Collagen Scaffolds Reinforced with Biomimetic Composite

Nano-Sized Carbonate-Substituted Hydroxyapatite Crystals and

Shaped by Rapid Prototyping to Contain Internal Microchannels

ELEFTHERIOS SACHLOS, D.Phil., DUCE GOTORA, D.Phil., and JAN T. CZERNUSZKA, Ph.D.

ABSTRACT

The next generation of tissue engineering scaffolds will bemade to accommodate blood vessels and nutrientchannels to support cell survival deep in the interior of the scaffolds. To this end, we have developed amethod that incorporates microchannels to permit the flow of nutrient-rich media through collagen-basedscaffolds. The scaffold matrix comprises nano-sized carbonate-substituted hydroxyapatite (HA) crystalsinternally precipitated in collagen fibers. The scaffold thereforemimicsmany of the features found in bone.A biomimetic precipitation technique is used whereby a collagen membrane separates reservoirs of cal-cium and phosphate solutions. The collision of calcium and phosphate ions diffusing from opposite di-rections results in the precipitation of mineral within the collagen membrane. Transmission electronmicroscopy analysis showed the dimension of the mineral crystals to be approximately 180�80�20 nm,indicating that the crystals reside in the intermicrofibril gaps. Electron diffraction indicated that themineral was in the HA phase, and infrared spectroscopy confirmed type A carbonate substitution. Thecollagen-HAmembrane is then used to make 3-dimensional (3D) scaffolds: the membrane is shredded andmixed in an aqueous-based collagen dispersion and processed using the critical point drying method.Adjusting the pH of the dispersion to 5.0 before mixing the composite component preserved the nano-sizedcarbonate-substituted HA crystals. Branching and interconnecting microchannels in the interior of thescaffolds aremade with a sacrificial moldmanufactured by using a 3Dwax printer. The 3Dwax printer hasbeen modified to print the mold from biocompatible materials. Appropriately sized microchannels withincollagen-HA scaffolds brings us closer to fulfilling the mass transport requirements for osteogenic cellsliving deep within the scaffold.

INTRODUCTION

THE EXTRACELLULAR MATRIX OF BONE is a composite ma-

terial composed of an organic phase reinforced by an

inorganic phase. The organic matrix, known as osteoid,

principally consists of collagen (approximately 90%), with

the remaining fraction completed by noncollagenous pro-

teins. The osteoid is mineralized by a calcium phosphate

described as carbonate-substituted hydroxyapatite (HA).1

Collagen is the main structural protein of vertebrates. As

many as 26 genetically distinct types of human collagen

have been identified, with subtle differences in their pri-

mary structure and molecular folding.2 Type I collagen is

the main collagen constituent of bone. Bone mineral, which

reinforces osteoid, consists of thin, flat plate crystals3–6

averaging 50 nm in length, 25 nm in width, and 2–5 nm in

thickness.5,6 These crystals are orientated with their long

crystallographic c-axis parallel to each other and aligned

with the collagen tropocollagen molecules.7,8 The exact

location of the crystals with respect to the Hodge and

Petruska model of collagen has not been precisely deter-

mined, but strong evidence supports their residence in the

Department of Medicine, University of Oxford, Oxford, United Kingdom.

TISSUE ENGINEERINGVolume 12, Number 9, 2006# Mary Ann Liebert, Inc.

2479

groove regions formed by the 3-dimensional (3D) organi-

zation of the microfibril.9–11

Bone mineral is similar to HA, which has the stoi-

chiometric formula of Ca10(PO4)6OH2. The calcium-to-

phosphorus molar ratio varies in bone mineral depending

on the species, age, and type of bone. This variation arises

because bone mineral is not pure HA but rather contains

impurities including carbonate, hydrogen phosphate, fluo-

rine, chlorine, magnesium, sodium, and potassium ions, and

traces of strontium and zinc elements.1 Carbonate ions, in

particular, may be present up to 8wt%.12 The carbonate

ions can substitute for hydroxide (type A substitu-

tion)13 or for phosphate ion (type B substitution), with

a consequent substitution of sodium for calcium ions to

maintain charge balance.14,15 The ionic substitution of HA

is important for destabilizing the crystal lattice and making

it easier to resorb in the body.16–18 HA must be resorbable

for the body to be able to remodel bone. Stoichiometric HA

has often been used as a scaffold or bone graft substitute,

but reports indicate that it is slow in resorbing.19,20

Bovine collagen is by far the most commonly used type

of collagen. It has been extensively used in many medical

devices and tissue-engineered products. For example, Or-

ganogenesis’s Apligraf and Ortec’s OrCel both use a matrix

based on bovine type I collagen. However, bovine collagen

may elicit an antigenic response21–23 and vary from batch

to batch. The antigenicity of collagen can be reduced by

crosslinking the collagen using physical or chemical tech-

niques.24,25 The risks of bovine spongiform encephalopathy

(BSE) transmission are minimized by using collagen ob-

tained from closed herds. Human recombinant collagen has

also become available and can be used in scaffold manu-

facturing.

Notwithstanding these issues, collagen arguably pos-

sesses the ideal surface for cell attachment. It is the major

structural protein of the human body’s extracellular matrices

and acts as the natural scaffold for cell attachment in the

body. Collagen type I contains the Arg-Gly-Asp (RGD)26

and Asp-Gly-Glu-Ala (DGEA)27 sequences that mediate

cell binding via integrin receptors.28 The cell attachment

protein fibronectin is also important in promoting binding of

cells to collagen. Fibronectin binds directly to collagen29–32

and possesses domains for the attachment of fibroblasts.33–35

Fibronectin binding also enhances cell migration.36 Colla-

gen is also chemotactic to fibroblasts.37 As a consequence of

these interactions, collagen scaffolds present a more native

surface to cells relative to synthetic polymer scaffolds for

tissue engineering purposes. These biological properties of

collagen emphasize its significance in tissue regeneration

and its value as a scaffold material.

Considering that the natural extracellular matrix (ECM)

of bone is made predominantly of nano-sized carbonate-

substituted HA crystals, which are internally precipitated in

a network of collagen fibers, we have developed a method

to fabricate biomimetic bone scaffolds based on this

structure. Other investigators38–41 have produced scaffolds

made from collagen-HA. We also believe that the incor-

poration of microchannels inside the scaffold that can

permit the flow of tissue culture media42 would be advan-

tageous in perfusing the scaffold and overcoming the dif-

fusion constraints arising from growing cells on relatively

homogeneous foam structures, which leads to cell survival

on the periphery of the scaffold.43–45

MATERIALS AND METHODS

Dispersion formulation

Collagen dispersions of 1%w/v and 2%w/v were pre-

pared. The 1%w/v dispersion was made by adding 1 g of

insoluble type I collagen derived from bovine Achilles

tendon (Sigma-Aldrich, Poole, UK) to 100mL of distilled

water adjusted to a pH of 3.2 by the dropwise addition of

analytical grade acetic acid (Sigma-Aldrich). The 2%w/v

dispersion was prepared in an identical manner but with 2 g

of collagen added to the solution. The mixtures were then

homogenized using a blender on a bed of ice to prevent

heat build-up for approximately 2min. Air bubbles were

removed by degassing in a bell jar. Dispersions were stored

at 4oC before use.

Collagen membrane fabrication

Collagen membranes were made by placing 10mL of

2%w/v collagen dispersion in a 90-mm-diameter polysty-

rene Petri dish under a laminar flow cabinet and allowing it

to air-dry for 24 h. The membranes were then peeled off the

Petri dish and cut to shape with a scalpel.

Biomimetic precipitation in collagen membranes

The experimental setup of a biomimetic precipitation

method46 capable of precipitating nano-sized calcium phos-

phate crystals inside a collagenmembrane is shown in Fig. 1.

A 20-mm-diameter collagen membrane was secured be-

tween 2 reservoirs, 1 filled with 35mM calcium solution

(CaCl2.2H2O) and the other with 50mM phosphate solution

(KH2PO4). Both solutions contain 0.1M potassium chloride

(KCl) to maintain ionic stability and were adjusted to a pH of

8.0. During precipitation, the solutions were stirred contin-

uously by using a magnetic stirrer and the temperature of

the reaction vessel was kept constant at 378C with a water

bath. The solutions were maintained at a pH of 8.0 by a

computer-controlled negative feedback system. An ionome-

ter (PHG201-8, Radiometer, Lyon, France) measures the pH

of the calcium solution and outputs these data every 30

seconds to a computer connected to a 2-flask autoburettes

system (ABU93, Radiometer). The flasks of the autoburettes

contain separated acid and alkaline solutions. Both solutions

were made from 35mM CaCl2.2H2O with 0.1M KCl, but

the acid solution was adjusted to a pH of 2.0–3.0 with

the addition of analytical grade hydrochloric acid (Fisons

2480 SACHLOS ET AL.

Chemicals, Loughborough, UK); in contrast, the alkaline

solution was adjusted to a pH of 12.0–13.0 with 5.0M po-

tassium hydroxide (KOH). Any drift from a pH of 8.0 was

recorded by the computer and counterbalanced by the in-

jection of the appropriate amount of acid or alkaline solution

from the autoburettes into the calcium solution. The negative

feedback system was sensitive to pH changes of 0.003 and

greater. The concentration of calcium was measured using

an ion meter (ISE 25Ca, Radiometer) and logged every

30 seconds on the computer. Precipitation was conducted for

24 h before the biomimetic composite membranes were al-

lowed to air-dry for another 24 h and then shredded into

submillimeter flakes with a razor blade.

Characterization of biomimetic composite

membranes

Microstructural examination. Membranes, flakes, and

scaffolds were prepared for scanning electron microscopy

(SEM) evaluation by gold-sputter (Biorad E5400, Polaron,

Hertfordshire, UK) coating and imaging in the secondary

electron mode the surface of the sample with a field-

emission gun SEM (JSM-840F, JEOL, Tokyo, Japan) at an

accelerating voltage of 2.5 or 10 kV.

Infrared spectroscopy. To assess for ionic substitution,

submillimeter flakes of biomimetic composite membranes

were diluted in ground potassium bromide powder, which

was then pressed into discs using a 13-mm-diameter

stainless steel die (Evacuable Pellet Die, Specac, Kent, UK)

and a 100-kN compaction force applied with a hydraulic

press (15011, Specac). Infrared spectra of the discs were

obtained using a Fourier transform infrared spectrometer

(Spectrum 2000, PerkinElmer, Wellesley, MA) in trans-

mission mode with a 4 cm�1 resolution and 64 scans per

spectrum. Spectra were obtained in the range of 4000–

400 cm-1. A background spectrumwas obtained every 30min

to take into account any possible environmental changes.

Transmission electronmicroscopy and electron diffraction.

Biomimetic composite membranes were characterized using

bright-field transmission electron microscopy to assess for

mineral crystal morphology. Electron diffraction of the

crystals was conducted to identify calcium phosphate phase.

Samples were embedded in epoxy resin (Agar 100, Agar

Scientific Inc., Essex, UK) and sectioned using an ultrami-

crotome (Nova Ultrotome, LKB Produkter, Bromma, Swe-

den) fitted with a glass blade. The cut sections were floated

on water and collected on 200 mesh copper grips (Agar

Scientific Inc.). The samples were viewed under a trans-

mission electron microscope ( JEM-200CX, JEOL) operated

at an accelerating voltage of 200 kV. Bright-field images and

diffraction patterns of the biomimetic membranes were ob-

tained. Photomicrographs with short and long exposure time

periods were acquired for the same diffraction patterns.

Mold fabrication

The process for fabricating the collagen scaffold using

sacrificial molds has been described elsewhere.42 Briefly,

sacrificial molds, which are soluble in ethanol, are used to

define the external shape and internal microchannel archi-

tecture of the scaffolds. The molds were designed using

commercial computer-aided design software (AutoCAD

2005, Autodesk, Hampshire, UK), converted to stl format

and fabricated using a rapid prototyping system based on

3D hot-melt ink jet printing. The 3D printer (T66, So-

lidscape Inc., Merrimack, NH) operates by printing 2 ma-

terials in their molten state that solidify on impact with the

substrate to form beads. Overlapping of beads forms a layer

that can be overprinted to gradually build up the mold.

Selective positioning of the beads allows for the formation

of any shape. The 3D printer has 2 printheads, 1 dedicated

to printing the build or mold material and 1 dedicated to

printing the sacrificial support necessary to generate over-

hanging features.

The Solidscape 3D printer has been modified for scaffold

fabrication. Specifically, it has been adapted to print 2

biocompatible materials: BioBuild and BioSupport, sup-

plied by TEOX Ltd. (Oxford, UK) act as the mold and

support materials, respectively. When the model is com-

pleted, it is immersed in water heated to 358C, which dis-

solves the BioSupport and relieves the mold.

FIG. 1. Schematic representation of biomimetic precipitation of

calcium phosphate in the interior of a collagen membrane. The pH

of the reaction vessel is maintained at a constant level of 8.0

through a negative feedback system with computer controlled auto-

burrettes.

COLLAGEN SCAFFOLDS REINFORCED WITH BIOMIMETIC COMPOSITE 2481

Scaffold fabrication

The biomimetic composite flakes were mixed with 1%w/v

collagen dispersion and cast into printed molds before

being placed in a freezer at �308C for at least 12 h. The

mold with frozen dispersion was then immersed in 3 sep-

arate hour-long baths of ethanol, which dissolved the ice

crystals and mold. The scaffold was then critical-point-

dried with liquid carbon dioxide (CO2) for 3 h, with fresh

liquid CO2 being exchanged every 15–20min. The flakes

were also mixed with 1%w/v collagen dispersion, which

was adjusted to a pH of 5.0 with the dropwise addition of

0.1M KOH before being cast into printed molds and pro-

cessed under identical conditions.

To assist with retrieval of the biomimetic composite

component from the scaffold in order to conduct trans-

mission electron microscopy and infrared spectroscopy

analysis, some scaffolds were made by adding biomimetic

composite membranes cut into 5�5mm segments with

1%w/v collagen dispersion at a pH of 3.2 or adjusted to a

pH of 5.0. After the scaffolds were critical-point-dried, the

square segments were retrieved using a pair of tweezers

under visual examination with a stereo-optical microscope

(MGG17; Wild, Heerbrugg, Switzerland). These samples

were then prepared for infrared spectroscopy and trans-

mission electron microscopy analysis following the same

methods described previously.

RESULTS

Following biomimetic precipitation, the smooth surface

of a collagen membrane was altered into a mineral coating

with rough crystal topography (Fig. 2). Infrared evaluation

of the biomimetic composite membrane revealed the main

amide I, II, and III bands characteristic of the collagen

component in addition to the phosphate bands attributed to

calcium phosphate (Fig. 3A). The band assignment for this

spectrum is listed in Table 147–49 and is indicative of HA.

Noteworthy is the band centered at 872 cm�1, which is

representative of type A carbonate substitution for hydroxyl

ions in the mineral.

The biomimetic composite membrane shows a large

quantity of tablet-shaped crystals in the interior of the

membrane (Fig. 4A). The smallest crystals were shown to

measure approximately 180�80�20 nm. Electron diffrac-

tion patterns of these crystals and the assigned Miller in-

dices are shown in Fig. 5A and B.

Flakes of biomimetic composite membranes shown

in Fig. 6 were used to create 3D composite scaffolds.

Fig. 7 shows the structure of these composite scaffolds and

reveals submillimeter flakes entrapped in a porous, fibrous

network of collagen.

The infrared spectrum of biomimetic composite mem-

brane retrieved from a scaffold processed using a collagen

dispersion with a pH of 3.2 is shown in Fig. 3B. The main

collagen amide I, II, and III bands are still evident, but

intensity is markedly reduced for the calcium phosphate

bands. The relative change in intensity is summarized in

Table 2. In contrast, the infrared spectrum of biomimeti-

cally precipitated membranes made using a collagen dis-

persion with a pH of 5.0 has a spectrum similar to that of

unprocessed biomimetic composite membrane (Fig. 3C).

The relative intensity for the phosphate bands of this

spectrum (summarized in Table 2) shows little change from

unprocessed biomimetic composite membrane. The band at

872 cm�1 indicates type A carbonate substitution in scaf-

folds made with collagen dispersion adjusted to a pH of 5.0.

The tablet-shaped crystals found in the interior of the

biomimetic composite membrane are still present when a

scaffold is made using a collagen dispersion with a pH of

5.0 (Fig. 4B). Electron diffraction of the crystals also

confirms that the HA phase is retained during scaffold

manufacturing (Fig. 5C and D).

By casting collagen dispersion containing flakes of bio-

mimetic composite membrane into molds that have been

printed with the 3D printer, scaffolds possessing internal

microchannels can be created (Fig. 8). In this particular

design, the microchannels are branched and interconnected

hexagonal patterns create island regions in the scaffold. The

microchannels measure approximately 800mm in width.

DISCUSSION

The rationale behind mimicking the compositional and

structural organization of the main organic and inorganic

components of bone in a single scaffold is to create an

environment that more closely resembles the natural ECM

of bone. In doing so, it is hypothesized that osteogenic cells

will recognize the surface and begin to proliferate, migrate,

and differentiate, and then stimulate the deposition of bone.

Collagen and HA both have merits as scaffold materials.

FIG. 2. Scanning electron micrograph of the surface of biomi-

metic composite membrane showing rough mineral coating.

2482 SACHLOS ET AL.

For example, collagen possesses attractive cell binding

properties required for seeded cells to attach to the scaffold,

whereas HA enhances the bioactivity of the scaffold by

providing a source of calcium and phosphate ions that can

be used by osteogenic cells to create their own bone.

We have developed a diffusion-based biomimetic pre-

cipitation method, schematically shown in Fig. 1, whereby

nano-sized crystals of carbonate-substituted HA can be

incorporated inside a membrane composed of a dense

network of collagen fibers. The reaction is conducted at

378C to mimic the natural biomineralization process, while

avoiding the denaturing of the collagen. The surface of the

collagen membrane is transformed to a mineral coating

with plate-like crystals when the membrane undergoes

biomimetic precipitation (Fig. 3). Infrared spectroscopy

analysis of the composite membranes revealed the presence

of major collagen peaks, namely amide A, I, II, and III, in

addition to major phosphate peaks indicative of HA

(Fig. 3A and Table 1). Of note is the presence of type A

carbonate substitution found in the HA. The relatively

moderate bands for type B substitution lie in the 1650–

1300 cm�1 range,50 which also features the strong amide I,

II, and II peaks of collagen. The possibility of type B

substitution cannot be ruled out because the collagen bands

could be masking the bands characteristic of this substitu-

tion. Carbonate substitution most probably arises from the

ion exchange with dissolved CO2 found in the solutions,

but further work is required to elucidate the exact mecha-

nism of substitution. Structural evaluation of the interior

of the composite membrane with the transmission elec-

tron microscope showed a high concentration of crystals,

with the smallest having dimensions of approximately

TABLE 1. INFRARED BAND ASSIGNMENT OF BIOMIMETIC

COLLAGEN-HYDROXYAPATITE MEMBRANES

Frequency (cm�1) Assignment

3290–3330 Amide A

2931 C-H stretch

2852 C-H stretch

1640–1660 C¼O stretch (amide I)

1535–1550 N-H in plane deformation

plus C-N stretch (amide II)

1445–1455 CH2 deformation and CH3

asymmetric deformation

1310–1340 CH2 wagging

1230–1270 C-N stretch plus N-H in-plane

deformation (amide III)

1200–1000 Phosphate v3 antisymmetric

961 Phosphate v1 symmetric

872 Carbonate v2 (type A substitution)

600 Phosphate v4 antisymmetric

bending mode

561 Phosphate v4471 Phosphate v1

FIG. 3. Infrared spectra of biomimetic collagen-hydroxyapatite membranes (A) unprocessed, (B) processed with collagen dispersion

at a pH of 3.2, and (C) processed with collagen dispersion at a pH of 5.0. The bands assigned to collagen amide I, II, and III and

phosphate v1, v3, and v4 are labeled accordingly. Note the reduction of the phosphate vibrations (v3 and v4) when processed at a pH of

3.2, but these bands are retained when processed at a pH of 5.0. Type A carbonate substitution is evident with the band at 872 cm�1

when unprocessed and processed at a pH of 5.0.

COLLAGEN SCAFFOLDS REINFORCED WITH BIOMIMETIC COMPOSITE 2483

180�80�20 nm (Fig. 4A). These crystals are larger than

the 50�25�2–5 nm found in bone. The implications of this

size discrepancy is that the biomimetic composite crystals

are most likely too large to fit into the 40-nm ‘‘grooves’’

believed to be the natural residence of bone mineral in the

collagen microfibril. It is more likely that the precipitated

crystals lie in intermicrofibrillar gaps of the membrane.

Electron diffraction of these internal crystals (Fig. 5 A andB)

confirms that they are of the HA phase. The data obtained

from infrared analysis, bright-field transmission electron

microscopy, and electron diffraction lead us to the premise

FIG. 5. Electron diffraction patterns of biomimetic composite

membrane at different exposure times (A and B) and electron

diffraction patterns of biomimetic composite membrane used

(C and D) make a scaffold with dispersion adjusted to a pH of 5.0.

The Miller indices for each corresponding ring are labeled ac-

cordingly in brackets.

FIG. 4. Bright-field transmission electron micrographs of the interior of (A) biomimetic composite membrane and (B) a biomimetic

composite membrane used to make a scaffold with collagen dispersion adjusted to a pH of 5.0. Note the dark tablet-like crystals present

throughout the cross-section in both micrographs.

FIG. 6. Scanning electron micrograph of flakes made by shred-

ding a biomimetic collagen-hydroxyapatite membrane.

2484 SACHLOS ET AL.

that the mineral present in the biomimetic composite

membranes is carbonate-substituted HA of nanometer di-

mensions that most likely resides in the intermicrofibrillar

gaps of the membrane.

The 2D biomimetic composite membrane is then used to

produce 3D scaffolds. The first step in the fabrication of

collagen-based scaffolds involves casting of collagen dis-

persion into a mold. At this step the biomimetic composite

membrane can be incorporated into the scaffold composi-

tion. The membrane is shred into submillimeter flakes (Fig.

6) and mixed with collagen dispersion, which is usually

maintained at a pH of 3.2, before casting into molds and

processing to form a scaffold. However, the infrared

spectra of scaffolds made with this method show a decrease

in intensity of the main phosphate bands relative to un-

processed biomimetic composite membrane (Fig. 3B and

Table 2). This relative reduction in phosphate band inten-

sity most likely arises from the dissolution of the HA

precipitate due to the acidic nature (i.e., pH of 3.2) of the

dispersion. The HA crystals probably start to rapidly dis-

solve before the dispersant is solidified during the freezing

step in scaffold manufacturing.

To decrease the dissolution rate of the HA component,

the pH of the collagen dispersion was adjusted to a pH of

5.0 before being mixed with the flakes. The infrared

spectrum of scaffolds made using this dispersion formula-

tion shows the relative retention in intensity of phosphate

peaks. Importantly, the HA in the scaffold is also found

to contain carbonate-substitution (Fig. 3C). Transmission

electron micrographs also revealed the preservation of

nano-sized crystals in the interior of the membrane used to

make scaffolds (Fig. 4B). Electron diffraction of the crys-

tals in these composite scaffolds verified the presence of the

HA phase (Fig. 5C and D). By adjusting the collagen dis-

persion to a pH of 5.0 before mixing the biomimetic

composite flakes, the HA component can be preserved and

scaffolds with nano-sized carbonate-substituted HA crys-

tals present in the intermicrofibrillar gaps can be entrapped

in the porous collagen matrix.

As shown in Fig. 7, these scaffolds are highly porous

matrices with interconnected pores. The pores of the scaf-

fold are formed when the aqueous component of the dis-

persion is frozen. The dendritic growth of ice crystals

aggregates and locks the collagen microfibrils and biomi-

metic composite flakes into the interstices of the inter-

connected ice crystal network. The collagen microfibrils

can weave through the rough surface topography of the

flakes and mechanically interlock or entrap them into the

porous matrix. As a consequence, the porous matrix con-

sists of a network of collagen fibers with flakes of biomi-

metic composite randomly suspended throughout.

Incorporating microchannels into the matrix of the

scaffold, which can allow for the flow of cell culture media,

FIG. 7. Scanning electron micrograph of a scaffold made with

biomimetic collagen-hydroxyapatite flakes. The flakes (F) are

found suspended in the porous collagen matrix.

TABLE 2. CHANGE IN PHOSPHATE BAND INTENSITY

OF BIOMIMETIC COLLAGEN-HYDROXYAPATITE PROCESSED

WITH PH 3.2 AND PH 5.0 DISPERSIONS*

Absorption

Ratio

Unpro-

cessed

Processed

at pH 3.2

Processed

at pH 5.0

Phosphate v3antisymmetric/

amide I 1.167 0.926 (�21%) 1.171 (�0.34%)

Phosphate v4antisymmetric

bending/

amide I 0.969 0.912 (�5%) 0.963 (�0.61%)

Phosphate

v4/amide I 1.010 0.921 (�8.8%) 0.988 (�2%)

*Percentage change relative to unprocessed biomimetic collagen-

hydroxyapatite membrane shown in parentheses.

FIG. 8. Scanning electron micrograph of scaffold with micro-

channels. The sample is tilted at 45 degrees to show the depth of the

island and channel features.

COLLAGEN SCAFFOLDS REINFORCED WITH BIOMIMETIC COMPOSITE 2485

has important implications in maintaining cell survival deep

within the scaffold. One such collagen-HA composite

scaffold is shown in Fig. 8. Future work will focus on using

such scaffolds to localize the relevant cells, osteogenic cells

in the case of bone tissue engineering, in the island regions

while cell culture media are pumped through the micro-

channels. Lining the lumen of the microchannels with an-

giogenic growth factors or endothelial and smooth muscle

cells will also be considered. Using the microchannels to

create artificial vasculature within the scaffold has great

advantages in overcoming the diffusion constraints facing

conventional foam scaffolds.

CONCLUSION

Composite scaffolds that closely resemble the composi-

tion and microstructural organization of collagen and HA in

bone can be fabricated by combining a diffusion-based

precipitation technique that creates biomimetic collagen-

HA membranes with a collagen scaffold manufacturing

technique. The biomimetic precipitation method creates

carbonate-substituted nano-sized HA crystals in the interior

of a collagen membrane. Flakes of this composite mem-

brane are mixed into collagen dispersion and then cast into

molds. To reduce the dissolution of the HA component due

to the acidic nature (pH of 3.2) of the collagen dispersion,

the dispersion needs to be adjusted to a pH of at least 5.0

before mixing of the composite membrane flakes.

Microchannel features can be added to the scaffold with a

sacrificial mold that has been fabricated with a 3D printer.

Such microchannels could assist in the perfusion of the

scaffold to increase the mass transport of essential meta-

bolic components and removal of waste products deep in-

side the scaffolds.

ACKNOWLEDGMENTS

The authors would like to acknowledge funding from the

Wellcome Trust under grant no. 074486. E. Sachlos thanks

the Bodossaki Foundation for financial support. D. Gotora

thanks the Rhodes Trust for financial support. The authors

thank TEOX Ltd for the supply of BioBuild and BioSup-

port materials, and Professor G.D.W. Smith for the provi-

sion of laboratory facilities.

REFERENCES

1. LeGeros, R.Z. Properties of osteoconductive biomaterials:

calcium phosphates. Clin. Orthop. Rel. Res. 395, 81, 2002.

2. Gelse, K., Poschl, E., and Aigner, T. Collagens—structure, func-

tion, and biosynthesis. Adv. Drug Deliv. Rev. 55, 1531, 2003.

3. Bocciare, D.S. Morphology of crystallites in bone. Calc.

Tissue Res. 5, 261, 1970.

4. Weiner, S., and Price, P.A. Disaggregation of bone into

crystals. Calc. Tissue Int. 39, 365, 1986.

5. Robinson, R.A. An electron-microscopic study of the crys-

talline inorganic component of bone and its relationship to the

organic matrix. J. Bone Joint Surg. Am. 34-A, 389, 1952.

6. Johansen, E., and Parks, H.F. Electron microscopic observa-

tions on the three-dimensional morphology of apatite crys-

tallites of human dentine and bone. J. Biophys. Biochem.

Cytol. 7, 743, 1960.

7. Weiner, S., and Traub, W. Crystal size and organization in

bone. Connect. Tissue Res. 21, 589, 1989.

8. Landis, W.J., Song, M.J., Leith, A., McEwen, L., and

McEwen, B.F. Mineral and organic matrix interaction in

normally calcifying tendon visualized in 3 dimensions by

high-voltage electron-microscopic tomography and graphic

image-reconstruction. J. Struct. Biol. 110, 39, 1993.

9. Katz, E.P., and Li, S.T. Intermolecular space of reconstituted

collagen fibrils. J. Molec. Biol. 73, 351, 1973.

10. Fraser, R.D.B., Macrae, T.P., Miller, A., and Suzuki, E.

Molecular-conformation and packing in collagen fibrils. J.

Molec. Biol. 167, 497, 1983.

11. Weiner, S., and Traub, W. Organization of hydroxyapatite

crystals within collagen fibrils. FEBS Lett. 206, 262, 1986.

12. Driessens, F.C.M. Formation and stability of calcium phos-

phates in relation to the phase composition of the mineral in

calcified tissue. In: deGroot, K., ed. Bioceramics of Calcium

Phosphates. Boca Raton, FL: 1983, pp. 1–32.

13. Rey, C., Renugopalakrishnan, V., Collins, B., and Glimcher,

M.J. Fourier-transform infrared spectroscopic study of the

carbonate ions in bone-mineral during aging. Calc. Tissue Int.

49, 251, 1991.

14. Legeros, R.Z. Apatites in biological-systems. Progr. Crystal

Growth Character. Mat. 4, 1, 1981.

15. Zapanta-LeGeros, R. Effect of carbonate on the lattice pa-

rameters of apatite. Nature 206, 403, 1965.

16. Baig, A.A., Fox, J.L., Young, R.A., Wang, Z., Hsu, J.,

Higuchi, W.I., Chhettry, A., Zhuang, H., and Otsuka, M.

Relationships among carbonated apatite solubility, crystallite

size, and microstrain parameters. Calc. Tissue Int. 64, 437,

1999.

17. Nelson, D.G.A., Barry, J.C., Shields, C.P., Glena, R., and

Featherstone, J.D.B. Crystal morphology, composition, and

dissolution behavior of carbonated apatites prepared at con-

trolled pH and temperature. J Colloid Interface Sci. 130, 467,

1989.

18. Okazaki, M., Moriwaki, Y., Aoba, T., Doi, Y., and Takahashi,

J. Solubility behavior Of Co3 apatites in relation to crystal-

linity. Caries Res. 15, 477, 1981.

19. Eggli, P.S., Muller, W., and Schenk, R.K. Porous

hydroxyapatite and tricalcium phosphate cylinders with 2

different pore-size ranges implanted in the cancellous bone of

rabbits: a comparative histomorphometric and histologic-

study of bony ingrowth and implant substitution. Clin Orthop

Rel. Res. 232, 127, 1988.

20. Barralet, J., Akao, M., Aoki, H., and Aoki, H. Dissolution of

dense carbonate apatite subcutaneously implanted in Wistar

rats. J. Biomed. Materials Res. 49, 176, 2000.

21. Lynn, A.K., Yannas, I.V., and Bonfield, W. Antigenicity and

immunogenicity of collagen. J. Biomed.Materials Res Part B

Appl. Biomaterials 71B, 343, 2004.

2486 SACHLOS ET AL.

22. Meade, K.R., and Silver, F.H. Immunogenicity of collagenous

implants. Biomaterials 11, 176, 1990.

23. Furthmayr, H., and Timpl, R. Immunochemistry of collagens

and procollagens. Int. Rev. Connect. Tissue Res. 7, 61, 1976.

24. HardinYoung, J., Carr, R.M., Downing, G.J., Condon, K.D.,

and Termin, P.L. Modification of native collagen reduces

antigenicity but preserves cell compatibility. Biotechnol.

Bioengin. 49, 675, 1996.

25. Obrien, T.K., Gabbay, S., Parkes, A.C., Knight, R.A., and

Zalesky, P.J. Immunological reactivity to a new glutaralde-

hyde tanned bovine pericardial heart-valve. Trans. Am. Soc.

Artificial Intern. Organs 30, 440, 1984.

26. Dedhar, S., Ruoslahti, E., and Pierschbacher, M.D. A cell-

surface receptor complex for collagen type-I recognizes the

arg-gly-asp sequence. J. Cell Biol. 104, 585, 1987.

27. Staatz, W.D., Fok, K.F., Zutter, M.M., Adams, S.P., Ro-

driguez, B.A., and Santoro, S.A. Identification of a tetrapep-

tide recognition sequence for the alpha-2-beta-1-integrin in

collagen. J. Biol. Chem. 266, 7363, 1991.

28. Tuckwell, D., and Humphries, M. Integrin-collagen binding.

Semin. Cell Develop. Biol. 7, 649, 1996.

29. Balian, G., Click, E.M., Crouch, E., Davidson, J.M., and

Bornstein, P. Isolation of a collagen-binding fragment from

fibronectin and cold-insoluble globulin. J. Biol. Chem. 254,

1429, 1979.

30. Hahn, L.H.E., and Yamada, K.M. Identification and isolation

of a collagen-binding fragment of the adhesive glycoprotein

fibronectin. Proc. Natl. Acad. Sci. U. S. A. 76, 1160, 1979.

31. Ruoslahti, E., Hayman, E.G., Kuusela, P., Shively, J.E., and

Engvall, E. Isolation of a tryptic fragment containing the

collagen-binding site of plasma fibronectin. J. Biol. Chem.

254, 6054, 1979.

32. Mcdonald, J.A., and Kelley, D.G. Degradation of fibronectin

by human-leukocyte elastase: release of biologically-active

fragments. J. Biol. Chem. 255, 8848, 1980.

33. Klebe, R.J. Isolation of a collagen-dependent cell attachment

factor. Nature 250, 248, 1974.

34. Pearlstein, E. Plasma-membrane glycoprotein which mediates

adhesion of fibroblasts to collagen. Nature 262, 497, 1976.

35. Kleinman, H.K., McGoodwin, E.B., Martin, G.R., Klebe,

R.J., Fietzek, P.P., and Woolley, D.E. Localization of binding-

site for cell attachment in alpha-1(i) chain of collagen. J. Biol.

Chem. 253, 5642, 1978.

36. Ali, I.U., and Hynes, R.O. Effects of lets glycoprotein on cell

motility. Cell 14, 439, 1978.

37. Postlethwaite, A.E., Seyer, J.M., and Kang, A.H. Chemotactic

attraction of human fibroblasts to type-i, type-2, and type-3

collagens and collagen-derived peptides. Proc. Natl. Acad.

Sci. U. S. A. 75, 871, 1978.

38. Liao, S.S., Cui, F.Z., Zhang, W., and Feng, Q.L. Hierarchi-

cally biomimetic bone scaffold materials: nano-HA/collagen/

PLA composite. J. Biomed. Materials Res. Part B Appl.

Biomaterials 69B, 158, 2004.

39. Liao, S., Watari, F., Uo, M., Ohkawa, S., Tamura, K., Wang,

W., and Cui, F.Z. The preparation and characteristics of a

carbonated hydroxyapatite/collagen composite at room tem-

perature. J. Biomed. Materials Res. Part B Appl. Biomaterials

74B, 817, 2005.

40. Zhao, F., Yin, Y.J., Lu, W.W., Leong, J.C., Zhang, W.J.,

Zhang, J.Y., Zhang, M.F., and Yao, K.D. Preparation and

histological evaluation of biomimetic three-dimensional

hydroxyapatite/chitosan-gelatin network composite scaffolds.

Biomaterials 23, 3227, 2002.

41. Du, C., Cui, F.Z., Zhang, W., Feng, Q.L., Zhu, X.D., and de

Groot, K. Formation of calcium phosphate/collagen com-

posites through mineralization of collagen matrix. J. Biomed.

Materials Res. 50, 518, 2000.

42. Sachlos, E., Reis, N., Ainsley, C., Derby, B., and Czernuszka,

J.T. Novel collagen scaffolds with predefined internal mor-

phology made by solid freeform fabrication. Biomaterials 24,

1487, 2003.

43. IshaugRiley, S.L., Crane, G.M., Gurlek, A., Miller, M.J.,

Yasko, A.W., Yaszemski, M.J., and Mikos, A.G. Ectopic

bone formation by marrow stromal osteoblast transplantation

using poly(DL-lactic-co-glycolic acid) foams implanted into

the rat mesentery. J. Biomed. Materials Res. 36, 1, 1997.

44. Freed, L.E., and Vunjak-Novakovic, G. Culture of organized

cell communities. Adv. Drug Deliv. Rev. 33, 15, 1998.

45. Schoeters, G., Leppens, H., Vangorp, U., and Vandenheuvel,

R. Hematopoietic long-term bone-marrow cultures from adult

mice show osteogenic capacity in vitro on 3-dimensional

collagen sponges. Cell Prolif. 25, 587, 1992.

46. Lawson, A.C., and Czernuszka, J.T. Collagen-calcium phos-

phate composites. Proc. Institution of Mechan. Engineers Part

H J. Engin. Med. 212, 413, 1998.

47. Rehman, I., and Bonfield, W. Characterization of hydroxyap-

atite and carbonated apatite by photo acoustic FTIR spectros-

copy. J. Materials Sci. Materials Med. 8, 1, 1997.

48. Yannas, I.V. Collagen and gelatin in solid-state. J. Macro-

molec. Sci. Rev. Macromolec. Chem. Physics C 7, 49, 1972.

49. Doyle, B.B., Bendit, E.G., and Blout, E.R. Infrared spec-

troscopy of collagen and collagen-like polypeptides. Bio-

polymers 14, 937, 1975.

50. Elliott, J.C., Holcomb, D.W., and Young, R.A. Infrared de-

termination of the degree of substitution of hydroxyl by car-

bonate ions in human dental enamel. Calc. Tissue Int. 37, 372,

1985.

Address reprint requests to:

Eleftherios Sachlos, D.Phil.

Department of Materials

University of Oxford

Oxford, United Kingdom

E-mail: eleftherios.sachlos@materials.ox.ac.uk

COLLAGEN SCAFFOLDS REINFORCED WITH BIOMIMETIC COMPOSITE 2487