poly(lactic-co-glycolic) acid composite scaffolds fabricated by thermally induced phase separation

poly(lactic-co-glycolic) acid composite scaffolds fabricated by thermally induced phase separation

ARTICLE IN PRESS Biomaterials 28 (2007) 2109–2121 www.elsevier.com/locate/biomaterials Polyurethane/poly(lactic-co-glycolic) acid composite scaffold...

2MB Sizes 0 Downloads 56 Views

ARTICLE IN PRESS

Biomaterials 28 (2007) 2109–2121 www.elsevier.com/locate/biomaterials

Polyurethane/poly(lactic-co-glycolic) acid composite scaffolds fabricated by thermally induced phase separation A.S. Rowlands, S.A. Lim, D. Martin, J.J. Cooper-White Tissue Engineering & Microfluidics Laboratory, Australian Institute for Bioengineering and Nanotechnology and the School of Engineering, The University of Queensland, Brisbane, Qld. 4072, Australia Received 23 August 2006; accepted 31 December 2006 Available online 16 January 2007

Abstract In this study, we present a novel composite scaffold fabricated using a thermally induced phase separation (TIPS) process from poly(lactic-co-glycolic) (PLGA) and biomedical polyurethane (PU). This processing method has been tuned to allow intimate (molecular) mixing of these two very different polymers, giving rise to a unique morphology that can be manipulated by controlling the phase separation behaviour of an initially homogenous polymer solution. Pure PLGA scaffolds possessed a smooth, directional fibrous sheetlike structure with pore sizes of 0.1–200 mm, a porous Young’s modulus of 93.5 kPa and were relatively brittle to touch. Pure PU scaffolds had an isotropic emulsion-like structure, a porous Young’s modulus of 15.7 kPa and were much more elastic than the PLGA scaffolds. The composite PLGA/PU scaffold exhibits advantageous morphological, mechanical and cell adhesion and growth supporting properties, when compared with scaffolds fabricated from PLGA or PU alone. This novel method provides a mechanism for the formation of tailored bioactive scaffolds from nominally incompatible polymers, representing a significant step forward in scaffold processing for tissue-engineering applications. r 2007 Published by Elsevier Ltd. Keywords: Composite; Scaffold; Polyurethane; Polylactic acid; Mechanical properties; Cell spreading

1. Introduction Three-dimensional polymeric scaffolds for tissue engineering hold much promise for the regeneration of damaged or diseased tissues. Regenerative therapies offer many advantages over existing replacement therapies, including potentially indefinite life-spans of implants, superior mechanical characteristics and better immunoacceptance. These artificial structures seek to provide a surrogate for the natural extracellular matrix (ECM) by directing the organisation, growth and differentiation of cells in the process of generating functional tissues and impart both chemical and physical cues [1,2]. A multitude of approaches to manufacturing threedimensional porous structures for tissue-engineering applications have been detailed in the literature [3-20]. In this particular work, we have employed a modified thermally Corresponding author. Tel.: +61 7 3346 3858; fax: +61 7 3346 3973.

E-mail address: [email protected] (J.J. Cooper-White). 0142-9612/$ - see front matter r 2007 Published by Elsevier Ltd. doi:10.1016/j.biomaterials.2006.12.032

induced phase separation (TIPS) method to fabricate scaffolds [21–23]. The TIPS process utilises changes in thermal energy to induce the demixing of a homogenous polymer-solvent or a polymer-solvent–non-solvent solution either by solid–liquid demixing or liquid–liquid phase separation. In the case of a liquid–liquid phase separation mechanism, two definitively different morphologies may be produced: (1) the solution separates into a polymer-rich and polymer-lean phase (giving rise to an emulsion-like morphology) when cooled below the binodal curve, or (2) the solution separates into a bicontinuous polymer-rich and polymer-lean phase when cooled below the spinodal curve. It is this second process that gives rise, by its very path of formation, to a highly interconnected polymer network, which inherently produces a highly porous structure once the solvent is removed via leaching or freeze-drying. One of the most attractive characteristics of TIPS over other scaffold fabrication techniques is the formation of not only an intrinsically interconnected polymer network, but also an interconnected porous space

ARTICLE IN PRESS 2110

A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

in one simple process that is scalable, fast and controllable. TIPS is thus a very convenient technique for fabricating porous scaffolds as many scaffold architectures can be formed with ease via the manipulation of various processing parameters and system properties. Based on previous work by our group [24,25], as well as that of others [26,27], it is known that dimethylsulphoxide (DMSO) is a suitable low-toxicity solvent in which to dissolve the known biostable, biocompatible polymers polyurethane (PU) and poly(lactic-co-glycolic acid) (PLGA) [28]. While both biocompatible and biodegradable, as well as FDA approved, PLGA is a fairly rigid solid that exhibits little to no elastic behaviour and is thus not suited for applications that involve high movement or shaping in situ. If blended successfully with a polymer such as polyurethane or poly(ethyleneglycol) (PEG), which have soft, elastic mechanical properties, a hybrid can be formed that exhibits much greater flexibility than a PLGA-only system. Wake et al. [29] described the enhanced pliability of three-dimensional foams made of PLGA/PEG blends and a particulate leaching method, evidenced by the ability to roll them into a tube without macroscopic damage to the scaffold—an impossibility with rigid PLGA-only structures. A PLGA/PU composite matrix is thought to have potential as an artificial ECM as the combination of biodegradable PLGA and a non-degradable PU will produce a unique structure that retains its overall mechanical strength as the PLGA degrades with the genesis of new tissues. However, little research has been carried out into the phase separation behaviour of a PLGA/PU/DMSO system, or its suitability in the fabrication of biocompatible scaffolds. It is was the intent of this study to investigate the use of a PLGA/PU/DMSO system to produce highly porous, biocompatible scaffolds of varying porosity, connectivity, size and shape by phase separation and solvent leaching and thereafter their utility in improving cell attachment to PU which is intrinsically non-cell adhesive.

were heated to 90 1C and stirred for 24 h to aid dissolution of the polymer to form a homogenous solution. Solutions were then pipetted into 13 mm diameter glass vials with snap-on lids and placed in a water bath at 5 1C for 24 h to freeze solutions. Vials were then removed from the water bath and placed onto a mesh support in a beaker filled with ice and cold water to leach the DMSO from the scaffold. Water and ice in the beaker was changed every hour for 3 h and then every 2 h until no visible solvent leaching was apparent. The final water/ice change was left for 24 h. This methodology has been proven to reduce DMSO residue in scaffolds to less than 1 ppm [24]. Scaffolds were removed from the leaching beaker and vacuum dried at 102 mbar in their vials for 4 h using a vacuum chamber and pump (Javac, DD300) to remove water. Scaffolds were then extracted from the vial and left to air-dry on tissue paper.

2.3. Determination of polymer solution thermal properties The thermal properties of the polymer solutions (prepared as described in Section 2.2) were determined by a cloud point (CP) analysis and differential scanning calorimetry (DSC). It was known from previous work by our group [24] that 5% PLGA/DMSO solutions crystallise at the freezing point of the solvent without exhibiting liquid–liquid phase separation behaviour and hence the cloud point measured is in fact a solid–liquid transition (i.e. crystallisation) and not explicitly a binodal phase separation, although this is often confused with such. Conversely, this study (to be discussed in detail in Section 3) has shown that turbidity changes of solutions of PU/DMSO (i.e. the cloud point) are representative of PU/DMSO homogenous solutions phase separating into a polymer-rich phase and a polymer-lean phase i.e. the cloud point represents a true binodal point. The effect of PU concentration on cloud point behaviour was determined for 0.5%, 1%, 2%, 4%, 5% and 6% PU/DMSO (w/w) solutions and a 2.5% PU/2.5% PLGA/DMSO blend (w/w). A 5% PLGA/ DMSO (w/w) solution was included in the cloud point experiment to determine the crystallisation temperature. The CP was determined as follows: polymer solutions in sealed glass vials were placed in a water bath initially set at 80 1C and cooled at a rate of 0.1 1C/min and the turbidity of the solutions was recorded by visual inspection until all cloud points and crystallisation temperatures had been determined. Modulated DSC (TA2920 Modulated DSC, TA instruments, USA) was used to verify the CP data and to more precisely identify accurately the temperatures at which respective phase changes occur. The following thermal procedure was used: sample equilibrated at 90 1C, held isothermal for 5 min, cooled at 5 1C/min to 90 1C, heated at 5 1C/min to 90 1C. Temperature was modulated 71 1C every 60 s.

2.4. Structural morphology and elemental analysis of scaffolds 2. Materials and methods 2.1. Materials 75/25 PLGA was supplied by Birmingham Polymers Inc (Birmingham, AL, USA). ElastEon PU 70A pellets were a gift to Dr. Darren Martin from Aortech Biomaterials in Melbourne. The PU used in this study incorporates hard segments of 4,40 -methylenediphenyl diisocyanate, a mixture of butanediol and a short siloxane chain extender. The soft segments are made up of a mixture of bis(hydroxyalkyl) polydimethyl siloxane (PDMS) and poly(hexamethylene oxide) (PHMO). PHMO compatibilises the PDMS with the more hydrophilic hard segments to give better overall strength and make this particular PU biostable in vivo. DMSO was supplied by Ajax Finechem (Seven Hills, NSW, Australia).

2.2. Fabrication of PLGA, PU and PLGA/PU scaffolds The polymer solutions were prepared by dissolving PLGA or PU or a PLGA/PU mixture in dry DMSO at a total polymer concentration of 5% w/w (2.5% w/w PLGA, 2.5% w/w PU for the mixed solution). Solutions

A scanning electron microscope (SEM) (JEOL 6400LA, USA) was used to determine the morphology of the scaffolds. Cross sections of the scaffolds were coated with titanium using a sputter coater (Eiko IB5 Ion Coater, Japan) under an argon atmosphere using a sputter current of 60 mA. Samples were observed under the SEM at an accelerating voltage of 5 kV and pressure of 1 Pa. An elemental analysis feature on the SEM was used in order to determine which polymer was giving rise to certain morphological features. This was possible due to the presence of silicon in the PDMS molecule in PU, which can be used to differentiate the PU in a blend from another polymer, e.g. PLGA.

2.5. Mechanical testing of scaffolds The Young’s modulus of circular scaffold Sections (13 mm diameter, 2 mm thick) was determined using dynamic mechanical thermal analysis (DMTA IV, Rheometric Scientific) at a frequency of 1 Hz and an initial static force of 25g force for PLGA and 10g force for PU and the PLGA/ PU composite scaffolds, these scaffolds being softer than the rigid PLGA scaffolds.

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2.6. Cell culture

3. Results and discussion

Mouse embryo fibroblast NIH-3T3 cells from a frozen stock were incubated at 37 1C and 5% CO2 using a medium containing Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% foetal bovine serum (FBS) and 50 mg/mL gentamycin to prevent microbial overgrowth. After 48 h (80% confluence) the growth medium was removed by aspiration, the cells washed with 10 mL of phosphate buffered saline (PBS) and then detached from the plate using trypsin for 3 min. Fresh medium was then added to the culture dish to inactivate the trypsin. The cell culture was split into a new culture plate using 1 mL of the original culture as the seeding volume and 9 mL of fresh medium. This was repeated until the cells were needed for seeding. The total number of passages before seeding was 4. A viable cell count was performed using trypan blue dye.

3.1. Thermal properties of polymer solutions

2111

3.1.1. Cloud point measurements The cloud point temperatures for the different concentration PU samples were found to be 59 1C72 1C (0.5%, 1.0% and 2.0%) and 66 1C72 1C (4.0%, 5.0% and 6.0%). The PU/PLGA blend had a cloud point temperature of 67 1C72 1C. The 5% PLGA solution was found to crystallise at 11 1C72 1C. Solvent freezing point depression was observed in the cloud point measurements and this is discussed in depth in more detail in the following section.

2.7. Scaffold seeding Two millimeter thick scaffold sections (diameter 13 mm) placed into a Petri dish were sterilised by dripping 70% ethanol onto them. The sections were thoroughly rinsed with PBS followed by DMEM. The scaffold sections were then transferred to a 6-well plate and a 100 mL suspension of 3T3 cells at a concentration of approximately 1  106 cells/mL was dripped onto each scaffold. The sections were then transferred to fresh wells, which were filled with DMEM/10% FBS/50 mg/mL gentamycin. The plates were then incubated at 37 1C and 5% CO2 for 48 h. After incubation cell fixation was achieved by soaking the scaffold sections in 4% paraformaldehyde in PBS for 30 min.

2.8. Cell morphology, adhesion and growth on scaffolds Laser scanning confocal imaging was used to obtain high-resolution images of the cell morphology. The fluorescent dye used was propidium iodide, which is a membrane-impermeant dye that stains by intercalating into nucleic acid molecules. It binds both DNA and RNA and is thus ideal to use to locate not only nuclei, but also to garner an idea of the spread of the cytoskeleton. Slices 0.2–0.3 mm thick were cut from the cell-cultured scaffold sections (Fig. 1). Slices were rinsed with 2  sodium saline citrate (SSC—0.3 M NaCl, 0.03 M sodium citrate at pH 7.0), a standard buffer used in the protocols given for the propidium iodide. Slices were then left for 5 min with 500 nM propidium iodide in 2  SSC and rinsed 3 times with fresh SSC buffer. The buffer was then replaced with a glycerol solution, which is used as a mountant and gives rise to a good contact angle between the sample and the cover-slip. This brings the refractive index closer to that of the oil used for the immersion lens and also makes the sample somewhat more stable over time. The slice was transferred to a microscope slide and covered with a cover-slip, which was secured to the slide with nail polish to prevent movement during imaging. Oil was placed on the cover slip and a 40  immersion lens was used to image the cells on the scaffold slices.

Slice Scaffold Disc

Fig. 1. Slice location for confocal imaging.

3.1.2. DSC measurements According to van Emmerik [30], when a homogenous solution demixes, an exotherm is observed in a DSC cooling scan. At a very high cooling rate, the onset of the exotherm is taken as the spinodal temperature as nucleation would not have taken place to a detectable amount prior to entering the spinodal region. When cooling at a lower rate, van de Witte reported that the binodal temperature instead of the spinodal temperature was detected as liquid–liquid phase separation occurred through nucleation and growth [31]. In our study, DSC analysis was thus carried out to see if, at these relatively low polymer concentrations, the change in enthalpy during phase separation of the solution into two distinct phases could be quantified, providing a greater level of detail above that provided by the cloud-point experiments. Most relevant, regardless of the selected cooling rate, Arnauts et al. reported for PMMA in n-butanol that at low polymer concentration (o10%), the DSC signal may be too small to be useful for any significant determination of either transition [32]. This sensitivity will however depend on the individual polymer/solvent system and hence, although our concentrations are lower than 10%, the technique was trialled. When cooling a system that liquid–liquid phase separates, if cooling at a slow enough rate, two distinct peaks should be observed (a solvent rich freezing peak and a solvent lean freezing peak). However, the most obvious result from Fig. 2 is the absence of two individual freezing and melting peaks corresponding to two distinct polymer/ DMSO phases. In fact, the only observable exotherm occurs during the freezing of what one would reasonably assume is a homogenous solution. The soft microphase Tg for the PDMS-based TPU used in this study is very low in a non-solvated form (50 to 100 1C [33]), so it is expected that it would be well below 100 1C when solvated by DMSO. It is also likely to be a substantially broader transition due to the phase mixing that occurs between hard and soft microphases in solution. This transition would therefore not be observable in the PU DSC results as the lowest temperature used within the protocol was 90 1C, which is the lowest practical

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2112

A

2 5% PLGA 5% PU 5% PLGA/PU Blend

1.75

Heat flow (W/g)

1.5

1.25

1

0.75

0.5

0.25

0 -10

-8

-6

-4

-2

0

2

4

Temperature (°C)

6

8

10

12

B -5

-0.1

0

5

10

15

20

25

30

Heat Flow (W/g)

-0.3

-0.5

-0.7

-0.9

-1.1

5% PLGA

-1.3

-1.5

5% PU 5% PLGA/PU Blend

Temperature (°C)

Fig. 2. Modulated DSC (A) freezing (top) and (B) melting (bottom) peaks.

operational temperature whilst retaining control over cooling rate for this particular DSC apparatus. For all three of the polymer solutions tested, the heat flow versus temperature data appears to be dominated by the solvent’s thermodynamic cycles only, which are the only clearly observable transitions. The heat evolved from liquid–liquid demixing are not seen in this data, even at an enlarged scale, and there are a few plausible reasons for this. Most likely, as the polymer concentration is relatively low (5%), there simply may not have been enough polymer in the polymer-rich phase of the sample to observe a enthalpy change of a sufficiently detectable magnitude, as previously noted by Arnauts et al. [32]. Further inspection of the thermal traces detailed in Fig. 2A and B have however provided critical temperatures and overall heat flow associated with the freezing and melting events (see Table 1). We note that while both the

freezing and melting temperature was lower for the blended system, the heat flow for the blended system is between that of the two individual polymer systems. This can be explained by the fact that the PU concentration in the blend is only half the concentration in the pure PU scaffold used as a reference here. Significant solvent freezing point depression is exhibited for all the three polymer solutions, considering that pure DMSO freezes at 18.4 1C. This behaviour was also observed in the cloud point experiments for PLGA. PLGA is easily dissolved in DMSO and therefore has a much higher level of interaction with the solvent than PU, which takes over 10 times longer than PLGA to fully dissolve at the same temperature and same mass of polymer. It is this level of interaction that probably explains the observed difference in the magnitude of freezing point depression. Polymer chains are effective energy stores when in solution

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2113

Table 1 Freezing (supercooling) and melting peak temperatures and heat flows 5% Polymer sample

Peak temp (oC)

Peak heat flow (J/g)

Onset of transition (1C)

Width of transition (1C)

Freezing PLGA PU PLGA/PU blend

8.31 4.34 4.10

144.0 130.6 133.6

12.03 9.08 8.83

26.97 41.56 44.42

Melting PLGA PU PLGA/PU blend

18.25 16.08 14.30

146.6 132.9 134.7

15.08 21.50 31.88

44.52 47.68 56.67

with a solvent, requiring a lower temperature to induce solvent crystallisation in these mixed systems resulting in the generic solvent freezing point depression behaviour exhibited for all three of the polymer solutions tested. The onset of crystallisation for both the PU and PLGA/ PU-blended systems was very close (9.08 and 8.83 1C, respectively) which is considerably lower than that of the PLGA-only system (12.03 1C). Similarly, the width of transition of both the PU containing systems were much closer to one another than that of a PLGA-only system. The closeness in crystallisation onset temperatures for the PU-containing systems indicates that PU plays a dominate role in determining the freezing point of the solution, despite the PU/PLGA blend containing only half the amount of PU as the pure PU sample. Furthermore, when the differential dH/dT (W/g/ 1C) is plotted against T ( 1C) (not shown), it becomes clear that the rate of both freezing and melting is much more rapid in the PLGA system than either the pure PU or PU/PLGA blended systems, again indicating that the presence of PU governs the kinetics of solvent crystallisation and melting in the mixed system. This dominance of the PU in solution can perhaps be explained by the differences in the heat capacity (Cp) of the polymers. The Cp of PLGA is 0.474 J/g 1C whereas for PU it is between 1.0 and 1.5 J/g 1C over the investigated temperature range. For the PU and PU/PLGA-blended systems that exhibit liquid–liquid phase separation behaviour when cooled, it is possible that the PU phase acts as an energy store, thus the entire solution has more energy and requires a lower temperature to induce solvent freezing. The fact that the blend depresses the freezing point most of all is likely due to both the higher heat capacity of PU than PLGA and polymer–polymer interchain interactions, which are known to affect the mobility of macromolecules and therefore affect processes such as interdiffusion, the miscibility of the components and phase separation [34,35]. 3.2. Mechanical properties Among the three kinds of scaffolds, the PLGA scaffold had the highest Young’s modulus (94 kPa) while the PU scaffold had the lowest (16 kPa). The blend exhibited a Young’s modulus in-between these two values (36 kPa).

The error associated with these measurements was approximately 75%. These moduli seem acceptable values when compared to others’ work [36], although a Young’s modulus as high as 6.6 MPa for a 10% 75/25 PLGA/ Chloroform scaffold has been reported [37]. It should be pointed out that the ‘modulus’ being reported here is not the bulk modulus, but a porous modulus as the scaffolds have porosities of up to 95%. These are far from the ideal solid samples that are usually measured with DMTA devices. However, there is good agreement between the measured moduli and the theoretical moduli, as calculated by using Eq. (1) to determine the Young’s modulus of a foam from bulk properties [38]. Scaffold density was calculated by assuming a negligible change in volume from the solution   2 E r ¼ (1) Es rs where E* is the scaffold modulus; Es the Young’s modulus of bulk polymer (PLGA ¼ 45 MPa, PU ¼ 5 MPa); r* the scaffold density (60 kg/m3); rs the density of bulk polymer (PLGA ¼ 1300 kg/m3, PU ¼ 1100 kg/m3). The calculated moduli (using a porosity of 95%) are 93, and 15 kPa for pure PLGA and PU foams, respectively. These numbers are extremely close to the measured moduli of the scaffolds fabricated in this study. If we assume a linear blending rule, we note that the predicted modulus for the blended scaffold should be around 54 kPa. The measured value was 36 kPa, suggesting that such a rule is not applicable and that the PU characteristics dominate the mechanical response of these mixed scaffolds. 3.3. Scaffold description and morphology Quantitatively, under compression, PLGA scaffolds were hard and brittle while PU scaffolds were soft and elastic. The blended scaffold exhibited properties somewhere in-between the two—not as rigid as pure PLGA, but not as elastic as pure PU although the mixed scaffold mechanical properties suggest a linear blending rule is not applicable. The morphological structure of the 5% polymer solutions is shown in Fig. 3. The PLGA scaffolds (A and B) possessed a directional radial structure as a result of dendritic crystallisation of the DMSO. The structure

ARTICLE IN PRESS 2114

A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

Fig. 3. Electron micrographs of scaffold morphology: PLGA (A and B); PU (C and D); PLGA/PU blend (E and F).

possessed sheet-like fibres and formed finer fibres around the edges where cooling was more rapid. The pore sizes ranged from 0.1 to 200 mm. The PU structure (C and D) consisted of random sheets of non-uniform size that were much smaller and more closely packed when compared with the PLGA scaffold, leading to a much denser structure. This structure is a result of the polymer solution undergoing liquid–liquid phase separation when cooled (prior to freezing of the solvent-rich phase). There are very few micropores on the surface of these sheets, and no porous fibres as observed for PLGA. The blended PLGA/ PU structure possesses a combination of the individual PLGA and PU morphologies. Interestingly, the blend had a much larger number of small pores on the surface than either of the pure polymer scaffolds, with a similar diameter of between 0.1 and 1 mm. Furthermore, the porous nature of the fibre bundles were equivalent to the small pores on the surface (0.1–1 mm). While fibres similar

to that observed in the pure PLGA samples are clearly evident, the most noticeable difference between the blended morphology and either of the pure samples is the presence of bundles of channels with structures that closely resemble the morphological structures found in the pure PU scaffold. These channels are approximately 10 mm in diameter, and the number of channels varies between bundles. In order to verify the hypothesis of the contributions of the individual polymers to the blend’s morphology, an elemental analysis was performed using energy dispersive spectroscopy (EDS) SEM. Two points were taken for spot elemental analysis of silicon, an element only present in PU, and carbon, which is present in larger amounts in PLGA than PU. The first spot was set on a fibrous structure, and the second on the internal walls of a bundle of channels. While no quantitative data could be obtained by the EDS, it was possible to determine the relative counts

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2115

PLGA-rich phase and a PU-rich phase. The PU-rich phase then phase separates into a PU-rich phase and a DMSOrich phase. When the freezing point of DMSO is reached, the DMSO in the PLGA-rich phase will start to crystallise causing the PLGA to envelop the PU phase. This means that PLGA will become the dominant interconnected structure. In order to test this hypothesis, isopropanol-wet scaffold sections were placed in 1.0 M NaOH solution for 15 min, which causes hydrolysis of PLGA but does not cause PU to break down. Post this treatment the scaffold sections had lost all mechanical integrity and broke apart. Where the PU had formed an interconnected network of polymer, clumps remained. These clumps were examined by SEM (see Fig. 7). The morphology is identical to the hypothesised PU region within the original scaffold. This proved conclusively that the PU was giving rise to the channel/ bundle like morphology found in the mixed polymer system and that PLGA in fact connected these bundles to form the composites scaffold in the latter stages of phase separation (i.e. after the PU came out of solution). The morphology of the single compound scaffolds can be controlled by altering the kinetics of phase separation through both changes in the initial and final temperatures and the rate of cooling. Thus, by controlling the thermal history of the solution explicitly prior to cooling, we hypothesised that we may be able to induce varying morphologies of the PU phase within the PLGA, as it is effectively the first polymer to come out of the solution as a separate phase. In the experiments previously described the temperature was reduced from 90 to 60 1C over 20 s and thereafter the solution quenched to 5 1C. In order to investigate the above hypothesis, the solution was kept at 90 1C prior to instantly decreasing the temperature to, ultimately, 5 1C by immersing the sample in the water bath. Fig. 8 below shows the resultant morphology of this

of elements found at each location. Fig. 4 below shows the location of the two elemental analysis points, and Figs. 5 and 6 the EDS results for points 1 and 2, respectively. The high platinum count was expected as the samples were initially coated in platinum to be analysed by SEM. Fig. 5 shows that the relative amount of carbon found at point 1 is some 11 times higher than the silicon content found at the same point. Conversely, Fig. 6 shows that at point 2 the relative silicon content is greater than twice the carbon. This suggests that point 1 consists of mainly PLGA while point 2 is made up of PU. This two-point EDS analysis was carried out at multiple locations with similar outcomes. The visual analysis and, more definitively, the EDS result suggests that it is indeed the PLGA giving rise to the fibrous morphologies found in the blend and the PU constituting the bundle of channels. Such a result suggests that the PLGA/PU/DMSO solution phase separates into a

Fig 4. Selected points for EDS.

007 3000 RM

3200 2800

C CKa

2000 1800

800

RM

Si SKa

1200

RM

CKa

Counts

2400

400 0 0.00

0.80

1.80

2.40

3.20

4.00

4.80

kV Fig 5. EDS Element counts for point 1 (fibre).

5.00

6.40

7.20

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2116 008

RM

1800 1600 1400

Si

1000

SKa

Counts

1200

800 600 RM

CKa CKa

200

RM

C 400

0 0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

10.00

kV Fig 6. EDS Element counts for point 2 (bundle wall).

Fig. 7. Electron micrographs of NaOH-treated scaffold sections showing remaining interconnecting PU structure.

Fig. 8. Electron micrographs of PU/PLGA scaffold produced via different quenching strategy.

blended scaffold. No large bundles or globules were observed and the structure is dominated by the aligned PLGA fibres. The surface of the PLGA fibres was however rough, most likely due to a coating of PU. This was confirmed to be the case via EDS and X-ray mapping was employed in order to verify the visual observations. Fig. 9 below shows the SEM morphology and the corresponding

X-ray map for both types of mixed scaffolds produced in this study. What is immediately apparent from comparing Fig. 9(B) and (D) is the difference in the distribution of silicon content for the two types of mixed scaffolds. Silicon is evenly distributed for the well-mixed scaffold (Fig. 9(D)), whereas for the less-well-mixed scaffold silicon is concentrated in the area of PU globules. This experiment showed

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2117

Fig. 9. Electron micrographs and corresponding X-ray maps of mixed scaffolds. Micrograph of mixed scaffold (A), X-ray map of mixed scaffold showing Si content (B), micrograph of molecularly mixed scaffold (C) and X-ray map of molecularly mixed scaffold showing Si content (D).

that by keeping the temperature of the solution constant until freezing, it is possible to control the presence or absence of PU globules and develop a truly well-mixed scaffold which, theoretically, would possess optimum mechanical properties. Upon testing of this scaffold, the modulus was in fact found to be almost exactly the same as the previously fabricated mixed scaffold.

3.4. Scaffold–cell interactions (cell adhesion and growth) 3.4.1. Cell attachment Cell attachment was performed qualitatively by looking at the morphology of the cells on the surface of each of the polymer scaffolds using confocal microscopy. From the confocal images it was clear that the cells were well spread throughout all scaffolds, indicating that cells were able to penetrate the entirety of the porous polymer network from top to bottom. The dye used, propidium iodide, stains both RNA and DNA, with DNA found exclusively in the nucleus while RNA is found within the nucleus and the cytoplasm of the cell. The imaging quite clearly shows the cytoskeleton of the cells and for all scaffolds the cells appeared to be relatively content growing on the surface of the polymers as they are well spread and showed multiple focal adhesion points (see Fig. 10). The multinucleated cell masses on the scaffolds show that the cells have divided, growing from the initial seeding point along the length of the pore network. While cell numbers were higher for PU (discussed below), the cytoplasm was more rounded than that of the

cells adhered to the PLGA and PU/PLGA blend scaffolds, indicating that cells were not quite as comfortable on PU as they were on PLGA. Fig. 10(A) is an extremely clear example of the well spread nature of the cell on a PLGA fibre. The morphology of the 3T3 fibroblasts on the PLGA/PU blend (Fig. 10(B)) is reminiscent of the morphology of this cell type in vivo, when compared to either PLGA (Fig. 10(A)) or PU (Fig. 10(C)) alone. It should be highlighted that obtaining detailed images of cells on scaffolds is not a simple task given the random orientation of the substrate in 3D space and consequently adhered cells reside spatially on multiple focal planes.

3.4.2. Cell growth To ascertain a cell count on the scaffolds, nuclei were counted in the confocal images for a range of focal slices. The Z slice stepping size was set to 4 mm, which is around the average diameter of a nucleus (3–6 mm). Therefore, it is unlikely that any one nucleus will appear in more than one slice. It was thus possible to quantify the number of cells per section volume of scaffold by counting the number of nuclei that appear in each Z slice of the section and then dividing by the section volume. Care was taken when counting nuclei in that if it was obvious that the same nucleus was appearing in two consecutive slices it was only counted once. Clearly, there is margin for error in this approach as it is only discriminatory judgement that distinguishes between what may or may not be the same nucleus rather than any definitive measurement. Table 2 shows the number of cells per unit volume for each of the

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2118

Fig. 10. Confocal images of cell growth on PLGA (A), PLGA/PU blend (B), and PU (C). Table 2 Middle and edge of slice cell density Scaffold type

Middle of slice (cells per mm3)

Edge of slice (cells per mm3)

PLGA Polyurethane PLGA/PU blend PLGA/PU ‘molecular’ blend

420740 710735 570735 490750

10707100 13607100 11507100 1030750

three scaffold types observed in the middle and at the edge of the sample slices. Interestingly, the PU scaffolds were shown to have the highest cell density, while the lowest cell density was recorded for the PLGA scaffolds even though it is known that PU, while having good biocompatibility, has been shown by others to have relatively poor cellular adhesion properties when untreated [39,40]. The mixed scaffolds were somewhere in-between the two. The scaffolds fabricated from PU (refer to Fig. 3) clearly have more available surface area, which would normally result in

greater cell attachment when compared, for example, to the PLGA scaffolds if the surface properties were identical, which of course they are not. However, these scaffolds also present a denser structure with smaller pores. It is thus thought that it is the structure of these PU scaffolds, which resulted in a higher cell count rather than favourable surface chemistries. In order to test the relative fibroblast attachment properties of PLGA and PU, fibroblasts were grown on films of PLGA and PU that had been cast onto glass coverslips. After 48 h the coverslips were gently rinsed with

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

2119

Fig. 11. Confocal images of cell growth on films of PLGA (A) and PU (B).

PBS and studied by confocal microscopy. As is clearly shown in Fig. 11(A) and (B), the number of cells adhered to the PLGA (A) surface is some 2.5 times higher than that of the PU (B) surface. Again, the cells on the PLGA surface were found to be well spread and those on the PU surface possessed a rounded morphology. These findings are inline with the results of a study by Hsu and Chen who, in a direct comparison to the polyester family to which PLGA belongs, showed that the number of both fibroblasts and endothelial cells adhering to an untreated PU (Pellethane 80A) surface was some 3 times lower than that of a PU surface grafted with L-lactide via plasma treatment [41]. This is most likely due to the lower degree of protein adsorption on the PU substrate [42]. However, in the case of the PU/PLGA-blended scaffold we note that the cells appear to be interacting very favourably with the surfaces presented to them. Thus it would seem that overall, the mixed scaffold provides the most favourable environment for cell expansion out of the three scaffold types. The available surface area for cell attachment has not been measured in this study, so while this reasoning alone (for higher cell numbers throughout the PU scaffold) is somewhat speculative, the confocal images agree with this hypothesis as the cells display a more spread morphology on PLGA than PU, indicating that PLGA has the more favourable surface chemistry for the development of focal adhesion complexes. It is well recognised that the chemistry of the surface is deterministic over the resultant absorption of selective proteins from serum. It is in fact these adsorbed proteins that the cells have specific integrins for allowing initial binding and thereafter focal adhesion development [43,44]. The presence of the hydrophobic PLGA would result in more protein adsorption and hence the resulting differences in cell morphology. For all scaffolds, cell density was higher at the edges than in the middle, which is understandable given that the scaffolds were seeded from the surface and cells have had to migrate into the bulk of the scaffold. Accuracy with regards to counting of cells in the slice volumes could have

been improved by pre-treating the scaffolds with the enzyme ribonuclease (RNAse) before dying with propidium iodide. RNAse breaks down the RNA in the cytoplasm, which would prevent the strong staining of the cytoplasm, thus the nucleus would stand out much clearer in the confocal images. However, we are confident that the numbers presented in Table 2 are representative of the actual cell numbers within each scaffold. Three-dimensional reconstructions were created for each of the samples using the Z-layering capabilities of the confocal microscope. Looking at these reconstructions (data not shown), it is apparent that for all three scaffolds there has been cell growth throughout the whole slice volume. Bearing in mind that the scaffolds were incubated for only 48 h after seeding, scaffolds made of these materials plainly support cellular growth and would be expected to continue to do so over longer time frames, although diffusion of nutrients and oxygen would slowly decrease as cell mass increased, so ultimately a point would be reached where cellular growth would be self limiting. Cell growth on these scaffolds is of a directional nature as the cells must grow along the pore channels. This directional growth is especially noticeable in the case of the PLGA scaffold. This result is not surprising given the morphology of the scaffolds observed in the electron micrographs (see Section 3.2). It is worth mentioning that depending on the intended application of the scaffold in an in vitro situation, a different morphology could prove useful to give rise to different structural tissues. For instance, in the regeneration of muscle tissues where a degree of orientation in a certain direction would be useful to encourage smooth muscle cell growth. This could easily be done using a matrix with a microstructure not unlike PLGA, whereas for more random bulk tissues that do not need such orientation such as fat for breast replacement purposes, a scaffold morphology more akin to that presented by the PU scaffold may be employed. It is also worth mentioning that the mechanical properties of the underlying substrate will ultimately effect changes in cell

ARTICLE IN PRESS 2120

A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

responses above and beyond the effects of surface chemistry in many cases [45], and given the significant difference in the mechanical properties of the these two materials, it is highly likely that some of the differences noted in cell morphology may be a result of mechanical differences in the substrates. 4. Conclusion In this work, we have presented a novel composite scaffold fabricated via a thermally induced phase separation process that exhibits morphological, mechanical and cell adhesion and growth supporting properties in-between that of scaffolds fabricated from the two individual polymers. While the blended PLGA/PU scaffold possessed architectural features similar to the PLGA scaffold, the cellular adhesion and growth properties of 3T3 fibroblasts seeded onto these scaffolds showed intermediate behaviour. The presence of PLGA throughout a PU scaffold gave rise to improved cell attachment and viability when compared to a scaffold fabricated from PU alone. Based on both cell number and morphology, the blended PLGA/PU scaffold was superior to the pure PLGA and PU scaffolds as both mechanical properties and cellular responses are important for tissue regeneration and this composite system showed improvement in both of these characteristics. A composite system made via this process is highly tuneable and can be tailored to possess more desirable characteristics overall when compared to scaffolds fabricated from single polymer systems. Acknowledgements This study was supported by the Australian Research Council Discovery Grants Scheme, The University of Queensland and The Australian Institute for Bioengineering and Nanotechnology. References [1] McIntire LV, Greisler HP, Griffith L, Johnson PC, Mooney DJ, Mrksich M, et al. WTEC panel report on tissue engineering research. Baltimore: International Technology Research Institute; 2002. [2] Kim BS, Mooney DJ. Development of biocompatible synthetic extracellular matrices for tissue engineering. Trends Biotechnol 1998;16(5):224–30. [3] Bhattarai SR, Bhattarai N, Yi HK, Hwang PH, Cha DI, Kim HY. Novel biodegradable electrospun membrane: scaffold for tissue engineering. Biomaterials 2004;25:2595–602. [4] Mo X, Weber HJ. Electrospinning P(LLA-CL) nanofiber: a tubular scaffold fabrication with circumferential alignment. Macromol Symp 2004;217:413–6. [5] Yang F, Murugan R, Wang S, Ramakrishna S. Electrospinning of nano/micro scale poly(L-lactic acid) aligned fibers and their potential in neural tissue engineering. Biomaterials 2005;26:2603–10. [6] Whang K, Thomas CH, Healy KE. A novel method to fabricate bioabsorbable scaffolds. Polymer 1995;36(4):837–42. [7] O’Brien FJ, Harley BA, Yannas IV, Gibson L. Influence of freezing rate on pore structure in freeze-dried collagen-GAG scaffolds. Biomaterials 2004;25:1077–86.

[8] Chen G, Ushida T, Tateishi T. Development of biodegradable scaffolds for tissue engineering. Mater Sci Eng 2001;C:63–9. [9] Chen VJ, Ma PX. Nano-fibrous poly(L-lactic acid) scaffolds with interconnected spherical macropores. Biomaterials 2004;25:2065–73. [10] Flynn L, Dalton PD, Shoichet MS. Fiber templating of poly (2-hydroxyethyl methacrylate) for neural tissue engineering. Biomaterials 2003;24:4265–72. [11] Tuzlakoglu K, Alves CM, Mano JF, Reis RL. Production and characterization of chitosan fibers and 3-D fiber mesh scaffolds for tissue engineering applications. Macromol Biosci 2004;4:811–9. [12] Cooper AI. Polymer synthesis and processing using supercritical carbon dioxide. J Mater Chem 1999;10:207–34. [13] Mooney DJ, Baldwin DF, Suh NP, Vacanti JP, Langer R. Novel approach to fabricate porous sponges of poly(D,L-lactic-co-glycolic acid) without the use of organic solvents. Biomaterials 1996;17:1417–22. [14] Yoon JJ, Kim JH, Park TG. Dexamethasone-releasing biodegradable polymer scaffolds fabricated by a gas-foaming/salt-leaching method. Biomaterials 2003;24:2323–9. [15] Dhariwala B, Hunt E, Boland T. Rapid prototyping of tissueengineering constructs, using photopolymerizable hydrogels and sterolithography. Tissue Eng 2004;10:1316–22. [16] Leong KF, Cheah CM, Chua CK. Solid freeform fabrication of three-dimensional scaffolds for engineering replacement tissues and organ. Biomaterials 2003;24:2363–78. [17] Sun W, Darling A, Starly B, Nam J. Computer-aided tissue engineering: overview, scope and challenges. Biotechnol Appl Biochem 2004;39:29–47. [18] Lu L, Peter SJ, Lyman MD, Lai HL, Leite SM, Tamada JA, et al. In vitro degradation of porous poly(L-lactic acid) foams. Biomaterials 2000;21:1595–605. [19] Ma PX, Choi JW. Biodegradable polymer scaffolds with well-defined interconnected spherical pore network. Tissue Eng 2001;7(1):23–33. [20] Barbetta A, Dentini M, Vecchis MSD, Filippini P, Formisano G, Caiazza S. Scaffolds based on biopolymeric foams. Adv Funct Mater 2006;15(1):118–24. [21] Nam YS, Park TG. Biodegradable polymeric microcellular foams by modified thermally induced phase separation method. Biomaterials 1999;20:1783–90. [22] Heijkants RGJC, Calck RVC, Groot JHD, Pennings AJ, Schouten AJ. Design, synthesis and properties of a degradable polyurethane scaffold for meniscus regeneration. J Mater Sci: Mater Med 2004; 15:423–7. [23] Lee SH, Kim BS, Kim SH, Kang SW, Kim YH. Thermally produced biodegradable scaffolds for cartilage tissue engineering. Macromol Biosci 2004;4:802–10. [24] Cao Y, Croll TI, O’Connor AJ, Stevens GW, Cooper-White JJ. Systematic selection of solvents for the fabrication of 3D PLGA scaffolds for tissue engineering. Transactions of the seventh world biomaterials congress, 2004. [25] Cao Y. Production of three dimensional polymeric scaffolds for use in tissue engineering. PhD. thesis, University of Melbourne, 2005. [26] Hol CE, Cheng C, Davies JE, Shoichet MS. Optimizing the sterilization of PLGA scaffolds for use in tissue engineering. Biomaterials 2001;22:25–31. [27] Guan J, Fujimoto KL, Sacks MS, Wagner WR. Preparation and characterization of highly porous, biodegradable polyurethane scaffolds for soft tissue applications. Biomaterials 2005;26:3961–71. [28] Wang M. Developing bioactive composite materials for tissue replacement. Biomaterials 2003;24:2133–51. [29] Wake MC, Gupta PK, Mikos AG. Fabrication of pliable biodegradable polymer foams to engineer soft tissues. Cell Transplant 1996; 5(4):465–73. [30] van Emmerik PT, Smolders CA. Differential scanning calorimetry of poly(2,6 dimethyl-1,4 phenylene-oxide)-toulene solutions. Eur Polym J 1973;9:293–300. [31] van de Witte P, Boorsma A, Esselbrugge H, Dijkstra PJ, van den Berg JWA, Feijen J. Differential scanning calorimetry study of phase

ARTICLE IN PRESS A.S. Rowlands et al. / Biomaterials 28 (2007) 2109–2121

[32]

[33]

[34]

[35]

[36]

[37]

transitions in poly(lactic)-chloroform-methanol systems. Macromolecules 1996;29:212–9. Arnauts J, De Cooman R, Vandeweerdt P, Koningsveld R, Berghmans H. Calorimetric analysis of liquid–liquid phase separation. Thermochim Acta 1994;238:1–16. Simmons A, Hyvarinen J, Odell RA, Martin DJ, Gunatillake PA, Noble KR, et al. Long-term in vivo biostability of poly(dimethylsiloxane)/poly(hexamethylene oxide) mixed macrodiol-based polyurethane elastomers. Biomaterials 2004;25:4887–900. Stepanek P, Morkved TL, Krishnan K, Lodge TP, Bates FS. Critical phenomena in binary and ternary polymer blends. Physica: Stat Mech Appl 2002;314(1-4):411–8. Litmanovich AD, Plate´ NA, Kudryavtsev YV. Reactions in polymer blends: interchain effects and theoretical problems. Prog Polym Sci 2002;27(5):915–70. Jun Jin Yoon TGP. Degradation behaviors of biodegradable macroporous scaffolds prepared by gas foaming of effervescent salts. J Biomed Mater Res 2001;55(3):401–8. Wu L, Ding J. In vitro degradation of three-dimensional porous poly(D,L-lactide-co-glycolide) scaffolds for tissue engineering. Biomaterials 2004;25(27):5821–30.

2121

[38] Gibson LJ, Ashby MF. Cellular solids. Exeter: A. Wheaton & Co. Ltd; 1988. [39] Lin HB, Cooper SL. Synthesis, surface and cell adhesion properties of polyurethanes containing covalently grafted RGD-peptides. Mater Res Soc 1994;331:105–13. [40] De S, Sharma R, Trigwell S, Laska B, Ali N, Mazumder MK, et al. Plasma treatment of polyurethane coating for improving endothelial cell growth and adhesion. J Biomater Sci Polym Ed 2005;16(8):973–89. [41] Hsu S, Chen WC. Improved cell adhesion by plasma-induced grafting of L-lactide onto polyurethane surface. Biomaterials 2000;21:359–67. [42] Serbetci AI, Piskin E. Proteins and cells on polyurethane surfaces. Clin Mater 1992;11:163–70. [43] Garcia AJ, Boettiger D. Integrin–fibronectin interactions at the cell–material interface: initial integrin binding and signalling. Biomaterials 1999;20:2427–33. [44] Miyamoto S, Teramoto H, Coso OA, Gutkind JS, Burbelo PD, Akiyama SK, et al. Integrin function: molecular hierarchies of cytoskeletal and signaling molecules. J Cell Biol 1995;131(3):791–805. [45] Yeung T, Georges PC, Flanagan LA, Marg B, Ortiz M, Funaki M, et al. Effects of substrate stiffness on cell morphology, cytoskeletal structure, and adhesion. Cell Motil Cytoskeleton 2005;60:23–34.