Collagen–GAG Materials

2.216.

Collagen–GAG Materials

S R Caliari and B A C Harley, University of Illinois at Urbana-Champaign, Urbana, IL, USA ã 2011 Elsevier Ltd. All rights reserved.

2.216.1. 2.216.1.1. 2.216.1.2. 2.216.1.3. 2.216.1.4. 2.216.2. 2.216.2.1. 2.216.2.2. 2.216.2.2.1. 2.216.2.2.2. 2.216.2.2.3. 2.216.2.3. 2.216.2.4. 2.216.2.5. 2.216.2.6. 2.216.3. 2.216.3.1. 2.216.3.2. 2.216.3.3. 2.216.3.4. 2.216.4. 2.216.4.1. 2.216.4.2. 2.216.4.3. 2.216.4.4. 2.216.4.5. 2.216.4.6. 2.216.4.7. 2.216.5. 2.216.5.1. 2.216.5.2. 2.216.5.3. 2.216.5.4. 2.216.5.5. 2.216.5.6. 2.216.5.7. 2.216.6. References

Introduction Collagen Glycosaminoglycans Collagen–GAG Material Development Antigenicity and Immunogenicity of Collagen–GAG Materials Fabrication of Collagen–GAG Materials CG Scaffold Fabrication: Lyophilization Modifications to CG Scaffold During Fabrication Relative density Pore size and shape Crosslinking CGCaP Scaffold Fabrication Multicompartment CG Scaffold Fabrication CG Hydrogels CG Membranes Characterization of Collagen–GAG Materials Microstructure: Experimental Measurement and Modeling Permeability: Experimental Measurement and Modeling Mechanical Properties: Experimental Measurement and Modeling CG Scaffold Degradation Kinetics and Byproducts In Vitro Applications Cell Attachment and Viability Cell Contraction Cell Motility Growth Factor and Gene Delivery Modifying Gene Expression Mechanical Stimulation Stem Cell Differentiation In Vivo Applications Dermal Regeneration Applications Peripheral Nerve Regeneration Applications Conjunctiva and Corneal Regeneration Applications Cartilage and Fibrocartilage Disk Tissue Engineering Applications Bone, Osteochondral Regeneration Applications Brain Tissue Engineering Applications Lung Tissue Engineering Applications Conclusions

Abbreviations CFD CFM CG CGCaP COLxAy

Computational fluid dynamics Culture force monitor Collagen–glycosaminoglycan coprecipitate Collagen–glycosaminoglycan–calcium phosphate triple coprecipitate Family of genes that encode for synthesizing distinct collagen subunit chains; x, y correspond to collagen type and subunit chain, respectively (e.g., COL1A2, COL3A2, COL2A1, COL4A4, COL4A5)

COX-2

DHT DRT ECM EDAC EDX ESEM FGF

280 281 281 282 283 283 283 283 283 283 284 285 286 286 287 287 287 288 288 291 292 292 292 294 296 296 296 297 297 297 298 298 298 299 300 300 300 300

Cyclooxygenase-2; key enzyme for the production of early bone formation marker prostaglandin E2 Dehydrothermal crosslinking Dermal regeneration template Extracellular matrix 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride Energy-dispersive X-ray spectroscopy Environmental scanning electron microscopy Fibroblast growth factor

279

280

Materials of Biological Origin

GAGs GDNF HA HGF

Glycosaminoglycans Glial cell line-derived neurotrophic factor Hyaluronic acid Hepatocyte growth factor; a paracrine signaling factor involved in cellular growth, motility, morphogenesis, and angiogenesis Heme oxygenase; an enzyme that catalyzes the degradation of heme and is critically involved in angiogenesis Interpenetrating network Mouse preosteoblast cell line

HMOX

IPN MC3T3E1 MMP

Matrix metalloproteinase; a family of enzymes that degrades extracellular matrix proteins. Key MMPs include: MMP-2, MMP-12, MMP-19 Mesenchymal stem cell or marrow stromal cell

MSC

Symbols Djxn

d Es E* E sI « Fc

2.216.1.

Strut junction spacing; defines the distance between points in the scaffold where multiple struts meet Scaffold pore size Elastic modulus of individual scaffold strut Elastic modulus of the entire cellular material Scaffold strut flexural rigidity Compressive strain Average contractile force generated by an individual cells within a 3D scaffold

Introduction

The extracellular matrix (ECM) is a complex macromolecular structure composed of a fibrillar network of structural proteins (collagens, elastin, etc.), proteoglycans (polypeptide core with attached polysaccharide chains called glycosaminoglycans (GAGs)), specialized proteins for cell adhesion (fibronectin, laminin, etc.), and other tissue-specific materials such as apatite in bone. The ECM defines the physical morphology of tissues and the local environment in which cells reside. In addition to serving as a physical support structure and insoluble regulator of cell activity, the ECM is a reservoir of numerous soluble regulators of cell behavior. Tissue engineering scaffolds are utilized as ECM analogs to heal or modify tissues in a defined manner. In order to be successful, these scaffolds must be able to mimic key aspects of the native ECM and support normal cell behaviors. As an analog of the native ECM, collagen–glycosaminoglycan (collagen– GAG) materials possess several vital characteristics that aid in the making of successful tissue engineering scaffolds, including three dimensionality with interconnected pores, tunable degradation and resorption rates, surface ligands for cell adhesion, and mechanical integrity.1,2 Since their inception, these materials have been utilized in a variety of tissue engineering studies, both in vivo as regenerative templates for skin, peripheral

NHS NRT OPN PGE2 SA/V TC TGF-b1 TIMP UC mCT VEGF

I k l ls s*el rjxn

r*/ rs

N-Hydroxysulfosuccinimide Nerve regeneration template Osteopontin; late marker of bone formation Prostaglandin E2; early bone formation marker Specific surface area; total surface area per unit volume for a given unit cell Tropocollagen; the fundamental organizational unit of collagen Transforming growth factor-b1; Tissue inhibitor of metalloproteinase. Key MMP inhibitors include: TIMP1, TIMP3 Unit cell Microcomputed tomography Vascular endothelial growth factor; a chemical signal produced by cells that stimulates the growth of new blood vessels (angiogenesis)

Second moment of inertia Scaffold permeability (m2) Edge length of unit cell Scaffold strut length Compressive plateau stress Strut junction density; the number of strut junctions per unit cell divided by the volume of the unit cell in a cellular material Relative density (1 – porosity)

nerves, conjunctiva, and cartilage3–6 and in vitro as 3D microenvironments to probe more fundamental questions about cell behaviors and cell–matrix interactions. Scaffold microstructure characteristics, such as porosity, mean pore size, pore shape, interconnectivity, specific surface area,4,5,7–15 and mechanical properties (Young’s modulus, yield stress),16–25 have been shown to be key factors that can influence cell behaviors such as adhesion, motility, contraction, stem cell differentiation, gene expression, and overall bioactivity.4 This chapter will provide an introduction to, and an overview of, the development and application of collagen–GAG (CG) materials. We begin with an overview of the structure and biochemistry of both collagen and GAGs and will consider questions concerning the immunogenicity, antigenicity, and regulatory compliance of CG biomaterials. We will continue with a summary of the fabrication techniques used to make a wide range of CG materials, including mineralized and nonmineralized (type I and II) scaffolds as well as membranes and hydrogels. Characterization of CG materials, notably microstructural, mechanical, and chemical properties, and the use of modeling approaches to describe these features will then be covered. We will detail in vitro applications of CG scaffolds for the investigation of cell adhesion, motility, contraction, stem cell differentiation, gene expression, and mechanotransduction. Finally, in vivo applications of CG materials as ECM

Collagen–GAG Materials

analogs for the regeneration of a wide variety of tissues, notably skin, peripheral nerves, conjunctiva, cartilage, and bone will be reviewed.

2.216.1.1. Collagen Collagen is the major organic component of the natural ECM.26–29 To date, 29 types of collagen have been identified,26,28 although 90% of collagen in the human body is fibrillar (main types: I, II, III, and V). Type I collagen is the major structural component of the ECM in a wide variety of tissues such as skin, tendon, ligament, and bone. Type II collagen is found principally in cartilage; type III is found in skin, blood vessels, and intestines; and type V is a component of bone, skin, cornea, and other tissues.28 Like all biological proteins, collagen is defined by four levels of organization: primary, secondary, tertiary, and quaternary. The primary structure of the collagen is highly conserved among mammals and follows a Gly-X-Y pattern characterized by the presence of a glycine every third residue with X and Y representing other amino acids (Figure 1(a)). Common residues in these X and Y positions are proline (28% of total residues) and hydroxyproline (38%), respectively.26–30 These amino acids form left-handed, polyproline II-type polypeptide chains that define the secondary structure. Three of these polypeptide chains come together in a coiled, helical manner to form a right-handed triple helix termed tropocollagen (TC). The TC molecule is the fundamental organizational unit of the collagen and defines its tertiary structure26,28,30 (Figure 1(b)), with dimensions of approximately 280 nm in length and 1.5 nm in diameter.28,30 Each of the three polypeptide units, or alpha units, contains 1050 amino acids.27 Typically, two of these units have very similar sequences and are termed a1 chains while the dissimilar third chain is denoted as a2. Glycine is the only amino acid with a side group (single hydrogen) small enough to fit

281

within the tight packing of the triple helix, thereby necessitating its presence at every third residue. While the glycines face inward because of steric constraints, the proline and hydroxyproline groups face outwards and help stabilize the overall triple helix structure via hydrogen bonding.27,28 For tissues composed of type I collagen (bone, tendon, ligament, etc.), TC molecules assemble into fibrils up to 500 nm in diameter and 1 cm in length (quaternary structure). Fibrillogenesis, and subsequent crosslinking, is aided by the presence of nonhelical residues on each end of the TC molecules called collagen telopeptides.27,28 These ends contain lysyl residues that promote covalent crosslinking between opposing ends of adjacent TC molecules. This crosslinking insures the stability of the fibrillar structure. Each TC molecule is displaced from its laterally adjacent partner by 67 nm. This periodic organization is the key feature of the quaternary structure (Figure 1(c)).27,28 Fibrils band together to form higher order fiber bundles that are tissue specific. For example, tendon is composed of uniaxially aligned, crimped fiber bundles. It is important to note that the banding (quaternary structure) of collagen is abolished as a result of acid coprecipitation during fabrication of CG materials, limiting platelet aggregation in vivo.4,31 The destruction of the quaternary structure is accomplished without denaturing the collagen to gelatin. Gelatin is the amorphous form of collagen, having the same amino acid sequence but lacking the triple helical character of collagen. Gelatin can be formed by heating hydrated collagen past a critical temperature of 75–105
C. The temperature varies according to collagen type, species, and hydration level. Dry collagen is denatured when heated over 150
C.30,31

2.216.1.2. Glycosaminoglycans GAGs are long, repeating disaccharide units that are linked to protein cores to form proteoglycans (Figure 2).32 GAGs serve as OH

H

O

O N

CH

H

CH

N

N

N

gly

O

H

O

H (a)

R1

pro

y

O

O

R2 N

CH CH

N

H

H

H

O x

gly

N

hyp

(b) 40 nm 67 nm

1.5 nm

300 nm (c)

Figure 1 Structural and chemical overview of collagen type I. (a) Typical primary amino acid sequence. (b) Secondary left-handed helix and tertiary right-handed triple-helix structure. (c) Staggered quaternary structure displaying characteristic 67 nm lateral offset between fibrils. Reprinted with permission from Friess, W. Eur. J. Pharm. Biopharm. 1998, 45, 113–136.

282

Materials of Biological Origin

NS 2S

6S

6S

2S

(a)

(b)

OO

S

O

O

y2

C

C2 C3

O

n = 2.0, h = 8.4 Å

-O O

C4

NS

C1 NH

HO O

S O-

HNS,6S

f1 O

O

O

OH

C4

O

C1

4

C1

2

SO

f2

C3

y1

O

O

I2S

O

S O-

C2 O 1C

4

O

S2

(c)

Figure 2 3D structure of GAG. (a) Top: ball-and-stick depiction of a heparin oligosaccharide with –[I2S-HNS,6S]– repeat unit (carbon, oxygen, nitrogen, and sulfur atoms are colored gray, red, blue, and yellow, respectively). Bottom: chair conformation of the expanded disaccharide unit showing backbone atoms (colored red) and glycosidic torsion angles (f1, c1) and (f2, c2). (b) Helical wheel projection of the sulfate groups of the structure shown in (a). (c) Commonly observed low-energy ring conformations of iduronic acid [atoms colored the same way as in (a)]. Reprinted with permission from Raman, R.; Sasisekharan, V.; Sasisekharan, R. Chem. Biol. 2005, 12, 267–277.

a crucial component of the native ECM and influence various biological functions including cell migration, division, angiogenesis, collagen fibrillogenesis, and the presentation of soluble factors.27,32–34 These macromolecules are strongly negatively charged because of the presence of sulfate groups, resulting in large associated osmotic pressures and allowing them to significantly swell in water. These properties are responsible for the high compressive modulus and excellent resistance to repeated deformation seen in GAG-rich tissues such as articular cartilage.35 The main GAG component of aggrecan, the chief proteoglycan in articular cartilage, is chondroitin sulfate, the most common GAG found in humans and also in CG materials.27,31,36 Other common classes of GAGs include dermatan sulfate (a derivative of chondroitin sulfate), heparin sulfate and heparin, keratan sulfate, and hyaluronic acid (HA).27,34,37 Chondroitin sulfate, like other GAGs, is composed of a disaccharide repeat unit consisting of a acidic sugar-like molecule and a sulfated amino sugar (the exception is HA, which is not sulfated).32 For chondroitin sulfate, these sugars are D-glucuronic acid and N-acetyl galactosamine respectively.27

2.216.1.3. Collagen–GAG Material Development CG materials were developed in the 1970s by a collaboration between Ioannis Yannas, a professor at Massachusetts Institute

of Technology, and John F. Burke, a surgeon at Massachusetts General Hospital. Burke’s search for a treatment for traumatic burns coupled with Yannas’ background in collagen biochemistry led to the development of a CG membrane that was one of the first bioactive materials designed for a tissue engineering application, in this case the regeneration of full-thickness skin wounds. Collagen is a well characterized biopolymer and was chosen as a model material in part because of its controllable biodegradability, relative weak antigenicity/ immunogenicity, and overall history in clinical applications (see Chapter 2.215, Collagen: Materials Analysis and Implant Uses).31,38 GAGs were incorporated into the collagen material for several reasons beyond that of ECM chemistry and biomimicry. First, although the biodegradation rate of collagen can be adjusted over a wide range of values via crosslinking, heavy crosslinking leaves the material brittle and stiff. The copolymerization of collagen with GAG results in a material more resistant to degradation, thereby reducing the need for heavy crosslinking.31 The addition of GAG was shown to significantly improve material mechanics, including elastic modulus and fracture energy.31 SEM imaging also showed that scaffolds made with collagen and GAG exhibit a more open pore structure than scaffolds made from collagen alone, a critical design requirement to insure adequate cell migration and permeability throughout the entire construct. The addition

Collagen–GAG Materials

283

of GAG also gives CG materials their unique hemostatic properties discussed earlier, limiting platelet aggregation normally associated with collagen.31

2.216.1.4. Antigenicity and Immunogenicity of Collagen–GAG Materials Collagen has been used in a variety of medical capacities, including sutures, skin and bone grafts, and for cosmetic purposes.5,38–40 The collagen source for these applications as well as that used for CG biomaterials has typically been bovine skin or tendon. Collagens from these sources have consistently displayed low antigenicity (ability to interact with antibodies) and immunogenicity (ability to induce immune response), with adverse immune responses occurring even less frequently than nickel or latex allergies.38 The majority of adverse responses arise from preexisting allergies (only 2–4% of population), but even in these cases, local inflammation and granulation subsides after several months to 1 year.38,41,42 GAGs on their own have been utilized in orthopedic implants and more recently as dietary supplements to improve joint health. They are weak antigens and break down into oligosaccharides that can be cleared by native biological processes.31 In addition, there is little evidence that the copolymerization of collagen and GAG increases antigenicity or immunogenicity of collagen on its own.38 With over 1500 human cases, CG skin grafts have not had an adverse response to date.5,38 With their long clinical history, CG materials are regulatory compliant in both the United States and Europe.

2.216.2.

Fabrication of Collagen–GAG Materials

2.216.2.1. CG Scaffold Fabrication: Lyophilization CG scaffolds are typically created via freeze-drying of a suspension consisting of coprecipitated type I collagen and the glycosaminoglycan chondroitin 6-sulfate in a solution of weak acetic acid.4 HA has been used as an alternative to chondroitin 6-sulfate in CG scaffolds for cartilage tissue engineering.43 Typical solids concentration for these suspensions are 0.5–1% (w/v) collagen and 0.05–0.1% (w/v) GAG. These values have been traditionally selected not only because the resultant scaffolds have high porosities to enable rapid cellular infiltration and diffusive transport processes, but also because of the inherent difficulty in processing CG suspensions with higher solids content without denaturing the collagen which occurs due to the increased temperature and shear stresses required for complete mixing of the CG content at high weight or volume percentages. The suspension is homogenized at 4
C to prevent collagen gelatinization, degassed under vacuum to remove air bubbles introduced during mixing, and then freeze-dried to produce the porous CG scaffold.1 During lyophilization, the CG suspension is frozen at a specified temperature,44 resulting in a continuous, interpenetrating network of ice crystals intertwined with CG coprecipitate. Sublimation of the ice crystals produces a highly porous scaffold whose microstructure is formed by individual fibers of CG called struts (Figure 3). These scaffolds resemble low-density, open-cell foams, with an interconnected network of struts and relative densities (r*/rs) typically significantly less than 10%. Relative density is calculated as the ratio of the

Acc.V Spot Magn 25.0 kV 3.0 250x

Det WD GSE 10.4 2.7 Torr

100 μm

Figure 3 ESEM image of the pore structure of the CG scaffold (Tf ¼ 40
C). Scale bar: 100 mm. Reprinted with permission from Harley, B. A.; Leung, J. H.; Silva, E.; Gibson, L. J. Acta Biomater. 2007, 3, 463–474.

density of the porous material (r*) to the density of the solid it is constructed from (rs). Porosity, another key parameter used to describe porous biomaterials, can be calculated from scaffold relative density: porosity ¼ 1  r*/rs.

2.216.2.2. Modifications to CG Scaffold During Fabrication 2.216.2.2.1.

Relative density

Scaffolds for certain applications, such as peripheral nerve and bone regeneration, have been created from CG suspension with considerably higher solids content (5–10%, w/v). Peripheral nerve scaffolds have been fabricated from high solids content CG suspension produced using plasticating extruders to homogenize the CG suspension.3 Vacuum filtration has also been used to remove solvent from homogenized CG (both mineralized and nonmineralized) suspensions, thus increasing scaffold relative density. Nonmineralized CG suspensions were created with relative densities as high as 2.5% (w/v) while still enabling fabrication of bioactive, porous CG scaffold structure;45 mineralized CG suspensions with relative densities as high as 18% (w/v) were also created in this manner.46 These densified scaffolds have been created in order to improve mechanical properties and slow biodegradation rates relative to nondensified suspensions.45,46

2.216.2.2.2.

Pore size and shape

During the freeze-drying process, modifying CG suspension thermal profiles enables fabrication of CG scaffolds with a wide range of pore sizes and shapes. Pore size is governed by the final freezing temperature, where lower temperatures lead to smaller pores.44,47 Final freezing temperature and degree of undercooling control the rate of ice crystal nucleation in the suspension. Longer solidification times lead to more significant coarsening, where ice crystals aggregate to reduce their surface energy, resulting in larger pores.9,10 Scaffolds with equiaxed pores can be fabricated using a constant cooling rate technique that results in homogenous cooling, solidification, and pore formation throughout the CG suspension.44 Using this method in combination with final freezing temperatures ranging from 10 to 60
C, CG scaffolds have been fabricated with pore sizes

284

Materials of Biological Origin

from 85 to 325 mm but with constant relative density (r*/ rs ¼ 0.006) (Table 1).9,10 More heterogeneous solidification processes at lower temperatures (80
C) have also been used to fabricate CG scaffolds with pore sizes as low as 10–20 mm.5,47,48 Scaffold microstructure can be further tuned by altering other aspects of the freezing profile. For example, the alignment and elongation of pores is controlled by the directionality of heat transfer during the freezing process. While the constant cooling method results in isotropic polyhedral pores by cooling the entire CG suspension homogenously, aligned tracks of ellipsoidal pores can be created using a thermally mismatched mold (polysulfone mold, copper base); the mismatch in thermal conductivity promotes unidirectional heat transfer through the copper bottom, resulting in aligned pores.49 Forthcoming work modifying this technique has allowed fabrication of CG scaffolds with pore anisotropies as great as 1.5:1 (aligned:unaligned direction) (Caliari and Harley, unpublished). Aligned features are a critical design criterion for scaffolds aimed towards regenerating directional tissues such as peripheral nerves,50 myocardium,51 and tendon.52 Rapid, unidirectional cooling has also been used to create aligned CG scaffolds for peripheral nerve regeneration.47,48 CG scaffolds with pore size and alignment gradients have been used to entubulate the transected ends of peripheral nerves and induce regeneration.3,48,53 Tubular CG scaffolds with uniform pore microstructures have been fabricated by injecting high solids content CG Table 1 Mean pore size of isotropic, equiaxed CG scaffolds produced using a variety of thermal treatments Freezing temperature (
C)

Mean pore size (mm)

References

10 10 20a 20b 20 30 40 40 60

325 151 190 164 121 110 120 96 85

9 10 9 9 10 10 9 10 9

Initial freezing temperature followed by 48 h anneal at 10
C. Initial freezing temperature followed by 24 h anneal at 10
C.

a

b

Mag = 20 x EHT = 20.00 kV

1 mm

Detector=BSE Date:15 Jan 2004

suspension into a tubular mold with a rotating mandrel. Recently, a spinning technique to create radially patterned CG scaffold tubes has been developed.54 The CG suspension is placed into a cylindrical copper mold and spun at high angular velocities (> 5000 rpm), resulting in radial sedimentation of the CG solids. The suspension is then flash frozen in liquid nitrogen, resulting in the creation of small ice crystals within the CG suspension that do not interrupt the sedimentation. By varying the angular velocity and spinning time, solid cylinders or hollow tubes of variable inner diameter and tube wall microstructure can be created. These tubes display a radially aligned pore structure and a gradient of porosity along the tube radius (Figure 4). A modeling framework that describes this sedimentation process using Lamm differential equations for collagen concentration and balances sedimentation and diffusion forces to determine scaffold tube geometry as a function of spinning conditions has also been recently developed.55 These scaffolds are currently being investigated as substrates to limit the inward migration of exogenous contractile cells into and through the tube wall for a peripheral nerve regeneration application.

2.216.2.2.3.

Crosslinking

Crosslinking improves the mechanical competence (e.g., Young’s modulus, yield stress) and decreases the degradation rate of CG scaffolds independent of the chemical and microstructural characteristics.3,56,57 The three most common crosslinking techniques are the physically based dehydrothermal (DHT) and ultraviolet (UV) processes and the chemically based carbodiimide (EDAC) process. DHT and EDAC in particular have been used extensively for in vitro and in vivo applications.5,56 DHT processing involves heating of the CG material under vacuum for a specified amount of time (typically 105–120
C, <50 mTorr, 24–48 h) to remove residual moisture and introduce covalent crosslinks within the CG struts. Increasing the DHT crosslinking temperature and duration has been shown to increase compressive modulus by up to twofold and tensile modulus by up to 3.8-fold.56,58 While increased crosslink density was correlated with increasing compressive modulus, higher denaturation levels led to higher tensile modulus, but also led to reduced ultimate stress and strain to failure.58 EDAC processing utilizes carbodiimide chemistry to translate carboxyl (COOH) groups into unstable amine-reactive esters. This intermediate can be stabilized by N-hydroxysulfosuccinimide

Mag = 100 x EHT = 20.00 kV

100 mm

Detector = BSE Date: 22 Jan 2004

Figure 4 ESEM image of the pore structure of the tubular CG scaffold fabricated via spinning method. Scale bars: 1 mm (tube), 100 mm (wall). Reprinted with permission from Harley, B. A.; Hastings, A. Z.; Yannas, I. V.; Sannino, A. Biomaterials 2006, 27, 866–874.

Collagen–GAG Materials (NHS), leading to the formation of stable amide crosslinks.59 Increasing the ratio of EDAC and NHS to COOH increases the degree of crosslinking.56,59 Both DHT and EDAC methods maintain the integrity of the open pore structure of the material by introducing crosslinks within the CG struts only.56

2.216.2.3. CGCaP Scaffold Fabrication In addition to solidification-induced structural modifications, the CG construct can be chemically modified to create collagen composites for specific regenerative medicine applications. Mineralized CG scaffolds are of particular interest for orthopedic applications because of their potential to mimic the native biochemistry of bone: interpenetrating collagenous matrix (organic) and CaP (mineral) content. The addition of a mineral phase to the classic CG scaffold archetype allows for the development of materials with the requisite biochemical and biomechanical properties for bone and osteochondral tissue engineering. Mineralized CG scaffolds have been created via two distinct mechanisms: coating a fully formed CG scaffold with a mineral shell, or synthesizing a mineralized CG chemistry in liquid phase prior to freeze drying. Mineralized CG materials have been fabricated via a surface coating process.60 Here, CG scaffolds were initially synthesized using conventional freeze-drying and then sequentially soaked in phosphate (NaNH4HPO4) and calcium (CaCl2) solutions. This treatment improved construct compressive modulus nearly 70-fold, from 0.3 to 31 kPa, while maintaining high porosity (95%).60 Mineralized CG (CGCaP) materials for bone and osteochondral tissue engineering can also be fabricated from a triple coprecipitate of collagen, GAG, and calcium phosphate (CaP). These scaffolds are synthesized from a suspension

(a)

Meg = 20 x EHT = 15.00 kV

SEM

1 mm

285

produced via a concurrent mapping technique that allows for simultaneous modulation of CG suspension pH as well as the CaP mass fraction (0–80 wt%), a range that covers a host of physiologically relevant tissues including subchondral bone (75%) and cartilage. This method is particularly powerful because it does not require the use of harmful titrants or additives that would be difficult to dissociate from the freezedried matrix.61–63 Concurrent mapping uses 3D maps of pH and CaP mass yield derived from experimental data to determine the concentration of CaP needed to achieve a specific mass fraction, given the concentrations of collagen and GAG as well as the pH of the suspension.62 This method yields nanocrystallite CaP content, precipitated within the CG struts, rather than coating the CaP on the CG content. This approach also enables the phase of the CaP mineral to be adjusted via hydrolytic conversion from the original brushite phase to either octacalcium phosphate or apatite, two biologically relevant CaP phases, without altering the pore structure of the scaffold.61 Conversion from brushite to apatite results in a scaffold that is less soluble and has altered bone-bonding effectiveness.64–66 Modified freeze-drying profiles have been developed to enable fabrication of CGCaP scaffolds with homogeneous microstructure (85  3% porosity), highly interconnected pores, and tunable mean pore sizes from 56 to 1085 mm (Figure 5).61 Mineralized CG scaffolds, produced by coating the CG scaffold with a mineral component or by precipitating nanocrystallites of CaP mineral into the CG content prior to fabrication, successfully mesh the advantageous microstructural and biochemical properties of standard CG scaffolds with the mechanical integrity and bone-bonding properties of calcium phosphate mineral in a manner reminiscent of the chemical composition of native bone67 to produce a promising class of scaffolds for orthopedic application.

Detector = BSE Date: 2 Jul 2004

Ca

P

(b)

Figure 5 (a) SEM and mCT micrographs of CGCaP scaffold microstructure displaying an open pore structure with interconnected pores. (b) EDX analysis of CGCaP scaffold shows uniform distribution of both calcium (Ca) and phosphorous (P) throughout the scaffold. Scale bars: 1 mm. Reprinted with permission from Harley, B. A.; Lynn, A. K.; Wissner-Gross, Z.; Bonfield, W.; Yannas, I. V.; Gibson, L. J. J. Biomed. Mater. Res. A 2010, 92, 1066–1077.

286

Materials of Biological Origin

freeze-drying mold, but then incorporating a processing step to enable partial diffusive mixing between the two suspensions near their interface. Once the region of interdiffusion was created between the suspension layers, freeze-drying was used to form the final multicompartment scaffold microstructure (Figure 6).69 The continuous interface between scaffold regions is critical for the recapitulation of native interfacial physiology for tissues such as cartilage and tendon.69 The differential chemistry, microstructure, and mechanics of the osseous and cartilaginous compartments enable these layered scaffolds to exhibit compressive deformation behavior that mimics behavior observed in natural articular joints.69 These layered scaffolds are currently the subject of numerous in vitro and in vivo experiments for various orthopedic tissue engineering applications.

2.216.2.4. Multicompartment CG Scaffold Fabrication The CGCaP scaffold technology has been used as the basis for creating multiphase collagen scaffolds for the repair of interfacial tissues, notably osteochondral defects. A liquid phase cosynthesis method was created to fabricate multicompartment CG scaffolds. The multicompartment scaffold was designed to be a single biomaterial construct containing multiple regions (‘compartments’), each with distinct microstructural, chemical, and mechanical properties that are connected via a continuous interface between regions. The initial multiphase scaffold developed via this approach contains an osseous compartment for subchondral bone regeneration (type I collagen, chondroitin 6-sulfate, and CaP) and a cartilagenous compartment for cartilage regeneration (type II collagen and chondroitin 6-sulfate). A key feature of this design is that by avoiding discrete interfaces between scaffold compartments that would require external fixation (sutures, glue, etc.) these materials reduce the potential for stress concentrations and resultant delamination at the interface.68 The continuous interface is created by layering the cartilaginous compartment and the osseous compartment suspensions in a conventional

2.216.2.5. CG Hydrogels CG hydrogels have been developed as alternative matrices for soft tissue engineering. CG hydrogels are typically composed of type I collagen, GAG, acetic acid, sodium hydroxide, and cell

Type II CG Conc

Type I CGCaP Interdiffusion

Conc

Freeze drying

•

•

• •

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

•

Porous, layered scaffold

Natural articular joint

Figure 6 Schematic of liquid phase cosynthesis method used to produce layered osteochondral CG scaffold with soft, continuous interface. Nonmineralized, type II collagen–glycosaminoglycan (Type II CG) suspension is layered on top of mineralized, type I collagen–glycosaminoglycan (CGCaP) suspension, allowed to interdiffuse, and then freeze-dried to produce a porous, layered scaffold that mimics natural articular joint physiology. Reprinted with permission from Harley, B. A.; Lynn, A. K.; Wissner-Gross, Z.; Bonfield, W.; Yannas, I. V.; Gibson, L. J. J. Biomed. Mater. Res. A 2010, 92, 1078–1093.

Collagen–GAG Materials culture media.70,71 As collagen fibers are more dispersed within a gel compared to a porous scaffold, the addition of GAG imparts additional mechanical strength, significantly increasing tensile modulus (56 kPa vs. 40 kPa) and maximum stress (7 kPa vs. 5 kPa).70 Other work has combined methacrylated HA hydrogels with semi-interpenetrating networks (semi-IPN) of collagen (see Chapter 2.214, Hyaluronic Acid).72,73 Both methacrylation of HA and addition of collagen significantly improve mechanical properties (compressive modulus and fracture stress) and bioactivity (fibroblast adhesion). In addition, these hydrogels can be modified with microscale features using lithographic techniques, demonstrating their potential as tissue engineering constructs.72 This type of hydrogel has also been explored as a neural tissue engineering template. Schwann cells remained viable in a collagen–HA hydrogel for 2 weeks and secreted nerve growth factor as well as brainderived neurotrophic factor. The addition of laminin to the hydrogel increased secretion of these factors.73

2.216.2.6. CG Membranes Nanoporous CG membranes have recently been developed as a bioactive, mechanically robust 2D substrate. CG membranes are synthesized by filtering and drying the classic type I CG suspension74 (Figure 7). CG membranes have been created with a range of thicknesses (25–240 mm) but with relatively consistent relative densities (0.68  0.06–0.91  0.09). These membranes have been found to be mechanically isotropic (in plane) and have demonstrated tensile moduli as high as 47 MPa (hydrated) (Caliari, Ramirez, and Harley, Unpublished data). Ongoing work is looking to improve fabrication processes to enable enhanced control over membrane thickness, isotropy/anisotropy, relative density, mechanical properties, and metabolite permeability.

2.216.3.

287

Characterization of Collagen–GAG Materials

Experimental characterization of properties such as chemistry, pore morphology (size/shape), relative density, permeability, and mechanics of a number of CG scaffold variants has been undertaken; such experimental characterization has also been supplemented by modeling approaches allowing for a more complete understanding of the microenvironmental cues presented within the CG structure to distinct cells or populations of cells.

2.216.3.1. Microstructure: Experimental Measurement and Modeling CG microstructure (pore size and morphology) is typically assessed using a stereological approach. Longitudinally and transversely oriented scaffold samples are embedded in glycolmethacrylate, sectioned, stained, and observed using optical microscopy. Using a linear intercept approach in any standard image analysis software package, a best-fit ellipse representative of the mean pore morphology is created. Dimensions of the ellipse (major and minor axis) are used to calculate an equivalent mean diameter and aspect ratio.10,44,75 Microcomputed tomography (mCT) has also been utilized to analyze pore microstructure,10,44,75 though often requires significant use of contrast agents to allow sufficient visualization of the nonmineralized CG content. Most recently, mCT analysis has been used to create a computational model of the CG microstructure for use in analysis of shear stresses on cells in dynamic culture conditions (Figure 8).76 Cellular solids modeling approaches have proven to be a useful tool in the description and characterization of CG scaffold pore geometry and microstructural properties (pore shape and specific surface area).77 The complex geometry of foams (and scaffolds) is difficult to model exactly; instead, dimensional arguments can be used to model salient microstructural features without incorporating exact cell geometries using the cellular solids modeling framework. CG scaffolds have been primarily modeled as a low density, open-cell foam using a

x = 31.8 mm

1 mm

´1000

10 mm

10

30

SE I

Figure 7 SEM of CG membrane microstructure. Mean thickness: 22.9  2.6 mm; mean porosity: 32.1  5.9%. Scale bar: 10 mm.

Figure 8 mCT scan of CG scaffold utilized in CFD simulations. Scale bar: 1 mm. Reprinted with permission from Jungreuthmayer, C.; Jaasma, M. J.; Al-Munajjed, A. A.; Zanghellini, J.; Kelly, D. J.; O’Brien, F. J. Med. Eng. Phys. 2009, 31, 420–427.

288

Materials of Biological Origin

tetrakaidecahedral (14-sided polyhedron) unit cell. Modeling microstructural features of CG scaffolds in this manner is possible because the mean pore structure of it and a variety of other low-density, open-cell foams has been observed to have a number of consistent features, notably approximately 14 faces per unit cell, 5.1 edges per face, and vertices that are nearly tetrahedral.77,78 The tetrakaidecahedron packs to fill space, nearly satisfies minimum surface energy requirements, and approximates the structural features of many experimentally characterized low density, open-cell foams (Figure 9).77,79 Application of modeling approaches using the tetrakaidecahedral unit cell with CG scaffolds has led to estimations of a number of key microstructural features of the CG scaffolds, many of which will be discussed throughout this chapter. The first microstructural element described using cellular solids approaches was the scaffold specific surface area (SA/V), the total surface area divided by the volume of the scaffold. Specific surface area describes the relative amount of surface available for cells to attach, and has been shown to be an integral factor affecting overall scaffold bioactivity.9,10 For a low density, open-cell foam with an interconnected pore structure and edges of circular cross-section modeled using the discussed approaches, specific surface area is related to the mean pore size (d) and the relative density (r*/rs):10,77   SA 10:15 r* 1=2 ¼ V d rs

[1]

2.216.3.2. Permeability: Experimental Measurement and Modeling Permeability of tissue engineering scaffolds is a critical design parameter to be controlled as it influences the diffusion of cytokines, nutrients, and waste throughout the scaffold prior to material vascularization; further, biomaterial permeability

t

I

has also been shown to significant impact cell migration processes and scaffold biodegradation rates, among others.80 Material permeability also affects fluid pressure fields and shear stresses within the construct, additional potential stimuli for functional adaptation.81 Permeability of tissue engineering scaffolds is dictated by a variety of microstructural characteristics including porosity, pore size and orientation, pore interconnectivity, fenestration size and shape, specific surface area, and applied strain. Permeability of low porosity foams (porosity <90%) has previously been modeled and measured experimentally.82 However, such models are not applicable to highly porous materials that are typically used in tissue engineering, such as CG scaffolds (porosity >99%); here, material permeability has been conventionally measured empirically with no associated modeling framework to describe the results. A series of CG scaffolds with constant relative density (r*/rs ¼ 0.006) were used not only to characterize both CG scaffold permeability as a function of pore size (96–151 mm) and applied compressive strain (0–40%), but also to calibrate a cellular solids-based model of the permeability of lowdensity, open-cell foams. As with specific surface area, a low density, open-cell foam cellular solids model was successfully created to describe the permeability (k) of the entire class of CG scaffolds as a function of the scaffold mean pore size (d), percent compression (applied strain: e), relative density (r*/rs), and a single dimensionless system constant (A0 ):83  2   d 1  r∗ 3=2 0 k¼A ð1  eÞ [2] rs 2:785 Experimental and modeling analyses showed that CG scaffold permeability increases with increasing pore size and decreases with increasing compressive strain (Figure 10). The experimentally measured and theoretically predicted scaffold permeability values showed excellent agreement with the need for only a single fitting constant (A0 ) to describe the entire class of scaffolds, again validating the use of the model tetrakaidecahedral unit cell for describing microstructural features of CG scaffolds.83 These results also suggested that cellular solids modeling techniques can be used to predict scaffold permeability under a wide range of physiological loading conditions and experimentally relevant pore microstructures (size and relative density).

2.216.3.3. Mechanical Properties: Experimental Measurement and Modeling

d Figure 9 Tetrakaidecahedral unit cell. d, pore diameter; l, strut length; t, strut thickness. Reprinted with permission from Harley, B. A. C.; Kim, H. D.; Zaman, M. H.; Yannas, I. V.; Lauffenburger, D. A.; Gibson, L. J. Biophys. J. 2008, 95, 4013–4024.

The mechanical properties of a range of biomaterials have been shown to significantly impact cell behaviors including adhesion, proliferation, differentiation, and overall bioactivity.4,7,16–25 The mechanical properties of a number of CG scaffold variants have been tested in tension and compression; further, the elastic modulus of the individual struts that define the scaffold network has also been characterized56 (Table 2). The mechanical behavior of CGCaP scaffolds under compression as well as the modulus of the individual mineralized struts has also been evaluated61,84 (Tables 2 and 3). These results offer insight to the mechanical properties of a standardized set of CG materials. Cellular solids modeling has been shown to be a powerful tool to aid the analysis of the mechanical properties of CG biomaterials. The cellular nature of CG scaffolds requires that

Collagen–GAG Materials

289

2E-10 1.8E-10

0% Meas

0% Calc

14% Meas

14% Calc

29% Meas

29% Calc

40% Meas

40% Calc

1.6E-10

K (m4 N-1s-1)

1.4E-10 1.2E-10 1E-10 8E-11 6E-11 4E-11 2E-11 0 96

110 121 Mean pore size (mm)

151

Figure 10 Comparison between experimental results (Kmeas, solid bars) and the predicted values obtained from the mathematical model (Kcalc, striped bars) for CG scaffold permeability under varying compressive strain (0, 14, 29, 40% strain, left to right for each pore size) for distinct pore sizes (96, 110, 121, 151 mm). Reprinted with permission from O’Brien, F. J.; Harley, B. A.; Waller, M. A.; Yannas, I. V.; Gibson, L. J.; Prendergast, P. J. Technol. Health Care 2007, 15, 3–17. Table 2 Average (mean  std. dev.) mechanical properties of the homogeneous CG scaffold variants (96–151 mm; 0.006 relative density; DHT crosslinking at 105
C for 24 h; hydrated) and the mineralized CGCaP scaffold (202 mm; 0.038 relative density; DHT crosslinking at 105
C for 24 h; hydrated) Property

Hydrated CG scaffold

Hydrated mineralized (50%) CG scaffold

E* sel eel Ds/De Es

0.208  0.041 kPa 0.021  0.008 kPa 0.10  0.04 0.092  0.014 kPa 5.28  0.25 MPa

6.20  1.51 kPa 0.547  0.089 kPa 0.0960  0.0302 3.05  0.392 kPa 78.8  8.5 MPa

Source: Harley, B. A.; Leung, J. H.; Silva, E.; Gibson, L. J. Acta Biomater. 2007, 3, 463–474; Kanungo, B. P.; Silva, E.; Van Vliet, K.; Gibson, L. J. Acta Biomater. 2008, 4, 490–503.

materials mechanical characterization must be performed at multiple scales: at the level of the whole material (macro) as well as at the level of the individual struts that define the scaffold structure (micro). The significant porosity of the CG scaffolds often results in multiple orders of magnitude difference in these mechanical properties. The theory predicts that stress–strain curves for low-density, elastomeric open cell foams in compression are characterized by three distinct regimes: a linear elastic regime (strut bending), a collapse plateau regime (struts buckling and pore collapse), and a densification regime (complete pore collapse) (Figure 11).77 In tension, the initial linear elastic response is typically the same as is observed in compression for small strains; however, as the strain increases, the struts become increasingly oriented in the direction of applied tension, increasing the material stiffness until tensile failure.77 Experimentally, the CG scaffold variants exhibited

Table 3 Elastic moduli of individual scaffold struts within hydrated CG and CGCaP scaffolds crosslinked via DHT and EDAC/NHS techniques Crosslinking treatment

Scaffold strut elastic moduli (MPa)

Relative elastic modulus

Mineralized (50%) scaffold strut elastic moduli (MPa)

Relative elastic modulus

Uncrosslinked DHT105/24 (Standard) DHT120/48 EDAC1:1:5 EDAC5:2:5 EDAC5:2:1

3.9  0.20 5.28  0.25

0.74 1.0

49.1  5.3 78.8  8.5

0.62 1.0

5.7  0.30 10.6  0.50 11.8  0.56 38.0  1.8

1.08 2.0 2.24 7.2

No data No data No data 193.2  20.7

No data No data No data 2.45

Source: Harley, B. A.; Leung, J. H.; Silva, E.; Gibson, L. J. Acta Biomater. 2007, 3, 463–474; Kanungo, B. P., Silva, E., Van Vliet, K. & Gibson, L. J. Acta Biomater. 2008, 4, 490–503.

the theoretically predicted stress–strain behavior with distinct linear elastic, collapse plateau, and densification regimes (Figure 12). Experimental results show that CG scaffolds with equiaxed pores (mean pore sizes: 96–151 mm) as well as CGCaP scaffolds (50 wt% CaP) are microstructurally and mechanically isotropic.56,84 CG materials demonstrated compressive moduli of order 30 kPa (dry) and 200 Pa (hydrated), independent of pore size, while CGCaP scaffold demonstrated compressive moduli of order 750 kPa (dry). The independent effects of variables including hydration level, pore size, and crosslink density on the mechanical properties were also determined (Tables 2 and 3).56,84 The Young’s modulus of the individual (CG and CGCaP) scaffold struts (Es) that define the scaffold microstructure was measured via AFM (Figure 13).56,84

290

Materials of Biological Origin

120 s (kPa)

Stress, s

Densification

E*

80

Densification

Linear elastic

160

Elastomeric foam Compression

Collapse plateau

40

s *el

Plateau (elastic buckling) 0

Linear elasticity (bending)

0.2

0

Strain, e

80 60 40

Data for a wide variety of open-cell foams give approximate values as C1 1 and C2  0.05.77 Interestingly, these equations show that the strain at which cell collapse by buckling occurs, e*el is a constant value equal to C2 =C1 ¼ 0:05 and that both E* and s*el are expected to be independent of the pore size.77 Experimental results of scaffold mechanical characterization and low density, open-cell foam model predictions for the scaffold E* and s*el (eqns [3] and [4]) for the standard series of isotropic CG scaffold variants (r*/rs: 0.006; mean pore size: 96–151 mm) showed good agreement. Notably, CG scaffolds showed distinct linear elastic, collapse plateau, and densification regimes, and CG scaffold modulus was found to be independent of scaffold pore size, but highly dependent on scaffold r*/rs.56 The effect of relative density on the mechanical properties of both mineralized and nonmineralized CG scaffolds has recently been evaluated.45,46,56,84 Nonmineralized CG scaffolds over a range of relative densities (0.0062–0.0239) showed significant increases in compressive modulus and strength as well as tensile modulus with increasing relative density (Table 4).45,56 While the model correctly predicted the relationship between Es and E*, it over-predicted E* at higher relative densities. This is likely due to difficulties in fabricating CG scaffolds with high solids content, which often leads to areas of microstructural heterogeneities.56

0.6

0.8

1

e 100

Figure 11 Uniaxial stress–strain curve for an elastomeric cellular solid in compression showing linear elastic, collapse plateau, and densification regions as well as the linear elastic modulus (E*) and elastic collapse stress (sel ). Reprinted with permission Gibson, L. J.; Ashby, M. F.; Harley, B. A. Cellular Materials in Nature and in Medicine; Cambridge University Press: Cambridge, 2010.

Collapse plateau E*

s*el

Δs/Δe

20 e*el 0 0

0.1

0.2

0.3 e

(b)

0.4

0.5

1800

s (Pa)

Using cellular solids theory, the Young’s modulus (E*) and elastic compressive strength (s*el , also called the compressive plateau stress) of low density, open-cell foams such as CG scaffolds are predicted to depend on the foam relative density, r*/rs, the Young’s modulus of the solid from which the foam is made, Es (termed the strut modulus), and a constant of proportionality (C1 and C2) related to the cell geometry:77 For elastomeric cellular solids, E* and s*el are  2 r* Es [3] E* ¼ C1 rs  2 r* s*el ¼ C2 Es [4] rs

0.4

(a)

Linear elastic

eD 1

0

Scaffold starts slipping from grips

1200

600

0 0 (c)

0.05

0.1

0.15

0.2

0.25

e

Figure 12 Characteristic stress–strain curves observed for the CG scaffold under compression for the entire tested strain range (a) e: 0–0.95, and for a subregion of a hydrated CG scaffold (b) e: 0–0.60. Linear regressions of the linear elastic and collapse plateau regimes are used to calculate the linear elastic modulus (E*) and the collapse plateau modulus (Ds=De).The elastic collapse stress and strain are sel , eel . (c) Characteristic stress–strain curve observed for the (hydrated) CG scaffold variants under uniaxial tensile testing in the plane of the scaffold sheet. Reprinted with permission from Harley, B. A.; Leung, J. H.; Silva, E.; Gibson, L. J. Acta Biomater. 2007, 3, 463–474.

Likewise, mineralized CG scaffolds with relative densities from 0.045 to 0.187 had significantly increased compressive modulus and strength (both dry and hydrated) with increasing relative density (Table 4).46 Meanwhile, modeling consistently over-predicted E* and s*el for the mineralized CG scaffolds. These discrepancies can be explained by the presence of voids, cracks, and disconnections within the scaffold microstructure.84 Experimentally, CGCaP scaffolds with higher

Collagen–GAG Materials

291

Mechanical properties (mean  standard deviation) of CG (relative density 0.0062–0.0239) and CGCaP (0.045–0.187) scaffolds in the dry

Table 4 state Scaffold

r*/rs

E* (kPa)

s* (kPa)

e* (%)

Ds/De (kPa)

CG 1X CG 2X CG 3X CG 4X CGCaP 1X CGCaP 2X CGCaP 3X CGCaP 4X

0.0062  0.0008 0.0120  0.0016 0.0198  0.0014 0.0239  0.0021 0.045  0.002 0.098  0.004 0.137  0.002 0.187  0.027

32  6 61  5a 97  17aa 127  23aa 780  95 3156  760b 6500  1270bb 3660  820b

5.0  1.1 11.8  2.4a 15.6  3.1aa 19.0  3.9a 39  10 132  14b 242  32bb 275  50bb

15.8  3.5 18.9  2.5a 15.9  1.2 14.6  3.5 5.00  1.00 4.00  1.41 4.00  1.00 4.38  1.15

10.6  1.0 43.6  2.5a 54.8  9.3aa 43.3  5.6aa 480  35 1300  200b 2680  610bb 1875  525bbb

Superscripts a, aa, aaa (CG) and b, bb, bbb (CGCaP) denote statistically significant differences between groups. For example, E* for the CG 1X, 2X, and 3X scaffolds are significantly different from each other while 3X and 4X are not. Source: Kanungo, B. P.; Gibson, L. J. Acta Biomater. 2009; Kanungo, B. P.; Gibson, L. J. Acta Biomater. 2009, 5, 1006–1018

AFM probe cantilever

X Super glue Scaffold strut

Glass slide

d1 D

mineral content (75 wt% CaP) displayed inferior mechanical properties to the 50 wt% CaP scaffolds.84 This behavior is possibly caused by the retention of residual moisture, even after extensive freeze-drying, due to the dehydration of 33% of the CaP mineral content from brushite to monetite in the 75 wt% CaP scaffold.84

(a)

d2

2.216.3.4. CG Scaffold Degradation Kinetics and Byproducts

600

nm

d

Unloa d

200

Loa

400

0

-1

0

1 mm

(b)

2

3

dt

70 dmax

Load, P (mN)

50

Pmax

30 S 10 -10 0

(c)

50 100 Displacement, d (nm)

150

Figure 13 (a) AFM experimental setup to assess individual CG scaffold strut modulus. (b) Characteristic load–unload curve for bending tests performed via AFM on individual CG scaffold struts (nonmineralized variant) and (c) (mineralized variant). Reprinted with permission from Harley, B. A.; Leung, J. H.; Silva, E.; Gibson, L. J. Acta Biomater. 2007, 3, 463–474; Kanungo, B. P.; Silva, E.; Van Vliet, K.; Gibson, L. J. Acta Biomater. 2008, 4, 490–503.

The degradation kinetics of tissue engineering scaffolds must be carefully tuned so that the scaffold structure remains intact throughout the duration of the regenerative process to prevent wound contraction, while degrading rapidly enough so as not to impede the healing process.1,2,4,85 This optimization has been termed the isomorphous tissue replacement model, where the biomaterial degradation half life of the scaffold (td) must equal the healing half life of the tissue (th) in which it is implanted. Scaffold degradation that happens too quickly (td > th) or too slowly (td < th) is detrimental to the tissue regeneration process.4 In addition to degradation kinetics, byproducts must also be carefully considered in scaffold design. Many synthetic scaffolds degrade to cytotoxic byproducts, limiting their in vivo applicability.1,2 However, CG scaffolds, like most scaffolds made of natural materials, break down into harmless byproducts already found in abundance in the body.31 The degradation kinetics of CG scaffolds can be tuned by altering chemical composition (collagen to GAG ratio), relative density, or most commonly through differential crosslinking (see previous section for more details), where increasing crosslinking densities are used to slow scaffold degradation.1,86 Pek et al. have evaluated the degradative process of CG scaffolds treated with one of four conventional crosslinking techniques: noncrosslinked (Nx), DHT (Dx), EDAC (Ex), and DHT/EDAC (DEx) in real time.86 Here, scaffolds displayed an inverse relationship between swelling ratio and crosslink density, as is characteristic of random polymeric materials. ESEM observation of Nx and Dx scaffolds degraded with collagenase showed micropitting and eventual collapse of scaffold struts, leading to complete loss of mechanical integrity after a week while Ex and DEx samples maintained their structure and showed only minor micropitting.86 Degradation by chondroitinase led to distinct structural (but not mechanical) changes for the Nx and

292

Materials of Biological Origin

Dx scaffolds that were predicted to be more prone to GAG removal: scaffold struts were observed to swell and thereby reduce the open pore interconnectivity of the microstructure. This effect was not nearly so pronounced for the Ex and DEx variants.86

2.216.4.

In Vitro Applications

2.216.4.1. Cell Attachment and Viability CG tissue engineering scaffolds must have a mean pore size that is large enough for efficient metabolite exchange and cell infiltration, but small enough to provide ample surface area (ligands) to achieve adequate cell attachment. Distinct optimal pore sizes have been hypothesized for different cells, biomaterial chemistries, and clinical applications. CG scaffolds with uniform, well characterized microstructures (pore sizes 85–325 mm) have been used as a platform to begin to investigate the relationship of pore microstructure and scaffoldspecific surface area with cell attachment, viability, and the uniformity of their distribution throughout the material.9,10,45 These scaffolds were seeded with MC3T3-E1 mouse clonal osteogenic cells and then cultured for anywhere between 1 and 7 days to determine viable cell number and distribution. Increased early cell adhesion (24–48 h) was correlated with increased scaffold specific surface area (Figure 14).10 Modulating scaffold relative density (0.0062–0.0239) to alter scaffold SA/V also gave a linear relationship between scaffold specific area and MC3T3-E1 cell attachment (R2 ¼ 0.97 at 24 h).45 These results indicate a strong correlation between specific surface area and early MC3T3-E1 cell attachment for this set of CG scaffolds. However, for longer culture periods (7 days), a higher number of cells are observed in scaffolds with larger pores (>300 mm). This result has been hypothesized to be due to the competing influence of scaffold proliferation and metabolic support via diffusion, both of which increase with increased scaffold pore size: larger pore sizes allow for easier cell proliferation and migration throughout the entire scaffold, especially at later time points (7 days). These results have helped confirm that scaffolds with larger pore sizes, 325 mm, may be more optimal for long-term bone tissue engineering applications.9

Cell attachment (%)

60

45

30

15

R2 = 0.95

R2 = 0.91

24 h postseeding 48 h postseeding

0 0.004

0.005

0.006

0.007

0.008

Specific surface area (mm-1)

Figure 14 MC3T3-E1 cell attachment plotted against specific surface area showing a strong linear relationship at 24 (gray line) and 48 h (black line) postseeding. Reprinted with permission from O’Brien, F. J.; Harley, B. A.; Yannas, I. V.; Gibson, L. J. Biomaterials 2005, 26, 433–441.

Fabrication flexibility also has enabled investigation of the effect of CG scaffold chemical composition on initial cell attachment and viability. MC3T3-E1 attachment and viability have been shown to depend on scaffold chemical composition (collagen and GAG concentration)12 and mechanical properties.11 Here, scaffolds were fabricated over a range of concentrations of collagen (0.25–1%, w/v) and GAG (0–0.088%, w/v), seeded with cells, and cultured for 6 h to 7 days. Lower concentrations of collagen and GAG showed high initial cell attachment, likely because of a temporary increase in surface area due to loss of structural integrity (pore collapse).12 However, the higher concentration of surface ligands on the 1% collagen scaffolds supported the highest cell attachment after 7 days; 0.088% GAG scaffolds also had better cell infiltration than their counterparts.

2.216.4.2. Cell Contraction Cell-mediated contraction plays an integral role in many physiological and pathological processes, notably organized contraction and scar formation during wound healing. One of the key functions of CG and other tissue engineering scaffolds is to block wound contraction to prevent scar tissue proliferation, instead allowing for regeneration of functional tissue.4 This objective requires a comprehensive understanding of the forces and kinetics associated with cell contractile processes within the scaffold microstructure so that scaffolds can be designed with the requisite mechanical integrity to withstand contraction for a critical period of time. Most studies of cell contraction in vitro have used 2D substrates. Here, contractile force is calculated as a function of substrate deformation due to a population of cells and the elastic modulus of the substrate.87–93 This technique has been used to show that substrate elastic modulus significantly affects cell behaviors such as proliferation, differentiation, migration speed, directional persistence, and applied traction forces.21,87,93–102 These methods yield estimates of fibroblast contractile forces on the order of hundreds of nanoNewtons.90,93,95 However, these results are not likely comparable to in vivo contractile forces, given the fact that the shape and cytoskeletal organization of cells on these 2D substrates (amorphous, spread) are significantly different than that of fibroblasts observed in wound sites within the natural ECM (elongated spindle-shaped).75,103–105 CG scaffolds have been used in a number of investigations to quantify the macroscopic (bulk construct) and microscale (individual strut) contractile behavior of cell populations as well as the relationship between distinct integrin–ligand interactions, cytoskeletal organization, and cell contraction. Macroscopically, the average contractile force generated by a known cell population can be calculated as a function of the gross deformation of the CG scaffold by that cell population and the scaffold’s elastic modulus. Total contractile force generated by a population of dermal fibroblasts within the CG scaffold was observed to reach an asymptote of 1.0  0.2 nN, independent of the number of cells seeded, with an associated time constant of 5.2  0.5 h.17 When the stiffness of the scaffold system was changed, fibroblasts were observed to apply differential levels of strain but a constant average force to the CG scaffold, suggesting that dermal fibroblasts apply contractile forces independent of the local microenvironment (Figure 15).106

293

Collagen–GAG Materials Similar analysis has been performed for chondrocytes107 and Schwann cells.108 Fibroblasts displayed significant cytoskeletal reorganization during contraction within the CG scaffold; initially rounded after seeding, they were observed to form attachments to single struts as well as between multiple CG struts and elongated over time into spindle-shaped cells. The average aspect ratio (maximal cell length/maximal cell thickness) of the cells increased asymptotically from 1.4 to 2.8 during the first 15 h in the scaffold (Figure 15) with a time constant similar to that observed for contractile force generation.17,109 In addition, the role of integrin–ligand complexes in the contraction of CG scaffolds has been examined using a culture force monitor (CFM) via specific inhibition of integrins and distinct cell adhesion proteins (fibronectin, vitronectin, and collagen).110 Initial dermal fibroblast attachment and force generation (0–5 h) was found to be sequentially mediated primarily by fibronectin followed by vitronectin. Attachment and force profiles at later time points (6–20 h) were shown to be dominated by cell–collagen interactions.110

However, two significant assumptions are made with this analysis. First, the deformation of the cell-seeded, rectangular CG scaffold sample was measured in only one direction. Second, the fraction of contractile cells versus the total cell population within the scaffold was not characterized. As the average force per cell was calculated using the assumptions that all cells were contracting in a single direction and that all cells were contracting at the same time, the calculated average contractile force of 1.0  0.2 nN is a lower bound. CG scaffolds have also been used to quantify the contractile forces generated by individual cells within a 3D ECM analog. Phase contrast microscopy has been used to generate timelapse images of dermal fibroblasts differentially migrating through the scaffold, or contracting the local scaffold microstructure, resulting in buckling of the strut(s) to which the cell was attached (Figure 16).17,106,109 Improved mechanical and

A

31 m

41 m

C

57 m

2 h 10 m

B

3 9

Extrapolated

2

6

Force (mN)

Average aspect ratio

12

C

3

C

Contractile force Average aspect ratio

1

0

12

(a)

24 Time (h)

36

48

0

(a)

Fc

Fc

Force per cell (nN)

3 l

t

1

0 (b)

Es

2

(b)

0.07 N m-1 1.4 N m-1 10.7 N m-1

0

6

12 Time (h)

13

24

Figure 15 (a) Plot of the fibroblast aspect ratio and generated contractile force with time in the CG scaffold. The average cell aspect ratio increased up to 15 h postseeding in a manner similar to the total force generated by the cell population. (b) Plot of force per cell over time for different system stiffnesses. The displacement developed per cell increased as the system stiffness decreased but the force developed per cell was independent of the system stiffness. Reprinted with permission from Gibson, L. J.; Ashby, M. F.; Harley, B. A. Cellular Materials in Nature and in Medicine; Cambridge University Press: Cambridge, 2010.

Fc

Fc

Figure 16 (a) Time lapse images of an individual dermal fibroblast within the CG scaffold. The sequence of images shows a dermal fibroblast (arrow A) elongating and deforming the surrounding scaffold struts (arrows B). Several struts are deformed over time (arrows C). Time, in hours and minutes, after cell seeding is indicated in the top right of corner of each image. Scale bar: 50 mm. (b) Schematic of a single cell applying a critical buckling load (Fc) to a scaffold strut within an idealized CG scaffold network (left). The surrounding struts inhibit rotation of the ends of the buckling strut (middle). A simplified model of CG scaffold strut buckling with the appropriate boundary conditions: the scaffold strut is restrained at its ends by a rotational spring that represents the surrounding strut network (right). Reprinted with permission from Harley, B. A.; Freyman, T. M.; Wong, M. Q.; Gibsony, L. J. Biophys. J. 2007, 93, 2911–2922.

294

Materials of Biological Origin

microstructural characterization of the individual CG scaffold struts75 and the application of cellular solids theory enabled the use of modified Euler column buckling relationships in order to quantify the magnitude of individual cell contractile forces within the CG scaffold.75 The magnitude of the cellmediated contraction forces generated by individual dermal fibroblasts within the CG scaffold was calculated from the observed strut deformation by individual dermal fibroblasts, strut dimensions and mechanical properties, and previous experimental and theoretical work describing the mechanics and collapse of open cell foams.77 The critical load (Fc) at which a scaffold strut of length ls, average pore size d, Young’s modulus Es, and second moment of area I (strut geometry was approximated as a cylindrical fiber, I ¼ pd4/64) buckles can be calculated by Euler’s formula and the hydrostatic compression end restraint (n2 ¼ 0.34):75,77,111 Fc ¼

n2 p2 Es I l2s

[5]

The contractile force generated by individual dermal fibroblasts that were able to buckle a CG scaffold strut was calculated to range between 11 and 41 nN, with an average contractile force (Fc) of 26  13 nN.75 The upper limit of fibroblast contractile capacity in CG scaffolds was established from data on a cell that was unable to buckle the strut it was attached to because that strut was significantly thicker than the average strut, thereby increasing its flexural rigidity (EsI) and buckling load (Fc). Analysis of this strut’s microstructure indicates that the force required to buckle the strut was approximately 450 nN. These results suggest that while dermal fibroblasts can easily develop the 25 nN force required to buckle conventional CG scaffold struts, they are unable to develop contractile forces at the level of 450 nN.75 While developed with a well-characterized CG scaffold system, this technique can be used to study individual cell contraction and cell– scaffold interactions within a wide variety of tissue engineering scaffolds.

2.216.4.3. Cell Motility Cell motility plays an integral role in many physiologic and pathologic processes, notably organized wound contraction and fibroblast and vascular endothelial cell migration during wound healing,4 metastatic tumor cell migration,112 stem cell mobilization and homing,113,114 and tissue remodeling.4 Cell migration is a complex process modulated by a range of spatiotemporally presented biochemical and biophysical signals, both extracellular and intracellular.115,116 Studies of cell motility on 2D substrates have led to an improved understanding of how surface features, particularly substrate stiffness, affect migration through changes in cytoskeletal organization and applied traction forces.98–100 However, 3D biomaterials enable the investigation of the complex set of biophysical and biochemical signals that can affect motility in a 3D environment that better resembles physiologically and pathologically relevant conditions. Understanding of the microenvironmental factors that govern cell motility is also critical for acellular biomaterials (like CG scaffolds) because they must be able to induce rapid cellular invasion.

Cell motility in CG scaffolds relies on a phenomenon known as contact guidance. As CG scaffold pore sizes are significantly larger than the characteristic length of the fibroblasts, cells are forced to migrate along scaffold struts. External microstructural and mechanical stimuli from the scaffold provide physical cues (contact guidance) to regulate cell motility and other cell behaviors.2 NR6 mouse fibroblasts were seeded into CG scaffolds with pore sizes ranging from 96 to 151 mm but with constant mechanical properties (E* ¼ 208  41 Pa), and single cell migration paths were tracked using 3D timelapse confocal microscopy.7 Cell dispersion (Wind-Rose plot, Figure 17(a)) and motile fraction (Figure 17(b)) significantly decrease with increasing pore size.7 Cell speed was also shown to significantly decrease with increasing scaffold pore size (Figure 17(b)); cell speed is reduced almost by half over the range from 96 to 151 mm, from about 12 to 6 mm h1. Further, as cell speed was calculated for only the motile population, which also decreased significantly with mean pore size, scaffold pore size has an even more significant influence on overall cell motility than is suggested by Figure 17(b) alone. The effect of scaffold stiffness on NR6 fibroblast motility was assessed in a series of DHT- and EDAC-crosslinked CG scaffolds of constant pore size (96 mm) but variable strut modulus (Es: 5.3–38 MPa). Cell migration speed exhibited a subtle biphasic behavior with strut modulus increasing (significantly) from 11 to 15 mm h1 for strut moduli between 5 and 12 MPa and then decreasing (significantly) back to 12 mm h1 for strut moduli of 38 MPa (Figure 17(b)). The effects of pore size and scaffold stiffness on cell migration speed correlated well with previous experimental and computational studies of cell motility in dense, 3D materials.25,117 Unlike with the dense materials used in those studies, cells were not exposed to significant steric hindrances in these porous, CG scaffolds, so the strong dependence of cell motility on pore size was not expected. In addition to satisfying the practical need to better understand cell motility in CG scaffolds to enable more efficient cellular invasion, the fabrication, characterization, and modeling tools recently developed for CG scaffolds enabled a rigorous analysis of the independent effect of microstructural and mechanical scaffold features on cell motility. Modeling techniques were then used in an attempt to better explain the initially inexplicable effect of pore size on cell motility. Cellular solids modeling suggested that while the strut moduli (Es) and scaffold relative density (r*/rs) are constant for the scaffolds of different pore size, scaffolds with a larger pore size have longer and thicker struts (Figure 9). So, while the moduli (Es) of these struts are identical, the second moment of inertia (I) increases with increasing pore size, translating to a larger strut flexural rigidity (EsI) with increasing pore size. Cells are known to apply a constant contractile force to the CG scaffold regardless of the system stiffness106 which suggests that cells probe their local mechanical environment by applying a constant traction force and measuring the resultant substrate deformation.118 These results suggested that strut flexural rigidity, not stiffness, is the more relevant mechanical signal to cells. So even though the struts have a constant elastic modulus, they may ‘feel’ stiffer in scaffolds with larger pore sizes because of an increased resistance to buckling. The ‘apparent’ stiffness, or flexural rigidity (EsI), of the scaffold with the largest pore size (151 mm) was calculated to be greater than that of the scaffold with the

Collagen–GAG Materials

295

Increasing pore size 96 mm

110 mm

121 mm

151 mm

0

−50 −50

(a)

50

y( 0 μm )

0 50−50

m)

x (μ

14

0.8 0.6

*

0.4

*

0.2

12

**

10

**

8

**

6 4 2 0

0 0 (b)

Cell speed (mm h-1)

Motile fraction

1.0

Cell speed (mm h-1)

z (μm)

50

100 120 140 160 Scaffold pore size (mm)

0

*

*

0 101 202 303 404 Scaffold strut stiffness (MPa)

100 Persistence time (min)

Cell speed (mm h-1)

*

100 120 140 160 Scaffold pore size (mm)

14 12 10 8 6 4 2

* 80 60

*

40 20 0

0 (c)

18 16 14 12 10 8 6 4 2 0

0 2 4 6 8 10 12 14 16 Junction density (103 mm-3)

0

100 120 140 160 Scaffold pore size (mm)

Figure 17 (a) 3D Wind-Rose plots of randomly chosen cell tracks showing cell dispersion; cell dispersion decreases as scaffold mean pore size increases. (b) Motile fraction and mean cell speed decrease as scaffold pore size increases. Cell speed shows a subtle biphasic relationship with scaffold strut modulus. (c) Cell speed increases proportionally with scaffold strut junction density. Cell persistence time increases with pore size (increasing distance between strut junctions). Reprinted with permission from Harley, B. A. C.; Kim, H. D.; Zaman, M. H.; Yannas, I. V.; Lauffenburger, D. A.; Gibson, L. J. Biophys. J. 2008, 95, 4013–4024.

smallest (96 mm) pore size by a factor of 6.1. Thus, if strut flexural rigidity was the mechanism by which pore size affected cell motility, cell motility would be expected to decrease for the series of scaffolds of constant pore size but increasing scaffold modulus (changing Es in the flexural rigidity term (EsI)). However, when strut modulus (Es) was increased over a range (between 5.3 and 38.0 MPa; Factor: 7.2; Table 3) that closely approximated the change in strut flexural rigidity with mean pore size (Factor: 6.1), motility did not decrease but instead displayed subtle biphasic behavior (Figure 17(b)).7 After attempting to explain the significant influence of CG scaffold pore size on cell motility using predictions made by the cellular solids modeling regarding local scaffold mechanics, geometric insights from the cellular solids model regarding

strut junctions (points in the scaffold where two or more struts meet) were investigated. The geometry of strut junctions within the scaffold was described as the mean spacing between junctions (Djxn) or the mean junction density (rjxn, the number of strut junctions per unit cell divided by the volume of a unit cell). Djxn and rjxn were calculated from the scaffold mean pore size (d) assuming a tetrakaidecahedral unit cell:7 d 2:785

[6]

11:459 d3

[7]

Djxn ¼ rjxn ¼

Replotting the pore-size-dependent cell speed data against strut junction density, an exceptionally strong correlation

296

Materials of Biological Origin

between cell speed and strut junction density was observed (Figure 17(c)). Further, cells migrating in scaffolds with larger pore sizes, and therefore larger Djxn and lower rjxn, were observed to exhibit greater persistence times, indicating more directional motion along a scaffold strut (Figure 17(c)). In contrast, persistence times of cells migrating in scaffolds with smaller pore sizes and greater rjxn are significantly lower, representative of erratic movement (time turning) that likely occurs more often at junctions when cells encounter higher surface ligand densities and can therefore move with increased speed. These results provided a mechanistic explanation for the initially counter-intuitive observation that cell motility decreases as scaffold pore size increases,7 as well as an excellent example of the types of mechanistic questions that can be answered using the CG scaffold system because of the capacity to create uniform scaffold variants with independently controlled microstructural and mechanical properties and to integrate rigorous characterization and modeling techniques.

2.216.4.4. Growth Factor and Gene Delivery Tissue engineering scaffolds are frequently enhanced through the addition of exogenous factors including growth factors, drugs, and genes. CG scaffolds have been used as a platform to study the combined effects of the scaffold and delivery of growth factors and genes. Recently, marrow stromal cells (MSCs) were transfected with a plasmid encoding for glial cell line-derived neurotrophic factor (pGDNF) and seeded onto standard type I CG scaffolds.119 For optimal plasmid loading, the level of GDNF remained six to seven times higher than a control after 2 weeks of culture, reaching previously described therapeutic levels.120 These results show promise for future brain tissue engineering studies with CG scaffolds. The potential for CG materials as cartilage and meniscus scaffolds has been enhanced using gene delivery. EDAC crosslinking of as little as 10 mg of IGF-1 plasmid to type II CG scaffolds resulted in sustained plasmid release that was controlled by the degradation rate of the scaffold. IGF-1 was overexpressed over a 2-week culture period, resulting in significantly higher amounts of chondrocyte-like cells, GAG, and type II collagen production compared to controls.121 Another study transfected MSCs with a plasmid encoding endostatin, a potent antiangiogenic factor that could potentially prevent unwanted vascularization of cartilage, and cultured on CG scaffolds.122 Endostatin levels, while negligible in control cultures, reached a peak of 13 ng ml1 after 3 days and approached therapeutic levels after 2 weeks.122 MSCs and meniscal cells have also been transfected with a vector for TGF-b1 and cultured in vitro on type I CG scaffolds. Implantation of transfected, cell-seeded constructs following culture led to increased matrix synthesis and meniscal gene expression over controls.123

2.216.4.5. Modifying Gene Expression CG scaffolds have been used to compare gene expression profiles of fibroblasts cultured on 2D and 3D constructs. The gene expression profiles investigated focused on genes involved in angiogenesis (VEGF, HMOX, and HGF) and matrix remodeling (matrix metalloproteinases and their inhibitors, ECM components), two critical processes that must be understood for

rational scaffold design. Genes for some matrix metalloproteinases (MMP-2, MMP-12, MMP-19), ECM protein synthesis (COL1A2, COL3A2, COL2A1, COL4A4, COL4A5), and proangiogenic factors were upregulated while some MMP inhibitors (TIMP1, TIMP3) were downregulated in the 3D CG scaffolds compared to the 2D CG surfaces.8 These results indicate that the geometric and topographical differences between 2D and 3D CG substrates can critically affect gene expression and possibly matrix remodeling.

2.216.4.6. Mechanical Stimulation Mechanical stimuli, derived from both physical extracellular binding sites and the movement of fluids within the ECM, is known to influence critical cell behaviors such as proliferation, differentiation, and gene expression.124–126 Experimental, computational, and modeling techniques have all been used with the CG scaffold system to improve our understanding of how these mechanical forces affect cell behavior. For example, the effects of pore size and mechanical stimulation on gene expression profiles for MSCs in CG scaffolds have recently been evaluated. Osteogenic markers collagen type I, osteocalcin, and osteopatin were all upregulated in MSCs seeded into CG scaffolds with larger pores (151 mm vs. 96 mm),127 indicating that larger pore sizes may be more optimal for bone tissue engineering. Also, scaffolds that were mechanically constrained to prevent contraction showed lower MSC RNA levels of osteocalcin, osteopatin, and bone sialoprotein, suggesting that prevention of cellmediated contraction may be detrimental to osteogenesis.127 CG scaffolds have also been used to investigate the role of direct mechanical stimulation on osteoblastic activity. Mechanical loading of MC3T3-E1 cells in culture is known to improve bone morphogenic activity within scaffolds. In addition, mechanical stimulation during in vitro culture improves nutrient and waste transport within the scaffold while promoting cellular infiltration, which should lead to more uniform tissue engineering constructs compared to static culture. Dynamic culture was shown to upregulate the gene for cyclooxygenase-2 (COX-2), a key enzyme for the production of early bone formation marker prostaglandin E2 (PGE2). The gene for osteopontin (OPN), a later marker of bone formation, was also upregulated.128 In addition, incorporation of rest periods during dynamic culture has been shown to significantly increase OPN expression over both static and dynamic conditions.129 This suggests that mechanical stimulation via flow perfusion could be a useful technique for encouraging osteoblastic activity and bone formation in CG scaffolds (see Chapter 5.501, Scaffolds: Flow Perfusion Bioreactor Design). 3D computational fluid dynamics (CFD) modeling has further been utilized to estimate shear stress forces on cells seeded within CG scaffolds for bone tissue engineering during in vitro culture in perfusion bioreactors.76,130 For bone, it is understood that applied wall shear stresses on cells are critical in activating matrix production and mineralization.81 Accurate simulations of applied shear stresses on cells would allow the selection of an optimal perfusion rate that correlates to physiologic shear stress levels. CFD showed that cells in CG scaffolds were exposed to a wide range of shear stresses while conventional analytical methods to estimate shear stress report a mean shear stress value only. The analytical methods also

Collagen–GAG Materials

predicted shear stress values 10–20% higher than CFD and do not take into account the type of secondary cell deformation that can take place on the basis of cell orientation within the scaffold.130 CFD analysis of fluid velocity, hydrostatic pressure, and wall shear stress on common cell attachment profiles within CG scaffolds (stretched along one strut or spread between two struts) shows that low wall shear stresses (20 mPa) are sufficient to activate bone growth mechanisms.76 While computational methods can be useful, the above analysis focused on small scaffold sections only as it was impractical to analyze an entire scaffold area.76,130,131 Recent work by Stops et al. has developed a finite element approach to predict cellular strains within CG scaffold systems (see Chapter 3.307, Finite Element Analysis in Bone Research: A Computational Method Relating Structure to Mechanical Function). Utilizing the tetrakaidecahedron unit cell and experimental data regarding scaffold deformation, cell attachment profiles, and cell sizes, a model was developed to predict individual cell strain as a function of pore size, cell length, and applied scaffold strain. This model correctly predicts individual cell strain within two standard deviations for 72% of cells.131 This approach has also been applied to predict the differentiation of MSCs in CG scaffolds as a function of Young’s modulus.132 The model takes into account pore size, cell proliferation, angiogenesis, and shear stress/strain among other factors. Results over a range of 0.001–1000 MPa show that higher Young’s modulus should lead to higher numbers of osteoblasts, indicating that stiffer CG scaffolds could be ideal for bone tissue engineering (Figure 18).132 All of these tools highlight the multi-scale experimental and analytical tools available when utilizing the CG scaffold system for studies of cellular mechanotransduction.

2.216.4.7. Stem Cell Differentiation CG scaffolds have recently been started to be used to explore extrinsic modulation of stem cell fate decisions

Predicted cell phenotype (%)

100

80

60

40

Fibroblasts Chondrocytes Osteoblasts

(differentiation, quiescence, and self-renewal) (see Chapter 2.209, Materials as Artificial Stem Cell Microenvironments). Adult mesenchymal stem cells (MSCs) can differentiate into a myriad of tissues, including bone, cartilage, and tendon, making them a powerful tool for orthopedic tissue engineering. Farrell et al. has recently shown that adult MSCs in CG scaffolds could be induced to differentiate along osteogenic or chondrogenic lineages in the presence of osteogenic (dexamethasone, ascorbic acid, and b-glycerophosphate) and chondrogenic (dexamethasone and TGF-b1) factors, respectively.133 The CG scaffold’s demonstrated ability to support both bone and cartilage growth makes it a promising material for the treatment of osteochondral defects. Subsequent work has compared the osteogenic potential of MSCs on 2D substrates versus 3D CG constructs. While MSCs initially expressed collagen type I quicker on 3D substrates, the delayed expression of osteocalcin indicates that osteogenic differentiation is slower in CG scaffolds compared to that in 2D substrates.134 MSCs cultured in CG materials have also shown variable differentiation and gene expression profiles under mechanical strain. High strains (15%) led to enhanced a-smooth muscle actin expression but poorer differentiation into type I collagen positive cells compared to lower strain levels (5%).135 Ongoing work is examining the utility of CG scaffolds to influence fate decisions of hematopoietic and cardiac stem cells.

2.216.5.

In Vivo Applications

Since the development of the first clinically viable artificial skin,5 CG scaffolds have been used in vivo to regenerate a wide variety of chronic and acute injuries involving skin (see Chapter 5.534, Skin Tissue Engineering), peripheral nerves (see Chapter 5.531, Peripheral Nerve Regeneration), conjunctiva, cartilage (see Chapter 5.515, Cartilage Tissue Engineering), and other tissues.4 CG scaffolds, like other tissue engineering scaffolds, serve a multitude of functions in vivo including physical inhibition of wound contraction, mechanical support for cells and neo-tissue growth, and carrier of biochemical ligands and signaling molecules. CG scaffold mechanical, chemical, and microstructural properties have been rigorously characterized in vitro, leading to independent control of pore size and shape, elastic moduli, biodegradation rate, chemical composition, and other scaffold parameters. These variables can be tuned to create scaffolds optimized for the regeneration of specific tissues.

2.216.5.1. Dermal Regeneration Applications

20

0 0.0001

297

1 100 0.01 Scaffold Young’s modulus (MPa)

10 000

Figure 18 Percentage of MSC-derived cell phenotypes inside CG scaffold as a function of Young’s modulus as predicted by finite element modeling. The fraction of osteoblasts compared to fibroblasts increases with increasing modulus, suggesting that stiffer CG scaffolds may be preferred for bone tissue engineering. Adapted from Khayyeri, H.; Checa, S.; Tagil, M.; O’Brien, F. J.; Prendergast, P. J. J. Mater. Sci. Mater. Med. 2010, 21, 2325–2330.

The CG scaffold is found to optimize skin regeneration as a monolithic, uniform structure (termed the dermal regeneration template (DRT)) that is fabricated from a CG copolymer with a 98:2 ratio of type I collagen to GAG (chondroitin 6-sulfate). This chemistry was developed in part because of the additional mechanical stability afforded by the GAG content as well as the processing to the collagen fibrils that allowed platelet attachment but prevented activation of platelet degranulation processes.5,136 The DRT has isotropic pores with a 20–125 mm size range and a degradation time of 5–15 days. The DRT is typically implanted acellularly (without cells) along

298

Materials of Biological Origin

with a thin silicone coating to control moisture loss and bacterial infection at the wound site. This bioactive scaffold recruits epidermal cells from the wound site and encourages spontaneous regeneration of skin tissue layer by layer. Sequential regeneration results in a mature epidermis and basement membrane as well as near physiologic dermis that lacks hair follicles. This scaffold has regenerated full-thickness skin wounds in animal models and human burn victims.4,5

2.216.5.3. Conjunctiva and Corneal Regeneration Applications CG scaffolds have demonstrated the ability to alter the typical contraction and scar synthesis healing mechanisms associated with full thickness lesions of the conjunctiva in a rabbit animal model. Scaffolds were able to significantly reduce wound contraction (6.8  3.2% fornix shortening vs. 26.4  5.0% for ungrafted wound sites) and promote the synthesis of a nearly physiologic stroma layer.140 CG scaffolds have also been shown to be promising substrates for the development of artificial corneas (see Chapter 6.631, Keratoprostheses). In vitro characterization of these scaffolds showed that stromal keratocytes, epithelial cells, and endothelial cells could be successively cocultured over a period of 12 weeks.141 The cells produced a new ECM, complete with an epithelium, basement membrane, and endothelial cell monolayer.141

2.216.5.2. Peripheral Nerve Regeneration Applications Peripheral nerve injuries are typically treated by entubulating transected nerve ends. Tubes made of type I collagen have been the most successful as defined by morphological and electrophysiological methods.3,4,48,137–139 While a tube is sufficient to induce regeneration across a gap of several millimeters, the addition of a porous material into the tube lumen can enhance the quality of regeneration and has been proven superior to other nonporous, nonpermeable lumen designs (Figure 19).4,139 This porous core acts as a physical support for the growing nerve while allowing migration of cells and cytokines across the nerve gap. In addition, this central scaffold core has also been shown to organize wound contraction and scar proliferation.54 The CG scaffold optimized for peripheral nerve regeneration, termed the nerve regeneration template (NRT), has axially elongated pores on the order of 10–20 mm that provide contact guidance for Schwann cell migration and axon formation between nerve stumps.137 These scaffolds, with an optimal degradation half life of 6 weeks,4,47 were found to regenerate peripheral nerves at the same level as autografts, the current gold standard in the industry.137

(a)

2.216.5.4. Cartilage and Fibrocartilage Disk Tissue Engineering Applications CG scaffolds have been applied to the regeneration of a range of other orthopedic tissues including articular cartilage, meniscus, and the intervertebral disk (see Chapters 5.524, Biomaterials for Replacement and Repair of the Meniscus and Annulus Fibrosus and 6.615, Intervertebral Disc). The effects of crosslinking density, chemical composition, and pore size of CG scaffolds, as well as the use of gene and growth factor seeded CG scaffolds, have been studied extensively in the context of articular cartilage regeneration.121,142–146 Recently, CG scaffolds populated with TGF-b1 transfected meniscus cells

(b)

(c)

(c)

(d)

(e)

Figure 19 Histomorphometic, cross-sectional images of the nerve trunk regenerated using collagen tubes with tailored biodegradation rates. The images are arranged in order of lowest to highest crosslink density, or fastest to slowest degradation rate from (a) to (e). Nerves trunks regenerated in devices (c) and (d), characterized by intermediate levels of the crosslink density and degradation rate (device half life: 2–3 weeks), showed superior morphology, with significantly larger axons, a more well-defined myelin sheath, and a significantly larger N ratio. Scale bar: 25 mm. Reprinted with permission from Harley, B. A.; Spilker, M. H.; Wu, J. W.; et al. Cells Tissues Organs 2004, 176, 153–165, S. Karger AG, Basel, Switzerland.

Collagen–GAG Materials

were used to successfully fill avascular zone meniscus lesions with repair tissue.123 Other work has compared type I and type II CG scaffolds for intervertebral disk tissue engineering applications, finding that type II CG scaffolds were preferential to type I on the basis of cell number as well as protein and GAG synthesis after 8 weeks.147

2.216.5.5. Bone, Osteochondral Regeneration Applications Single phase and layered, multiphase CGCaP scaffolds have recently been developed for in vivo bone and osteochondral tissue engineering applications, respectively (see Chapter 6.617, Bone Tissue Grafting and Tissue Engineering Concepts).61–63,69,148 These materials are currently undergoing in vivo examination as both bone scaffolds (single phase CGCaP form) and as full osteochondral scaffolds. CGCaP bone regeneration scaffolds are composed of type I collagen,

299

chondroitin 6-sulfate, and calcium phosphate. Initial in vitro and in vivo experiments have demonstrated that these scaffolds can successfully integrate into bone defects and show preliminary bony substitution and mineralization as early as 6 weeks post implantation.149 The biphasic layered, gradient scaffold developed for the treatment of osteochondral defects has been shown to mimic aspects of the natural structure of articular joints,69 notably a continuous boundary that should induce the formation of interfacial tissue as seen in healthy joints between articular cartilage and subchondral bone (Figure 20). Utilization of this fabrication method also prevents complications often observed in layered scaffolds with abrupt interfaces including delamination, foreign body contamination (from glue or other adhesives), and inefficient cellular transport between scaffold phases.69 Under mechanical loading, the osteochondral scaffolds perform as expected for a biphasic material, with the

Articular cartilage: Unmineralized Type II collagen-rich Tide mark: Interfacial region Continuity of collagen fibers Subchondral bone: Highly mineralized Type I collagen rich (a)

Type II CG scaffold: Cartilagenous compartment Type I CGCaP scaffold: Osseous compartment

(b)

(c)

Figure 20 (a) Structure of the natural articular joint showing articular cartilage and subchondral bone joined by a continuous interfacial region. (b) X-ray mCT image of the layered osteochondral scaffold showing distinct cartilaginous and osseous compartments (scale bar 1 mm). (c) SEM images of the osteochondral scaffold showing the complete scaffold microstructure (left; scale bar 500 mm), and the interfacial region (middle; scale bar 200 mm) showing continuity between the osseous (tan dashed arrow) and cartilaginous (blue solid arrow) compartments including collagen struts extending across the transition (white arrows). No regional areas of delamination or debonding are observed between the compartments. Distribution of Ca mineral (P similar but not shown) content (red shading) superimposed over an SEM image of the osteochondral scaffold showed distinct mineralized (high CaP content, tan dashed arrow) and nonmineralized (low/zero CaP content, blue solid arrow) layers (right; black scale bar 400 mm). Reprinted with permission from Harley, B. A.; Lynn, A. K.; Wissner-Gross, Z.; Bonfield, W.; Yannas, I. V.; Gibson, L. J. J. Biomed. Mater. Res. A 2010, 92, 1078–1093.

300

Materials of Biological Origin

majority of deformation confined to the cartilagenous compartment and no evidence of delamination between the scaffold compartments during or following loading.69 This scaffold is currently being developed as a clinical product that can be implanted using a mosaicplasty approach without the use of sutures, glue, or screws.69 This scaffold system is currently undergoing Phase I clinical trials for primary and secondary (backfill of traditional mosaicplasty harvest sites) osteochondral defects in the knee. The developed technologies and techniques hold promise for the regeneration of not only osteochondral defects, but also other physiological interfaces such as the tendon to bone insertion site.

2.216.5.6. Brain Tissue Engineering Applications CG scaffolds are currently being explored as substrates for neural defects in the brain. These scaffolds are fabricated via the conventional freeze-drying approach from type I or type II collagen and HA, a GAG that is the chief component of the brain ECM.150 Although the brain is almost entirely composed of HA, the collagen component of these CG scaffolds allows additional control of mechanical properties and biodegradation rates.151 Altering the ratio of collagen to HA allowed manipulation of porosity over a wide range (75–91%, brain ¼ 76%) and compressive modulus (1.33–6.31 kPa, brain ¼ 1.06 kPa).151 Neuronal stem cells showed the capacity to differentiate into neuronal cells in these HA–collagen scaffolds,151 and ongoing studies are continuing to optimize scaffold properties via a series of in vitro and in vivo experiments.

2.216.5.7. Lung Tissue Engineering Applications The cellular nature of CG scaffolds and the ability to modify scaffold porosity and pore size have made them candidate materials for regenerative medicine studies in the lung. CG scaffolds have been utilized as prospective materials to investigate design parameters for in vivo healing of lung tissue damage. Lung cells from Sprague-Dawley rats were seeded onto standard CG scaffolds (type I collagen and chondroitin 6-sulfate) and cultured for 2–21 days. This in vitro culture led to the development of alveolar-like structures as well as cellmediated contraction, possibly as a result of expression of a-smooth muscle actin.152 These results suggest that CG scaffolds hold promise as an in vivo treatment for lung defects and that they also might be interesting model systems for ex vivo culture of cells associated with a wide range of lung-tissue abnormalities and diseases.

2.216.6.

Conclusions

Porous, 3D biomaterials have been used extensively for a variety of tissue engineering applications, primarily as analogs of the ECM capable of inducing regeneration of damaged tissues and organs. An important evolving application is their use as constructs to quantitatively study cell behavior and cell– scaffold interactions. Scaffold material, microstructural, and mechanical properties have all been observed to significantly affect individual cell behavior as well as overall scaffold bioactivity and regenerative capacity. CG scaffolds are comprised of

naturally derived ECM and allow investigations of highly porous and fibrous ECM systems relevant to physiology and tissue engineering applications. Fabricating CG scaffolds via freeze-drying allows independent modulation of scaffold mean pore size, pore shape and orientation, relative density, mechanical properties, chemical composition, and available ligands. CG scaffolds and triple coprecipitated CGCaP nanocomposite scaffolds have been utilized as regenerative templates for a wide range of tissues following injury, notably of skin, peripheral nerves, brain, lung, cartilage, bone, fibrocartilage disks, and the conjunctiva. CG scaffolds have been shown to not only be a bioactive substrate for regenerative medicine applications, but are also well-suited to answer biologically relevant questions regarding cell–material interactions. As CG scaffolds as well as many other tissue engineering scaffolds resemble low-density, open-cell foams, with an interconnected network of struts, they can be modeled as cellular solids. Models for cellular solids contribute to our understanding of cell– scaffold interactions and hold promise for future studies of in vitro cell mechanobiology and in vivo tissue engineering.153,154

References 1. Harley, B. A. C.; Gibson, L. J. Chem. Eng. J. 2008, 137, 102–121. 2. Liu, Y.; Ramanath, H. S.; Wang, D. A. Trends Biotechnol. 2008, 26, 201–209. 3. Harley, B. A.; Spilker, M. H.; Wu, J. W.; et al. Cells Tissues Organs 2004, 176, 153–165. 4. Yannas, I. V. Tissue and Organ Regeneration in Adults; Springer: New York, 2001. 5. Yannas, I. V.; Lee, E.; Orgill, D. P.; Skrabut, E. M.; Murphy, G. F. Proc. Natl. Acad. Sci. USA 1989, 86, 933–937. 6. Yannas, I. V. Angew. Chem. Int. Ed. Engl. 1990, 29(1), 20–35. 7. Harley, B. A. C.; Kim, H. D.; Zaman, M. H.; Yannas, I. V.; Lauffenburger, D. A.; Gibson, L. J. Biophys. J. 2008, 95, 4013–4024. 8. Jaworski, J.; Klapperich, C. M. Biomaterials 2006, 27, 4212–4220. 9. Murphy, C. M.; Haugh, M. G.; O’Brien, F. J. Biomaterials 2010, 31, 461–466. 10. O’Brien, F. J.; Harley, B. A.; Yannas, I. V.; Gibson, L. J. Biomaterials 2005, 26, 433–441. 11. Tierney, C. M.; Haugh, M. G.; Liedl, J.; Mulcahy, F.; Hayes, B.; O’Brien, F. J. J. Mech. Behav. Biomed. Mater. 2009, 2, 202–209. 12. Tierney, C. M.; Jaasma, M. J.; O’Brien, F. J. J. Biomed. Mater. Res. A 2009, 91, 92–101. 13. Zeltinger, J.; Sherwood, J. K.; Graham, D. A.; Mueller, R.; Griffith, L. G. Tissue Eng. 2001, 7(5), 557–572. 14. Nehrer, S.; Breinan, H. A.; Ramappa, A.; et al. Biomaterials 1997, 18(11), 769–776. 15. Wake, M. C.; Patrick, C. W., Jr.; Mikos, A. G. Cell Transplant. 1994, 3(4), 339–343. 16. Engler, A.; Bacakova, L.; Newman, C.; Hategan, A.; Griffin, M.; Discher, D. Biophys. J. 2004, 86, 617–628. 17. Freyman, T. M.; Yannas, I. V.; Yokoo, R.; Gibson, L. J. Biomaterials 2001, 22, 2883–2891. 18. Grinnell, F.; Ho, C. H.; Lin, Y. C.; Skuta, G. J. Biol. Chem. 1999, 274, 918–923. 19. Grinnell, F.; Ho, C. H.; Tamariz, E.; Lee, D. J.; Skuta, G. Mol. Biol. Cell 2003, 14, 384–395. 20. Jiang, H.; Grinnell, F. Mol. Biol. Cell 2005, 16, 5070–5076. 21. Pelham, J.; Robert, J.; Wang, Y.-L. Proc. Natl. Acad. Sci. USA 1997, 9, 13661–13665. 22. Peyton, S. R.; Putnam, A. J. J. Cell Physiol. 2005, 204, 198–209. 23. Schulz-Torres, D.; Freyman, T. M.; Yannas, I. V.; Spector, M. Biomaterials 2000, 21, 1607–1619. 24. Yeung, T.; Georges, P. C.; Flanagan, L. A.; et al. Cytoskeleton 2005, 60, 24–34. 25. Zaman, M. H.; Trapani, L. M.; Sieminski, A. L.; et al. Proc. Natl. Acad. Sci. USA 2006, 103, 10889–10894. 26. Carter, E. M.; Raggio, C. L. Curr. Opin. Pediatr. 2009, 21, 46–54. 27. Lodish, H., Berk, A., Zipursky, S. L., Matsudaira, P., Baltimore, D., Darnell, J. E., Eds.; In Molecular Cell Biology; W.H. Freeman: New York, 2000. 28. Shoulders, M. D.; Raines, R. T. Annu. Rev. Biochem. 2009, 78, 929–958.

Collagen–GAG Materials

29. Berisio, R.; De Simone, A.; Ruggiero, A.; Improta, R.; Vitagliano, L. J. Pept. Sci. 2009, 15(3), 131–140. 30. Yannas, I. V. J. Macromol. Sci. Rev. Macromol. Chem. Phys. 1972, C7, 49. 31. Yannas, I. V.; Burke, J. F. J. Biomed. Mater. Res. 1980, 14, 65–81. 32. Raman, R.; Sasisekharan, V.; Sasisekharan, R. Chem. Biol. 2005, 12, 267–277. 33. BI, Y. M.; Ehirchiou, D.; Kilts, T. M.; et al. Nat. Med. 2007, 13, 1219–1227. 34. Lamari, F. N.; Karamanos, N. K. Adv. Pharmacol. 2006, 53, 33–48. 35. Kiani, C.; Chen, L.; Wu, Y. J.; Yee, A. J.; Yang, B. B. Cell Res. 2002, 12, 19–32. 36. Roughley, P. J. Eur. Cell Mater. 2006, 12, 92–101. 37. Yoon, J. H.; Halper, J. J. Musculoskelet. Neuronal. Interact 2005, 5, 22–34. 38. Lynn, A. K.; Yannas, I. V.; Bonfield, W. J. Biomed. Mater. Res. B Appl. Biomater. 2004, 71, 343–354. 39. Mehlisch, D. R. Oral Surg. Oral Med. Oral Pathol. 1989, 68, 505–514; discussion 514–516. 40. Snyder, C. C. Plast. Reconstr. Surg. 1976, 58, 401–406. 41. Charriere, G.; Bejot, M.; Schnitzler, L.; Ville, G.; Hartmann, D. J. J. Am. Acad. Dermatol. 1989, 21, 1203–1208. 42. Cooperman, L.; Michaeli, D. J. Am. Acad. Dermatol. 1984, 10, 647–651. 43. Tang, S. Q.; Vickers, S. M.; Hsu, H. P.; Spector, M. J. Biomed. Mater. Res. A 2007, 82A, 323–335. 44. O’Brien, F. J.; Harley, B. A.; Yannas, I. V.; Gibson, L. Biomaterials 2004, 25, 1077–1086. 45. Kanungo, B. P.; Gibson, L. J. Acta. Biomater. 2009, 5, 1006–1018. 46. Kanungo, B. P.; Gibson, L. J. Acta Biomater. 2009. 47. Chang, A. S.; Yannas, I. V.; Perutz, S.; et al. In Progress in Biomedical Polymers; Dunn, R. L., Ed.; Plenum Press: New York, 1990. 48. Chamberlain, L. J.; Yannas, I. V.; Hsu, H. P.; Strichartz, G.; Spector, M. Exp. Neurol. 1998, 154, 315–329. 49. Madaghiele, M.; Sannino, A.; Yannas, I. V.; Spector, M. J. Biomed. Mater. Res. Part A 2008, 85A, 757–767. 50. Kim, Y. T.; Haftel, V. K.; Kumar, S.; Bellamkonda, R. V. Biomaterials 2008, 29, 3117–3127. 51. Engelmayr, G. C.; Cheng, M.; Bettinger, C. J.; Borenstein, J. T.; Langer, R.; Freed, L. E. Nat. Mater. 2008, 7, 1003–1010. 52. Moffat, K. L.; Kwei, A. S.; Spalazzi, J. P.; Doty, S. B.; Levine, W. N.; Lu, H. H. Tissue Eng. Part A 2009, 15, 115–126. 53. Chamberlain, L. J.; Yannas, I. V. In Methods of Molecular Medicine; Morgan, J. R., Yarmush, M. L., Eds.; Humana Press: Totowa, NJ, 1998. 54. Harley, B. A.; Hastings, A. Z.; Yannas, I. V.; Sannino, A. Biomaterials 2006, 27, 866–874. 55. Sannino, A.; Silvestri, L.; Madaghiele, M.; Harley, B.; Yannas, I. V. J. Appl. Polym. Sci. 2010, 116(4), 1879–1888. 56. Harley, B. A.; Leung, J. H.; Silva, E.; Gibson, L. J. Acta. Biomater. 2007, 3, 463–474. 57. Yannas, I. V.; Tobolsky, A. V. Nature 1967, 215, 509–510. 58. Haugh, M. G.; Jaasma, M. J.; O’Brien, F. J. J. Biomed. Mater. Res. A 2009, 89, 363–369. 59. Olde Damink, L. H.; Dijkstra, P. J.; Van Luyn, M. J.; Van Wachem, P. B.; Nieuwenhuis, P.; Feijen, J. Biomaterials 1996, 17, 765–773. 60. Al-Munajjed, A. A.; O’Brien, F. J. J. Mech. Behav. Biomed. Mater. 2009, 2, 138–146. 61. Harley, B. A.; Lynn, A. K.; Wissner-Gross, Z.; Bonfield, W.; Yannas, I. V.; Gibson, L. J. J. Biomed. Mater. Res. A 2010, 92, 1066–1077. 62. Lynn, A. K.; Best, S. M.; Cameron, R. E.; et al. J. Biomed. Mater. Res. A 2010, 92, 1057–1065. 63. Lynn, A. K.; Bonfield, W. Acc. Chem. Rev. 2005, 38, 202–207. 64. Clarke, K. I.; Graves, S. E.; Wong, A. T. C.; Triffitt, J. T.; Francis, M. J. O.; Czernoszka, J. T. J. Mater. Sci. Mater. Med. 1993, 4, 107–110. 65. Itoh, S.; Kikuchi, M.; Takakuda, K.; et al. J. Biomed. Mater. Res. 2001, 54, 445–453. 66. Sasano, Y.; Kamakura, S.; Nakamura, M.; et al. Anat. Rec. 1995, 242, 40–46. 67. Kikuchi, M.; Itoh, S.; Ichinose, S.; Shinomiya, K.; Tanaka, J. Biomaterials 2001, 22, 1705–1711. 68. Genin, G. M.; Kent, A.; Birman, V.; et al. Biophys. J. 2009, 97, 976–985. 69. Harley, B. A.; Lynn, A. K.; Wissner-Gross, Z.; Bonfield, W.; Yannas, I. V.; Gibson, L. J. J. Biomed. Mater. Res. A 2010, 92, 1078–1093. 70. Osborne, C. S.; Barbenel, J. C.; Smith, D.; Savakis, M.; Grant, M. H. Med. Biol. Eng. Comput. 1998, 36, 129–134. 71. Saddiq, Z. A.; Barbenel, J. C.; Grant, M. H. J. Biomed. Mater. Res. A 2009, 89, 697–706. 72. Brigham, M. D.; Bick, A.; Lo, E.; Bendali, A.; Burdick, J. A.; Khademhosseini, A. Tissue Eng. Part A 2009, 15, 1645–1653. 73. Suri, S.; Schmidt, C. E. Tissue Eng. A 2010, 16, 1703–1716.

301

74. Green, J.; Kienitz, B.; Baskaran, H. In AIChE Annual Meeting, San Francisco, CA, 2006. 75. Harley, B. A.; Freyman, T. M.; Wong, M. Q.; Gibsony, L. J. Biophys. J. 2007, 93, 2911–2922. 76. Jungreuthmayer, C.; Jaasma, M. J.; Al-Munajjed, A. A.; Zanghellini, J.; Kelly, D. J.; O’Brien, F. J. Med. Eng. Phys. 2009, 31, 420–427. 77. Gibson, L. J.; Ashby, M. F. Cellular Solids: Structure and Properties; Cambridge University Press: Cambridge, UK, 1997. 78. Williams, R. E. Science 1968, 161, 276–277. 79. Thompson, W. Philos. Mag. 1887, 24, 503. 80. Agrawal, C. M.; Mckinney, J. S.; Lanctot, D.; Athanasiou, K. A. Biomaterials 2000, 21, 2443–2452. 81. Prendergast, P. J.; Huiskes, R.; Soballe, K. J. Biomech. 1997, 30, 539–548. 82. Levick, J. R. Q. J. Exp. Physiol. 1987, 72, 409–437. 83. O’Brien, F. J.; Harley, B. A.; Waller, M. A.; Yannas, I. V.; Gibson, L. J.; Prendergast, P. J. Technol. Health Care 2007, 15, 3–17. 84. Kanungo, B. P.; Silva, E.; Van Vliet, K.; Gibson, L. J. Acta. Biomater. 2008, 4, 490–503. 85. Hutmacher, D. W. Biomaterials 2000, 21, 2529–2543. 86. Pek, Y. S.; Spector, M.; Yannas, I. V.; Gibson, L. J. Biomaterials 2004, 25, 473–482. 87. Dembo, M.; Wang, Y.-L. Biophys. J. 1999, 76, 2307–2316. 88. Harris, Science 1980, 208, 177–179. 89. Lee, J.; Leonard, M.; Oliver, T.; Ishihara, A.; Jacobson, K. J. Cell. Biol. 1994, 127, 1957–1964. 90. Lemmon, C. A.; Sniadecki, N. J.; Ruiz, S. A.; Tan, J. L.; Romer, L. H.; Chen, C. S. Mech. Chem. Biosyst. 2005, 2, 1–16. 91. Oliver, T.; Dembo, M.; Jacobson, K. Cytoskeleton 1995, 31, 225–240. 92. Roy, P.; Petroll, W. M.; Chuong, C. J.; Cavanagh, H. D.; Jester, J. V. Ann. Biomed. Eng. 1999, 27, 721–730. 93. Tan, J. L.; Tien, J.; Pirone, D. M.; Gray, D. S.; Bhadriraju, K.; Chen, C. S. Proc. Natl. Acad. Sci. USA 2003, 100, 1484–1489. 94. Beningo, K. A.; Dembo, M.; Kaverina, I.; Small, J. V.; Wang, Y.-L. J. Cell Biol. 2001, 153, 881–887. 95. Beningo, K. A.; Wang, Y.-L. Trends Cell Biol. 2002, 12, 79–84. 96. Chen, C. S.; Ingber, D. E. Osteoarthr. Cartil. 1999, 7, 81–94. 97. Engler, A. J.; Sen, S.; Sweeney, H. L.; Discher, D. E. Cell 2006, 126, 677–689. 98. Lo, C.-M.; Wang, H.-B.; Dembo, M.; Wang, Y.-L. Biophys. J. 2000, 79, 144–152. 99. Munevar, S.; Wang, Y.-L.; Dembo, M. Mol. Biol. Cell 2001, 12, 3947–3954. 100. Wang, H.-B.; Dembo, M.; Hanks, S. K.; Wang, Y.-L. Proc. Natl. Acad. Sci. USA 2001, 98, 11295–11300. 101. Wang, H.-B.; Dembo, M.; Wang, Y.-L. Am. J. Physiol. Cell Physiol. 2000, 279, C1345–C1350. 102. Wang, N.; Butler, J. P.; Ingber, D. Science 1993, 260, 1124–1127. 103. Guilak, F.; Erickson, G. R.; Ting-Beall, H. P. Biophys. J. 2002, 82, 720–727. 104. Marquez, J. P.; Genin, G. M.; Zahalak, G. I.; Elson, E. L. Biophys. J. 2005, 88, 778–789. 105. Zahalak, G. I.; Wagenseil, J. E.; Wakatsuki, T.; Elson, E. L. Biophys. J. 2000, 79, 2369–2381. 106. Freyman, T. M.; Yannas, I. V.; Yokoo, R.; Gibson, L. J. Exp. Cell Res. 2002, 272, 153–162. 107. Zaleskas, J. M.; Kinner, B.; Freyman, T. M.; Yannas, I. V.; Gibson, L. J.; Spector, M. Biomaterials 2004, 25, 1299–1308. 108. Spilker, M. H.; Asano, K.; Yannas, I. V.; Spector, M. Biomaterials 2001, 22, 1085–1093. 109. Freyman, T. M.; Yannas, I. V.; Pek, Y.-S.; Yokoo, R.; Gibson, L. J. Exp. Cell Res. 2001, 269, 140–153. 110. Sethi, K. K.; Yannas, I. V.; Mudera, V.; Eastwood, M.; Mcfarland, C.; Brown, R. A. Wound Repair Regen. 2002, 10, 397–408. 111. Triantafillou, T. C.; Zhang, J.; Shercliff, T. L.; Gibson, L. J.; Ashby, M. F. Int. J. Mech. Sci. 1989, 31, 665–678. 112. Condeelis, J.; Segall, J. E. Nat. Rev. Cancer 2003, 3, 921–930. 113. Lapidot, T.; Dar, A.; Kollet, O. Blood 2005, 106, 1901–1910. 114. Wilson, A.; Trumpp, A. Nat. Rev. Immunol. 2006, 6, 93–106. 115. Friedl, P.; Zanker, K. S.; Brocker, E. B. Microsc. Res. Tech. 1998, 43, 369–378. 116. Lauffenburger, D. A.; Horwitz, A. F. Cell 1996, 84, 359–369. 117. Zaman, M. H.; Kamm, R. D.; Matsudaira, P.; Lauffenburger, D. A. Biophys. J. 2005, 89, 1389–1397. 118. Vogel, V.; Sheetz, M. Nat. Rev. Mol. Cell. Biol. 2006, 7, 265–275. 119. Bolliet, C.; Bohn, M. C.; Spector, M. Tissue Eng. Part C Methods 2008, 14, 207–219. 120. Price, T. J.; Louria, M. D.; Candelario-Soto, D.; et al. BMC Neurosci. 2005, 6, 4. 121. Capito, R. M.; Spector, M. Gene Ther. 2007, 14, 721–732.

302

Materials of Biological Origin

122. Sun, X. D.; Jeng, L.; Bolliet, C.; Olsen, B. R.; Spector, M. Biomaterials 2009, 30, 1222–1231. 123. Steinert, A. F.; Palmer, G. D.; Capito, R.; et al. Tissue Eng. 2007, 13, 2227–2237. 124. Geris, L.; Vandamme, K.; Naert, I.; Vander Sloten, J.; Duyck, J.; Van Oosterwyck, H. J. Biomech. 2008, 41, 145–154. 125. Mcmahon, L. A.; Reid, A. J.; Campbell, V. A.; Prendergast, P. J. Ann. Biomed. Eng. 2008, 36, 185–194. 126. Wu, Q. Q.; Chen, Q. Exp. Cell Res. 2000, 256, 383–391. 127. Byrne, E. M.; Farrell, E.; Mcmahon, L. A.; et al. J. Mater. Sci. Mater. Med. 2008, 19, 3455–3463. 128. Jaasma, M. J.; O’Brien, F. J. Tissue Eng. Part A 2008, 14, 1213–1223. 129. Partap, S.; Plunkett, N. A.; Kelly, D. J.; O’Brien, F. J. J. Mater. Sci. Mater. Med. 2010, 21, 2325–2330. 130. Jungreuthmayer, C.; Donahue, S. W.; Jaasma, M. J.; et al. Tissue Eng. Part A 2009, 15, 1141–1149. 131. Stops, A. J.; Mcmahon, L. A.; O’mahoney, D.; Prendergast, P. J.; Mchugh, P. E. J. Biomech. Eng. 2008, 130, 061001. 132. Khayyeri, H.; Checa, S.; Tagil, M.; O’Brien, F. J.; Prendergast, P. J. J. Mater. Sci. Mater. Med. 2010, 21, 2331–2336. 133. Farrell, E.; O’Brien, F. J.; Doyle, P.; et al. Tissue Eng. 2006, 12, 459–468. 134. Farrell, E.; Byrne, E. M.; Fischer, J.; et al. Technol. Health Care 2007, 15, 19–31. 135. Kobayashi, M.; Spector, M. Mol. Cell Biomech. 2009, 6, 217–227. 136. Sylvester, M. F.; Yannas, I. V.; Salzman, E. W.; Forbes, M. J. Thromb. Res. 1989, 55, 135–148. 137. Chamberlain, L. J.; Yannas, I. V.; Arrizabalaga, A.; Hsu, H. P.; Norregaard, T. V.; Spector, M. Biomaterials 1998, 19, 1393–1403.

138. Chamberlain, L. J.; Yannas, I. V.; Hsu, H. P.; Strichartz, G. R.; Spector, M. J. Neurosci. Res. 2000, 60, 666–677. 139. Harley, B. A.; Yannas, I. V. Minerva Biotecnologica 2006, 18, 97–120. 140. Hsu, W. C.; Spilker, M. H.; Yannas, I. V.; Rubin, P. A. Invest. Ophthalmol. Vis. Sci. 2000, 41, 2404–2411. 141. Vrana, N. E.; Builles, N.; Justin, V.; et al. Invest. Ophthalmol. Vis. Sci. 2008, 49, 5325–5331. 142. Capito, R. M.; Spector, M. IEEE Eng. Med. Biol. Mag. 2003, 22, 42–50. 143. Kinner, B.; Capito, R. M.; Spector, M. Adv. Biochem. Eng. Biotechnol. 2005, 94, 91–123. 144. Lee, C. R.; Grodzinsky, A. J.; Hsu, H.-P.; Spector, M. J. Orthop. Res. 2003, 21, 272–281. 145. Samuel, R. E.; Lee, C. R.; Ghivizzani, S. C.; et al. Hum. Gene Ther. 2002, 13, 791–802. 146. Vickers, S. M.; Squitieri, L. S.; Spector, M. Tissue Eng. 2006, 12. 147. Saad, L.; Spector, M. J. Biomed. Mater. Res. A 2004, 71, 233–241. 148. Harley, B. A.; Lynn, A. K.; Wissner-Gross, Z.; Bonfield, W.; Yannas, I. V.; Gibson, L. J. In Design and Fabrication of a Multiphase Osteochondral Scaffold, Proceedings of the First International Conference on Mechanics of Biomaterials & Tissues, 2005. 149. Lynn, A. K.; Harley, B. A.; Wissner-Gross, Z.; Yannas, I. V.; Gibson, L. J.; Bonfield, W. In Design and Fabrication of a Mineralized, Triple-co-Precipitate Collagen–Glycosaminoglycan Scaffold, Proceedings of the First International Conference on Mechanics of Biomaterials & Tissues, 2005. 150. Bignami, A.; Hosley, M.; Dahl, D. Anat. Embryol. (Berl) 1993, 188, 419–433. 151. Wang, T. W.; Spector, M. Acta Biomater. 2009, 5, 2371–2384. 152. Chen, P.; Marsilio, E.; Goldstein, R. H.; Yannas, I. V.; Spector, M. Tissue Eng. 2005, 11, 1436–1448.