- Email: [email protected]
Solid lipid templating of macroporous tissue engineering scaffolds
ARTICLE IN PRESS
Biomaterials 28 (2007) 3497–3507 www.elsevier.com/locate/biomaterials
Solid lipid templating of macroporous tissue engineering scaffolds Michael Hackera,d, Michael Ringhoferb, Bernhard Appela, Markus Neubauera, Thomas Vogela,c, Simon Youngd, Antonios G. Mikosd, Torsten Blunka, Achim Go¨pfericha, Michaela B. Schulza,c,e, a
Department of Pharmaceutical Technology, University of Regensburg, Universita¨tsstrasse 31, 93040 Regensburg, Germany b Anton Paar GmbH, Anton Paar Strasse 20, 8054 Graz, Austria c Department of Pharmaceutical Technology, University of Graz, Schubertstrasse 6, 8010 Graz, Austria d Department of Bioengineering, Rice University, 6100 Main Street, Houston, TX 77005-1892, USA e Department of Pharmaceutical Technology, University of Leipzig, Schoenauer Strasse 160, 04207 Leipzig, Germany Received 20 December 2006; accepted 11 April 2007 Available online 18 April 2007
Abstract Macroporous biodegradable cell carriers (scaffolds) provide the three-dimensional matrix for tissue formation in vitro. In this study, we present the fabrication of macroporous scaffolds with high inter-pore connectivity from different biodegradable polymers using the recently developed solid lipid templating technique. Starting from a polymer solution and solid lipid microparticles, a dispersion is prepared and subsequently transferred into molds, which are finally submerged in warm hexane to precipitate the polymer and extract the porogens. The study shows how to control pore structure, pore size and porosity of the scaffold using this technique. The process parameters dispersion viscosity, porogen size and type of polymer are considered. Limits of viscosity are examined by macroscopic and microstructure evaluation of the scaffolds prepared at different viscosities. An approach to rationalize these data by oscillation rheometry is shown. Pore size can be controlled by porogen particle size and adaptation of the viscosity of the polymer solution. Porosity can be modified by changing the ratio of porogen to polymer. The suitability of the resulting scaffolds was shown using an established cartilage cell culture model. r 2007 Elsevier Ltd. All rights reserved. Keywords: Scaffold; Lipid; Polylactic acid; Polyethylene oxide; Cartilage tissue engineering
1. Introduction A typical approach to engineer functional tissue for medical and basic science applications combines biocompatible, biodegradable, polymeric cell carriers (scaffolds) as a temporary template in combination with mammalian cells and signaling molecules (e.g. growth, differentiation and adhesion factors) to guide and promote threedimensional tissue development [1–4]. In such strategies, the scaffolds play a pivotal role with regard to cell seeding, proliferation, and new tissue formation [5]. Biodegradable Corresponding author. Department of Pharmaceutical Technology, University of Leipzig, Schoenauer Strasse 160, 04207 Leipzig, Germany. Tel.: +49 341 4229747; fax: +49 341 4123007. E-mail address: [email protected] (M.B. Schulz).
0142-9612/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2007.04.018
polyesters, particularly poly(D,L-lactic acid) (PLA) and its copolymers, are attractive and widely applied scaffolding materials with advantages over ceramics, metals, and natural polymers because they degrade as the new tissue forms and can be synthesized in reproducible quality and free of pathogenic or immunogenic organic residue [6,7]. Depending on the mechanical properties of the tissue to be engineered and the extent of cell–material interaction desired, a suitable polymer has to be selected with regard to its degradation kinetic, composition and molecular weight. Consequently, there is a need to process polymers with different physicochemical characteristics into macroporous scaffolds with reproducible microstructure. Various techniques such as salt leaching [8,9], fibrous fabric processing [6], gas foaming [10], thermally induced phase separation [11] and several solid freeform fabrication
ARTICLE IN PRESS 3498
M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
techniques [12,13] have been developed to generate highly porous polymer scaffolds from biodegradable polymers. Despite the remarkable potential that some of these scaffolds have shown, existing scaffold fabrication techniques lack the ability to control critical scaffold properties and allow for the processing of hydrolytically sensitive or water soluble materials [14,15]. Architectural features, namely pore size and shape, pore wall morphology, porosity, surface area and pore interconnectivity, are probably the most critical parameters as those have been shown to directly impact cell seeding, cell migration, tissue differentiation, transport of oxygen, nutrients and wastes, and new tissue formation in three dimensions [16,17]. As some applications use experimental polymers that may contain hydrolysable functional groups [18] or require the encapsulation of growth factors during scaffold fabrication, an anhydrous fabrication technique is generally preferred. Moreover, the technique should allow for labscale processing, but not be limited to a certain scaffold size. A previously published hydrocarbon templating technique [19], which met most of the described demands, was further improved with regard to biocompatibility and adaptability to low molecular weight block copolymers by our group [20]. The developed solid lipid templating technique employs solid lipid microparticles as porogens and non-halogenated polymer solvents to process biodegradable polymers into macroporous scaffolds, providing an interconnected pore structure. Pore interconnectivity is obtained by precipitating the polymer from its solution at the phase boundary with a continuous molten lipid phase that is formed during porogen extraction in warm n-hexane, which is a non-solvent for the polymer [20]. In this study, we investigated the parameters critical for the control of pore structure and size. In particular, composition and viscoelastic properties of the paste-like dispersion from polymer solution and porogen particles were varied and effects on pore morphology were determined. Scaffolds with variable pore sizes and open pore structure were fabricated and the potential of solid lipid templating to process different polymers and to yield scaffolds with varying porosities was shown. The suitability of the fabricated scaffolds for in vitro tissue engineering was tested using an established cartilage cell culture model [21]. 2. Materials and methods 2.1. Materials PLA, Resomers R207, Mw ¼ 160 kDa (Mw/Mn ¼ 1.4), Tg ¼ 54.8 1C, poly(D,L-lactic-co-glycolic) (PLGA), Resomers RG756, synthesized from 75% lactic acid (LA) and 25% glycolic acid (GA), Mw ¼ 90 kDa (Mw/ Mn ¼ 1.5), Tg ¼ 50.7 1C, were kindly provided by Boehringer Ingelheim (Ingelheim, Germany). Monomethyl ether-poly(ethylene glycol)-blockPLA (MeO-PEG2PLA40; MeO-240), Mw ¼ 69 kDa (Mw/Mn ¼ 1.9), Tg ¼ 43.0 1C, which consists of a 2 kDa MeO-PEG block covalently bound to a 40 kDa PLA block, was synthesized and characterized as previously described [22]. The solid lipids that were used as porogen materials, namely Softisans 154 and Witepsols H42, were kindly provided by SASOL Germany
GmbH (Witten, Germany). Methyl ethyl ketone (MEK), tetrahydrofuran (THF), ethyl acetate and n-hexane were purchased in analytical grade from Merck (Darmstadt, Germany).
2.2. Methods 2.2.1. Preparation of solid lipid microparticles Microparticles were prepared using a melt dispersion technique. Ten grams of the solid lipid mixture prepared from Softisans 154 und Witepsols H42 (SH), either SH 1:1 or SH 2:1, was weighed in a poly(propylene) tube. After the addition of 7.5 mL water, the tube was heated to 65 1C. In a normal batch, eight tubes were processed in parallel. The mixture was emulsified by turning the tube upside down 10 times and subsequently cast into a larger volume (600 mL) of stirred, cold (15 1C) water. After 10 min of constant stirring, the hardened particles were collected by filtration, rinsed with cold water and dried under laminar air flow for 3 days. Finally, the microparticles were separated into different size ranges by sifting through sieves (mesh sizes: 100, 300, 500, 710 mm) (Retsch GmbH & Co. KG, Haan, Germany). 2.2.2. Particle size analysis The size distribution of the prepared lipid microparticles was investigated using laser diffraction (Mastersizer 2000 Hydro 2000S, Malvern Instruments, Herrenberg, Germany) as follows: approximately 100 mg of lipid microparticles were directly added to the dispersion unit (Hydro 2000S), which was filled with an ethanol-water-mixture (68.2% (v/v), r ¼ 0.89 g/cm3, refractive index: 1.36). The particles were dispersed by stirring at 3000 rpm for 5 min. The volume-based particle size distribution was calculated using the Fraunhofer approximation (Malvern Software V5.1). Further values provided by the software included the average particle size (d3,2) (surface weighted mean diameter) and specific surface area. The measurements were repeated in triplicate. 2.2.3. Scaffold fabrication procedure To fabricate macroporous scaffolds from biodegradable polymers, we recently developed an anhydrous technique that uses solid lipid microparticles as porogens [20]. Briefly, scaffold fabrication started with a polymer solution. Appropriate solvents and polymer concentrations are described in the Results section. Under ice cooling, the polymer solution was mixed with the solid lipid microparticles. A polymer to porogen ratio of 1:4 (w/w) was used if not otherwise stated. The homogeneous dispersion was transferred into Teflons molds (1.9 cm 1.9 cm 1.2 cm) with a cylindrical cavity of 0.8 cm in diameter. After a pre-extraction treatment step in n-hexane at 0 1C (MeO-PEG2PLA40: 90 min; other polymers: 15 min), the molds were submerged in warm n-hexane to induce solvent extraction followed by the precipitation of the polymer and extraction of the lipid porogen. This procedure was carried out in two separate n-hexane baths of different temperatures T1 and T2 for t1 and t2 with t1+t2 ¼ 30 min. The resulting porous, cylindrical polymer constructs were allowed to cool in cold (0 1C) n-hexane and were removed from the molds. After drying under vacuum for 48 h, the constructs were cut into 2 mm slices, which were then described as scaffolds. Table 1 gives a summary of the applied processing conditions. Solvent residues were removed by a 48 h period under vacuum and subsequent 24 h incubation under seeding conditions in primary medium in a spinner flask at 37 1C. Thermogravimetric analysis (TGA 1000, Polymer Laboratories Thermal Sciences) was used to quantify solvent residues. A heating rate of 5 K/min was applied to analyze scaffolds after vacuum drying and after additional incubation. 2.2.4. Microscopic assessment of scaffold macro- and microstructure The macrostructure of the porous polymer cylinders and scaffolds was examined using a zoom stereo microscope (Wild M7A, Wild Heerbrugg Ltd., Heerbrugg, Switzerland). The scaffold microstructure was visualized by scanning electron microscopy (SEM). For this procedure, samples were mounted on aluminum stubs with conductive carbon tape and coated with
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3499
Table 1 Porogen characteristics and processing parameters applied for the fabrication of scaffolds from different polymers Polymer
Porogen material
Size range (mm)
d3,2 (mm)
Polymer content (w/w) (%)
Conc. (polymer solution) (mg/mL)
Extraction conditions
Step 1 (T1; t1)
Softisan 154: Witepsol H42 Pore size/rheology PLGA (RG756) PLGA (RG756) PLGA (RG756) MeO-240
2:1 2:1 2:1 1:1
100–300 300–500 500–710 100–300
200 360 590 180
20 20 20 25
330 350 365 435
52 1C; 52 1C; 52 1C; 45 1C;
Porosity PLGA (RG756) PLGA (RG756) PLGA (RG756)
2:1 2:1 2:1
100–300 100–300 100–300
200 210 210
20 15 10
330 265 185
52 1C; 100 52 1C; 100 52 1C; 100
Polymer type MeO-240 MeO-240 PLA (R207) PLA (R207)
1:1 1:1 2:1 2:1
100–300 300–500 100–300 300–500
200 400 180 255
20 20 20 20
400 410 230 250
45 1C; 45 1C; 52 1C; 52 1C;
100 100 100 100
100 100 100 100
Porosity (%)
Step 2 (T2; t2)
200 200 200 200
85.7#71.2 87.3#71.0 87.1#70.9 —
40 1C; 200 40 1C; 200 40 1C; 200
90.4*70.5 91.5*70.4 94.1*70.4
40 1C; 40 1C; 40 1C; 35 1C;
35 1C; 35 1C; 40 1C; 40 1C;
200 200 200 200
— — — —
Porosity data of different PLGA (RG756) scaffolds was determined gravimetrically (*) or by micro-CT (#) as means7standard deviation for n ¼ 3.
gold-palladium. All micrographs were obtained at 10 kV on a DSM 950 (Zeiss, Oberkochen, Germany). 2.2.5. Computer tomographic assessment of scaffold microstructure and interconnectivity Micro-computed tomography (micro-CT) (Desktop Micro CT1172, SkyScan, Aartselaar, Belgium) was employed to non-destructively determine the microstructure, interconnectivity and porosity of PLGA scaffolds fabricated with porogen microparticles of different size ranges. Each scaffold (cylindrical discs: 8 mm diameter 2 mm height) was mounted vertically on a sample holder using double adhesive tape. The resolution was set to 8 mm. The samples were scanned at a source voltage of 42 kV and a current of 250 mA. Following image acquisition, transverse slices were generated with SkyScan reconstruction software (NRecon) and resliced twice to yield coronal sections of the scaffolds. Image analysis was performed using the standard SkyScan software (CTAn). Images were loaded and resized by half resulting in pixel and voxel sizes of (16 mm)2 and (16 mm)3, respectively. A volume of interest (VOI) was defined by superimposing a cylinder just enclosing the scaffold volume. For each resulting dataset the VOI and the object volume (OV) (polymer volume of scaffold) were quantified (VOIpre and OVpre). In order to determine pore interconnectivity, a three-dimensional ‘shrink wrap’ was performed using the SkyScan software plugin. In this operation, the VOI is reduced to the boundary of the binarized solid object from outside the object to the inside eliminating all void space that is accessible through interconnects of a predefined size (even multiples of the voxel size). This analysis was performed for 2, 4, 6, 8, 10, 12, 16, 20, 30, and 40 times the voxel size [(16 mm)3] and the shrunk VOI (VOIpost) was determined. OVpost equaled OVpre. The relative inaccessible volume was determined as (VOIpost OVpost)/(VOIpreOVpre). Scaffold porosity was calculated as (VOIpreOVpre)/ VOIpre. 2.2.6. Oscillatory rheological measurements The dispersions of porogen particles in polymer solution prepared during scaffold fabrication were characterized using a Physica MCR 300 rheometer (Anton Paar GmbH, Graz, Austria) with a 2.5 cm sandblasted steel plate geometry according to the following protocol: under ice cooling, the polymer solution was mixed with the solid lipid microparticles
for 3 min. The mixture was transferred to the rheometer using a polypropylene syringe and equilibrated at 5 1C. The measurement gap was set to 2 mm. After 60 s, the first frequency sweep (100–0.1 Hz) was measured at a strain of 0.01% and a temperature of 5 1C. A second frequency sweep (100–0.1 Hz at 0.01% strain) at 5 1C followed another equilibration period of 3 min. The parameters recorded during the second frequency sweep were analyzed to determine storage (G0 ) and loss modulus (G00 ), loss factor (tan d ¼ G00 /G0 ), and complex viscosity (|Z*|) of the dispersion. To compare different formulations, the values of the complex viscosity at 1 Hz were interpolated. 2.2.7. Determination of residual porogen material The detection of triglyceride residues inside the scaffolds was realized through modulated differential scanning calorimetry (MDSC). The samples were precisely weighed in non-hermetic AutoDSC aluminum sample pans (TA Instruments, Alzenau, Germany). The sample pans were sealed using the sample encapsulating press (TA Instruments, Alzenau, Germany) and analyzed on a DSC 2920 equipped with a refrigerated cooling system and an autosampler (TA Instruments, Alzenau, Germany). An empty, sealed pan served as reference. All measurements were carried out between 20 and 120 1C. Typically, samples were equilibrated at 20 1C for 15 min and heated to 120 1C at a heating rate of 5 1C/min. After an isothermal phase of 15 min, samples were cooled to 20 1C at 5 1C/min. Finally, after another 15 min isothermal phase, samples were again heated to 120 1C at 5 1C/min. A sinusoidal temperature modulation with a period of 60 s and temperature amplitude of 0.8 1C was applied to both heating cycles. The resulting thermograms (total heat flow of the second heating cycle) were analyzed for melting enthalpy (peak area) of the peak attributed to the melting lipid residuals located near the glass transition step of the polymer with the Universal Analysis for NTs software provided with the DSC system. 2.2.8. Determination of scaffold porosity The porosity of the scaffolds was determined by measuring the dimensions and the mass of the porous polymer cylinders as obtained from fabrication (cylinder ends were cut away) [23]. The skeletal density of the cylinder (r) was calculated from the mass (m), the diameter (d) and the
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3500 height (h) of the cylinder: 4m r¼ 2 . pd h
ltop (1)
l = lbottom bottom - ltop
The porosity (e) was calculated from the density of the construct (r) and density (rp) of the polymer (rp(RG756) ¼ 1.26 g/cm3): r . rp
lbottom
(2)
2.2.9. Chondrocyte cell culture Primary chondrocytes were isolated from full-thickness bovine articular cartilage by digestion with type II collagenase (CellSystem, St. Katharinen, Germany) as previously described [24] and resuspended in culture medium (DMEM) containing 4.5 g/L glucose (Life Technologies GmbH, Karlsruhe, Germany), 584 mg/L glutamine, 10% FBS (Life Technologies GmbH, Karlsruhe, Germany), 50 U/mL penicillin, 50 mg/mL streptomycin (Sigma-Aldrich Steinheim, Germany), 10 mM HEPES (Sigma-Aldrich Steinheim, Germany), 0.1 mM non-essential amino acids (Life Technologies GmbH, Karlsruhe, Germany), 0.4 mM proline (Sigma-Aldrich), and 50 mg/mL ascorbic acid (Sigma-Aldrich). This medium was also used for cell seeding and cultivation. The cells were seeded in spinner flasks, as described elsewhere [21]. In brief, after preincubation of the scaffolds for 24 h in medium, a cell suspension containing 5 106 chondrocytes per scaffold was stirred at 80 rpm in a humidified (37 1C/5% CO2) incubator for 2 days to allow for cell attachment to the polymers. Cell-polymer constructs were transferred into 6-well plates (one construct and 6 mL culture medium per well) and cultured for 3 weeks on an orbital shaker at 50 rpm. Medium supplemented with different amounts of insulin (0 mg/mL, 0.05 mg/mL, 2.5 mg/mL) was completely exchanged three times per week. At the end of the culture period, each cell-polymer construct was weighed ( ¼ wet weight) and cut in half. One half was prepared as histological sample by fixing in 2% glutaraldehyde in PBS for 30 min and then in 10% formalin. The samples were embedded in paraffin and crosssectioned (5 mm thick). Deparaffinized sections were stained with hematoxylin and eosin (H&E) (C.I.: 45380, Sigma-Aldrich) for cells and with safranin-O for glycosaminoglycans (GAG). The other half of the construct was used for biochemical analysis of collagen content and GAG content [21].
3. Results 3.1. Control of pore structure during scaffold fabrication 3.1.1. Influence of polymer concentration on the scaffold structure In solid lipid templating, solid lipid microparticles are dispersed into an organic solution of the polymer. Suitable polymer solvents are non-solvents for the lipid microparticles. The composition of the paste-like dispersion from porogen particles and polymer solution has been found to be decisive with regard to shape, pore structure and porosity of the resulting scaffolds. The dispersion’s viscosity is mainly controlled by the ratio of polymer to porogen particles, the porogen particle size and the amount of polymer solvent used. Since polymer to porogen ratio and porogen size are preset in order to control scaffold porosity and pore size, respectively, the concentration of the polymer solution is the key parameter to control dispersion properties.
3
Length difference [mm]
¼1
2
1
0 380
330
300
Concentration of polymer solution [mg/ml] Fig. 1. (a) Macroscopic view of a deformed scaffold as a result of low dispersion viscosity. To characterize the deformation the length difference (Dl) is calculated. (b) Length difference measured on PLGA (RG756) scaffolds fabricated from differently concentrated polymer solutions. Columns represent average7standard deviation (n ¼ 4). Statistical significance (po0.01) is denoted by a *.
Fig. 1 shows a gross approach to find limits for dispersion viscosity. Low polymer concentrations (300 mg/mL) resulted in a paste with low viscosity that partly ran out of the mold and caused deformation of the polymer cylinder (Fig. 1a). This deformation can be described by the difference in length between the top and the bottom of the resulting scaffold cylinder (Fig. 1b). Minor deformations of the forming polymer cylinder were observed when the polymer concentration was increased to 330 mg/mL. The remaining deformation was not significantly improved with a high concentration of 380 mg/mL. In order to correlate this gross approach to the scaffold microstructure, SEM images of the corresponding scaffolds were taken. Polymer processing at the highest concentration (380 mg/mL) resulted in a microstructure characterized by pores in the shape of the porogen microparticles showing inter-pore connections of less than 100 mm diameter (Fig. 2). Lower polymer concentrations (330 and 300 mg/mL), in contrast, resulted in a highly permeable network of well-condensed polymer sponges with large pore interconnections. The structure achieved with these lower concentrations closely resembles natural spongy bone [25]. Based on the data shown in Figs. 1
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3501
Fig. 2. Scanning electron micrographs of PLGA (RG756) scaffolds fabricated using different polymer concentrations. (a) 380 mg/mL, (b) 330 mg/mL, (c) 300 mg/mL. Porogen particles were 100–300 mm in size. Scale bars represent 100 mm.
and 2, a polymer concentration of 330 mg/mL is suitable to process PLGA (RG756) with microparticles of 100–300 mm into macroporous scaffolds with a spongy microstructure. In order to rationalize the empirical approach by rheological data, we characterized the scaffolding dispersion using oscillation rheology. 3.1.2. Rheological characterization of the scaffolding dispersions On a controlled stress rheometer equipped with a plate–plate geometry, frequency sweep measurements were performed. The complex viscosity (|Z*|) and the complex moduli, storage (G0 ) and loss modulus (G00 ), were recorded over a frequency range of 0.1–100 Hz. Fig. 3a shows the frequency sweeps of the three dispersions of 300, 330 and 380 mg/mL PLGA. All systems displayed storage and loss moduli that were frequency dependent. Typically, increasing the oscillatory frequency increased both the storage and loss modulus at frequencies higher than 1 Hz, whereas the complex viscosity decreased. In all three formulations, the storage modulus (G0 ) exceeded the loss modulus (G00 ) resulting in a loss factor (tan d ¼ G00 /G0 ) smaller 1 (Fig. 3b). Accordingly the systems can be addressed as gels. We observed that the loss factor first decreased between 0.1 and 1 Hz, indicating an increase in elastic properties. At frequencies higher then 1 Hz viscous properties became stronger in dispersions with low (300 mg/mL) to intermediate (330 mg/mL) concentrations, whereas the highly concentrated dispersions (380 mg/mL) showed only small changes over the frequency range and the lowest loss factor indicating a more stable system. These differences are also reflected by the larger standard deviations found for dispersion with low polymer concentrations. Dispersions that remained stable and predominantly elastic over the frequency range such as the 380 mg/mL PLGA system resulted in non-desired pore structures as shown in Fig. 2a. Systems showing more inhomogeneity in their loss factor seem to allow for the necessary coalescence of the melting microparticles while systems embedding their microparticles perfectly in a polymer gel lead to pores reflecting the outer shape of the porogen particles.
The dispersion’s viscosity also depends on the size distribution of the porogen particles and the corresponding surface area of the microparticles that may vary depending on the microparticle batch. More definite results may therefore be generated by use of monodisperse porogen particles. Furthermore, a change in polymer type and molecular weight will change the required frequency-dependent gel state of the dispersion. The intended rationalization of the dispersions composition by rheology was, therefore, only possible to a limited degree. 3.2. Control of pore size 3.2.1. Porogen microparticle preparation The solid lipid templating technique described here uses triglyceride microparticles prepared from a mixture of two solid lipids, Softisans 154 und Witepsols H42, as pore forming templates [20]. Despite the principle of melting the porogen particles during extraction and simultaneous polymer precipitation to generate interconnected pore networks, the pore forming particles are intended as a means to control the pore size distribution in the resulting scaffolds. Solid lipid microparticles ranged in size from 50 to 1000 mm and were fractionated by sieving. The following size ranges were collected for scaffold fabrication: 100–300, 300–500 and 500–710 mm. Fig. 4 shows the particle size distribution within the three porogen classes, as determined by laser diffraction. The representative graphs depict narrow distributions characterized by D(0.5) values (the diameter below which 50% of the volume of particles are found) of about 210 mm (fraction: 100–300 mm), 365 mm (fraction: 300–500 mm) and 580 mm (fraction: 500–710 mm) that corresponded to the d3,2 (surface weighted mean) values (Table 1). 3.2.2. Processing of dispersions with different porogen sizes For polymer processing using larger porogen particles, the concentration of the polymer solution was adapted to the changed conditions. Larger particles result in less surface area per weight unit that need to be coated and bridged by the viscous polymer solution. Hence, we found
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3502
4) 5) 7) 6) 8)
105
106
9)
104 105 103
Differential size distribution
1) 2) 3)
Complex moduli: G’ (open symbols) and G’’ (filled symbols) [Pas]
Complex viscosity (|η*|) [Pas]
106
107
50
10 102 0.1
1
104 100
10 Frequency [Hz]
1000
0 10000
Particle size [µm] Fig. 4. Differential and cumulative porogen microparticle size distribution of different sieve fractions. K: 100–300 mm, J: 300–500 mm, : 500–710 mm. Particles were prepared from the 2:1 mixture of Softisans 154 and Witepsols H42.
0.8
Loss factor (tanδ)
100
Cumulative size distribution [%]
100 107
highly interconnected microstructure were fabricated from all three different porogen size ranges.
0.6
0.4
0.2 0.1
1
10
100
Frequency [Hz]
Fig. 3. Rheological measurements of dispersions prepared from differently concentrated RG756 solutions and SH 2:1 microparticles (100–300 mm). (a) Frequency dependence of storage (G0 ) and loss (G00 ) modulus, and the complex viscosity (|Z*|) of the dispersions at 5 1C. Rheograms originate from dispersions with polymer concentrations of 300 mg/mL (squares), 330 mg/mL (triangles), and 380 mg/mL (circles): (1) |Z*| for 380 mg/mL, (2) |Z*| for 330 mg/mL, (3) |Z*| for 300 mg/mL (1–3: left y-axis), (4) G0 for 380 mg/mL, (5) G0 for 330 mg/mL, (6) G0 for 300 mg/mL, (7) G00 for 380 mg/mL, (8) G00 for 330 mg/mL, (9) G00 for 300 mg/mL (4–9: right y-axis). Data are represented as means and standard deviations of separately prepared dispersions (300 mg/mL: n ¼ 3; 330 and 380 mg/mL: n ¼ 4). (b) Frequency dependence of the loss factors of the different dispersions at 5 1C. Rheograms originate from dispersions with polymer concentrations of (’) 380 mg/mL, (m) 330 mg/mL, and (K) 300 mg/mL. Data are represented as means and standard deviations of separately prepared dispersions (300 mg/mL: n ¼ 3; 330 and 380 mg/mL: n ¼ 4).
empirically that the solvent volume needed to be decreased with increasing porogen size, resulting in an increase in polymer concentration. The sensitivity of the rheological procedure, however, was not high enough to detect these differences most likely as a consequence of the large gap size of 2 mm between the measuring plates that had to be used to hold the dispersion containing the increasingly macroscopic particles. The corresponding microstructures of the fabricated scaffolds are shown in Fig. 5. Scaffolds with a spongy,
3.2.3. Determination of scaffold interconnectivity PLGA (RG 756) scaffolds fabricated with porogen microparticles of different size ranges were scanned by micro-CT. The reconstructed models were analyzed for pore interconnectivity by determining the pore volume inaccessible for a virtual object of varying sizes. Fig. 6 displays the inaccessible volumes as a function of the size of the permeating object for scaffolds fabricated with porogen microparticles of different size ranges. The data illustrates large interconnect sizes that increase with increasing porogen size. For a cubic object of an edge length of 96 mm (6 fold voxel size), for example, only 9.274.7% of the pore volume of a scaffold prepared with the 100–300 mm porogen particles was inaccessible. Correspondingly, more than 90% of the pore volume was accessible for the object. The same object could access more than 94% of the pore volume in scaffolds fabricated with 300–500 mm porogens and more than 97% when porogen particles of the 500–710 mm size range were used. 3.2.4. Determination of lipid and solvent residuals in the scaffolds The solid lipid microparticles were extracted successfully with the described extraction conditions (Table 1). Independent of the porogens’ size distribution, a residual content of less than 1% (lipid per scaffold (w/w)) porogen material was found per scaffold by DSC analysis. The residual lipid contents in PLGA (RG756) scaffolds were 0.3470.08% for the 100–300 mm porogen fraction, 0.2870.06% for the 300–500 mm fraction and 0.6270.20% for the 500–710 mm fraction. TGA of the scaffolds revealed that about 3–4% of solvent residuals remained in the scaffolds after fabrication and subsequent vacuum drying for 48 h. After the pre-seeding
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3503
Fig. 5. Scanning electron micrographs of PLGA (RG756) scaffolds fabricated using porogen microparticles (SH 2:1) of different size ranges. (a) 100–300 mm, (b) 300–500 mm, (c) 500–710 mm. Scale bars represent 500 mm (large pictures) and 200 mm (close-ups).
such scaffolds are listed in Table 1. The increase of the porogen concentration necessitated a relative increase in the amount of polymer solvent to homogeneously disperse the porogen microparticles, resulting in decreasing concentrations of the polymer solutions.
Fraction of inaccessible volume
0.75
0.5
3.4. Processing of different polymers
0.25 PLGA 100-300 PLGA 300-500 PLGA 500-710
0 0
5
10
15 20 25 30 35 Object size [fold voxel size]
40
45
Fig. 6. Fraction of inaccessible volume of PLGA (RG756) scaffolds relative to the size of a penetrating virtual object as determined from micro-CT data. Scaffolds fabricated with porogen microparticles (SH 2:1) of different size ranges were analyzed. &100–300 mm, B 300–500 mm, n 500–710 mm.
incubation with cell culture media in a spinner flask at 37 1C for 24 h, however, no residual solvent was detected in the freeze-dried samples by TGA. 3.3. Control of scaffold porosity The use of different size porogen microparticles had no effect on the bulk porosity of PLGA (RG756) scaffolds (Table 1). Micro-CT data revealed a porosity of 85–87% for PLGA scaffolds fabricated with a polymer to porogen ratio of 1:4 (w/w) independent of the porogen size distribution. A significant increase in porosity was achieved by reducing the polymer content during processing (Table 1). PLGA (RG756) scaffolds fabricated from 20% (w/w) polymer and 80% (w/w) solid lipid microparticles displayed a porosity of 90.470.5%. An increase in porogen content to 85% resulted in a porosity of 91.570.4%. A porosity of 94.170.4% was found for scaffolds fabricated with 90% (w/w) pore forming microparticles (Table 1). The suitable processing conditions for
Processing parameters were determined and optimized for a variety of biodegradable polymers to demonstrate the versatility and adaptability of the solid lipid templating technique. As a representative for high molecular weight biodegradables, PLA (R207, Mw ¼ 160 kDa) was processed. A sufficient dispersion viscosity was achieved with comparably low polymer concentrations (Table 1). Lipid microparticles with a melting point of 49 1C (SH 2:1) were used as porogen material and extracted under the corresponding conditions (step 1: 52 1C for 100 ; step 2: 40 1C for 200 ). Control experiments (data not shown) revealed that the microstructure was not improved by using solid lipid microparticles with a higher melting point in combination with higher extraction temperatures. Control over the pore size was again achieved by varying the size of the porogen particles and adapting the amount of polymer solvent (Table 1 and Fig. 7a, b). Processing of a low molecular weight polymer with increased hydrophilicity, MeO-PEG2PLA40, was realized by using a 20% (w/w) solution of polymer in a MEK–THFmixture (59:41 (v/v)) and 80% (w/w) lipid microparticles with a low melting point (SH1:1, Tm ¼ 44 1C) adapted to the relatively low glass transition temperature of the polymer. Correspondingly, the extraction temperatures were adjusted (T1 ¼ 45 1C, T2 ¼ 35 1C) [22]. As described for PLA and PLGA, the solvent volume was adjusted to the size range of the porogen particles (Table 1). This way, porous scaffolds with large interconnects were fabricated with the pore sizes being dependent on the porogen size range (Fig. 7c, d) 3.5. Scaffold testing: engineering of cartilaginous tissue An established, insulin-dependent three-dimensional culture system to engineer cartilage from bovine chondrocytes
ARTICLE IN PRESS 3504
M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
[21] was transferred from fiber meshes to solid lipid templated PLGA (RG756) scaffolds to test the applicability of the scaffolds. The scaffolds that were used in this experiment were fabricated with a polymer content of 17.5% (w/w) using a porogen size range of 100–300 mm. Insulin-dependent tissue development and extracellular matrix (ECM) production was indicated by increasing wet weights along with the dose of exogenously administered insulin (Table 2). After 3 weeks, the wet weight of scaffolds that received no exogenous insulin was 88.3 mg718.9. The dry weight of an empty scaffold after fabrication was 10 mg. Scaffolds that received exogenous insulin had a significantly higher weight. The addition of 0.5 mg/mL insulin resulted in constructs of almost double the wet weight of control
samples. Supplementation with 2.5 mg/mL insulin further increased the wet weight by 50%. In all constructs, the cells deposited collagen and GAG, two major ECM components of cartilaginous tissue. The ECM production was stimulated by exogenous insulin in a dose-dependent manner. The collagen fraction (amount of collagen per wet weight) was increased by insulin supplementation. No difference in the amount of collagen per wet weight was found for the two insulin concentrations (Table 2). The GAG fraction also increased with the dose of supplemented insulin. The constructs of the insulin (2.5 mg/mL) group showed an approximately 2-fold higher GAG fraction than constructs receiving no insulin (Table 2). The different amounts of GAG were also reflected in histological sections of the cell-polymer constructs, stained red with safranin-O for GAG. Staining of the ECM of control constructs receiving no insulin revealed an uneven GAG distribution (Fig. 8a). In constructs that received insulin, even in those treated with the low insulin concentration (0.5 mg/mL), the ECM stained evenly positive for GAG (Fig. 8b). The ECM in constructs that received 2.5 mg/mL insulin showed strong positive staining for GAG up to the edge of the tissue (Fig. 8c). 4. Discussion
Fig. 7. Scanning electron micrographs of scaffolds fabricated from two different polymers using porogen particles of different composition and sizes. (a) PLA (R207)/SH 2:1 (100–300 mm), (b) PLA (R207)/SH 2:1 (300–500 mm), (c) MeO–PEG2PLA40/SH 1:1 (100–300 mm), (d) MeO– PEG2PLA40/SH 1:1 (300–500 mm). Scale bars represent 100 mm.
Various biodegradable polymers that differ in molecular weight, glass transition temperature, composition and hydrophilicity were processed into scaffolds with a microstructure similar to that of spongy bone [25] by controlled adaptation of the processing parameters (Table 1, Figs. 2 and 7). Porogen materials with melting points a few degrees below the glass transition temperature of the polymers were chosen. For polymer processing, the concentration of the polymer solution was adapted to type and molecular weight of the polymer and to amount and average particle size of the solid lipid porogen microspheres. The dispersion prepared from the porogen particles and the polymer solution optimally contains a sufficient amount of solvent to prolong the phase separation period until polymer precipitation and final solidification in order to create highly interconnected frameworks (Fig. 2). At the same time, the dispersion has to be viscous enough to maintain the shape of the mold, which limits the amount of solvent (Fig. 1).
Table 2 Construct characteristics of chondrocytes cultured on PLGA scaffolds for 21 days with and without supplementation of insulin Control
Wet weight (ww) (mg) Collagen (mg) Collagen/ww (%) GAG (mg) GAG/ww (%)
88.3718.9 1.970.3 2.270.0 2.170.5 2.470.2
Data represents the means7standard deviation of four independent measurements.
Insulin (mg/mL) 0.05
2.5
143.578.9 3.870.4 2.770.0 5.370.8 3.770.4
211.4719.8 5.670.5 2.770.0 9.071.0 4.370.1
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3505
Fig. 8. Histological cross sections of 3-week constructs (bovine chondrocytes on PLGA (RG756) scaffolds) cultured with (0.05 and 2.5 mg/mL) and without the supplementation of insulin. GAG in extracellular matrix was stained red with safranin–O. Top: gross view (scale bar: 4 mm). Bottom: magnified view (scale bar: 500 mm).
Oscillatory rheological measurements were conducted to characterize the optimum viscosity range (Fig. 3). A complex viscosity of 105 Pas at 1 Hz was found to be optimal for the processing of PLGA (RG756) (Fig. 3a). Different viscosity values were found to be best when polymers of different composition and/or molecular weight were processed. A higher dispersion viscosity than that described for PLGA (RG756) was required for the processing of a low molecular weight polymer. For the processing of high molecular weight polymers, however, a lower dispersion viscosity is suggested (data not shown). A fine tuning of the scaffold properties purely based on the viscosity of the scaffolding dispersion, however, appears to be inapplicable. More detailed data may be provided by frequency sweeps and the analysis of the frequencydependent loss factor. Non-linear changes in the loss factor indicate a non-homogeneous dispersion system, while a strictly linear correlation seems to be a sign for dispersions with an excessively high polymer content resulting in honeycomb-like scaffold pore structures (Figs. 2 and 3b). With the objective of controlling the pore size of the scaffolds by varying the size of the porogen microparticles, different sieve fractions were processed (Fig. 4). The concentration of the polymer that was mixed with the microparticles was adapted to the increasing porogen size, so as to keep the dispersion viscosity constant and the rheological properties unchanged. This adaptation enabled the fabrication of spongy scaffolds with pore sizes ranging from 100 mm to more than 700 mm (Figs. 5 and 7). Control over this architectural property offers the possibility for systematic pore size testing and optimization focused on the individual application. For special applications, this technique allows one to fill a mold with dispersions
containing different porogen size fractions to fabricate scaffolds with a pore size gradient [26]. In addition to the pore size, the scaffold porosity is another key architectural property, which could be influenced by changing the polymer to porogen ratio. Porosities higher than 90% are achievable employing this protocol (Table 1). By means of a MDSC technique, less than 1% (w/w) of residual porogen material was found entrapped in the polymeric scaffold illustrating the effectiveness of the extraction protocol. The few triglyceride remnants will finally undergo metabolism in a biological system, and thus not compromise the biocompatibility of the cell carrier. Polymer solvent residuals were effectively extracted during the pre-seeding incubation with cell culture medium for 24 h. No signs of scaffold cytotoxicity became obvious during the chondrocyte differentiation study and in an adipose tissue engineering study using rat marrow stromal cells [27]. PLGA (RG756) scaffolds fabricated by solid lipid templating were tested using an established insulindependent cartilage cell culture system [21]. Bovine chondrocytes could be homogeneously seeded on the highly permeable polymeric scaffolds. After 3 weeks of in vitro culture, considerable amounts of ECM were synthesized by the chondrocytes. Histological staining with safranin-O confirmed the presence of GAG, a key component of cartilaginous ECM, in the scaffolds, albeit unevenly distributed (Fig. 8). The supplementation of exogenous insulin dose-dependently increased the wet weight of the constructs and GAG production. Histological cross-sections of the constructs showed a strong and homogeneous staining for GAG throughout the entire scaffold. Thus, the scaffolds proved to be sufficiently permeable to cells, nutrients and wastes and allowed for
ARTICLE IN PRESS 3506
M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
cartilage tissue development throughout the whole scaffold. In fact, the results were comparable to those obtained on fiber meshes, which represent the ‘‘gold-standard’’ for porosity and permeability [21]. No signs of dedifferentiation were observed, but rather the chondrocytes reacted strongly to the exogenous insulin stimuli, resulting in an improved tissue development (Table 2). In summary, the generated microstructure proved to be suitable for cartilage tissue engineering and was recently used for the generation of adipose tissue derived from marrow stromal cells [27]. As shown in this study, solid lipid templating is a technique that can be adapted to a variety of different polymers, offers easy control of architectural properties, such as pore size and porosity, and generates spongy frameworks characterized by high pore interconnectivity (Fig. 6). In addition, this technique avoids aqueous media so that experimental polymers bearing hydrolysable functional groups or water-soluble polymers can be processed [20]. Further theoretical advantages of an anhydrous templating technique include: first, bioactive proteins, such as growth factors, can be directly encapsulated as solids during scaffold fabrication and will be released when the scaffold is brought into contact with body fluids or cell culture medium [19]. Second, there is no requirement for sophisticated equipment, unlike the textile technology, solid free-form fabrication or three-dimensional printing. Third, the porogen extraction is completed after 30 min, while conventional porogen-leaching techniques require extraction times from hours up to days. Fourth, an expansion of the process and even automation for largescale production is easily conceivable. Solid lipid templating is a quick lab scale scaffold fabrication technique that offers the advantages of traditional porogen-leaching techniques, such as control over pose size and porosity [28], and allows for the generation of highly interconnected spongy microstructures traditionally obtained by phase separation [29], emulsion templating [30], or solid freeform fabrication techniques [12]. 5. Conclusions In conclusion, the solid lipid templating technique is capable of producing tailored cell carriers for a variety of tissue engineering and biomedical applications. Processing conditions can be adapted to a variety of polymers offering different physical, mechanical and degradation properties. Architectural properties, such as pore size and porosity can be easily controlled and pore interconnectivity can be guaranteed. Acknowledgments The authors thank Kristina Ambrosch (Department of Pharmaceutical Technology, University of Leipzig) for conducting the thermogravimetric analysis. Thanks go also to Boehringer Ingelheim for providing us with polymers. Financial support for this project was in part provided by
the Bundesministerium fuer Bildung und Forschung (BMBF), Germany and the National Institutes of Health (R01 DE15164) (A.G.M.).
References [1] Langer R, Vacanti JP. Tissue engineering. Science 1993;260:920–6. [2] Cima LG, Vacanti JP, Vacanti C, Ingber D, Mooney D, Langer R. Tissue engineering by cell transplantation using degradable polymer substrates. J Biomed Eng 1991;113:143–51. [3] Mooney DJ, Mikos AG. Growing new organs. Sci Am 1999; 280:60–5. [4] Tabata Y. The importance of drug delivery systems in tissue engineering. Pharm Sci Technol Today 2000;3:80–9. [5] Liu X, Ma PX. Polymeric scaffolds for bone tissue engineering. Ann Biomed Eng 2004;32:477–86. [6] Freed LE, Vunjak NG, Biron RJ, Eagles DB, Lesnoy DC, Barlow SK, et al. Biodegradable polymer scaffolds for tissue engineering. Biotechnology (NY) 1994;12:689–93. [7] Agrawal CM, Ray RB. Biodegradable polymeric scaffolds for musculoskeletal tissue engineering. J Biomed Mater Res 2001;55: 141–50. [8] Mikos AG, Thorsen AJ, Czerwonka LA, Bao Y, Langer R, Winslow DN, et al. Preparation and characterization of poly(L-lactic acid) foams. Polymer 1994;35:1068–77. [9] Mikos AG, Sarakinos G, Leite SM, Vacanti JP, Langer R. Laminated three-dimensional biodegradable foams for use in tissue engineering. Biomaterials 1993;14:323–30. [10] 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. [11] Nam YS, Park TG. Biodegradable polymeric microcellular foams by modified thermally induced phase separation method. Biomaterials 1999;20:1783–90. [12] Leong KF, Cheah CM, Chua CK. Solid freeform fabrication of three-dimensional scaffolds for engineering replacement tissues and organs. Biomaterials 2003;24:2363–78. [13] Hollister SJ. Porous scaffold design for tissue engineering. Nat Mater 2005;4:518–24. [14] Ma PX, Choi JW. Biodegradable polymer scaffolds with well-defined interconnected spherical pore network. Tissue Eng 2001;7:23–33. [15] Sachlos E, Czernuszka JT. Making tissue engineering scaffolds work, Review on the application of solid freeform fabrication technology to the production of tissue engineering scaffolds. Eur Cell Mater 2003; 5:29–40. [16] Hutmacher DW. Scaffolds in tissue engineering bone and cartilage. Biomaterials 2000;21:2529–43. [17] Hutmacher DW. Scaffold design and fabrication technologies for engineering tissues—state of the art and future perspectives. J Biomater Sci Polym Ed 2001;12:107–24. [18] Tessmar JK, Mikos AG, Goepferich A. Amine-reactive biodegradable diblock copolymers. Biomacromolecules 2002;3:194–200. [19] Shastri VP, Martin I, Langer R. Macroporous polymer foams by hydrocarbon templating. Proc Natl Acad Sci USA 2000;97:1970–5. [20] Hacker M, Tessmar J, Neubauer M, Blaimer A, Blunk T, Gopferich A, et al. Towards biomimetic scaffolds: anhydrous scaffold fabrication from biodegradable amine-reactive diblock copolymers. Biomaterials 2003;24:4459–73. [21] Kellner K, Schulz MB, Gopferich A, Blunk T. Insulin in tissue engineering of cartilage: a potential model system for growth factor application. J Drug Targeting 2001;9:439–48. [22] Lucke A, Tessmar J, Schnell E, Schmeer G, Gopferich A. Biodegradable poly(,-lactic acid)-poly(ethylene glycol)-monomethyl ether diblock copolymers: structures and surface properties relevant to their use as biomaterials. Biomaterials 2000;21:2361–70.
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507 [23] Ma PX, Zhang R. Synthetic nano-scale fibrous extracellular matrix. J Biomed Mater Res 1999;46:60–72. [24] Freed LE, Marquis JC, Nohria A, Emmanual J, Mikos AG, Langer R. Neocartilage formation in vitro and in vivo using cells cultured on synthetic biodegradable polymers. J Biomed Mater Res 1993; 27:11–23. [25] Nazarian A, Muller R. Time-lapsed microstructural imaging of bone failure behavior. J Biomech 2004;37:55–65. [26] Sherwood JK, Riley SL, Palazzolo R, Brown SC, Monkhouse DC, Coates M, et al. A three-dimensional osteochondral composite scaffold for articular cartilage repair. Biomaterials 2002;23: 4739–51.
3507
[27] Neubauer M, Hacker M, Bauer-Kreisel P, Weiser B, Fischbach C, Schulz MB, et al. Adipose tissue engineering based on mesenchymal stem cells and basic fibroblast growth factor in vitro. Tissue Eng 2005;11:1840–51. [28] Mikos AG, Temenoff JS. Formation of highly porous biodegradable scaffolds for tissue engineering. Electron J Biotechnol 2000;3:114–9. [29] Nam YS, Park TG. Porous biodegradable polymeric scaffolds prepared by thermally induced phase separation. J Biomed Mater Res 1999;47:8–17. [30] Barbetta A, Dentini M, De Vecchis MS, Filippini P, Formisano G, Caiazza S. Scaffolds based on biopolymeric foams. Adv Funct Mater 2005;15:118–24.
Biomaterials 28 (2007) 3497–3507 www.elsevier.com/locate/biomaterials
Solid lipid templating of macroporous tissue engineering scaffolds Michael Hackera,d, Michael Ringhoferb, Bernhard Appela, Markus Neubauera, Thomas Vogela,c, Simon Youngd, Antonios G. Mikosd, Torsten Blunka, Achim Go¨pfericha, Michaela B. Schulza,c,e, a
Department of Pharmaceutical Technology, University of Regensburg, Universita¨tsstrasse 31, 93040 Regensburg, Germany b Anton Paar GmbH, Anton Paar Strasse 20, 8054 Graz, Austria c Department of Pharmaceutical Technology, University of Graz, Schubertstrasse 6, 8010 Graz, Austria d Department of Bioengineering, Rice University, 6100 Main Street, Houston, TX 77005-1892, USA e Department of Pharmaceutical Technology, University of Leipzig, Schoenauer Strasse 160, 04207 Leipzig, Germany Received 20 December 2006; accepted 11 April 2007 Available online 18 April 2007
Abstract Macroporous biodegradable cell carriers (scaffolds) provide the three-dimensional matrix for tissue formation in vitro. In this study, we present the fabrication of macroporous scaffolds with high inter-pore connectivity from different biodegradable polymers using the recently developed solid lipid templating technique. Starting from a polymer solution and solid lipid microparticles, a dispersion is prepared and subsequently transferred into molds, which are finally submerged in warm hexane to precipitate the polymer and extract the porogens. The study shows how to control pore structure, pore size and porosity of the scaffold using this technique. The process parameters dispersion viscosity, porogen size and type of polymer are considered. Limits of viscosity are examined by macroscopic and microstructure evaluation of the scaffolds prepared at different viscosities. An approach to rationalize these data by oscillation rheometry is shown. Pore size can be controlled by porogen particle size and adaptation of the viscosity of the polymer solution. Porosity can be modified by changing the ratio of porogen to polymer. The suitability of the resulting scaffolds was shown using an established cartilage cell culture model. r 2007 Elsevier Ltd. All rights reserved. Keywords: Scaffold; Lipid; Polylactic acid; Polyethylene oxide; Cartilage tissue engineering
1. Introduction A typical approach to engineer functional tissue for medical and basic science applications combines biocompatible, biodegradable, polymeric cell carriers (scaffolds) as a temporary template in combination with mammalian cells and signaling molecules (e.g. growth, differentiation and adhesion factors) to guide and promote threedimensional tissue development [1–4]. In such strategies, the scaffolds play a pivotal role with regard to cell seeding, proliferation, and new tissue formation [5]. Biodegradable Corresponding author. Department of Pharmaceutical Technology, University of Leipzig, Schoenauer Strasse 160, 04207 Leipzig, Germany. Tel.: +49 341 4229747; fax: +49 341 4123007. E-mail address: [email protected] (M.B. Schulz).
0142-9612/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2007.04.018
polyesters, particularly poly(D,L-lactic acid) (PLA) and its copolymers, are attractive and widely applied scaffolding materials with advantages over ceramics, metals, and natural polymers because they degrade as the new tissue forms and can be synthesized in reproducible quality and free of pathogenic or immunogenic organic residue [6,7]. Depending on the mechanical properties of the tissue to be engineered and the extent of cell–material interaction desired, a suitable polymer has to be selected with regard to its degradation kinetic, composition and molecular weight. Consequently, there is a need to process polymers with different physicochemical characteristics into macroporous scaffolds with reproducible microstructure. Various techniques such as salt leaching [8,9], fibrous fabric processing [6], gas foaming [10], thermally induced phase separation [11] and several solid freeform fabrication
ARTICLE IN PRESS 3498
M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
techniques [12,13] have been developed to generate highly porous polymer scaffolds from biodegradable polymers. Despite the remarkable potential that some of these scaffolds have shown, existing scaffold fabrication techniques lack the ability to control critical scaffold properties and allow for the processing of hydrolytically sensitive or water soluble materials [14,15]. Architectural features, namely pore size and shape, pore wall morphology, porosity, surface area and pore interconnectivity, are probably the most critical parameters as those have been shown to directly impact cell seeding, cell migration, tissue differentiation, transport of oxygen, nutrients and wastes, and new tissue formation in three dimensions [16,17]. As some applications use experimental polymers that may contain hydrolysable functional groups [18] or require the encapsulation of growth factors during scaffold fabrication, an anhydrous fabrication technique is generally preferred. Moreover, the technique should allow for labscale processing, but not be limited to a certain scaffold size. A previously published hydrocarbon templating technique [19], which met most of the described demands, was further improved with regard to biocompatibility and adaptability to low molecular weight block copolymers by our group [20]. The developed solid lipid templating technique employs solid lipid microparticles as porogens and non-halogenated polymer solvents to process biodegradable polymers into macroporous scaffolds, providing an interconnected pore structure. Pore interconnectivity is obtained by precipitating the polymer from its solution at the phase boundary with a continuous molten lipid phase that is formed during porogen extraction in warm n-hexane, which is a non-solvent for the polymer [20]. In this study, we investigated the parameters critical for the control of pore structure and size. In particular, composition and viscoelastic properties of the paste-like dispersion from polymer solution and porogen particles were varied and effects on pore morphology were determined. Scaffolds with variable pore sizes and open pore structure were fabricated and the potential of solid lipid templating to process different polymers and to yield scaffolds with varying porosities was shown. The suitability of the fabricated scaffolds for in vitro tissue engineering was tested using an established cartilage cell culture model [21]. 2. Materials and methods 2.1. Materials PLA, Resomers R207, Mw ¼ 160 kDa (Mw/Mn ¼ 1.4), Tg ¼ 54.8 1C, poly(D,L-lactic-co-glycolic) (PLGA), Resomers RG756, synthesized from 75% lactic acid (LA) and 25% glycolic acid (GA), Mw ¼ 90 kDa (Mw/ Mn ¼ 1.5), Tg ¼ 50.7 1C, were kindly provided by Boehringer Ingelheim (Ingelheim, Germany). Monomethyl ether-poly(ethylene glycol)-blockPLA (MeO-PEG2PLA40; MeO-240), Mw ¼ 69 kDa (Mw/Mn ¼ 1.9), Tg ¼ 43.0 1C, which consists of a 2 kDa MeO-PEG block covalently bound to a 40 kDa PLA block, was synthesized and characterized as previously described [22]. The solid lipids that were used as porogen materials, namely Softisans 154 and Witepsols H42, were kindly provided by SASOL Germany
GmbH (Witten, Germany). Methyl ethyl ketone (MEK), tetrahydrofuran (THF), ethyl acetate and n-hexane were purchased in analytical grade from Merck (Darmstadt, Germany).
2.2. Methods 2.2.1. Preparation of solid lipid microparticles Microparticles were prepared using a melt dispersion technique. Ten grams of the solid lipid mixture prepared from Softisans 154 und Witepsols H42 (SH), either SH 1:1 or SH 2:1, was weighed in a poly(propylene) tube. After the addition of 7.5 mL water, the tube was heated to 65 1C. In a normal batch, eight tubes were processed in parallel. The mixture was emulsified by turning the tube upside down 10 times and subsequently cast into a larger volume (600 mL) of stirred, cold (15 1C) water. After 10 min of constant stirring, the hardened particles were collected by filtration, rinsed with cold water and dried under laminar air flow for 3 days. Finally, the microparticles were separated into different size ranges by sifting through sieves (mesh sizes: 100, 300, 500, 710 mm) (Retsch GmbH & Co. KG, Haan, Germany). 2.2.2. Particle size analysis The size distribution of the prepared lipid microparticles was investigated using laser diffraction (Mastersizer 2000 Hydro 2000S, Malvern Instruments, Herrenberg, Germany) as follows: approximately 100 mg of lipid microparticles were directly added to the dispersion unit (Hydro 2000S), which was filled with an ethanol-water-mixture (68.2% (v/v), r ¼ 0.89 g/cm3, refractive index: 1.36). The particles were dispersed by stirring at 3000 rpm for 5 min. The volume-based particle size distribution was calculated using the Fraunhofer approximation (Malvern Software V5.1). Further values provided by the software included the average particle size (d3,2) (surface weighted mean diameter) and specific surface area. The measurements were repeated in triplicate. 2.2.3. Scaffold fabrication procedure To fabricate macroporous scaffolds from biodegradable polymers, we recently developed an anhydrous technique that uses solid lipid microparticles as porogens [20]. Briefly, scaffold fabrication started with a polymer solution. Appropriate solvents and polymer concentrations are described in the Results section. Under ice cooling, the polymer solution was mixed with the solid lipid microparticles. A polymer to porogen ratio of 1:4 (w/w) was used if not otherwise stated. The homogeneous dispersion was transferred into Teflons molds (1.9 cm 1.9 cm 1.2 cm) with a cylindrical cavity of 0.8 cm in diameter. After a pre-extraction treatment step in n-hexane at 0 1C (MeO-PEG2PLA40: 90 min; other polymers: 15 min), the molds were submerged in warm n-hexane to induce solvent extraction followed by the precipitation of the polymer and extraction of the lipid porogen. This procedure was carried out in two separate n-hexane baths of different temperatures T1 and T2 for t1 and t2 with t1+t2 ¼ 30 min. The resulting porous, cylindrical polymer constructs were allowed to cool in cold (0 1C) n-hexane and were removed from the molds. After drying under vacuum for 48 h, the constructs were cut into 2 mm slices, which were then described as scaffolds. Table 1 gives a summary of the applied processing conditions. Solvent residues were removed by a 48 h period under vacuum and subsequent 24 h incubation under seeding conditions in primary medium in a spinner flask at 37 1C. Thermogravimetric analysis (TGA 1000, Polymer Laboratories Thermal Sciences) was used to quantify solvent residues. A heating rate of 5 K/min was applied to analyze scaffolds after vacuum drying and after additional incubation. 2.2.4. Microscopic assessment of scaffold macro- and microstructure The macrostructure of the porous polymer cylinders and scaffolds was examined using a zoom stereo microscope (Wild M7A, Wild Heerbrugg Ltd., Heerbrugg, Switzerland). The scaffold microstructure was visualized by scanning electron microscopy (SEM). For this procedure, samples were mounted on aluminum stubs with conductive carbon tape and coated with
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3499
Table 1 Porogen characteristics and processing parameters applied for the fabrication of scaffolds from different polymers Polymer
Porogen material
Size range (mm)
d3,2 (mm)
Polymer content (w/w) (%)
Conc. (polymer solution) (mg/mL)
Extraction conditions
Step 1 (T1; t1)
Softisan 154: Witepsol H42 Pore size/rheology PLGA (RG756) PLGA (RG756) PLGA (RG756) MeO-240
2:1 2:1 2:1 1:1
100–300 300–500 500–710 100–300
200 360 590 180
20 20 20 25
330 350 365 435
52 1C; 52 1C; 52 1C; 45 1C;
Porosity PLGA (RG756) PLGA (RG756) PLGA (RG756)
2:1 2:1 2:1
100–300 100–300 100–300
200 210 210
20 15 10
330 265 185
52 1C; 100 52 1C; 100 52 1C; 100
Polymer type MeO-240 MeO-240 PLA (R207) PLA (R207)
1:1 1:1 2:1 2:1
100–300 300–500 100–300 300–500
200 400 180 255
20 20 20 20
400 410 230 250
45 1C; 45 1C; 52 1C; 52 1C;
100 100 100 100
100 100 100 100
Porosity (%)
Step 2 (T2; t2)
200 200 200 200
85.7#71.2 87.3#71.0 87.1#70.9 —
40 1C; 200 40 1C; 200 40 1C; 200
90.4*70.5 91.5*70.4 94.1*70.4
40 1C; 40 1C; 40 1C; 35 1C;
35 1C; 35 1C; 40 1C; 40 1C;
200 200 200 200
— — — —
Porosity data of different PLGA (RG756) scaffolds was determined gravimetrically (*) or by micro-CT (#) as means7standard deviation for n ¼ 3.
gold-palladium. All micrographs were obtained at 10 kV on a DSM 950 (Zeiss, Oberkochen, Germany). 2.2.5. Computer tomographic assessment of scaffold microstructure and interconnectivity Micro-computed tomography (micro-CT) (Desktop Micro CT1172, SkyScan, Aartselaar, Belgium) was employed to non-destructively determine the microstructure, interconnectivity and porosity of PLGA scaffolds fabricated with porogen microparticles of different size ranges. Each scaffold (cylindrical discs: 8 mm diameter 2 mm height) was mounted vertically on a sample holder using double adhesive tape. The resolution was set to 8 mm. The samples were scanned at a source voltage of 42 kV and a current of 250 mA. Following image acquisition, transverse slices were generated with SkyScan reconstruction software (NRecon) and resliced twice to yield coronal sections of the scaffolds. Image analysis was performed using the standard SkyScan software (CTAn). Images were loaded and resized by half resulting in pixel and voxel sizes of (16 mm)2 and (16 mm)3, respectively. A volume of interest (VOI) was defined by superimposing a cylinder just enclosing the scaffold volume. For each resulting dataset the VOI and the object volume (OV) (polymer volume of scaffold) were quantified (VOIpre and OVpre). In order to determine pore interconnectivity, a three-dimensional ‘shrink wrap’ was performed using the SkyScan software plugin. In this operation, the VOI is reduced to the boundary of the binarized solid object from outside the object to the inside eliminating all void space that is accessible through interconnects of a predefined size (even multiples of the voxel size). This analysis was performed for 2, 4, 6, 8, 10, 12, 16, 20, 30, and 40 times the voxel size [(16 mm)3] and the shrunk VOI (VOIpost) was determined. OVpost equaled OVpre. The relative inaccessible volume was determined as (VOIpost OVpost)/(VOIpreOVpre). Scaffold porosity was calculated as (VOIpreOVpre)/ VOIpre. 2.2.6. Oscillatory rheological measurements The dispersions of porogen particles in polymer solution prepared during scaffold fabrication were characterized using a Physica MCR 300 rheometer (Anton Paar GmbH, Graz, Austria) with a 2.5 cm sandblasted steel plate geometry according to the following protocol: under ice cooling, the polymer solution was mixed with the solid lipid microparticles
for 3 min. The mixture was transferred to the rheometer using a polypropylene syringe and equilibrated at 5 1C. The measurement gap was set to 2 mm. After 60 s, the first frequency sweep (100–0.1 Hz) was measured at a strain of 0.01% and a temperature of 5 1C. A second frequency sweep (100–0.1 Hz at 0.01% strain) at 5 1C followed another equilibration period of 3 min. The parameters recorded during the second frequency sweep were analyzed to determine storage (G0 ) and loss modulus (G00 ), loss factor (tan d ¼ G00 /G0 ), and complex viscosity (|Z*|) of the dispersion. To compare different formulations, the values of the complex viscosity at 1 Hz were interpolated. 2.2.7. Determination of residual porogen material The detection of triglyceride residues inside the scaffolds was realized through modulated differential scanning calorimetry (MDSC). The samples were precisely weighed in non-hermetic AutoDSC aluminum sample pans (TA Instruments, Alzenau, Germany). The sample pans were sealed using the sample encapsulating press (TA Instruments, Alzenau, Germany) and analyzed on a DSC 2920 equipped with a refrigerated cooling system and an autosampler (TA Instruments, Alzenau, Germany). An empty, sealed pan served as reference. All measurements were carried out between 20 and 120 1C. Typically, samples were equilibrated at 20 1C for 15 min and heated to 120 1C at a heating rate of 5 1C/min. After an isothermal phase of 15 min, samples were cooled to 20 1C at 5 1C/min. Finally, after another 15 min isothermal phase, samples were again heated to 120 1C at 5 1C/min. A sinusoidal temperature modulation with a period of 60 s and temperature amplitude of 0.8 1C was applied to both heating cycles. The resulting thermograms (total heat flow of the second heating cycle) were analyzed for melting enthalpy (peak area) of the peak attributed to the melting lipid residuals located near the glass transition step of the polymer with the Universal Analysis for NTs software provided with the DSC system. 2.2.8. Determination of scaffold porosity The porosity of the scaffolds was determined by measuring the dimensions and the mass of the porous polymer cylinders as obtained from fabrication (cylinder ends were cut away) [23]. The skeletal density of the cylinder (r) was calculated from the mass (m), the diameter (d) and the
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3500 height (h) of the cylinder: 4m r¼ 2 . pd h
ltop (1)
l = lbottom bottom - ltop
The porosity (e) was calculated from the density of the construct (r) and density (rp) of the polymer (rp(RG756) ¼ 1.26 g/cm3): r . rp
lbottom
(2)
2.2.9. Chondrocyte cell culture Primary chondrocytes were isolated from full-thickness bovine articular cartilage by digestion with type II collagenase (CellSystem, St. Katharinen, Germany) as previously described [24] and resuspended in culture medium (DMEM) containing 4.5 g/L glucose (Life Technologies GmbH, Karlsruhe, Germany), 584 mg/L glutamine, 10% FBS (Life Technologies GmbH, Karlsruhe, Germany), 50 U/mL penicillin, 50 mg/mL streptomycin (Sigma-Aldrich Steinheim, Germany), 10 mM HEPES (Sigma-Aldrich Steinheim, Germany), 0.1 mM non-essential amino acids (Life Technologies GmbH, Karlsruhe, Germany), 0.4 mM proline (Sigma-Aldrich), and 50 mg/mL ascorbic acid (Sigma-Aldrich). This medium was also used for cell seeding and cultivation. The cells were seeded in spinner flasks, as described elsewhere [21]. In brief, after preincubation of the scaffolds for 24 h in medium, a cell suspension containing 5 106 chondrocytes per scaffold was stirred at 80 rpm in a humidified (37 1C/5% CO2) incubator for 2 days to allow for cell attachment to the polymers. Cell-polymer constructs were transferred into 6-well plates (one construct and 6 mL culture medium per well) and cultured for 3 weeks on an orbital shaker at 50 rpm. Medium supplemented with different amounts of insulin (0 mg/mL, 0.05 mg/mL, 2.5 mg/mL) was completely exchanged three times per week. At the end of the culture period, each cell-polymer construct was weighed ( ¼ wet weight) and cut in half. One half was prepared as histological sample by fixing in 2% glutaraldehyde in PBS for 30 min and then in 10% formalin. The samples were embedded in paraffin and crosssectioned (5 mm thick). Deparaffinized sections were stained with hematoxylin and eosin (H&E) (C.I.: 45380, Sigma-Aldrich) for cells and with safranin-O for glycosaminoglycans (GAG). The other half of the construct was used for biochemical analysis of collagen content and GAG content [21].
3. Results 3.1. Control of pore structure during scaffold fabrication 3.1.1. Influence of polymer concentration on the scaffold structure In solid lipid templating, solid lipid microparticles are dispersed into an organic solution of the polymer. Suitable polymer solvents are non-solvents for the lipid microparticles. The composition of the paste-like dispersion from porogen particles and polymer solution has been found to be decisive with regard to shape, pore structure and porosity of the resulting scaffolds. The dispersion’s viscosity is mainly controlled by the ratio of polymer to porogen particles, the porogen particle size and the amount of polymer solvent used. Since polymer to porogen ratio and porogen size are preset in order to control scaffold porosity and pore size, respectively, the concentration of the polymer solution is the key parameter to control dispersion properties.
3
Length difference [mm]
¼1
2
1
0 380
330
300
Concentration of polymer solution [mg/ml] Fig. 1. (a) Macroscopic view of a deformed scaffold as a result of low dispersion viscosity. To characterize the deformation the length difference (Dl) is calculated. (b) Length difference measured on PLGA (RG756) scaffolds fabricated from differently concentrated polymer solutions. Columns represent average7standard deviation (n ¼ 4). Statistical significance (po0.01) is denoted by a *.
Fig. 1 shows a gross approach to find limits for dispersion viscosity. Low polymer concentrations (300 mg/mL) resulted in a paste with low viscosity that partly ran out of the mold and caused deformation of the polymer cylinder (Fig. 1a). This deformation can be described by the difference in length between the top and the bottom of the resulting scaffold cylinder (Fig. 1b). Minor deformations of the forming polymer cylinder were observed when the polymer concentration was increased to 330 mg/mL. The remaining deformation was not significantly improved with a high concentration of 380 mg/mL. In order to correlate this gross approach to the scaffold microstructure, SEM images of the corresponding scaffolds were taken. Polymer processing at the highest concentration (380 mg/mL) resulted in a microstructure characterized by pores in the shape of the porogen microparticles showing inter-pore connections of less than 100 mm diameter (Fig. 2). Lower polymer concentrations (330 and 300 mg/mL), in contrast, resulted in a highly permeable network of well-condensed polymer sponges with large pore interconnections. The structure achieved with these lower concentrations closely resembles natural spongy bone [25]. Based on the data shown in Figs. 1
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3501
Fig. 2. Scanning electron micrographs of PLGA (RG756) scaffolds fabricated using different polymer concentrations. (a) 380 mg/mL, (b) 330 mg/mL, (c) 300 mg/mL. Porogen particles were 100–300 mm in size. Scale bars represent 100 mm.
and 2, a polymer concentration of 330 mg/mL is suitable to process PLGA (RG756) with microparticles of 100–300 mm into macroporous scaffolds with a spongy microstructure. In order to rationalize the empirical approach by rheological data, we characterized the scaffolding dispersion using oscillation rheology. 3.1.2. Rheological characterization of the scaffolding dispersions On a controlled stress rheometer equipped with a plate–plate geometry, frequency sweep measurements were performed. The complex viscosity (|Z*|) and the complex moduli, storage (G0 ) and loss modulus (G00 ), were recorded over a frequency range of 0.1–100 Hz. Fig. 3a shows the frequency sweeps of the three dispersions of 300, 330 and 380 mg/mL PLGA. All systems displayed storage and loss moduli that were frequency dependent. Typically, increasing the oscillatory frequency increased both the storage and loss modulus at frequencies higher than 1 Hz, whereas the complex viscosity decreased. In all three formulations, the storage modulus (G0 ) exceeded the loss modulus (G00 ) resulting in a loss factor (tan d ¼ G00 /G0 ) smaller 1 (Fig. 3b). Accordingly the systems can be addressed as gels. We observed that the loss factor first decreased between 0.1 and 1 Hz, indicating an increase in elastic properties. At frequencies higher then 1 Hz viscous properties became stronger in dispersions with low (300 mg/mL) to intermediate (330 mg/mL) concentrations, whereas the highly concentrated dispersions (380 mg/mL) showed only small changes over the frequency range and the lowest loss factor indicating a more stable system. These differences are also reflected by the larger standard deviations found for dispersion with low polymer concentrations. Dispersions that remained stable and predominantly elastic over the frequency range such as the 380 mg/mL PLGA system resulted in non-desired pore structures as shown in Fig. 2a. Systems showing more inhomogeneity in their loss factor seem to allow for the necessary coalescence of the melting microparticles while systems embedding their microparticles perfectly in a polymer gel lead to pores reflecting the outer shape of the porogen particles.
The dispersion’s viscosity also depends on the size distribution of the porogen particles and the corresponding surface area of the microparticles that may vary depending on the microparticle batch. More definite results may therefore be generated by use of monodisperse porogen particles. Furthermore, a change in polymer type and molecular weight will change the required frequency-dependent gel state of the dispersion. The intended rationalization of the dispersions composition by rheology was, therefore, only possible to a limited degree. 3.2. Control of pore size 3.2.1. Porogen microparticle preparation The solid lipid templating technique described here uses triglyceride microparticles prepared from a mixture of two solid lipids, Softisans 154 und Witepsols H42, as pore forming templates [20]. Despite the principle of melting the porogen particles during extraction and simultaneous polymer precipitation to generate interconnected pore networks, the pore forming particles are intended as a means to control the pore size distribution in the resulting scaffolds. Solid lipid microparticles ranged in size from 50 to 1000 mm and were fractionated by sieving. The following size ranges were collected for scaffold fabrication: 100–300, 300–500 and 500–710 mm. Fig. 4 shows the particle size distribution within the three porogen classes, as determined by laser diffraction. The representative graphs depict narrow distributions characterized by D(0.5) values (the diameter below which 50% of the volume of particles are found) of about 210 mm (fraction: 100–300 mm), 365 mm (fraction: 300–500 mm) and 580 mm (fraction: 500–710 mm) that corresponded to the d3,2 (surface weighted mean) values (Table 1). 3.2.2. Processing of dispersions with different porogen sizes For polymer processing using larger porogen particles, the concentration of the polymer solution was adapted to the changed conditions. Larger particles result in less surface area per weight unit that need to be coated and bridged by the viscous polymer solution. Hence, we found
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3502
4) 5) 7) 6) 8)
105
106
9)
104 105 103
Differential size distribution
1) 2) 3)
Complex moduli: G’ (open symbols) and G’’ (filled symbols) [Pas]
Complex viscosity (|η*|) [Pas]
106
107
50
10 102 0.1
1
104 100
10 Frequency [Hz]
1000
0 10000
Particle size [µm] Fig. 4. Differential and cumulative porogen microparticle size distribution of different sieve fractions. K: 100–300 mm, J: 300–500 mm, : 500–710 mm. Particles were prepared from the 2:1 mixture of Softisans 154 and Witepsols H42.
0.8
Loss factor (tanδ)
100
Cumulative size distribution [%]
100 107
highly interconnected microstructure were fabricated from all three different porogen size ranges.
0.6
0.4
0.2 0.1
1
10
100
Frequency [Hz]
Fig. 3. Rheological measurements of dispersions prepared from differently concentrated RG756 solutions and SH 2:1 microparticles (100–300 mm). (a) Frequency dependence of storage (G0 ) and loss (G00 ) modulus, and the complex viscosity (|Z*|) of the dispersions at 5 1C. Rheograms originate from dispersions with polymer concentrations of 300 mg/mL (squares), 330 mg/mL (triangles), and 380 mg/mL (circles): (1) |Z*| for 380 mg/mL, (2) |Z*| for 330 mg/mL, (3) |Z*| for 300 mg/mL (1–3: left y-axis), (4) G0 for 380 mg/mL, (5) G0 for 330 mg/mL, (6) G0 for 300 mg/mL, (7) G00 for 380 mg/mL, (8) G00 for 330 mg/mL, (9) G00 for 300 mg/mL (4–9: right y-axis). Data are represented as means and standard deviations of separately prepared dispersions (300 mg/mL: n ¼ 3; 330 and 380 mg/mL: n ¼ 4). (b) Frequency dependence of the loss factors of the different dispersions at 5 1C. Rheograms originate from dispersions with polymer concentrations of (’) 380 mg/mL, (m) 330 mg/mL, and (K) 300 mg/mL. Data are represented as means and standard deviations of separately prepared dispersions (300 mg/mL: n ¼ 3; 330 and 380 mg/mL: n ¼ 4).
empirically that the solvent volume needed to be decreased with increasing porogen size, resulting in an increase in polymer concentration. The sensitivity of the rheological procedure, however, was not high enough to detect these differences most likely as a consequence of the large gap size of 2 mm between the measuring plates that had to be used to hold the dispersion containing the increasingly macroscopic particles. The corresponding microstructures of the fabricated scaffolds are shown in Fig. 5. Scaffolds with a spongy,
3.2.3. Determination of scaffold interconnectivity PLGA (RG 756) scaffolds fabricated with porogen microparticles of different size ranges were scanned by micro-CT. The reconstructed models were analyzed for pore interconnectivity by determining the pore volume inaccessible for a virtual object of varying sizes. Fig. 6 displays the inaccessible volumes as a function of the size of the permeating object for scaffolds fabricated with porogen microparticles of different size ranges. The data illustrates large interconnect sizes that increase with increasing porogen size. For a cubic object of an edge length of 96 mm (6 fold voxel size), for example, only 9.274.7% of the pore volume of a scaffold prepared with the 100–300 mm porogen particles was inaccessible. Correspondingly, more than 90% of the pore volume was accessible for the object. The same object could access more than 94% of the pore volume in scaffolds fabricated with 300–500 mm porogens and more than 97% when porogen particles of the 500–710 mm size range were used. 3.2.4. Determination of lipid and solvent residuals in the scaffolds The solid lipid microparticles were extracted successfully with the described extraction conditions (Table 1). Independent of the porogens’ size distribution, a residual content of less than 1% (lipid per scaffold (w/w)) porogen material was found per scaffold by DSC analysis. The residual lipid contents in PLGA (RG756) scaffolds were 0.3470.08% for the 100–300 mm porogen fraction, 0.2870.06% for the 300–500 mm fraction and 0.6270.20% for the 500–710 mm fraction. TGA of the scaffolds revealed that about 3–4% of solvent residuals remained in the scaffolds after fabrication and subsequent vacuum drying for 48 h. After the pre-seeding
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3503
Fig. 5. Scanning electron micrographs of PLGA (RG756) scaffolds fabricated using porogen microparticles (SH 2:1) of different size ranges. (a) 100–300 mm, (b) 300–500 mm, (c) 500–710 mm. Scale bars represent 500 mm (large pictures) and 200 mm (close-ups).
such scaffolds are listed in Table 1. The increase of the porogen concentration necessitated a relative increase in the amount of polymer solvent to homogeneously disperse the porogen microparticles, resulting in decreasing concentrations of the polymer solutions.
Fraction of inaccessible volume
0.75
0.5
3.4. Processing of different polymers
0.25 PLGA 100-300 PLGA 300-500 PLGA 500-710
0 0
5
10
15 20 25 30 35 Object size [fold voxel size]
40
45
Fig. 6. Fraction of inaccessible volume of PLGA (RG756) scaffolds relative to the size of a penetrating virtual object as determined from micro-CT data. Scaffolds fabricated with porogen microparticles (SH 2:1) of different size ranges were analyzed. &100–300 mm, B 300–500 mm, n 500–710 mm.
incubation with cell culture media in a spinner flask at 37 1C for 24 h, however, no residual solvent was detected in the freeze-dried samples by TGA. 3.3. Control of scaffold porosity The use of different size porogen microparticles had no effect on the bulk porosity of PLGA (RG756) scaffolds (Table 1). Micro-CT data revealed a porosity of 85–87% for PLGA scaffolds fabricated with a polymer to porogen ratio of 1:4 (w/w) independent of the porogen size distribution. A significant increase in porosity was achieved by reducing the polymer content during processing (Table 1). PLGA (RG756) scaffolds fabricated from 20% (w/w) polymer and 80% (w/w) solid lipid microparticles displayed a porosity of 90.470.5%. An increase in porogen content to 85% resulted in a porosity of 91.570.4%. A porosity of 94.170.4% was found for scaffolds fabricated with 90% (w/w) pore forming microparticles (Table 1). The suitable processing conditions for
Processing parameters were determined and optimized for a variety of biodegradable polymers to demonstrate the versatility and adaptability of the solid lipid templating technique. As a representative for high molecular weight biodegradables, PLA (R207, Mw ¼ 160 kDa) was processed. A sufficient dispersion viscosity was achieved with comparably low polymer concentrations (Table 1). Lipid microparticles with a melting point of 49 1C (SH 2:1) were used as porogen material and extracted under the corresponding conditions (step 1: 52 1C for 100 ; step 2: 40 1C for 200 ). Control experiments (data not shown) revealed that the microstructure was not improved by using solid lipid microparticles with a higher melting point in combination with higher extraction temperatures. Control over the pore size was again achieved by varying the size of the porogen particles and adapting the amount of polymer solvent (Table 1 and Fig. 7a, b). Processing of a low molecular weight polymer with increased hydrophilicity, MeO-PEG2PLA40, was realized by using a 20% (w/w) solution of polymer in a MEK–THFmixture (59:41 (v/v)) and 80% (w/w) lipid microparticles with a low melting point (SH1:1, Tm ¼ 44 1C) adapted to the relatively low glass transition temperature of the polymer. Correspondingly, the extraction temperatures were adjusted (T1 ¼ 45 1C, T2 ¼ 35 1C) [22]. As described for PLA and PLGA, the solvent volume was adjusted to the size range of the porogen particles (Table 1). This way, porous scaffolds with large interconnects were fabricated with the pore sizes being dependent on the porogen size range (Fig. 7c, d) 3.5. Scaffold testing: engineering of cartilaginous tissue An established, insulin-dependent three-dimensional culture system to engineer cartilage from bovine chondrocytes
ARTICLE IN PRESS 3504
M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
[21] was transferred from fiber meshes to solid lipid templated PLGA (RG756) scaffolds to test the applicability of the scaffolds. The scaffolds that were used in this experiment were fabricated with a polymer content of 17.5% (w/w) using a porogen size range of 100–300 mm. Insulin-dependent tissue development and extracellular matrix (ECM) production was indicated by increasing wet weights along with the dose of exogenously administered insulin (Table 2). After 3 weeks, the wet weight of scaffolds that received no exogenous insulin was 88.3 mg718.9. The dry weight of an empty scaffold after fabrication was 10 mg. Scaffolds that received exogenous insulin had a significantly higher weight. The addition of 0.5 mg/mL insulin resulted in constructs of almost double the wet weight of control
samples. Supplementation with 2.5 mg/mL insulin further increased the wet weight by 50%. In all constructs, the cells deposited collagen and GAG, two major ECM components of cartilaginous tissue. The ECM production was stimulated by exogenous insulin in a dose-dependent manner. The collagen fraction (amount of collagen per wet weight) was increased by insulin supplementation. No difference in the amount of collagen per wet weight was found for the two insulin concentrations (Table 2). The GAG fraction also increased with the dose of supplemented insulin. The constructs of the insulin (2.5 mg/mL) group showed an approximately 2-fold higher GAG fraction than constructs receiving no insulin (Table 2). The different amounts of GAG were also reflected in histological sections of the cell-polymer constructs, stained red with safranin-O for GAG. Staining of the ECM of control constructs receiving no insulin revealed an uneven GAG distribution (Fig. 8a). In constructs that received insulin, even in those treated with the low insulin concentration (0.5 mg/mL), the ECM stained evenly positive for GAG (Fig. 8b). The ECM in constructs that received 2.5 mg/mL insulin showed strong positive staining for GAG up to the edge of the tissue (Fig. 8c). 4. Discussion
Fig. 7. Scanning electron micrographs of scaffolds fabricated from two different polymers using porogen particles of different composition and sizes. (a) PLA (R207)/SH 2:1 (100–300 mm), (b) PLA (R207)/SH 2:1 (300–500 mm), (c) MeO–PEG2PLA40/SH 1:1 (100–300 mm), (d) MeO– PEG2PLA40/SH 1:1 (300–500 mm). Scale bars represent 100 mm.
Various biodegradable polymers that differ in molecular weight, glass transition temperature, composition and hydrophilicity were processed into scaffolds with a microstructure similar to that of spongy bone [25] by controlled adaptation of the processing parameters (Table 1, Figs. 2 and 7). Porogen materials with melting points a few degrees below the glass transition temperature of the polymers were chosen. For polymer processing, the concentration of the polymer solution was adapted to type and molecular weight of the polymer and to amount and average particle size of the solid lipid porogen microspheres. The dispersion prepared from the porogen particles and the polymer solution optimally contains a sufficient amount of solvent to prolong the phase separation period until polymer precipitation and final solidification in order to create highly interconnected frameworks (Fig. 2). At the same time, the dispersion has to be viscous enough to maintain the shape of the mold, which limits the amount of solvent (Fig. 1).
Table 2 Construct characteristics of chondrocytes cultured on PLGA scaffolds for 21 days with and without supplementation of insulin Control
Wet weight (ww) (mg) Collagen (mg) Collagen/ww (%) GAG (mg) GAG/ww (%)
88.3718.9 1.970.3 2.270.0 2.170.5 2.470.2
Data represents the means7standard deviation of four independent measurements.
Insulin (mg/mL) 0.05
2.5
143.578.9 3.870.4 2.770.0 5.370.8 3.770.4
211.4719.8 5.670.5 2.770.0 9.071.0 4.370.1
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
3505
Fig. 8. Histological cross sections of 3-week constructs (bovine chondrocytes on PLGA (RG756) scaffolds) cultured with (0.05 and 2.5 mg/mL) and without the supplementation of insulin. GAG in extracellular matrix was stained red with safranin–O. Top: gross view (scale bar: 4 mm). Bottom: magnified view (scale bar: 500 mm).
Oscillatory rheological measurements were conducted to characterize the optimum viscosity range (Fig. 3). A complex viscosity of 105 Pas at 1 Hz was found to be optimal for the processing of PLGA (RG756) (Fig. 3a). Different viscosity values were found to be best when polymers of different composition and/or molecular weight were processed. A higher dispersion viscosity than that described for PLGA (RG756) was required for the processing of a low molecular weight polymer. For the processing of high molecular weight polymers, however, a lower dispersion viscosity is suggested (data not shown). A fine tuning of the scaffold properties purely based on the viscosity of the scaffolding dispersion, however, appears to be inapplicable. More detailed data may be provided by frequency sweeps and the analysis of the frequencydependent loss factor. Non-linear changes in the loss factor indicate a non-homogeneous dispersion system, while a strictly linear correlation seems to be a sign for dispersions with an excessively high polymer content resulting in honeycomb-like scaffold pore structures (Figs. 2 and 3b). With the objective of controlling the pore size of the scaffolds by varying the size of the porogen microparticles, different sieve fractions were processed (Fig. 4). The concentration of the polymer that was mixed with the microparticles was adapted to the increasing porogen size, so as to keep the dispersion viscosity constant and the rheological properties unchanged. This adaptation enabled the fabrication of spongy scaffolds with pore sizes ranging from 100 mm to more than 700 mm (Figs. 5 and 7). Control over this architectural property offers the possibility for systematic pore size testing and optimization focused on the individual application. For special applications, this technique allows one to fill a mold with dispersions
containing different porogen size fractions to fabricate scaffolds with a pore size gradient [26]. In addition to the pore size, the scaffold porosity is another key architectural property, which could be influenced by changing the polymer to porogen ratio. Porosities higher than 90% are achievable employing this protocol (Table 1). By means of a MDSC technique, less than 1% (w/w) of residual porogen material was found entrapped in the polymeric scaffold illustrating the effectiveness of the extraction protocol. The few triglyceride remnants will finally undergo metabolism in a biological system, and thus not compromise the biocompatibility of the cell carrier. Polymer solvent residuals were effectively extracted during the pre-seeding incubation with cell culture medium for 24 h. No signs of scaffold cytotoxicity became obvious during the chondrocyte differentiation study and in an adipose tissue engineering study using rat marrow stromal cells [27]. PLGA (RG756) scaffolds fabricated by solid lipid templating were tested using an established insulindependent cartilage cell culture system [21]. Bovine chondrocytes could be homogeneously seeded on the highly permeable polymeric scaffolds. After 3 weeks of in vitro culture, considerable amounts of ECM were synthesized by the chondrocytes. Histological staining with safranin-O confirmed the presence of GAG, a key component of cartilaginous ECM, in the scaffolds, albeit unevenly distributed (Fig. 8). The supplementation of exogenous insulin dose-dependently increased the wet weight of the constructs and GAG production. Histological cross-sections of the constructs showed a strong and homogeneous staining for GAG throughout the entire scaffold. Thus, the scaffolds proved to be sufficiently permeable to cells, nutrients and wastes and allowed for
ARTICLE IN PRESS 3506
M. Hacker et al. / Biomaterials 28 (2007) 3497–3507
cartilage tissue development throughout the whole scaffold. In fact, the results were comparable to those obtained on fiber meshes, which represent the ‘‘gold-standard’’ for porosity and permeability [21]. No signs of dedifferentiation were observed, but rather the chondrocytes reacted strongly to the exogenous insulin stimuli, resulting in an improved tissue development (Table 2). In summary, the generated microstructure proved to be suitable for cartilage tissue engineering and was recently used for the generation of adipose tissue derived from marrow stromal cells [27]. As shown in this study, solid lipid templating is a technique that can be adapted to a variety of different polymers, offers easy control of architectural properties, such as pore size and porosity, and generates spongy frameworks characterized by high pore interconnectivity (Fig. 6). In addition, this technique avoids aqueous media so that experimental polymers bearing hydrolysable functional groups or water-soluble polymers can be processed [20]. Further theoretical advantages of an anhydrous templating technique include: first, bioactive proteins, such as growth factors, can be directly encapsulated as solids during scaffold fabrication and will be released when the scaffold is brought into contact with body fluids or cell culture medium [19]. Second, there is no requirement for sophisticated equipment, unlike the textile technology, solid free-form fabrication or three-dimensional printing. Third, the porogen extraction is completed after 30 min, while conventional porogen-leaching techniques require extraction times from hours up to days. Fourth, an expansion of the process and even automation for largescale production is easily conceivable. Solid lipid templating is a quick lab scale scaffold fabrication technique that offers the advantages of traditional porogen-leaching techniques, such as control over pose size and porosity [28], and allows for the generation of highly interconnected spongy microstructures traditionally obtained by phase separation [29], emulsion templating [30], or solid freeform fabrication techniques [12]. 5. Conclusions In conclusion, the solid lipid templating technique is capable of producing tailored cell carriers for a variety of tissue engineering and biomedical applications. Processing conditions can be adapted to a variety of polymers offering different physical, mechanical and degradation properties. Architectural properties, such as pore size and porosity can be easily controlled and pore interconnectivity can be guaranteed. Acknowledgments The authors thank Kristina Ambrosch (Department of Pharmaceutical Technology, University of Leipzig) for conducting the thermogravimetric analysis. Thanks go also to Boehringer Ingelheim for providing us with polymers. Financial support for this project was in part provided by
the Bundesministerium fuer Bildung und Forschung (BMBF), Germany and the National Institutes of Health (R01 DE15164) (A.G.M.).
References [1] Langer R, Vacanti JP. Tissue engineering. Science 1993;260:920–6. [2] Cima LG, Vacanti JP, Vacanti C, Ingber D, Mooney D, Langer R. Tissue engineering by cell transplantation using degradable polymer substrates. J Biomed Eng 1991;113:143–51. [3] Mooney DJ, Mikos AG. Growing new organs. Sci Am 1999; 280:60–5. [4] Tabata Y. The importance of drug delivery systems in tissue engineering. Pharm Sci Technol Today 2000;3:80–9. [5] Liu X, Ma PX. Polymeric scaffolds for bone tissue engineering. Ann Biomed Eng 2004;32:477–86. [6] Freed LE, Vunjak NG, Biron RJ, Eagles DB, Lesnoy DC, Barlow SK, et al. Biodegradable polymer scaffolds for tissue engineering. Biotechnology (NY) 1994;12:689–93. [7] Agrawal CM, Ray RB. Biodegradable polymeric scaffolds for musculoskeletal tissue engineering. J Biomed Mater Res 2001;55: 141–50. [8] Mikos AG, Thorsen AJ, Czerwonka LA, Bao Y, Langer R, Winslow DN, et al. Preparation and characterization of poly(L-lactic acid) foams. Polymer 1994;35:1068–77. [9] Mikos AG, Sarakinos G, Leite SM, Vacanti JP, Langer R. Laminated three-dimensional biodegradable foams for use in tissue engineering. Biomaterials 1993;14:323–30. [10] 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. [11] Nam YS, Park TG. Biodegradable polymeric microcellular foams by modified thermally induced phase separation method. Biomaterials 1999;20:1783–90. [12] Leong KF, Cheah CM, Chua CK. Solid freeform fabrication of three-dimensional scaffolds for engineering replacement tissues and organs. Biomaterials 2003;24:2363–78. [13] Hollister SJ. Porous scaffold design for tissue engineering. Nat Mater 2005;4:518–24. [14] Ma PX, Choi JW. Biodegradable polymer scaffolds with well-defined interconnected spherical pore network. Tissue Eng 2001;7:23–33. [15] Sachlos E, Czernuszka JT. Making tissue engineering scaffolds work, Review on the application of solid freeform fabrication technology to the production of tissue engineering scaffolds. Eur Cell Mater 2003; 5:29–40. [16] Hutmacher DW. Scaffolds in tissue engineering bone and cartilage. Biomaterials 2000;21:2529–43. [17] Hutmacher DW. Scaffold design and fabrication technologies for engineering tissues—state of the art and future perspectives. J Biomater Sci Polym Ed 2001;12:107–24. [18] Tessmar JK, Mikos AG, Goepferich A. Amine-reactive biodegradable diblock copolymers. Biomacromolecules 2002;3:194–200. [19] Shastri VP, Martin I, Langer R. Macroporous polymer foams by hydrocarbon templating. Proc Natl Acad Sci USA 2000;97:1970–5. [20] Hacker M, Tessmar J, Neubauer M, Blaimer A, Blunk T, Gopferich A, et al. Towards biomimetic scaffolds: anhydrous scaffold fabrication from biodegradable amine-reactive diblock copolymers. Biomaterials 2003;24:4459–73. [21] Kellner K, Schulz MB, Gopferich A, Blunk T. Insulin in tissue engineering of cartilage: a potential model system for growth factor application. J Drug Targeting 2001;9:439–48. [22] Lucke A, Tessmar J, Schnell E, Schmeer G, Gopferich A. Biodegradable poly(,-lactic acid)-poly(ethylene glycol)-monomethyl ether diblock copolymers: structures and surface properties relevant to their use as biomaterials. Biomaterials 2000;21:2361–70.
ARTICLE IN PRESS M. Hacker et al. / Biomaterials 28 (2007) 3497–3507 [23] Ma PX, Zhang R. Synthetic nano-scale fibrous extracellular matrix. J Biomed Mater Res 1999;46:60–72. [24] Freed LE, Marquis JC, Nohria A, Emmanual J, Mikos AG, Langer R. Neocartilage formation in vitro and in vivo using cells cultured on synthetic biodegradable polymers. J Biomed Mater Res 1993; 27:11–23. [25] Nazarian A, Muller R. Time-lapsed microstructural imaging of bone failure behavior. J Biomech 2004;37:55–65. [26] Sherwood JK, Riley SL, Palazzolo R, Brown SC, Monkhouse DC, Coates M, et al. A three-dimensional osteochondral composite scaffold for articular cartilage repair. Biomaterials 2002;23: 4739–51.
3507
[27] Neubauer M, Hacker M, Bauer-Kreisel P, Weiser B, Fischbach C, Schulz MB, et al. Adipose tissue engineering based on mesenchymal stem cells and basic fibroblast growth factor in vitro. Tissue Eng 2005;11:1840–51. [28] Mikos AG, Temenoff JS. Formation of highly porous biodegradable scaffolds for tissue engineering. Electron J Biotechnol 2000;3:114–9. [29] Nam YS, Park TG. Porous biodegradable polymeric scaffolds prepared by thermally induced phase separation. J Biomed Mater Res 1999;47:8–17. [30] Barbetta A, Dentini M, De Vecchis MS, Filippini P, Formisano G, Caiazza S. Scaffolds based on biopolymeric foams. Adv Funct Mater 2005;15:118–24.