartic acid, and therefore turned to zetapotential measurements. To investigate the interfacial charge of the precursor, the zeta-potential was measured by the indirect method using polystyrene beads f unctionalized with amine (PS-amine) or carboxyl groups (PS-COOH), because the size of the precursor particles alone is too small to be detected by the zeta-potential meter. The results of th e zeta-potential measurement are summarized in table 6-2. In the case of polystyrene beads functionalized with carboxyl groups (PS-COOH), the surface of the PS-COOH bead was negatively ch arged (sample 1, about -30 mV) even though the tris-buffer solution contained sodium ions. This implies that the monovalent ion (Na+) is not
178 effective in forming a positive compact stern layer on the negatively charged polystyrene bead surface. When the PS-COOH beads were adde d to 4.5 mM of calcium solution, the zetapotential of PS-COOH bead was -10.72 1.29 mV. Normally, the divalent ion (Ca2+) can form a very compact stern layer on a negatively charge d surface, leading to a positive zeta-potential. However, in this case, because the amount of cal cium ion (which was based on our PILP reaction conditions) was insufficient to fully cover th e entire surface of the PS-COOH bead, the zetapotential of the PS-COOH bead was still nega tive, but the negative value was dramatically reduced by the adsorption of calcium ions. When poly-aspartic acid was added to the tris-buffer solution (sample 2, about -35 mV), the zeta-pot ential of the PS-COOH bead did not change much because of the repulsion between the ne gatively charged bead and poly-aspartic acid, preventing the adsorption of poly-aspartic acid on the beads. However, when the PS-COOH beads were dispersed in the calcium solution co ntaining poly-aspartic acid, the zeta potential drops significantly (sample 3, about -20 mV), as compared to that of PS-COOH beads in calcium solution (about -11 mV). In this case, it appears that the poly-aspa rtic acid is incorporated with the PS-COOH beads. In the case of sample 4, the poly-aspartic acid was added to the calcium solution to induce the polymer-calcium complex prior to the addition of PS-COOH bead. We expected a zeta-potential varia tion from sample 4 due to the formation of polymer-calcium ion complex, which may prevent the incorporation between PS-COOH and calcium ions. However, there was no significant difference between sample 3 (-19.87 1.45 mV) and 4 (-19.15 1.41 mV) indicating very little differe nce in interfacial conditions betw een the two samples. An EDS analysis of both PS-COOH beads (Sample 3 and 4) shows calcium ion incorporation, as shown in Figure 6-5 A and B. When the zeta-potenti al of sample 3 (about -20 mV) and 4 (about -19 mV) are compared to that of the PS-COOH beads in calcium so lution alone (about -11 mV), the
179 additional negative readings can be attributed to negatively charge d poly-aspartic acid that is incorporated onto the PS-COOH beads by the formation of calcium ion bridges. In the case of sample 5 (-18 mV), the order of species addition in this sample preparation was designed to form the amorphous calcium ph osphate precursor before the addition of PSCOOH beads. We speculated that the negatively charged poly-asp artic acid, which is added to the calcium solution before mixing in the phos phate solution, may fo rm a calcium-polymer complex, to which phosphate ions can sequentially be attracted to positively charged complex. Therefore, the interfacial charge of precursor is thought to be slightly negative even though the total net charge of precursor ma y be neutral. Electrostatic repulsion from a negatively charged surface of the precursor is thought to prevent the agglomeration of precursor. On the other hand, there should be free calcium ions in the mineraliz ation solution after the formation of precursor. When it is assumed that two carboxyl groups can attr act one calcium ion (due to charge balance), only 2.35 mM of calcium ions are consumed by th e formation of precursor, leaving 2.15 mM of free calcium ions that could be incorporated with the PS-COOH beads, leading to partially positively charged PS-COOH beads. Therefore, it is thought that the nega tive charge character of the PS-COOH bead in sample 5 is originated by the adsorption of pr ecursor on the partially positively charged PS-COOH beads created by the adso rption of free calcium ions. In the case of sample 6, in which the calcium ions were ex pected to adsorb onto the PS-COOH beads before the formation of precursor, the negative zeta-p otential value is slight ly reduced (-15 mV) in comparison to the sample 5 (-18 mV). It is thou ght that because the calcium ions are initially incorporated with the PS-COOH beads, a smaller amount of calcium ions are available to be involved in the formation of precursor, and the PS-COOH beads may have more positive calcium ions in comparison to sample 5. Howe ver, because the difference of zeta-potential
180 values between sample 5 and 6 is not significant, it is hard to argue that the interf acial conditions are significantly different between samples. In the case of EDS analysis of sample 5 and 6 (Figure 6-5 C and D), EDS results showed th at the surface of the PS-COOH beads were incorporated with calcium and phosphate ions wh ich are thought to have originated from the amorphous calcium phosphate precursor. Wh en the polymer was not present in the mineralization solution (sample 7), the zeta-poten tial value (-7 mV) was far less negative in comparison to sample 5 (-18 mV) and 6 (-15 mV). In this case, EDS analysis showed that most of the ions in the mineralization solution were incorporated with the PS-COOH beads (Figure 65 E). However, it is not fully understood why those results were obtained, and more investigation may be need ed to understand it. The possible interfacial conditi ons of the PS-COOH beads in various solutions, which are deduced from the zeta-potential measurement, ar e illustrated in Figure 6-6. When the PS-COOH beads are dispersed in tris-buffer solution (sam ple 1), the surface of the PS-COOH beads were negatively charged due to the de protonation of carboxyl groups. Sodium ions in tris-buffer solution are thought not to fully neutralize the surface charge of the PS-COOH beads. In the case of the PS-COOH beads in calcium solution (PS+Ca), calcium ions adsorb on the negatively charged PS-COOH beads leading to a decrease in the negative value of the surface. When the poly-aspartic acid was added to the calcium solu tion (sample 3), the calcium ions adsorb on the PS-COOH beads, and the poly-aspartic acid inco rporates with calcium ions on the surface of PSCOOH bead. In the case of sample 5 in whic h the amorphous precursor is formed before adding the PS-COOH beads, the negatively charged amorphous precursor adsorbs on the positive regions of the PS-COOH beads that are created by the adsorption of the free calcium ions. When the PS-COOH beads are added before the formati on of amorphous precursor (sample 6), most of
181 calcium ions adsorb on the PS-COOH beads surface initially. However, when the poly-aspartic acid and phosphate ion are added to the solu tion containing the PS-COOH beads and calcium ions, some of calcium ions are detached from the surface and those free calcium ions involve in the formation of amorphous precursor which la ter incorporates with the PS-COOH bead via calcium ion bridging. In the case of polystyrene beads functionali zed with amine groups (PS-amine bead), the beads should be positively charged via protonation of the amine group (NH3 +) at pH 7.4. When the PS-amine beads were dispersed in tris-buffe r solution (sample 1), th e zeta-potential of PSamine bead was around -14 mV, as summarized in table 6-2. Even though the tris-buffer solution only contained sodium chloride, there was a charge conversion of the PS-amine bead surface from positive to negative. As mentioned prev iously, it is hard to covert the positive zetapotential to negative zeta-potential with monovalent ion (Cl-). However, when the PS-amine beads were examined with EDS (Figure 6-7), ever y sample showed the chloride ion incorporated on polystyrene bead, which was not observed from PS-COOH bead (Figure 6-5) Therefore, it is thought that the surface of PS-amine bead is sligh tly positive due to the neutral pH of solution (pH 7.4), and the adsorption of chloride ions on the positive domain may produce the negative value of zeta-potential. This positive characte r of PS-amine bead was highlighted with the negatively charged poly-aspartic acid. When the poly-aspartic acid was added to tris-buffer solution (sample 2), the positively charged PS-amine bead attracted the negatively charged polyaspartic acid, leading to a highly negative zeta-potential (about -31 mV). This indicates that the surface of PS-amine has dominant positively ch arged amine groups and positively charged amine groups can attract the negatively charged poly-aspartic acid which may have more affinity to positive amine groups in comparison to chloride ions. In the case of PS-amine beads in the
182 solution containing poly-aspartic acid and 4.5 mM of calcium ions (sample 3 and 4), the zetapotential value was about -18 mV for both sample s. Because the calcium ions in the solution make a complex with the poly-aspartic acid ad sorbed on the PS-amine bead, the negative zetapotential value was reduced in comparison to that of sample 2 (about -31mV). As shown in Figure 6-7 A and B, when the PS-amine beads of sample 3 and 4 were examined by EDS, the surface of PS-amine bead was incorporated with calcium ions via the negatively charged polyaspartic acid. A small amount of chloride i ons, which were thought to adsorb on the positively charged PS-amine bead surface, were also detected by EDS. In the case of sample 5 in which the amorphous precursor was formed before the addition of PS-amine beads, the zeta-potential was more ne gative (about -19mV) than that of sample 1 (14 mV), and calcium and phosphate ions were de tected from the PS-amine bead surface by EDS (Figure 6-7 C). We thought that because of the adsorption of negatively charged amorphous precursor on the PS-amine bead, the negative value of zeta-potential is sli ghtly increased. When the PS-amine beads were dispersed in the solution before the amorphous precursor was formed (sample 6), the PS-amine bead seemed to interact with chloride ions in itially. As shown in Figure 6-7 D, the surface contains a large amount of chloride ion which is thought to adsorb on the positively charged surface. However, when the intensity of the calcium and phosphate ion peaks in sample 6 (Figure 6-7 D) was compared to the intensity of the calcium and phosphate peaks in sample 5 (Figure 6-7 C), the amounts of calcium and phosphate ions detected from both samples were similar. Therefore, even though sample 6 contained more chloride ions, which may increase the negative value of zeta potential to about -20 mV, the interfacial charge of both samples (sample 5 and 6) seems to be similar. In the case of sample 7, which contained no poly-
183 aspartic acid, the results of zeta-potential measur ement and EDS analysis were almost identical to those of PS-COOH bead. More inves tigation may be needed for this system. Schematic diagrams illustrating possible in terfacial conditions of the PS-amine beads in various solutions are shown in Figure 6-6. When the PS-amine beads are dispersed in the trisbuffer solution containing sodium chloride (sam ple 1), the positively ch arged surface attracts the chloride ions creating a negatively charged surfa ce. In the case of the negatively charged polyaspartic acid, it adsorbs on the positively charged PS-amine bead, which was occupied by chloride ions in the tris-buffer solution (sampl e 1) and induces the high negative zeta-potential on the PS-amine bead (sample 2). In the case of sample 3 containing poly-aspartic acid and 4.5 mM of calcium ions, the calcium ions prefer entially combine with the poly-aspartic acid adsorbed on the PS-amine bead, lowering the ne gative zeta-potential of the PS-amine bead. When amorphous precursor is formed in the solu tion before adding the PS-amine bead (sample 5), the negatively charged precursors adsorb on the positively charged PS-amine beads. However, when the PS-amine beads are added before the formation of amorphous precursor (sample 6), large amounts of chloride ions inco rporate with the beads initially and then, the negatively charged precursor adsorb s onto the beads, either directly or via chloride ion bridging. Conclusions When high molecular w eight poly-L-aspartic aci d was used as a process-directing agent for the PILP process in the calcium phosphate system, it induces stable amorphous calcium phosphate precursor droplets, whic h are not agglomerated for up to 1 month. When the collagen sponges were mineralized at different heights in the reaction vessel, ther e was no difference in the degree of mineralization between the collag en sponges mineralized at the bottom versus those near the air/solution interface. Moreove r, when the size of precursor droplets was measured by the particle size analyzer, neither si gnificant growth of precursor nor agglomeration
184 of precursor was observed, even after 3 days of reaction. We thought that the charged interface of the precursor may keep the precursor stable and dispersed in the mineralization solution. Zeta-potential measurements were performed to i nvestigate the charge character of the precursor interface. The differences between samples, wh ich were designed to deduce the electrostatic characteristic of the precursor indirectly by ad sorption to charged polystyrene beads, were not significant enough to verify the exac t interfacial charge of precur sor. However, the precursor phase was very stable and well-dispersed for long times and seemed to act in a similar fashion to the charged particles in the zeta-potential meas urement. Therefore, we suggest that the amorphous calcium phosphate precursor may have a negatively charged interface, even though the total charge of precursor, which is formed by the sequent ial adsorption of calcium and phosphate ions on the negativel y charged poly-aspartic acid core, may maintain charge neutrality.
185 Figure 6-1. Electrical double layer formed on a charged solid surface. ( 0 = electrical potential on solid surface, s = electrical potentia l of stern layer, = zeta-potential; electrical potential of shear plane) Diagram is re -illustrated from the images in [227, 229]. + + + + + + + + + + + + + + 0 s Stern layer Diffuse layer
186 Figure 6-2. Thermogravimetry (TG) analysis of collagen sponges which were mineralized by the PILP process at different heights in th e reaction vessel. Collagen sponges were mineralized with 50 g/ml of poly-L-aspa rtic acid for 16 days and samples were positioned at the bottom of vessel (red cu rve) and 7 cm above the bottom (black curve). Pictures of collagen mineralized at different heights are inserted to the right. The sample is half of a Collagen Matrix sponge of diameter 1.2 cm. The collagen sponge is translucent and floppy until it minera lizes, which can be seen by the white patches. 20040060080010001200 40 60 80 100 7 cm from bottom bottom 73 % 76 % Weight Loss(%)Temperature (oC)
187 46810121416182022 0 5 10 15 20 25 30 35 Cumulative percent (%) Bar graph (%)Size (nm)0 20 40 60 80 100 MV(nm): 13.64 46810121416182022 0 5 10 15 MV(nm): 8.41Cumulative percent (%) Bar graph (%)Size (nm)0 20 40 60 80 100 46810121416182022 0 5 10 15 20 25 30 35 40 Cumulative percent (%) Bar graph (%)Size (nm)0 20 40 60 80 100 MV(nm): 14.80 46810121416182022242628 0 5 10 15 20 25 30 Cumulative percent (%) Bar graph (%)Size (nm)0 20 40 60 80 100 MV(nm): 13.38 Figure 6-3. Particle size analysis of poly-L-aspartic acid in va rious solutions. The shape of poly-L-aspartic acid is assumed to be s pherical (a random coil) for analysis. The concentration of poly-L-aspartic acid (Mw: 32,200 Da) was 1 mg/ml and polymer was dissolved in A) tris-buffer, B) tris-buffer containing 4.5 mM of calcium chloride or C) 2.1 mM of potassium phosphate, D) tris-buffer containing 4.5 mM of calcium chloride and 2.1 mM of potassium phosphate A) B) C) D)
188 46810121416182022242628 0 5 10 15 20 25 30 Cumulative percent (%) Bar graph (%)Size (nm)0 20 40 60 80 100 MV(nm): 13.38 46810121416182022242628 0 5 10 15 20 25 MV(nm): 16.32Cumulative percent (%) Bar graph (%)Size (nm)0 20 40 60 80 100 Figure 6-4. Particle size analysis of poly-L-aspartic acid (Mw: 32,200 Da) in the mineralization solution of the PILP process for 1 hour (A ) and 3 days (B). Mineralization solution contained 4.5 mM of calcium chloride, 2.1 mM of potassium phosphate and 1 mg/ml of poly-L-aspartic acid. A) B)
189 Table 6-1. Sample preparation fo r zeta-potential analysis of functionalized polystyrene beads with poly-L-aspartic acid or precurs ors formed by the PILP process. Sample number Sample preparation (Order of addition is from left to right) 1 Tris-buffer PS 2 Tirs-buffer pAsp PS 3 Tris-buffer PS pAsp incubation Ca 4 Ca pAsp Tris-buffer incubation PS 5 Ca pAsp P incubation PS 6 Ca PS pAsp P 7 Ca PS P PS: 0.025 mg/ml of polystyrene bead, Ca : 9 mM of calcium chloride in tris-buffer, P: 4.2 mM of potassium phosphate in tris-buffer, pAsp: 100 g/ml of poly-L-aspartic acid (Mw: 10,300 Da), incubation : 1 hour Final concentration of Ca and P were adjusted to 4.5 mM and 2.1 mM with/without th e addition of tris-buffer. For tris-buffer solution, 6.4 g of tris-H Cl, 1.1 g of tris-base and 8.77 g of NaCl were dissolved in 1000ml of D.I. water. Polystyrene beads reacted with solution for about 3 hours.
190 Table 6-2. Zeta-potential (mV) of polystyrene beads functionaliz ed with carboxyl group (COOH) or amine group (NH2), which arenegatively (COO-) or positively (NH3 +) charged at pH 7.4. Zeta potential average and standard deviation values were obtained from one hundred measurements. Zeta-potential of PS-COOH in 4.5 mM of calcium solution was -10.72 1.29 mV Sample number Zeta-potential of PS-COOH Zeta-potential of PS-NH2 1 -29.83 1.64 -14.19 0.67 2 -34.75 3.94 -30.74 1.49 3 -19.87 1.45 -17.72 1.19 4 -19.15 1.41 -18.01 1.87 5 -17.89 0.81 -18.93 0.91 6 -14.96 1.09 -19.58 1.35 7 -7.07 0.82 -8.19 1.07
191 Figure 6-5. The EDS analysis of COOH functionalized polystyrene beads reacted with the various solutions indicated at table 6-1. S pot EDS analysis was done for an isolated polystyrene bead (white circle in SEM micrograph) during SEM examination. Scale bar is 5 m. A) sample 3 TrisPS Pasp Ca D) sample 6 Ca PS Pasp P E) sample 7 Ca PS P C) sample 5 Ca Pasp P PS B) sample 4 Ca Pasp Tris PS S Ca Ca P Ca S Ca P Cl Na
192 0 -5 -10 -15 -20 -25 -30 -35 -40 Ca+PS +Pasp+P 6 Ca+Pasp +P+PS 5 PS+Pasp+Ca 3 PS+Ca PS 1Zeta-potential (mV) of Polystyrene beads Figure 6-6. Interfacial condition of polystyrene beads functi onalized with carboxyl (-COOH) groups. The sample numbers indicated on the x-axis correspond to the sample numbers in tables 6-1 and 6-2. PS Ca2+ Ca2+ Ca2+ Ca2+ PS Na+ ClPS Ca2+Ca2+ Ca2+ Ca2+ PS Ca2+Ca2+Ca2+ Ca2+Ca2+Ca2+ PS Ca2 + Ca2+ Ca2+ Ca2+ Ca2+ Ca2+ Ca2+ Ca2+Ca2+ Ca2+ Na+ Ca2+Ca2+ Ca2+ Ca2+ Ca2+ Ca2+Ca2+Ca2+Ca2+ Ca2+Ca2+Ca2+Pol y -as p artic acid Negatively charged precursors
193 Figure 6-7. The EDS analysis of PSNH2 functionalized polystyrene bead reacting with various solutions indicated in table 6-1. Spot EDS analysis was done for polystyrene beads during SEM examination. E) sample 7 Ca PS P Ca Cl Ca Cl Ca P Cl Cl Ca P A) sample 3 TrisPS Pasp Ca B) sample 4 Ca Pasp Tris PS C) sample 5 Ca Pasp P PS D) sample 6 Ca PS Pasp P
194 0 -5 -10 -15 -20 -25 -30 -35 -40 Ca+PS +Pasp+P 6 Ca+Pasp +P+PS 5 PS+Pasp+Ca 3 PS+Pasp 2 PS 1Zeta-potential (mV) of Polystyrene beads Figure 6-8. Interfacial condition of polystyrene beads which we re functionalized with amine (NH2) groups. The sample numbers indicated on the x-axis correspond to the sample numbers in tables 6-1 and 6-2. PS + + + + + + + ClClClClClClClNa+ Na+ PS + + + + + + + + + ClClClClClClNa+ Na+ PS + + + + + + + + + Cl-Cl-Cl-ClNa+ Ca2+Ca2+Ca2+Ca2+Ca2+Ca2+ Ca2+ PS + + + + + + + + + Cl-Cl-ClPS + + + + + + + + + ClCl-ClClClClClCa2+ Ca2+ Ca2+ Ca2+ Pol y -as p artic acid Ne g ativel y char g ed p recursorCa2+
195 CHAPTER 7 CONCLUSIONS The objective of the work pres ented within this dissertati on is to develop a bone-like collagen/hydroxyapatite com posite via a biomim etic approach called the PILP (PolymerInduced-Liquid-Precursor) pr ocess. Using the process-directi ng agent, poly-aspartic acid, which induces the amorphous calcium phosphate pr ecursors and may mimic the function of noncollagenous proteins in bone formation, we have successfully mimicked the nanostructure of secondary bone. Specifically, we have synthe sized a collagen/hydroxyapatite composite in which the hydroxyapatite platelets are embedde d in the interstices between tropocollagen molecules, and the  direction of hydroxyapatite is oriented parallel to the long axis of the collagen fibril. In addition, we also carefully suggest that the biomimetic PILP process may aid in the understanding of the intrafibrillar mine ralization of natural bone. Bone is the most frequently replaced huma n organ, and the demand for the development of synthetic bone grafts has been increasing over th e last several decades. While the ultimate goal of developing synthetic bone grafts is to fa bricate large-scale bone substitutes which have structural and compositional similarity to natural bone, this is difficult in practice due to the lack of dense and well-organized collagen scaffolds. However, mimicking the nanostructure of secondary bone should be considered as a prerequisi te step, in this stage, to the development of the ultimate synthetic bone grafts because the interpenetrating structur e of the mineralized collagen fibril, which can be achieved by intrafib rillar mineralization, is thought to cause the unique physical and chemical properties of secondary bone. Mineralized collagen fibrils are composed of collagen fibrils, in which tropocollagen is aligned with a quarter-staggered arrangement leaving a gap zone and grooves, and hydroxyapatite platelets, which ar e embedded in the interstices between tropocollagen molecules
196 with uniaxial orientati on. We tried to mimic this nanos tructure using an amorphous calcium phosphate precursor, which was induced by negatively charged poly-L-as partic acid. The strategy of our process is to form a fluidic amorphous calcium phosphate precursor which can infiltrate the interstices of the collagen fibrils via the gap zone by capillary forces. The poly-Laspartic acid is thought to successfully sequest er the calcium and phosphate ions, along with water molecules, forming the hydrated amorphous calcium phosphate precursor. Even though the amorphous precursor is thought to be ther modynamically unstable in the calcium phosphate system, the precursor phase produced by high mo lecular weight poly-L-aspartic acid (Mw: 10,300 or 32,200 Da) was surprisingly very stable in solution, and the crystallization of the precursor occurred only on the collagen substrate without any agglomeration of the precursor in solution, or any precipitation of crystalline or non-crystalline pha ses. The crystals originating from the amorphous calcium phosphate precursor we re examined by x-ray diffraction analysis of the mineralized collagen scaffolds. The mineral phase was determined to be hydroxyapatite when the collagen was mineralized using the hi gher molecular weights of poly-L-aspartic acid (Mw: 10,300 or 32,200 Da). TEM analysis of mi neralized collagen fibr ils provided visible evidence of oriented hydroxyapat ite platelets embedded in the collagen fibrils. The oriented hydroxyapatite along the  direction of collage n fibrils created the (002) and (004) arcs generated by the slightly misori ented hydroxyapatite platelets. In addition to the SAED patterns of the mineralized collagen fibril s, the dark field image of the mineralized collagen fibrils also confirmed that oriented hydroxyapat ite platelets are embedded in the fibrils. When the dark field image of collagen fibrils was constructed by the electron beam diffracted from the (002) planes of hydroxyapatite, the side view of hydroxyapatite platelets was projected, creating the needleshape spots aligned along the  direction of th e collagen fibril.
197 The micro and nano-scale observations of mine ralized collagen fibrils done using x-ray diffraction, SEM and TEM analysis enable us to confidently argue that the intrafibrillar mineralization of collagen fibrils can be successfu lly achieved by the PILP process, and that the physical and chemical characteristics of mineralized collagen are e xpected to be similar to those of natural bone due to the struct ural and compositional similarities of mineralized collagen to the nanostructure of bone. We discovered that the PILP process also provides the ability to fabricate high mineral loading composites. As mentioned in Chapter 2, bone is composed of 60-70 wt% inorganic mineral phase, 20-30 wt % organic substances and 10 % water. TG analysis showed that the mineralized collagen scaffold contained about 70 wt% of hydroxyapa tite after 16 days of mineralization. High molecular we ight poly-L-aspartic acid produ ced scaffolds with the highest degree of mineralization. For example, coll agen mineralized with 10,300 Da poly-L-aspartic acid contained 60 wt% of hydroxyapatite after 16 days of mineraliza tion, while the collagen scaffold mineralized with 32,200 Da poly-L-aspa rtic acid contained 58 wt% of hydroxyapatite after only 8 days of mineralizati on, and the mineral content increa sed to 74 wt% after 16 days of mineralization. The PILP process is also expected to enable the fabrication of large-scale bulk composites suitable for bone grafts. Using a porous collagen sponge with 1 cm diameter and 3 mm thickness, we fabricated a mineralized collagen sponge containing about 70 wt% of hydroxyapatite after 16 days of mineralization. However, when the mine ralization solution was replaced after 8 days of mineralization, the mineral content was increase d to 84 wt% after 16 days of mineralization. Therefore, we expect that a continuous supply of amorphous ca lcium phosphate precursor via a constant composition technique, or the freque nt replacement of solution, may enable the mineralization of a large-scale collage scaffold.
198 For the development of synthetic bone grafts, th e ability of the PILP process to mineralize a collagen scaffold with densely-packed collag en fibers will be required if load-bearing properties are to be achieved. Un fortunately, this type of synt hetic scaffold was difficult to obtain. Therefore, we used bovine achilles tendon as a scaffold, even though the bovine tendon is not naturally mineralized and apparently co ntains calcification inhibitors, to examine the ability of the PILP process to mineralize dense collagen scaffolds. Poly-aspartic acid with various molecular weights was used to investigat e the influence of molecular weight of polymer on the degree of intrafibrillar mineralization of bovine tendon. When low molecular weight poly-aspartic acid (Mw: 5,500 or 10,300 Da) was used for the mineralization, hydroxyapatite peaks were not observed by x-ray diffraction anal ysis. However, the intensity of the peak created by the oriented collagen fibrils was incr eased after the mineraliz ation process. We thought that chelation between certain functional groups of the amino acids and the amorphous precursor might enhance the intensity of the collagen peak. When bovine tendon mineralized with low molecular weight polymer was observed using TEM, the collagen fibrils showed the typical 67 nm periodic banding pa tterns without any staining process, but the SAED patterns of the fibrils indicated that collagen fibrils had no crystalline phase. We thought that the banding patterns from the unstained fibrils may be created by the amorphous precursor incorporating with colla gen fibrils. The banding pattern could be enhanced via z-contrast, because the amorphous mineral precursor is more electron dense than the collagen phase. EDS analysis in the SEM of an individual collagen fibril also confirmed the presence of calcium and phosphate ions in fibrils mineralized using the same conditions as those used in the TEM experiment. However, the collagen fibrils containing the amorphous precursors did not crystallize into hydroxyapatite. We suspect that the calcification inhibitors in bovine
199 tendon may retard or inhibit the crystallizat ion of the amorphous precursor. When the calcification inhibitors were removed by EDTA tr eatment, intrafibrillar mineralization of bovine tendon was achieved by the intermediate molecular weight poly-aspartic acid (Mw: 10,300 Da). The SAED patterns of mineralized collagen fibrils that were pretreated with EDTA showed the typical diffraction patterns of oriented hydroxya patite ((002) arcs), indicating successful intrafibrillar mineraliza tion of bovine tendon. In the case of high molecular wei ght poly-aspartic acid (Mw: 32,200 Da), the bovine tendon, whic h was not pre-treated with EDTA, was intrafibrillarly mineraliz ed with the PILP process. The mineral content of bovine tendon mineralized with high molecular weight polymer was around 24 wt% after 8 days of mineralization, even though the bovine tendon contai ned the calcification inhibitors. On the other hand, the bovine tendon mineralized with low molecular weight polymer (Mw: 10,300 Da) contained only 6 wt% of hydroxyapatite. The hydroxyapatite content of mineralized bovine tendon that was pre-treated with EDTA was increas ed to 12 wt%. This sample also had the SAED pattern of hydroxyapatite. Th ose results suggest that the cal cification inhibitors in bovine tendon prevent the infiltration and crystallization of the amorphous precursor, and that intrafibrillar mineralization of densely-packed collagen scaffolds is enhanced when the high molecular weight poly-aspartic acid is used. Although the bovine tendon has densely packed collagen fibers, the mineral content of bovine tendon was only half of the mineral content of the loosely-packed synthetic collagen sponge mineralized under the same conditions. The difference of collagen fiber densities between bovine tendon and collagen sponge and th e sheath surrounding collagen fibers in tendon may cause the difference of mineral contents. In either case, the PILP process is required to fabricate practical synthetic bone grafts. R ecent research by another member of the group
200 showed that the degree and rate of intrafibrillar mineralization could be enhanced by agitating the solution during the mineralization. We believe that the agitation of solution may increase the number of contacts between disper sed precursor droplets and the collagen scaffold. Therefore, a new experimental setup which can enhance the contact between scaffold and precursors should be considered for the development of dense bone substitutes. The degree of intrafibrillar mineralization of collagen fibrils via the PILP process depends on the molecular weight of poly-aspartic acid. In the case of high molecular weight poly-aspartic acid, the precursor was stable and well-dispersed without precipita tion or agglomeration. On the other hand, with low molecular weight poly-as partic acid, the precip itation of an amorphous phase caused by the agglomeration of precurs or droplets was observed. Even though the crystallization of the amorphous phase was inhi bited by the addition of low molecular weight poly-aspartic acid, little to no intr afibrillar mineralization of the collagen fibrils was achieved (some amorphous phase may be present, as indi cated by the enhanced collagen peak). We hypothesize that the rapid precip itation or agglomeration of pr ecursor droplets may hinder the infiltration of the precursor vi a the narrow gap zones in the co llagen fibril. Therefore, the formation of a stable precurso r phase may be required to accomplish the high degree of intrafibrillar minerali zation of collagen. Although the re ason why a high molecular weight polymer produces a more stable amorphous precurso r than low molecular weight polymer is not fully understood, we suspect that the precursors formed by the high molecular weight polymer may act as charged particles in a stable colloidal system. On the other hand, in the case of calcium carbonate system, the intrafibrillar minera lization of collagen sponge was achieved even with the rapid preci pitation of precursor. It is thought that the difference in the physical and electrostatic characteristics of precursor in calcium carbonate syst em may allow the intrafibrillar
201 mineralization of collagen sponge. However, more investigation will be needed to verify the difference between precursors in calcium phosphate system and calcium carbona te system. To determine the veracity of our stable colloi dal system hypothesis, ze ta-potential analysis of the precursors via an indirect measurement us ing charged polystyrene be ads as a substrate was done. This analysis showed that the beads ( PS-amine and PS-COOH) were incorporating with the charged precursor, and EDS analysis confir med that the calcium and phosphate ions were adsorbed on the polystyrene bead s. Although the differences of zeta-potential values between samples, which were designed to verify the el ectrostatic character of precursor, were not significant, we could deduce the interfacial conditions of the pr ecursors from the zeta-potential data with some speculation. We suspect that because the high mo lecular weight polymer can be more negatively charged than the low molecu lar weight polymer, the high molecular weight polymer may more efficiently seque ster the calcium ions. Phospha te ions are then attracted by the sequestered calcium ions. This sequential attraction of cations and anions may form the amorphous precursor whose interfa ce is slightly negatively charge d, and the charged interface of precursors may cause repulsion between precursor drop lets, leading to the stable colloidal system. However, more investigation us ing an alternative method is re quired to verify the underlying cause of the stable precursor formati on by high molecular weight polymer. As mentioned in the previous paragraphs, the in trafibrillar mineraliza tion of collagen fibril is successfully achieved by the PILP process. The addition of acidic polymer can induce the fluidic amorphous calcium phosphate precursors, which is thought to be drawn into the narrow interstices between tropocollagen molecules by cap illary force. Then, the amorphous precursors drawn into collagen fibril solidif y and crystallize into hydroxyapatite crystals through metastable phases such as amorphous calcium phosphate solid, and maybe ev en octacalcium phosphate or
202 brushite. Although the capillary action of i ndividual fluidic precurs ors was not examined, confocal microscopy observations showed that the fluidic precursor, which incorporated the labeled poly-aspartic acid, was infiltrated into the densely-packed collagen scaffold (turkey tendon) more than 6 or 7 times further than the labeled polymer alone (via diffusive transport). Therefore, we suspect that capillary action may force the fluidic precursor, which is distinguished from the soluti on by a phase boundary, to migrate into turkey tendon over a long range. This issue of the presence of a phase bou ndary is less resolved here than the in the calcium carbonate system, where the precursor dr oplets become large enough to visually observe them as being a distinct phase. In the calci um phosphate system, the droplets are extremely small, and apparently only consist of one or tw o polymer chains, and they dont settle or collect in the vessel or by centrifugation; therefore, the on ly evidence that there is a phase boundary for the calcium phosphate precursor is indirect, and that is the (potential) capill ary action mechanism. In the case of thermal analysis (TG/DTA) of mineralized collagen, we could observe the decrease in thermal stability of collagen molecule as intrafibrillar mineralization progressed. As the intrafibrillar mineralization progressed, the thermal stability of collagen was dramatically decreased, which was observed in human bone. We thought that the decrease of thermal stability of collagen may be used as a simple characte rization technique for determining whether the mineral is intrafibrillar or interfibrillar (surf ace mineral). The presence of the metastable phase in the collagen scaffold mineralized via the PI LP process was also examined by TG/DTA. The DTA analysis of mineralized co llagen showed that th e collagen sponge at the early stage of mineralization contained amorphous calcium phosphate solid. An exothermic reaction was observed at around 540-550oC corresponding to the transition of amorphous calcium phosphate to hydroxyapatite without si gnificant weight loss.
203 Now, we can confidently state that the PILP pr ocess is a reliable method for the fabrication of bone-like composites at the nano-scale. Howeve r, more research may be required to optimize the PILP process for practical bone-like composites. First, more investig ation into the innate characteristics of the amorphous precursor, su ch as ionic concentration, viscosity, charge character and molecular structure of the precursor s, are required to optimize the PILP process for the mineralization of collagen. In addition, those types of st udies will help to expand the applications of the PILP process to various area s, including electronic, optical and structural materials, as well as biomedical materials. Second, various experiment al setups should be explored to scale up the PILP process to make bone substitu tes which are large enough for implantation, and commercially viable. A lthough several ideas we re introduced in the dissertation, a systematic set of experiments is re quired to develop a practical setup that can be utilized by the industrial area. The immobilization of growth fact or on the mineralized collagen is also needed to enhance the osteoinductivity of the bone-like composites. We hope that in the near future, the PILP process is highlighted as a new and remarkable technique for the forth generation bone grafts. Results of our research successfully demons trate that a collagen scaffold can be mineralized by the PILP process, and that the na nostructure of collagen fibrils mineralized by the PILP process is almost identical to that of na tural bone. We are also confident that the PILP process may provide a new way to develop synthetic bone-like collagen/hydroxyapatite composites whose physical and chemical properties are very close to the properties of natural bone. Moreover, we hope that our successful in trafibrillar mine ralization of co llagen via the precursor mechanism will revive discussions of bone mineralization via the amorphous calcium phosphate phase hypothesis.
204 LIST OF REFERENCES  C. J. Da mien, J. R. Parsons, Journa l of Applied Biomaterials 2 (1991) 187-208.  S. R. R. Murugan, Bioac tive Nanomaterials in Bone Grafting and Tissue Engineering, in: H. S. Nalwa, Handbook of nanostructured biomaterials and their applications in nanobiotechnology, vol 2, Americal Scientific Publishers, California, 2005, p. 141-168.  R. R. Betz, Orthope dics 25 (2002) S561-S570.  R. Murugan, S. Ramakrishna, Composites Science and Technology 20th Anniversary Special Issue 65 (2005) 2385-2406.  E. C. Shors, Orthopedic Clinic s of North America 30 (1999) 599-613.  T. J. Webster, American Cerami c Society Bulletin 82 (2003) 23-28B.  K. U. Lewandrowski, J. D. Gresser, D. L. Wise, D. J. Trantolo, Biomaterials 21 (2000) 757-764.  E. Baer, J. J. Cassidy, A. Hiltner, Hierarch ical structure of collagen composite systems: Lessons from biology, in: M. Sarikaya, I. A. Aksay, BiomimeticsDesign and Processing of Materials, AIP Press, Woodbury, NY, 1995, p. 13-34.  J. D. Currey, Journal of E xperimental Biol ogy 202 (1999) 3285-3294.  S. Weiner, H. D. Wagner, Annual Re view of Materials Science 28 (1998) 271-298.  S. Weiner, W. Traub, H. D. Wagner, Journal of Structural Biology 126 (1999) 241255.  A. Veis, in: P. M. Dove, De Yoreo, J. Y., Weiner, S., In Biom ineralization, vol 54, The Mineralogical Society of Am erica, Washington DC, 2003, p. 249.  J. D. Currey, Osteoporosis International 14 (2003) S29-S36.  R. B. Martin, D. Burr, N. Sharkey, Sk eletal Tissue Mechanics, Springer-Verlag New York Incorporated, New York, 1998, p. 406.  M. J. Olszta, X. Cheng, S. S. Jee, R. Kumar, Y.-Y. Kim, M. J. Kaufman, E. P. Douglas, L. B. Gower, Materials Science a nd Engineering: R: Re ports 58 (2007) 77-116.  J. D. Currey, P. Zioupos, P. Davies, A. Casinos, Proceedings of the Royal Society of London Series B-Biological Sciences 268 (2001) 107-111.  L. W. Fisher, J. D. Termine, Clin. Orthop. 200 (1985) 362.  S. V. Dorozhkin, M. Epple, Angewa ndte Chemie-International Edition 41 (2002) 3130-3146.
205  J. E. Eastoe, B. Eastoe, Biochemical Journa l 57 (1954) 453-459.  S. Weiner, W. Traub, Orga nization of Crystals in Bone, in: S. Suga, H. Nakahara, Mechanisms and Phylogeny of Minerali zation in Biological Systems, vol Biomineralization '90, Springer-Verlag, N.Y., 1991, p. 247-253.  S. Weiner, W. Traub, Fe bs Letters 206 (1986) 262-266.  H. R. Wenk, F. Heid elbach, Bone 24 (1999) 361-369.  E. R. Wise, S. Maltsev, M. E. Davies, M. J. Duer, C. Jaeger, N. Loveridge, R. C. Murray, D. G. Reid, Chemistry of Materials 19 (2007) 5055-5057.  P. S. Stayton, G. P. Drobny, W. J. Shaw, J. R. Long, M. Gilbert, Critical Reviews in Oral Biology & Medi cine 14 (2003) 370-376.  S. Weiner, T. Arad, I. Saba nay, W. Traub, Bone 20 (1997) 509-514.  M. J. Olszta, A New Paradigm for Bi omineral Formation: Mineralization via AN Amorphous Liquid-Phase Precursor Process, Materials Science and Engineering, University of Florida, Gainesville Fl, 2004, p. 96-113.  J. F. Vincent, Structur al Biomaterials, Princeton Univ ersity Press, Princeton, 1990, p. 179-196.  H. W. Kim, J. C. Knowles, H. E. Kim, Journal of Biomedical Ma terials Research Part A 70A (2004) 467-479.  W. D. Callister, Materials Scinecne and Engineering An Introduction, John Wiley & Sons, Inc., Hoboken, NJ, 2003, p. 527-568.  M. M. GiraudGuille, Twisted liquid cr ystalline supramolecular arrangements in morphogenesis, in: International Review of Cytology a Survey of Cell Biology, Vol 166, vol 166, 1996, p. 59-101.  C. R. Barrett, Nix, W. D ., Tetelman, A. S., The Principl es of Engineering Materials, Prentice-Hall Inc., Englewood Cliffs, 1973, p. p 554.  Q. Zhang, J. Chen, J. Feng, Y. Cao, C. Deng, X. Zhang, Biom aterials 24 (2003) 47414748.  A. R. Vaccaro, Orth opedics 25 (2002) S571-S578.  R. Murugan, S. Ramakrishna, Applied Physics Letters 88 (2006).  T. J. Webster, J. U. Ejiofo r, Biomaterials 25 (2004) 4731-4739.  P. Tambasco de Oliveira, A. Nanci, Biomaterials 25 (2004) 403-413.
206  Y. Yang, J. Tian, Bone 28 (2001) S153-S153.  C. Y. Kim, S. S. Jee, Journal of the European Ceramic So ciety 23 (2003) 1803-1811.  L. L. Hench, Journal of the Amer ican Ceramic Society 74 (1991) 1487-1510.  L. L. Hench, Science 208 (1980) 826-831.  Y. L. Liu, P. Layrolle, J. de Bruijn, C. van Blitterswijk, K. de Groot, Journal of Biomedical Materials Re search 57 (2001) 327-335.  Y. J. Yin, F. Ye, J. F. Cui, F. J. Zha ng, X. L. Li, K. D. Yao, Journal of Biomedical Materials Research Pa rt A 67A (2003) 844-855.  R. J. Dekker, J. D. de Bruijn, M. Stigter, F. Barrere, P. Layrolle, C. A. van Blitterswijk, Biomaterials 26 (2005) 5231-5239.  R. De Santis, F. Mollica, F. Zarone, L. Ambrosio, L. Nicolais, Acta Biomaterialia 3 (2007) 121-126.  F. Z. Cui, Y. Li, J. Ge, Materials Sc ience & Engineering R-Re ports 57 (2007) 1-27.  R. G. T. Geesink, N. H. M. Hoefnagels Journal of Bone and Joint Surgery-British Volume 77B (1995) 534-547.  R. Kumar, K. H. Prakash, K. G. Lim, N. H. Kiat, K. Yennie, P. Cheang, K. A. Khor, Hydroxyapatite-chitosan hybrid nano-materia ls, in: Bioceramics, Vol 17, vol 17, 2005, p. 721-724.  H. W. Kim, Journal of Biomedical Materials Resear ch Part A 83A (2007) 169-177.  S. S. Liao, F. Z. Cui, W. Zhang, Q. L. Feng, Journal of Biomedical Materials Research Part B-Applied Bi omaterials 69B (2004) 158-165.  M. Kikuchi, S. Itoh, S. Ichinose, K. Shinomiya, J. Tanaka, Biomaterials 22 (2001) 1705-1711.  M. Kikuchi, S. Itoh, H. N. Matsumot o, Y. Koyama, K. Takakuda, K. Shinomiya, J. Tanaka, Bioceramics 15 240-2 (2003) 567-570.  M. Kikuchi, H. N. Matsumoto, T. Ya mada, Y. Koyama, K. Takakuda, J. Tanaka, Biomaterials 25 (2004) 63-69.  A. Yokoyama, M. Gelinsky, T. Kawasaki T. Kohgo, U. Konig, W. Pompe, F. Watari, Journal of Biomedical Materials Research Part B-Applied Biomaterials 75B (2005) 464472.  J. C. Goes, S. D. Figueiro, A. M. Oliveira, A. A. M. Macedo, C. C. Silva, N. M. P. S. Ricardo, A. S. B. Sombra, Acta Biomaterialia 3 (2007) 773-778.
207  E. K. Girija, Y. Yokogawa, F. Nagata, Journal of Materials Science-Materials in Medicine 15 (2004) 593-599.  E. K. Girija, Y. Yokogawa, F. Nagata, Chemistry Letters (2002) 702-703.  W. Traub, T. Arad, S. Weiner, V. Ziv, Ab stracts of Papers of the American Chemical Society 212 (1996) 172-INOR.  A. Oyane, K. Onuma, A. Ito, T. Fu ruya, H. M. Kim, T. Kokubo, T. Nakamura, Variation of clusters in simulated body flui ds with time, in: Bioceramics 14, vol 218-2, 2002, p. 629-632.  S. H. Rhee, J. Tanaka, Journal of the American Ceramic Society 81 (1998) 3029-3031.  S. H. Rhee, J. Tanaka, Biomaterials 20 (1999) 2155-2160.  J. F. V. Vincent, O. A. Bogatyreva, N. R. Bogatyrev, A. Bowyer, A. K. Pahl, Journal of the Royal Society Interface 3 (2006) 471-482.  S. Mann, Angewandte Chemie -International Edition 39 (2000) 3393-3406.  A. W. Xu, Y. R. Ma, H. Colfen, Journal of Mate rials Chemistry 17 (2007) 415-449.  E. DiMasi, M. J. Olszta, V. M. Pate l, L. B. Gower, Crystengcomm 5 (2003) 346-350.  E. DiMasi, V. M. Patel, M. Sivakumar, M. J. Olszta, Y. P. Yang, L. B. Gower, Langmuir 18 (2002) 8902-8909.  S. Mann, Structur e and Bonding 54 (1983) 125-174.  S. Weiner, L. Addadi, Trends in Biochemical Sciences 16 (1991) 252-256.  L. Addadi, S. Weiner, Angew. Chem. Int. Ed. Engl. 31 (1992) 153-169.  R. J. Davey, N. Blagden, G. D. Potts, R. Docherty, Journal of the American Chemical Society 119 (1997) 1767-1772.  L. A. Gower, D. A. Tirrell, Jour nal of Crystal Growth 191 (1998) 153-160.  L. B. Gower, D. J. Odom, Jour nal of Crystal Growth 210 (2000) 719-734.  X. Cheng, L. B. Gower, Biotechnol. Prog. 22 (2006) 141-149.  L. Dai, E. P. Douglas, L. B. Gower, Journal of Non-Crystalline Solids 354 (2008) 1845-1854.  Y.-Y. Kim, E. P. Douglas, L. B. Gower, Langmuir 23 (2007) 4862-4870.
208  M. D. McKee, W. N. Addison, M. T. Kaartinen, Cells Tissues Organs 181 (2005) 176-188.  E. Bonucci, Calcifica tion in Biological Systems, C R C Press LLC, Boca Raton, 1992, p. 432 p.  D. A. Cameron, Clin Orthop 26 (1963) 199-228.  G. J. Tortora, K. Petti, Principles of human anatomy, John W iley & Sons, New York, 2002, p. xxiii, 918.  H. A. Lowenstam, S. Weiner, On biom ineralization, Oxford University Press, New York, 1989, p. 135-188.  M. J. Glimcher, Reviews of Modern Physics 31 (1959) 359-393.  J. Orgel, A. Miller, T. C. Irving, R. F. Fischetti, A. P. Hammersley, T. J. Wess, Structure 9 (2001) 1061-1069.  A. L. Boskey, ELEMENTS 3 (2007) 385-391.  F. H. Silver, D. Christiansen, P. B. Snowhill, Y. Chen, W. J. Landis, Biomacromolecules 1 (2000) 180-185.  F. H. Silver, J. W. Freeman, G. P. Seehra, Journal of Bi omechanics 36 (2003) 15291553.  D. L. Helseth, J. H. Lechner, A. Veis, Biopolymers 18 (1979) 3005-3014.  D. L. Helseth, A. Veis, Fede ration Proceedings 38 (1979) 819-819.  A. J. Hodge, J. A. Petruska, Recent st udies with the electron microscope on ordered aggreagates of the tropocollagen molecule, in : G. N. Ramanchandran (Ed.), Aspects of Protein Structure, vol Academic Press, London, 1963, p.^pp. 289-300.  A. J. Hodge, Connective Tissue Research 21 (1989) 467-477.  K. E. Kadler, D. F. Holmes, J. A. Tr otter, J. A. Chapman, Biochemical Journal 316 (1996) 1-11.  W. J. Landis, M. J. Song, A. Leith, L. McEwen, B. F. McEwen, Journal of Structural Biology 110 (1993) 39-54.  K. N. An, Y. L. Sun, Z. P. Luo, Biorheology 41 (2004) 239-246.  K. Takaku, T. Ogawa, T. Kuriyama, I. Narisawa, Journal of Applied Polymer Science 59 (1996) 887-896.
209  G. D. Pins, F. H. Silver, Materials Science & Engineering C-Biomimetic Materials Sensors and Systems 3 (1995) 101-107.  L. Yang, K. O. van der Werf, B. Koop man, V. Subramaniam, M. L. Bennink, P. J. Dijkstra, J. Feijen, Journal of Biomedical Materials Research Part A 82A (2007) 160-168.  R. Vanderby, Journal of Biomechanics 36 (2003) 1523-1527.  A. K. Shah, J. Lazatin, R. K. Sinha, T. Lennox, N. J. Hickok, R. S. Tuan, Biology of the Cell 91 (1999) 131-142.  S. J. Xiao, M. Textor, N. D. Sp encer, H. Sigrist, Langmuir 14 (1998) 5507-5516.  R. G. Lebaron, K. A. Athana siou, Tissue Engineering 6 (2000) 85-103.  H. Aoki, Medical Applic ation of Hydroxyapatite, Ishi yaku EuroAmerica, Inc., Tokyo, St. Louis, 1994, p. 1-326.  K. Sato, Inorganic-organi c interfacial intera ctions in hydroxyapa tite mineralization processes, in: Biomineralizati on I: Crystallization and Self-O rganization Process, vol 270, 2007, p. 127-153.  K. Sato, Journal of the Cera mic Society of Japan 115 (2007) 124-130.  C. Rey, J. L. Miquel, L. Facchini, A. P. Legrand, M. J. Glimcher, Bone 16 (1995) 583-586.  S. Weiner, T. Arad, W. Traub, Febs Letters 285 (1991) 49-54.  W. Traub, T. Arad, S. Weiner, Proceedi ngs of the National Academy of Sciences of the United States of America 86 (1989) 9822-9826.  R. A. Robinson, M. L. Watson, An atomical Record 114 (1952) 383-&.  S. A. Jackson, A. G. Cartwright, D. Lewis, Calcified Tissue Research 25 (1978) 217222.  E. N. Marierb, Human Anatomy & P hysiology, Addison Wesley Longman, Inc., Menlo Park, California, 1998, p. 164-186.  W. J. Landis, K. J. Hodgens, J. Ar ena, M. J. Song, B. F. McEwen, Microscopy Research and Technique 33 (1996) 192-202.  D. B. Williams, Carter, C. B., Transm ission Electron Microsc opy, Plenum Press, New York, 1996, p. 177-345.  J. Ge, F. Z. Cui, X. M. Wang, Y. Wang, Materials Science & Engineering CBiomimetic and Supramolecular Systems 27 (2007) 46-50.
210  R. Baron, ANATOMY AN D ULTRASTRUCTURE OF BONE HISTOGENESIS, GROWTH AND REMODELING, Dec 2007, http://www.endotext.org/parathyr oid/parathyroid1/index.html.  F. Borovecki, N. Pecina-Slaus, S. V ukicevic, International Orthopaedics 31 (2007) 799-805.  A. M. Parfitt, Bone 30 (2002) 5-7.  E. Bonucci, Role of colla gen fibrils in calcification, in: E. Bonucci '(Ed.)'^'(Eds.)', Calcification of biological systems, vol CRC Press, Boca Raton, 1992, p.^pp. 19-39.  N. Shepard, Role of prot eoglycans in calcification, in: E. Bonucci '(Ed.)'^'(Eds.)', Calcification in biologic al systems, vol CRC Press, Boca Raton, 1992, p.^pp. 41-58.  A. M. Parfitt, Calcified Ti ssue International 36 (1984) S37-S45.  A. Michael Parfitt, Bone 39 (2006) 1170-1172.  A. Parfitt, Calc ified Tissue International 36 (1984) S37-S45.  Top40-Chart.com, Osteoc last, Mar 2008, http://www.top40charts.info/?title=Osteoclast.  J. Davies, J. Warwick, N. Totty, R. Ph ilp, M. Helfrich, M. Horton, Journal of Cell Biology 109 (1989) 1817-1826.  R. Riihonen, C. T. Supuran, S. Parkkila, S. Pastorekova, H. K. Vaananen, T. LaitalaLeinonen, Bone 40 (2007) 1021-1031.  M. R. Badger, G. D. Price, Annual Re view of Plant Physiol ogy and Plant Molecular Biology 45 (1994) 369-392.  J. Morko, R. Kiviranta, S. Hurme, J. Rantakokko, E. Vuorio, Bone 36 (2005) 854865.  E. Britannica, Osteoblast, Ma y 2008, http://www.britannica.com/eb/topic434269/osteoblast.  D. R. Ganta, M. B. McCarthy, G. A. Gronowicz, Endocrinology 138 (1997) 36063612.  M. KlapiszWolikow, J. L. Saffar, J ournal of Cellular Phys iology 167 (1996) 359-368.  Dictionary of Cell and Molecular. Biology, Osteoid, May 2008, http://cancerweb.ncl.ac.uk /cgi-bin/omd?osteoid.  J. H. Martin, J. L. Matthews, Calcified Tissu e Research 3 (1969) 184-&.
211  H. C. Anderson, Journal of Cell Biology 41 (1969) 59-&.  A. L. Arsenault, F. P. Ottensmeye r, Journal of Cell Biology 98 (1984) 911-921.  A. L. Arsenault, F. P. Ottensmeye r, Proceedings of th e National Academy of Sciences of the United States of Amer ica-Biological Sciences 80 (1983) 1322-1326.  D. M. Kohler, M. A. Crenshaw, A. L. Arsenault, Matrix Biology 14 (1995) 543-552.  W. J. Landis, K. J. Hodgens, M. D. Mc Kee, A. Nanci, M. J. Song, S. Kiyonaga, J. Arena, B. McEwen, Bone and Mineral 17 (1992) 237-241.  John D. Termine, Non-Collagen Proteins in Bone, in: S. H. David Evered Ciba Foundation Symposium 136 Cell and Molecula r Biology of Vertebrate Hard Tissues, 2007, p. 178-206.  A. Nanci, Journal of St ructural Biology 126 (1999) 256-269.  D. A. Wahl, E. Sachlos, C. Z. Liu, J. T. Czernuszka, Journal of Materials ScienceMaterials in Medici ne 18 (2007) 201-209.  D. A. Wahl, J. T. Czernuszka, Eu ropean Cells & Materials 11 (2006) 43-56.  R. A. Harper, A. S. Posner, Proceedings of the Society for Experimental Biology and Medicine 122 (1966) 137-142.  J. D. Termine, A. S. Posner, Science 153 (1966) 1523-1525.  A. S. Posner, N. C. Blumenthal, A. L. Boskey, F. Betts, Journal of Dental Research 54 (1975) B88-B93.  E. D. Eanes, J. D. Termine, M. U. Nylen, Calcified Tissue Research 12 (1973) 143158.  E. D. Eanes, Gill esse.Ih, A. S. Posner, Nature 208 (1965) 365-367.  M. J. Glimcher, L. C. Bonar, M. D. Gr ynpas, W. J. Landis, A. H. Roufosse, Journal of Crystal Growth 53 (1981) 100-119.  E. D. Eanes, J. D. Termine, M. U. Nylen, Calcified Tissue Research 12 (1973) 143158.  W. Traub, T. Arad, S. Weiner, Matrix 12 (1992) 251-255.  M. J. Glimcher, S. M. Krane, Biology of Collagen, Academic Press, London, 1968, p.  E. P. Katz, S. T. Li, Journa l of Molecular Biology 73 (1973) 351-369.
212  E. P. Katz, E. Wachtel, M. Yamauchi G. L. Mechanic, Connective Tissue Research 21 (1989) 479-488.  W. J. Landis, K. J. Hodgens, M. J. Song, J. Arena, S. Kiyonaga, M. Marko, C. Owen, B. F. McEwen, Journal of St ructural Biology 117 (1996) 24-35.  K. Sato, T. Kogure, Y. Kumagai, J. Ta naka, Journal of Colloid and Interface Science 240 (2001) 133-138.  W. Zhang, Z. L. Huang, S. S. Liao, F. Z. Cui, Journal of the American Ceramic Society 86 (2003) 1052-1054.  S. Weiner, Bone 39 (2006) 431-433.  S. Weiner, L. Adda di, Science 298 (2002) 375-376.  K. M. Towe, H. A. Lowenstam, Journal of Ultrastructure Res earch 17 (1967) 1-13.  N. T. Ingrid Maria We iss, Lia Addadi, Steve Weiner ,, Journal of Experimental Zoology 293 (2002) 478-491.  Y. Politi, T. Arad, E. Klein, S. We iner, L. Addadi, Science 306 (2004) 1161-1164.  M. Faatz, F. Grohn, G. Wegner, Advanced Materials 16 (2004) 996-1000.  M. J. Olszta, D. J. Odom, E. P. Dougl as, L. B. Gower, Connective Tissue Research 44 (2003) 326-334.  B. Guillemet, M. Faatz, F. Gro hn, G. Wegner, Y. Gnanou, Langmuir 22 (2006) 1875-1879.  S. Weiner, I. Sagi, L. Addadi, Science 309 (2005) 1027-1028.  N. J. Crane, V. Popescu, M. D. Morris, P. Steenhuis, M. A. Ignelzi, Bone 39 (2006) 434-442.  L. Addadi, S. Weiner, Proceedings of the National Academy of Sciences of the United States of America 82 (1985) 4110-4114.  J. P. Gorski, Calcified Ti ssue Internationa l 50 (1992) 391-396.  B. Constantz, S. Weiner, Journa l of Experimental Zoology 248 (1988) 253-258.  E. M. Greenfield, M. A. Crenshaw, Mi neral Induction by the Soluble Matrix from Molluscan Shells, Plenum Press, New York, 1990, p. 303-308.  T. Saito, A. L. Arsenault, M. Yamauchi, Y. Kuboki, M. A. Crenshaw, Bone 21 (1997) 305-311.
213  A. Tsortos, G. H. Nancollas, Journa l of Colloid and Interface Science 250 (2002) 159-167.  J. D. Termine, A. B. Belcourt, K. M. Conn, H. K. Kleinman, Journal of Biological Chemistry 256 (1981) 403-408.  D. Zaffe, Micron 36 (2005) 583-592.  L. L. Hench, J. M. Po lak, Science 295 (2002) 1014-1017.  I. A. Karnezis, Clinic al Biomechanics 20 (2005) 270-276.  S. Nishiguchi, H. Kato, H. Fujita, M. Oka, H. M. Kim, T. Kokubo, T. Nakamura, Biomaterials 22 (2001) 2525-2533.  P. V. Pavoor, B. P. Gearing, O. Muratogl u, R. E. Cohen, A. Bellare, Biomaterials 27 (2006) 1527-1533.  L. L. Hench, Chemis try & Industry (1995) 547-550.  L. L. Hench, H. A. Paschall, Journal of Biomedical Materials Research 7 (1973) 2542.  T. Kokubo, Anales De Quimica 93 (1997) S49-S55.  M. van der Elst, C. P. A. T. Klein, J. M. de Blieck-Hogervorst, P. Patka, H. J. T. M. Haarman, Biomaterials 20 (1999) 121-128.  H. M. Kim, F. Miyaji, T. Kokubo, S. Nishiguchi, T. Nakamura, Journal of Biomedical Materials Re search 45 (1999) 100-107.  R. G. Courteneyharris, M. V. Kayser, S. Downes, Biomaterials 16 (1995) 489-495.  M. Morra, C. Cassinelli, G. Cascardo, P. Cahalan, L. Cahalan, M. Fini, R. Giardino, Biomaterials 24 (2003) 4639-4654.  M. Morra, C. Cassinelli, G. Cascardo, A. Carpi, M. Fini, G. Giavaresi, R. Giardino, Biomedicine & Pharmacotherapy 58 (2004) 418-422.  D. Bakos, M. Soldan, I. Hernand ez-Fuentes, Biomaterials 20 (1999) 191-195.  S. H. Rhee, J. D. Lee, J. Tanaka, J ournal of the American Ceramic Society 83 (2000) 2890-2892.  X. Y. Lin, X. D. Li, H. S. Fan, X. T. Wen, H. Lu, X. D. Zh ang, Materials Letters 58 (2004) 3569-3572.  D. Scharnweber, R. Born, K. Flad e, S. Roessler, M. Stoelzel, H. Worch, Biomaterials 25 (2004) 2371-2380.
214  M. Tanahashi, T. Matsuda, Journal of Biomedical Materials Research 34 (1997) 305315.  K. Kato, Y. Eika, Y. Ikada, Journal of Biomedical Materials Research 32 (1996) 687-691.  Y. Doi, T. Horiguchi, Y. Moriwa ki, H. Kitago, T. Kajimoto, Y. Iwayama, Biomedical Materials Re search 31 (1996) 43-49.  R. Z. Wang, F. Z. Cui, H. B. Lu, H. B. Wen, C. L. Ma, H. D. Li, Journal of Materials Science Letters 14 (1995) 490-492.  M. L. Oyen, MRS Bulletin 33 (2008) 49-55.  M. J. Olszta, S. Gajjeraman, M. Kaufma n, L. B. Gower, Chemistry of Materials 16 (2004) 2355-2362.  L. A. Gower, The Influence of Po lyaspartate Additive on the Growth and Morphology of Calcium Carbonate Crystals, Polymer Science & Engineering, University of Massachusetts at Amherst, 1997, p. 119.  T. Ono, T. K. Nemoto, Matrix Biology 24 (2005) 239-244.  D. Toroian, J. E. Lim, P. A. Price, Journal of Biological Chemistry 282 (2007) 22437-22447.  L. W. Fisher, D. A. Torchia, B. Fohr M. F. Young, N. S. Fedarko, Biochemical and Biophysical Research Communi cations 280 (2001) 460-465.  J. L. Chen, C. Burger, C. V. Krishnan, B. Chu, B. S. Hsiao, M. J. Glimcher, Macromolecular Chemistry and Physics 206 (2005) 43-51.  W. J. Landis, F. H. Silver, Compara tive Biochemistry and Physiology Part A: Molecular & Integrative P hysiology 133 (2002) 1135-1157.  M. J. Olszta, E. P. Douglas, L. B. Go wer, Calcified Tissue In ternational 72 (2003) 583-591.  K. Yamauchi, T. Goda, N. Takeuchi, H. Einaga, T. Tanabe, Biomaterials 25 (2004) 5481-5489.  E. Banks, S. Nakajima, I. C. Shapiro, O. Tilevitz, J. R. Alonzo, R. R. Chianelli, Science 198 (1977) 1164-1166.  I. J. Lorch, Journal of Bone and Jo int Surgery-British Volume 31 (1949) 94-99.  Y. Doi, T. Horiguchi, S. H. Kim, Y. Moriwaki, N. Wakamatsu, M. Adachi, K. Ibaraki, K. Moriyama, S. Sasaki, H. Shim okawa, Archives of Oral Biology 37 (1992) 1521.
215  Y. F. Chou, W. A. Chi ou, Y. H. Xu, J. C. Y. Dunn, B. M. Wu, Biomaterials 25 (2004) 5323-5331.  L. G. Wade, Organic Chemistry, Pears on Education, Inc., Upper Saddle River, NJ, 2003, p. 405-444.  Zirchrom, Dissociation Constants of Organic Acids And Bases, May 2008, http://www.zirchrom.com/organic.htm.  W. Traub, T. Arad, S. Weiner, C onnective Tissue Research 28 (1992) 99-111.  L. C. Labaki, I. L. To rriani, J. R. Grigera, Brazi lian Journal of Medical and Biological Research 24 (1991) 115-122.  R. Murugan, S. Ramakrishna Biomaterials 25 (2004) 3829-3835.  R. Murugan, S. Ramakrishna, K. P. Rao, Materials Lette rs 60 (2006) 2844-2847.  B. D. Cullity, Elements of X-ray Diffraction, Bando, Seoul, 1997, p. 672.  C. G. Darwin, Philosophi cal Magazine 27 (1914) 675-690.  C. G. Darwin, Philosophi cal Magazine 27 (1914) 315-333.  Innovagen, Peptide property calculator, May 2008, http://www.innovagen.se/custom-peptide-synthe sis/peptide-propertycalculator/peptideproperty-calculator.asp.  K. I. Kimiko Maruta, Hisao Matsui, Tohru Yamagami, Akimi Sano, Haruo Tsuji,, Journal of Orthopaedic Re search 11 (1993) 92-103.  L. Cohen-Solal, J. B. Lian, D. Kossi va, M. J. Glimcher, Biochem. J. 177 (1979 ) 8198.  A. C. E. Strings, Collage n and Natural Gut Strings, May 2008, http://web.mit.edu/3.082/www/team1_f02/collagen.htm.  C. H. Lee, A. Singla, Y. Lee, International Journal of Pharmaceutics 221 (2001) 1-22.  V. C. A. Martins, G. Goissis, A. C. Ribeiro, E. Marcantonio, M. R. Bet, Artificial Organs 22 (1998) 215-221.  G. D. Pins, D. L. Christiansen, R. Pa tel, F. H. Silver, Bi ophysical Journal 73 (1997) 2164-2172.  M. R. Urist, Sc ience 150 (1965) 893-899.
216  D. Olsen, C. L. Yang, M. Bodo, R. Cha ng, S. Leigh, J. Baez, D. Carmichael, M. Perala, E. R. Hamalainen, M. Jarvinen, J. Polarek, Advanced Drug Delivery Reviews 55 (2003) 1547-1567.  R. Murugan, S. Ramakrishna, Crys tal Growth & Design 5 (2005) 111-112.  C. Y. Ooi, M. Hamdi, S. Rames h, Ceramics Interna tional 33 (2007) 1171-1177.  T. Sakae, H. Mishima, Y. Kozawa, R. Z. LeGeros, Connective Tissue Research 32 (1995) 515-518.  L. F. Lozano, M. A. Pena-Rico, A. Heredi a, J. Ocotlan-Flores, A. Gomez-Cortes, R. Velazquez, I. A. Belio, L. Bucio, Jour nal of Materials Sc ience 38 (2003) 4777-4782.  R. Kumar, P. Cheang, K. A. Khor, Acta Materialia 52 (2004) 1171-1181.  M. N. Rahaman, Ceramic Processing and Sintering, Marcel Dekker, Inc., New York, 1995, p. 146-200.  A. Sze, D. Erickson, L. Q. Ren, D. Q. Li, Journal of Colloid and Interface Science 261 (2003) 402-410.  E. C. ORHAN, Electrical D ouble Layer (Counter-Ion Layer), May 2008, http://www.geocities.com/CapeCa naveral/Hangar/5555/zeta.htm.  A. Szymczyk, P. Fievet, B. A oubiza, Desalination 151 (2003) 177-184.  V. Knecht, H. J. Risselada, A. E. Ma rk, S. J. Marrink, Journal of Colloid and Interface Science 318 (2008) 477-486.  H. M. Kim, T. Himeno, M. Kawashita T. Kokubo, T. Nakamura, Journal of the Royal Society Interface 1 (2004) 17-22.  M. Alkan, O. Demirbas, M. Dogan, J ournal of Colloid and Interface Science 281 (2005) 240-248.  J.-H. Zhu, B. Zhang, W.-W. Fang, X.-J Lao, H. Yu, Colloids and Surfaces B: Biointerfaces 43 (2005) 1-6.  E. DiMasi, Olszta, M and Gower, L, Materials Research Society Symposium Proceedings 873E (2005) K10.16.11-K10.16.17.  Y. T. Wu, C. Gran t, Langmuir 18 (2002) 6813-6820.
217 BIOGRAPHICAL SKETCH Sang Soo Jee was born in Seoul, South Ko rea in 1975. He was raised with a high expectation from his parents and also had to lead his younger brother and younger sister. He decided to study science when he was a sophomor e in high school. Based on his enthusiasm for chemistry, he applied to the Department of I norganic Materials in Inha University (Incheon, Korea) and was accepted. He worked at the glass science lab in Inha University for a while and he decided to join graduate school to study mo re about glass science and engineering. During his graduate school period, he studied bioactive glasses and bi oactive glass coat on bioinert materials. He obtained several successful results from his res earch and decided to present his work in the meeting of American Ceramic So ciety in 2001 (Indianapo lis, USA). During the meeting, he was impressed by numerous works from entire world and decided to study abroad to open his eyes to the high level of research. He applied to the departme nt of materials science and engineering in University of Florida, which was one of top 10 materials science and engineering departments in USA, and got an admission. He joined the biomimetics lab at his first year and has worked with Dr. Laurie Go wer who has showed huge generosity to him and supported him during his tenure. Even though he had to work at a new filed, polymer science and biomimetics, he could manage his work which was big challenge for him over 30. Based on the huge amount of support from Dr. Gower, lab member and family member, he finally passed the Ph. D defense on May 7th, 2008 and he is waiting for the new worl d in front of him.