IOP PUBLISHING

BIOMEDICAL MATERIALS

doi:10.1088/1748-6041/2/2/013

Biomed. Mater. 2 (2007) 142–150

An electrospun triphasic nanofibrous scaffold for bone tissue engineering S A Catledge1, W C Clem2, N Shrikishen1,4 , S Chowdhury1, A V Stanishevsky1, M Koopman3 and Y K Vohra1 1

Department of Physics, University of Alabama at Birmingham, Birmingham, AL 35294-1170, USA Department of Biomedical Engineering, University of Alabama at Birmingham, Birmingham, AL 35294-4440, USA 3 Department of Materials Science and Engineering, University of Alabama at Birmingham, Birmingham, AL 35294-4461, USA 2

E-mail: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected] and [email protected]

Received 14 March 2007 Accepted for publication 17 April 2007 Published 8 May 2007 Online at stacks.iop.org/BMM/2/142 Abstract A nanofibrous triphasic scaffold was electrospun from a mixture of polycaprolactone (PCL), type-I collagen and hydroxyapatite nanoparticles (nano-HA) with a mixture dry weight ratio of 50/30/20, respectively. Scaffolds were characterized by evaluating fiber morphology and chemical composition, dispersion of HA particles and nanoindentation. Scanning electron microscopy revealed fibers with an average diameter of 180 ± 50 nm, which coincides well with the collagen fiber bundle diameter characteristic of the native extracellular matrix of bone. The triphasic fibers, stained with calcein and imaged with confocal microscopy, show a uniform dispersion of apatite particles throughout their length with minor agglomeration. Scaffold fibers of triphasic (50/30/20), collagen/nano-HA (80/20), PCL/nano-HA (80/20), pure PCL and pure collagen were each pressure consolidated into non-porous pellets for evaluation by transmission electron microscopy and nanoindentation. While the majority of apatite particles are uniformly dispersed having an average size of 30 nm, agglomerated particles as large as a few microns are sparsely distributed. Nanoindentation of the pressure-consolidated scaffolds showed a range of Young’s modulus (0.50–3.9 GPa), with increasing average modulus in the order of (PCL < PCL/nano-HA < collagen < triphasic < collagen/nano-HA). The modulus data emphasize the importance of collagen and its interaction with other components in affecting mechanical properties of osteoconductive scaffolds. (Some figures in this article are in colour only in the electronic version)

1. Introduction Bone tissue repair accounts for approximately 500 000 surgical procedures per year in the United States alone [1]. The search for successful bone analogue materials has led many researchers to prepare porous scaffolds with the intent to mimic as closely as possible the composition and/or structure of the extracellular matrix (ECM) of natural bone. The major solid components of human bone are collagen (type I) 4

Current address: 120 Sarah Hughes Drive, Madison, AL 35758, USA.

1748-6041/07/020142+09$30.00

and a biological apatite that differs in composition from the stoichiometric hydroxyapatite [HA, Ca10(PO4)6(OH)2] by the presence of other ions, of which carbonate is the most abundant species (∼8 wt%). The biological apatite is nanostructured, with plate-like crystals approximately 2– 7 nm thick, 15–200 nm long and 10–80 nm wide. Compared with conventionally crystallized HA, nanocrystalline HA (nano-HA) has been shown to promote greater osteoblast adhesion and proliferation, improve osseointegration and increase deposition of calcium-containing minerals on its surface [2]. Although HA may provide strength to a bone

© 2007 IOP Publishing Ltd Printed in the UK

142

An electrospun triphasic nanofibrous scaffold for bone tissue engineering

analogue material, monolithic HA is inherently brittle and has therefore been limited to non-load-bearing applications. The nanostructure of bone is also manifested in the form of mineralized collagen fibrils of about 100 nm in diameter. Collagen, a major component of the extracellular matrix, is easily degraded and resorbed by the body and allows good attachment to cells. Since collagen proteins are a major structural element in so many of the body’s tissues and organs, collagen fibers are a logical choice for scaffolds. In bone, apatite nanocrystals are grown in intimate contact with the collagen fibers, and this may explain why a simple ‘rule of mixtures’ approach has been only moderately successful in describing the mechanical properties of bone [3–5]. In fact, the relationship between bone stiffness and mineral content is nonlinear [6]. The mechanical behavior depends not only on relative mineral/organic content, degree of porosity and degree of crosslinking, but also on the physical and chemical interaction between its components. For example, mineral crystals in bone grow with a specific crystalline orientation— the c-axis of the crystals is roughly parallel to the long axis of the collagen fibrils, and this orientation relationship is expected to influence mechanical properties [7]. Generally, the strengthening effect of HA can be explained by the fact that the collagen matrix is a load transfer medium and therefore transfers the load to the intrinsically rigid apatite crystals. In addition, the apatite deposits between tangled fibrils ‘cross-link’ the fibers through mechanical interlocking or by forming calcium ion bridges, thus increasing the resistance to deformation of the collagenous fiber network [8]. Since apatite is a rigid material not capable of dissipating much energy, collagen is thought to play a major role in providing resistance to fracture (i.e. high toughness). Regardless of the final tissue application, scaffolds are a requisite for tissue growth by providing a temporary, artificial extracellular matrix or structural support framework, onto which bone cells can attach and proliferate to form tissues. Therefore, scaffolds should ideally be mechanically and biologically similar to natural bone and should exhibit controlled degradation as a new bone matrix is formed. In order to mimic bone structure and composition, researchers have prepared collagen/HA bone analogue composites using a variety of conventional methods such as direct mineralization or chemical precipitation [9–17]. Elastic modulus is a property of interest for such composites since the scaffold stiffness should be mechanically similar to the surrounding bone. Animal skeletal bones have been shown to exhibit elastic modulus over a wide range up to 50 GPa, depending on several factors such as anatomical location and porosity [18]. Values of Young’s modulus up to approximately 3 GPa have been reported for some collagen/HA composites, such as cross-linked multilayer sheets prepared by enzymatic mineralization [16] or those made with up to 40% HA by direct mineralization [17], both composites having nearly zero porosity. In another study, a collagen/HA composite showing a bone-like self-organized nanostructure (with the c-axis alignment of HA nanocrystals along fibers) was made by a co-precipitation technique [13]. The measured Young’s modulus was reported to be 2.5 GPa after consolidating

the composite by uniaxial and isostatic compression. A triphasic (nano-HA/collagen/poly(lactic acid)) 3D porous bone scaffold material prepared by biomimetic synthesis was reported by Liao et al [19]. The 3D porous scaffold showed features of natural bone in terms of primary composition and hierarchical microstructure, but was not fibrous and had a measured compressive modulus of only 50–60 MPa. In the present study, we use the technique of electrospinning to prepare a nanostructured triphasic (polycaprolactone/collagen/nano-HA) fibrous composite and compare to electrospun collagen, polycaprolactone (PCL), collagen/HA and PCL/HA composites. The electrospinning technique is capable of producing a nanofibrous structure with interconnected pores and can result in a high surface area-tovolume ratio for enhanced cell attachment compared to other techniques [20, 21]. Poly(ε-caprolactone) is a semicrystalline, resorbable aliphatic polyester that has a much lower rate of degradation than collagen and may be useful in a composite scaffold to fine tune this rate of degradation. Collagen/PCL electrospun scaffolds have been reported to be useful in matching both initial mechanical properties and degradation rates, particularly for vascular tissue engineering applications [22–24]. Collagen/HA scaffolds prepared with up to 20 wt% HA having a 100–150 nm particle size and 500–700 nm average fiber diameter have previously been reported using a co-electrospinning technique [25]. Recently, we have used sol-gel methods to incorporate nano-HA particles (15– 80 nm particle size) into the solution mixture for coelectrospinning [26]. In this study, we report on triphasic fibrous nanocomposite scaffolds with average fiber diameter below 200 nm. In addition, we show that the polymer/HA composites have an average HA particle size of 30 nm and exhibit an elastic modulus up to 3.9 GPa. The agglomeration of HA particles is discussed.

2. Materials and methods Collagen type I (lyophilized, from calf skin) was purchased from MP Biomedicals. PCL pellets were purchased from Absorbable Polymers, Pelham, AL. The intrinsic viscosity of the PCL in CH3Cl at 30 ◦ C is 1.08 dL g−1, as listed by the manufacturer. The solvent used to dissolve both collagen and PCL was 1,1,1,3,3,3-hexafluoro-2-propanol (HFP), obtained from Sigma-Aldrich. The precursors for the hydroxyapatite nanoparticles’ synthesis (Ca(OH)2, H3PO4 and NH4OH) were purchased from Alfa Aesar and used as received. 2.1. Preparation of hydroxyapatite nanoparticles In a general approach, hydroxyapatite nanoparticles can be prepared by the chemical reaction of several calcium and phosphorus compounds [11, 27, 28]. The size and shape of formed hydroxyapatite nanoparticles crucially depend on the procedure and parameters such as precursor stoichiometry, solution temperature, pH and aging time. Hydroxyapatite nanoparticles, in our case, were synthesized by chemical precipitation at room temperature using Ca(OH)2 and H3PO4 as starting materials. The amounts 143

S A Catledge et al

of both chemicals were chosen to obtain the Ca/P ratio of 1.67 in the reaction product. The solution was stirred during the reaction while maintaining the pH in the range 10–11 using NH4OH. The reaction was performed in open air. The assynthesized hydroxyapatite nanoparticles were aged for 24 h, then washed several times in deionized water and methanol using a centrifuge and finally dispersed in HFP (1 g nanoHA per 25 ml HFP) for further use. The nano-HA solution was ultrasonically agitated for 1 h prior to its addition to the polymer/HFP mixture in order to provide good particle dispersion. 2.2. Preparation of electrospun scaffolds The following composite mixtures were made (percentage of total solid weight): PCL (80%) + nano-HA (20%), collagen (80%) + nano-HA (20%) and PCL (50%) + collagen (30%) + nano-HA (20%). Composite solutions were made by adding HFP to each mixture such that the solid weight ranged from 5 to 10% of the total solution weight. Pure collagen and pure PCL solutions were also made with HFP as a solvent and the same 5–10% solid weight content. The solution mixtures were magnetically stirred at room temperature for 1 h. For each mixture, the solution was drawn into a 5 mL plastic syringe with a 3.8 cm long, 27 gauge blunt-tip needle. The loaded syringe was placed in an electrically grounded syringe pump (model PHD2000, Harvard Apparatus, MA) and the needle was connected to a dc voltage power supply (Model ES60P-20W/DAM, Gamma High Voltage Research Inc., FL). An electrically grounded aluminum foil collector sheet (approximately 8 cm × 8 cm square) was placed 12–13 cm from the tip of the spray needle. The tip of the electrically charged 27-gauge needle and the grounded collector sheet were fully enclosed within a clear Lexan box in order to minimize influence from air currents. An exhaust hose pulling a weak vacuum was attached to the Lexan box to help remove solvent vapors during electrospinning. Fibrous, porous scaffolds were produced by electrospinning the solutions using a voltage of 15 kV (for PCL, PCL/HA and triphasic scaffolds) and 22 kV (for collagen and collagen/HA scaffolds) at a flow rate of 3 ml h−1. Higher voltage was found to be necessary in order to effectively spin the collagen-based mixtures without fiber beading. The resulting samples were randomly arranged fibers deposited as a sheet with estimated maximum thickness between 50 and 100 µm on an aluminum foil. No chemicalor radiation-induced cross-linking of collagen fibers was performed in this study. The scaffold sheets were refrigerated until further study. For each scaffold sheet, approximately half of the scaffold was peeled from the aluminum foil and uniaxially pressed (at 1000 psi) into a consolidated cylindrical pellet (approximately 3 mm diameter and 3 mm long) using a hydraulic press.

highest. Scanning electron microscopy (SEM, Philips 515) was performed at an accelerating voltage of 10 kV on these areas after first sputter coating with gold-palladium. The average fiber diameter was determined from SEM images taken up to 10 000× magnification by measuring the width across several fibers at locations closely intercepting equally spaced points on a 5 × 5 grid overlaid onto the images. Asspun and pressure-consolidated scaffolds were imaged using both secondary and back-scattered electron detection in order to obtain topography and atomic number contrast, respectively. Energy dispersive x-ray microanalysis (EDX) was also used to determine relative amounts of Ca, P, C and O atomic species at select locations in order to detect HA-rich or polymer-rich regions on the surface. Fiber chemical bonding structure was analyzed using Fourier transform infrared spectroscopy (FTIR, Bruker Optics Tensor 27 with a CsI beamsplitter and variable angle reflection unit). Square areas (approximately 10 mm on edge) were cut from near the edge of the electrospun sheet deposits where thickness was low in order to allow for measurement in the reflectance mode. The fiber crystalline structure was analyzed before and after uniaxial compression using x-ray diffraction (XRD) (Philips MPD, Cu-Kα anode). The as-spun samples were 10 mm square sheets cut from the center region of the electrospun sheet deposit where the fiber collection density was the highest. The nano-HA particle size was estimated from the Scherrer equation [29] by using the full-width at half-maximum (FWHM) of the (0 0 2) reflection at 2θ = 26◦ . In order to observe the distribution of nano-HA particles within the triphasic scaffold, the fibers were stained overnight with 5 µg mL−1 calcein. Imaging was performed on a Leica DMIRBE inverted epifluorescence/Nomarski microscope outfitted with Leica TCS NT SP1 Laser Confocal optics (Leica. Inc., Exton, PA). The system is equipped with an argon-ion laser for the imaging of calcein (green fluorescence). Precise control of excitation (488 nm) and emission (500–550 nm) was respectively afforded by an acousto-optical tunable filter and a TCS SP1 prism spectrophotometer. Optical sections through the Z-axis were generated using a stage galvanometer. During image acquisition, the format size was set to 1024 × 1024 pixels for high resolution, and the pinhole aperture was set to 200 µm. A control scaffold (calcein-stained, but without nano-HA) emitted only a very weak green fluorescence, with the intensity being similar to the autofluorescence emitted by an unstained scaffold. In order to determine HA particle size and distribution, the pressure-consolidated collagen/HA scaffold was prepared by sectioning the pellet using a Reichert Jung ultra-microtome to approximately 100 nm thickness, placed on a carboncoated formvar copper grid and bright field imaged in an FEI TecnaiTM T12 transmission electron microscope (TEM). The accelerating voltage was 60 kV.

2.3. Fiber morphology and structure analysis For each scaffold type, a square area (approximately 10 mm on edge) was cut from the center region of the electrospun sheet deposit where the fiber collection density was the 144

2.4. Elastic modulus by nanoindentation Nanoindentation (Nanoindenter XP, MTS Instruments) was used to measure Young’s modulus of uniaxially pressed

An electrospun triphasic nanofibrous scaffold for bone tissue engineering

(a )

(a )

(c )

(b )

(d )

(b )

Figure 2. Higher magnification scanning electron microscopy images of (a) PCL, (b) triphasic, (c) PCL/HA and (d) collagen/HA electrospun nanofibrous scaffolds.

Figure 1. Scanning electron microscopy image typical of the electrospun polymer/HA composite scaffolds imaged with (a) secondary electrons and (b) back-scattered electrons.

scaffold samples. The as-deposited samples were 10 mm square sheets cut from near the center of the electrospun sheet deposits. The indenter system was calibrated using a fused silica standard sample (measured modulus of 70.5 GPa). A Berkovich diamond indenter with a total included angle of 142.3◦ and 50 nm nominal radius was used for all measurements. A 10 s hold time at maximum load and 50 s hold time during unloading (10% of maximum load) were used in order to minimize thermal drift. We have used this approach successfully in characterizing other bio-ceramic scaffold materials [30]. The sample stiffness was measured continuously during loading and unloading segments, allowing for determination of elastic modulus continuously as a function of indenter depth displacement. The data sets were processed using proprietary software to produce load-displacement curves, from which the elastic modulus was calculated as a function of indentation depth. The values for elastic modulus of pressed scaffolds were averaged from depth segments of 500–1500 nm for between 20 and 40 indents per sample, thus providing a strong statistical base.

3. Results and discussion 3.1. Structure and morphology of electrospun fibers Figure 1 shows SEM images taken by detection of (a) secondary electrons and (b) back-scattered electrons (BSE) for a typical polymer/HA electrospun scaffold in this study. Magnification was chosen to be low enough to observe large HA particles (up to several microns in diameter) sparsely distributed throughout the fibers. The BSE detection (showing

a brighter contrast in areas of a higher average atomic number) was useful to emphasize not only the presence of these large particles, but also to show the relatively uniform dispersion of nanosize HA particles within the fibers. The presence of calcium and phosphorous, attributed to hydroxyapatite, was confirmed by energy dispersive x-ray microanalysis (EDX, not shown) for the beam focused on the large particles as well as on the fibers. The EDX calcium/carbon intensity ratio was a factor of 2.6 higher for the beam focused on the large particle shown in figure 1(a) than for the beam focused on the surrounding fibers. The presence of calcium, phosphorous, oxygen and carbon was confirmed for all locations probed on the scaffold surface. It is clear from figure 1 that agglomeration of nanoHA into large micron-size particles (although somewhat sparsely dispersed) remains an issue to be resolved. Higher concentrations of HA in the polymer/HA electrospinning solution mixture will likely be more challenging as particle settling and agglomeration may occur. Continuous agitation/stirring of the solution during electrospinning may help in this regard. Figure 2 shows higher magnification SEM images for (a) PCL, (b) triphasic, (c) PCL/HA and (d) collagen/HA electrospun scaffolds. All scaffolds show a random, nonwoven porous mat of fibers characteristic of that produced by the electrospinning process [20, 21]. The average fiber diameter for the triphasic scaffold is 180 ± 50 nm, which coincides well with the protein fiber diameter of up to several hundred nanometers, characteristic of the native extracellular matrix (ECM) of bone. The average fiber diameter varies between scaffold samples with the triphasic scaffold having the smallest average diameter, followed by PCL, collagen+HA and then PCL+HA. The triphasic, collagen/HA and PCL scaffolds have a relatively narrow distribution in fiber diameter compared to the PCL/HA scaffold. The triphasic fiber diameter reported here approaches the lower limit of diameters reported in the literature (∼100 nm) for collagen and collagen-based 145

S A Catledge et al

(d)

(c)

(b)

(a)

Figure 3. Confocal micrograph of triphasic scaffold after staining with calcein to emphasize the presence of HA (shown as green fluorescence).

electrospun fibers [31]. Solution viscosity (as controlled in part by polymer concentration) has been found to be one of the most influential parameters on fiber size and morphology [21]. Generally, for polymer fibers, higher polymer concentrations lead to larger fiber diameter. For our case of composite scaffolds, solution viscosity may be a more complicated interplay between overall polymer concentration and different blending ratios between components. Accelerating voltage and solution conductivity also influence fiber diameter, although the correlation is not universal for all polymers [32–35]. Figure 3 shows a confocal micrograph for the triphasic scaffold after staining the fibers with calcein to reveal green fluorescence attributed to hydroxyapatite particles. The overall distribution of HA particles observed on this scale is fairly uniform throughout the fibers, with only minor agglomeration of micron-size particles. A similar scaffold was made without nano-HA as a control and stained with calcein. The control scaffold (not shown) emitted only a very weak green fluorescence, with the intensity being similar to the auto-fluorescence emitted by an unstained scaffold. Infrared spectroscopic analysis was done to characterize functional groups in the fibers in order to confirm the presence of the scaffold component phases and to discern any possible chemical modification or interaction between phases. Figure 4 shows FTIR spectra for (a) PCL, (b) triphasic, (c) PCL/HA and (d) collagen/HA scaffolds (corresponding to figure 2). Typical infrared bands for PCL-related stretching modes are observed for the PCL and PCL/HA scaffolds. These include 2949 cm−1 (asymmetric CH2 stretching), 2865 cm−1 (symmetric CH2 stretching), 1727 cm−1 (carbonyl stretching), 1293 cm−1 (C–O and C–C stretching in the crystalline phase) and 1240 cm−1 (asymmetric COC stretching). Collagen amide I (C=O stretching at 1648 cm−1) and amide II (N–H deformation at 1542 cm−1) bands are seen for the triphasic and collagen/HA scaffolds, respectively. Collagen amide A (N–H stretching at 3310 cm−1) and amide B (C–H stretching at 3068 cm−1) were also observed (but not shown here). Finally, characteristic PO3− 4 absorption bands attributed to HA nanoparticles are seen 146

Figure 4. FTIR spectra of (a) PCL, (b) triphasic, (c) PCL/HA and (d) collagen/HA electrospun nanofibrous scaffolds.

for each of PCL/HA, collagen/HA and triphasic scaffolds. −1 −1 −1 These PO3− 4 bands are seen at 562 cm , 602 cm , 961 cm , −1 −1 1029 cm and 1104 cm . In a previous FTIR study comparing ‘bulk’ collagen with electrospun collagen scaffolds, Stanishevsky et al [36] reported a shift to lower wavenumbers in electrospun collagen of the major peak in the amide I band from 1667 cm−1 (fibers) to 1642 cm−1 (bulk), as well as the peak of amide II band from 1574 cm−1 (fibers) to 1544 cm−1 (bulk). We also observe similar amide I and amide II peak shifts for our collagen/HA and triphasic scaffolds. It was suggested that changes in the triple helix structure of collagen fibrils during the fiber drawing in the electrified jet may account for this infrared shift. Other changes in the infrared spectra were reported (such as a blue shift of the amide I band) upon addition of HA nanoparticles in electrospun collagen fibers. The observed changes in the infrared spectra were associated with a certain degree of interaction between the HA nanoparticles and collagen [36]. To investigate the HA crystalline phase structure and particle size, XRD was performed. Figure 5 shows XRD patterns for (a) pure PCL, (b) PCL/HA, (c) triphasic and (d) collagen/HA scaffolds. The two most intense peaks at 21.4◦ and 23.8◦ were indexed to PCL. For reference, figure 5(e) shows the XRD pattern of the as-synthesized nano-HA particles, which was indexed to the hexagonal hydroxyapatite. The hydroxyapatite phase in the assynthesized nano-HA and in the electrospun scaffolds is poorly crystalline and shows broad XRD peaks indicative of nanosize particles. The most intense HA peak corresponds to the (2 1 1) reflection at 2θ = 32◦ and is not well resolved from nearby HA reflections. The HA mean particle size was estimated for each scaffold using Lorentzian fitting of the well-resolved (0 0 2) reflection at 2θ = 26◦ and the well-known Scherrer equation [29]. The estimated mean particle size averaged over each scaffold was 31 ± 2 nm. The HA peaks present in the as-synthesized powder are also present with similar relative intensities in the collagen/HA, PCL/HA and triphasic electrospun scaffolds. As mentioned earlier, the electrospun scaffolds were consolidated into non-porous pellets by uniaxial compression

An electrospun triphasic nanofibrous scaffold for bone tissue engineering

(a )

(a) (b) (d) (e)

Figure 5. XRD spectra for (a) PCL, (b) PCL/HA, (c) triphasic, (d) collagen/HA and (e) as-synthesized nano-HA powder. The hydroxyapatite phase in the as-synthesized nano-HA and in the electrospun scaffolds shows broad XRD peaks indicative of nanosize particles. (b)

(a) (b)

Figure 6. XRD of (a) collagen/HA and (b) PCL/HA after consolidating fibers by uniaxial compression. No apparent change in crystallinity of the polymer or HA phases is observed as a result of compression.

to facilitate nanoindentation testing. In order to confirm that no substantial crystalline modifications occurred during this consolidation, XRD was performed on the scaffolds after compression. Figure 6 shows two XRD patterns from pressed (a) PCL/HA and (b) collagen/HA scaffolds. No changes in phase structure or relative peak intensities were detected for the pressed scaffolds when compared to the as-spun fiber scaffolds. Figure 7 shows bright-field TEM images of the pressure-consolidated collagen/HA scaffold at low and high magnifications, revealing the size and dispersion of HA particles. Although the majority of HA particles are well dispersed with an average diameter of 30 nm, some agglomeration of particles (up to several hundred nanometers diameter) can also be seen on this scale. The size of the majority of HA particles imaged by TEM is consistent with the average particle size estimated from XRD data using the Scherrer equation (31 ± 2 nm). 3.2. Nanoindentation Difficulties in obtaining nanomechanical property data from fibrous networks with high porosities led to the use of consolidated pellets for comparative analyses. The indenter approach segment of the test sequence must accurately

Figure 7. Bright-field transmission electron microscopy images of the pressure-consolidated collagen/HA scaffold shown at (a) lower and (b) higher magnifications.

determine the location of the surface based on the measured stiffness. Ideally, the test sample is flat (on the scale of the indenter tip radius of 50 nm) and fixed in space so that the indenter makes an unambiguous surface find and that the initial contact stiffness and contact area are accurately determined. This makes indentation of a porous nanofiber matrix quite challenging since the indenter tip area may not make full contact with fibers, may slip off fibers into a nearby pore or may simply ‘push’ the fibers into the surrounding empty space. In looking at the scale of figure 2, one can imagine that reliable indentation of this porous surface using a 50 nm radius indenter tip would be challenging. To consistently hit a nanofiber or a group of nanofibers ‘square-on’ with sufficient support underneath to prevent fiber ‘pushing’ or to prevent the non-homogeneous influence of ‘pinning’ forces from several nearby cross-linked fibers is not likely. Indeed, the number of indents that succeeded in generating data from the porous nanofiber samples was very low, and resulted in a large scatter in measured values. This was likely due to the inconsistent contact mechanics for these samples and to the ambiguity of surface find test segments as examined from the stiffness versus displacement data. This problem led us to attempt 147

S A Catledge et al

Figure 8. Young’s modulus as measured by nanoindentation for pressure-consolidated scaffolds. Table 1. Values of elastic modulus for scaffold materials. Elastic modulus (GPa)

PCL

Collagen

Hydroxyapatite

Reported in the literature

0.12–0.31 [38–40]

39–125 [47–49]

Measured in this study

0.5 ± 0.1

0.1–1.5 (fiber) 2.9–9.0 (molecule) [37, 41–46] 2.4 ± 0.2

Not measured

nanoindentation on consolidated samples, made by uniaxial compression of the porous scaffold into dense pellets. Consolidation of the electrospun fibers into dense pellets by uniaxial compression was therefore deemed necessary in order to obtain an accurate and reproducible surface find segments for determination of the contact area and stiffness during nanoindentation. Table 1 shows values of elastic modulus for PCL, hydroxyapatite and collagen as quoted in the literature, along with the values measured in this study by nanoindentation. The wide variation of collagen modulus from the literature is due to the variation in source (calf skin versus rat tendon, for example) as well as the scale of measurement (individual molecule versus fibrils). Our measured modulus value lies between that reported for fibers and that of individual molecules, suggesting that the scale of nanoindentation (i.e. indenter tip radius of ∼50 nm) may be useful in bridging this measurement gap. The collagen molecule is approximately 300 nm long and only about 1.5 nm wide. Therefore, in order to make full surface contact, the indenter tip would have to span the width of many molecules (about 120), approaching the width of an individual fiber (about 180 nm). The large variation in the reported hydroxyapatite modulus is likely due to the various methods of measurement, as well as due to a variety of processing methods giving rise to samples with a different porosity and purity/composition. The measured elastic (Young’s) modulus for each of the scaffold systems prepared in this study is shown graphically in figure 8. The order of increasing measured elastic modulus is (PCL < PCL/HA < collagen < triphasic < collagen/HA). It should be emphasized that the pellets are non-porous and 148

do not contain fibers. Therefore, the measured mechanical properties for these samples are not influenced by the porosity or fiber diameter, as they would be for as-spun fibers. The main source of spread in the data shown in figure 8 is believed to be due to the local sample inhomogeneity on the scale of the nanoindenter tip (tip radius ∼50 nm). A non-uniform size and dispersion of HA particles is expected to influence the nanoscale mechanical properties measured in this study as some regions will be polymer rich and others will be HA rich underneath the nanoindenter tip. For this reason, many indents (up to 40 per sample) were performed on each sample with data averaged over a range of depths (from 500–1500 nm). As can be seen from the error bars in figure 8, statistical significance in the data is achieved in comparing PCL, PCL/HA, triphasic and collagen/HA scaffolds. The only overlap in the range of modulus values occurs between triphasic and pure collagen samples. Although the measured mechanical properties of pressureconsolidated samples would not be representative of in vivo conditions, it is useful to establish how these properties are changing from one material system to another. Since Young’s modulus is a true material property, on the atomic and molecular scales it is independent of factors such as strain hardening or porosity and will provide a direct measure of bond stiffness. In fact, techniques such as x-ray diffraction have been used to determine the elastic modulus of individual molecules such as collagen [37]. The composite scaffolds in this study are best described as a dispersion-strengthened material with reinforcing nanoparticles of HA acting to hinder plastic deformation. As mentioned previously, the apatite deposits between tangled fibrils may also help ‘cross-link’ the fibers through mechanical interlocking or by forming calcium ion bridges, thus increasing the resistance to deformation of the polymer fiber network. The nano-HA is expected to provide extra stiffness to the matrix (PCL/HA > PCL and collagen/HA > collagen), but one cannot rule out the importance of collagen and collagen/PCL polymer blend components and their interaction with nano-HA. Although collagen/HA and triphasic scaffolds have the same nano-HA content as the PCL/HA scaffold, their Young’s modulus is higher. The PCL/collagen (50/30) blend making up the remaining 80% of the triphasic scaffold clearly has improved the stiffness. Finally, it is noteworthy that the measured collagen/HA (80/20) scaffold modulus of 3.9 ± 0.4 GPa is higher than that for other collagen/HA scaffolds reported in the literature having a similar (or even higher) HA content [9].

4. Conclusion In conclusion, a nanofibrous triphasic scaffold was electrospun from a mixture of polycaprolactone (PCL), collagen type I and hydroxyapatite nanoparticles. The mean fiber diameter for the triphasic scaffold was 180 ± 50 nm, a value that coincides well with the collagen fiber bundle diameter characteristic of the native extracellular matrix of bone. FTIR spectroscopy confirmed the presence of the various amide, PO43−, and PCL-related absorption bands expected

An electrospun triphasic nanofibrous scaffold for bone tissue engineering

in the scaffold. The triphasic fibers show a fairly uniform dispersion of apatite particles throughout their length with minor agglomeration. While the majority of apatite particles are uniformly dispersed having an average size of 30 nm, agglomerated particles as large as a few microns are sparsely distributed. Nanoindentation of consolidated (pressed) fibrous scaffolds containing 20 wt% nano-HA revealed an elastic modulus as high as 3.9 GPa (for the collagen/HA scaffold). The importance of collagen in improving scaffold stiffness is evident from the order of increasing scaffold Young’s modulus (PCL < PCL/HA < collagen < triphasic < collagen/HA). The composite scaffolds in this study may be approximated as dispersion-strengthened materials in which apatite deposits between tangled fibrils may also help ‘cross-link’ the fibers through mechanical interlocking or by forming calcium ion bridges, thus increasing the resistance to deformation of the polymer fiber network.

Acknowledgments The authors acknowledge support from the National Science Foundation (NSF) under a Nanoscale Interdisciplinary Research Team (NIRT) program grant no. DMR-0402891. Neel Shrikishen acknowledges support from a NSF-Research Experiences for Undergraduates (REU)-site program at UAB under grant no. DMR-0243640. The authors also thank Leigh Millican and Albert Tousson at the UAB High Resolution Imaging Facility for their assistance with TEM and confocal microscopy measurements, respectively.

References [1] Geiger M, Li R H and Friess W 2003 Collagen sponges for bone regeneration with rhBMP-2 Adv. Drug Deliv. Rev. 55 1613–29 [2] Webster T J, Ergun C, Doremus R H, Siegel R W and Bizios R 2000 Enhanced functions of osteoblasts on nanophase ceramics Biomaterials 21 1803–10 [3] Currey J D 1964 Three analogies to explain to explain the mechanical properties of bone Biorheology 2 1–10 [4] Piekarski K 1973 Analysis of bone as a composite material Int. J. Eng. Sci. 11 557 [5] Katz J L 1971 Hard tissue as a composite material: I. Bounds on the elastic behavior J. Biomech. 4 455–73 [6] Gupta H S, Schratter S, Tesch W, Roschger P, Berzlanovich A, Schoeberl T, Klaushofer K and Fratzl P 2005 Two different correlations between nanoindentation modulus and mineral content in the bone-cartilage interface J. Struct. Biol. 149 138–48 [7] Rho J Y, Kuhn-Spearing L and Zioupos P 1998 Mechanical properties and the hierarchical structure of bone Med. Eng. Phys. 20 92–102 [8] Hellmich C and Ulm F J 2002 Are mineralized tissues open crystal foams reinforced by crosslinked collagen? Some energy arguments J. Biomech. 35 1199–212 [9] Wahl D A and Czernuszka J T 2006 Collagen-hydroxyapatite composites for hard tissue repair Eur. Cell Mater. 11 43–56 [10] Itoh S, Kikuchi M, Takakuda K, Koyama Y, Matsumoto H N, Ichinose S, Tanaka J, Kawauchi T and Shinomiya K 2001 The biocompatibility and osteoconductive activity of a novel hydroxyapatite/collagen composite biomaterial, and its function as a carrier of rhBMP-2 J. Biomed. Mater Res. 54 445–53

[11] Tampieri A, Celotti G, Landi E, Sandri M, Roveri N and Falini G 2003 Biologically inspired synthesis of bone-like composite: self-assembled collagen fibers/hydroxyapatite nanocrystals J. Biomed. Mater. Res. A 67 618–25 [12] Roveri N F G, Sidoti M C, Tampieri A, Landi E, Sandri M and Parma B 2003 Biologically inspired growth of hydroxyapatite nanocrystals inside self-assembled collagen fibers Mater. Sci. Eng. C 23 441–6 [13] Kikuchi M, Itoh S, Ichinose S, Shinomiya K and Tanaka J 2001 Self-organization mechanism in a bone-like hydroxyapatite/collagen nanocomposite synthesized in vitro and its biological reaction in vivo Biomaterials 22 1705–11 [14] Chang M C, Ikoma T, Kikuchi M and Tanaka J 2001 Preparation of a porous hydroxyapatite/collagen nanocomposite using glutaraldehyde as a crosslinkage agent J. Mater. Sci. Lett. 20 1199–201 [15] Du C, Cui F Z, Zhang W, Feng Q L, Zhu X D and de Groot K 2000 Formation of calcium phosphate/collagen composites through mineralization of collagen matrix J. Biomed. Mater. Res. 50 518–27 [16] Yamauchi K, Goda T, Takeuchi N, Einaga H and Tanabe T 2004 Preparation of collagen/calcium phosphate multilayer sheet using enzymatic mineralization Biomaterials 25 5481–9 [17] Lawson A C and Czernuszka J T 1998 Collagen–calcium phosphate composites Proc. Inst. Mech. Eng. H 212 413–25 [18] Clarke K I, Graves S E, Wong A T C, Triffitt J T, Francis M J O and Czernuszka J T 1993 Investigation into the formation and mechanical-properties of a bioactive material based on collagen and calcium-phosphate Mater. Sci. Mater. Med. 4 107–10 [19] Liao S S, Cui F Z, Zhang W and Feng Q L 2004 Hierarchically biomimetic bone scaffold materials: nano-HA/collagen/ PLA composite J. Biomed. Mater. Res. B. Appl. Biomater. 69 158–65 [20] Liao S, Li B, Ma Z, Wei H, Chan C and Ramakrishna S 2006 Biomimetic electrospun nanofibers for tissue regeneration Biomed. Mater. 1 R45–53 [21] Pham Q P, Sharma U and Mikos A G 2006 Electrospinning of polymeric nanofibers for tissue engineering applications: a review Tissue Eng. 12 1197–211 [22] Venugopal J, Ma L L, Yong T and Ramakrishna S 2005 In vitro study of smooth muscle cells on polycaprolactone and collagen nanofibrous matrices Cell Biol. Int. 29 861-67 [23] Venugopal J, Zhang Y Z and Ramakrishna S 2005 Fabrication of modified and functionalized polycaprolactone nanofibre scaffolds for vascular tissue engineering Nanotechnology 16 2138–42 [24] Zhang Y Z, Venugopal J, Huang Z M, Lim C T and Ramakrishna S 2005 Characterization of the surface biocompatibility of the electrospun PCL-collagen nanofibers using fibroblasts Biomacromolecules 6 2583–9 [25] Thomas V, Dean D R, Jose M V, Mathew B, Chowdhury S and Vohra Y K 2007 Nanostructured biocomposite scaffolds based on collagen coelectrospun with nanohydroxyapatite Biomacromolecules 8 631–7 [26] Stanishevsky A, Chowdhury P C S, Thomas V, Catledge A and Dean D 2006 Compositionally modified hydroxyapatite nanocrystals for polymer/ceramic scaffold applications Biomimetic Polymers and Gels ed N Langrana, F Horkay and B Yurke (Mater. Res. Soc. Symp. Proc.) vol 897E pp 0897-J03-18 [27] Tanaka Y, Hirata Y and Yoshinaka R 2003 Synthesis and characteristics of ultrafine hydroxyapatite particles J. Ceram. Process. Res. 4 197–201 [28] Bigi A, Boanini E and Rubini K Hydroxyapatite gels and nanocrystals prepared through a sol–gel process Solid State Chem. 177 3092–8

149

S A Catledge et al

[29] Birks L S and Friedman H 1946 Particle size determination from x-ray line broadening J. Appl. Phys 17 687–92 [30] Chowdhury S, Thomas V, Dean D, Catledge S A and Vohra Y K 2005 Nanoindentation on porous bioceramic scaffolds for bone tissue engineering J. Nanosci. Nanotechnol. 5 1816–20 [31] Thomas V, Dean D R and Vohra Y K 2006 Nanostructured biomaterials for regenerative medicine—A review Curr. Nanosci. 2 155–77 [32] Zong X, Fang D, Kim K-S, Kim J, Cruz S, Hsiao B S and Chu B 2002 Structure and process relationships in bioabsorbable nanofiber membranes by electrospinning Polymer 43 4403–12 [33] Zhang C, Yuan X, Wu L, Han Y and Sheng J 2005 Study on morphology of electrospun poly(vinyl alcohol) mats Eur. Polym. J. 41 423–32 [34] Buchkoa C J, Chena L C, Shena Y and Martin D C 1999 Processing and microstructural characterization of porous biocompatibleprotein polymer thin films Polymer 40 7397–407 [35] Yuan X, Zhang Y, Dong C and Sheng J 2004 Morphology of ultrafine polysulfone fibers prepared by electrospinning Polym. Int. 53 1704–10 [36] Stanishevsky A V, Chinoda P, Chowdhury S, Thomas V, Catledge S A and Dean D 2006 Compositionally modified hydroxyapatite nanocrystals for polymer/ceramic scaffold applications, in biomimetic polymers and gels Mater. Res. Soc. Symp. Proc. 897E, Warrendale, PA, 2006 Paper # 18.1 [37] Sasaki N and Odajima S 1996 Stress-strain curve and Young’s modulus of a collagen molecule as determined by the X-ray diffraction technique J. Biomech. 29 655–8 [38] Williamson M R, Woolard K J, Griffiths H R and Coombes A G A 2006 Gravity spun polycaprolactone fibers for applications in vascular tissue engineering: proliferation

150

[39]

[40]

[41] [42] [43] [44] [45] [46] [47] [48] [49]

and function of human vascular endothelial cells Tissue Eng. 12 45–51 Williams J M, Adewunmi A, Scheka R M, Flanagan C L, Krebsbach P H, Feinberg S E, Hollister S J and Das S 2005 Bone tissue engineering using polycaprolactone scaffolds fabricated via selective laser sintering Biomaterials 26 4817–27 Williamson M R, Adams E F and Coombes A G A 2002 Cell attachment and proliferation on novel polycaprolactone fibers having application in soft tissue engineering Eur. Cells Mater. 4 62–3 Dobrin P B 1978 Mechanical properties of arteries Physiol. Rev. 58 397–460 Akiva U, Wagner H D and Weiner S 1998 Modelling the three dimensional elastic constants of parallel-fibred and lamellar bone J. Mater. Sci. 33 1497–509 Humphrey J D 2002 Continuum biomechanics of soft biological tissues Proc. R. Soc. A 459 3–46 Cusack S and Miller A 1979 Determination of the elastic constants by Brillouin light scattering J. Mol. Biol. 135 39–51 Harley R, James D, Miller A and White J W 1977 Phonons and the elastic moduli of collagen and muscle Nature 267 285–7 Canfield T R and Dobrin P B 1987 Static elastic properties of blood vessels Handbook of Bioengineering ed R Skalak and S Chien (New York: McGraw-Hill) chapter 16 Currey J D 1969 The relationship between the stiffness and the mineral content of bone J. Biomech. 2 477–80 Torgalkar A M 1979 A resonance frequency technique to determine elastic modulus of hydroxyapatite J. Biomed Mater. Res. 13 907–20 Kumar R R and Wang M 2002 Modulus and hardness evaluations of sintered bioceramic powders and functionally graded bioactive composites by nano-indentation technique Mater. Sci. Eng. A 338 230–6