Porous methacrylate scaffolds: supercritical fluid fabrication and in vitro chondrocyte responses

ARTICLE IN PRESS

Biomaterials 25 (2004) 3559–3568

Porous methacrylate scaffolds: supercritical fluid fabrication and in vitro chondrocyte responses J.J.A. Barrya, H.S. Giddab, C.A. Scotchfordb,c, S.M. Howdlea,* a School of Chemistry, University of Nottingham, University Park, Nottingham NG7 2RD, UK School of Biomedical Sciences, University of Nottingham Medical School, Queen’s Medical Centre, Nottingham NG7 2UH, UK c School of Mechanical, Materials, Manufacturing Engineering and Management, University of Nottingham, Nottingham NG7 2RD, UK b

Received 20 December 2002; accepted 25 September 2003

Abstract This paper describes a method of foaming a polymer system comprising poly(ethyl methacrylate)/tetrahydrofurfuryl methacrylate (PEMA/THFMA), characterisation of the resulting porosity and use of the foam for chondrocyte culture. The potential for this polymer system to support cartilage repair has been investigated previously, both in vivo and in vitro. PEMA/THFMA foamed created using supercritical carbon dioxide were characterised using scanning electron microscopy, mercury intrusion porosimetry and helium pycnometry. Foams were found to be 82% porous with open porosities of 57%. The mean pore diameter was found to be 99+60 mm. Bovine chondrocytes seeded directly onto the surface of the foamed and unfoamed PEMA/THFMA demonstrated lower proliferation on the foamed material, greater retention of the rounded cell morphology and increased glycosaminoglycan synthesis. In conclusion, this study has shown that a porous PEMA/THFMA system can further enhance the ability of the material to support chondrocytes in vitro. However, further modifications in processing are necessary to determine optimum conditions for cartilage tissue formation. r 2003 Elsevier Ltd. All rights reserved. Keywords: Chondrocyte; Scaffold; Polymer; Biomaterial; Methacrylate; Porosity

1. Introduction The repair of damaged cartilage remains one of the major challenges of orthopaedics. Once damaged, cartilage has only limited potential to repair itself, because of the low mitotic activity of chondrocytes and the absence of vascularisation and innervation in cartilage. Current therapies for treatment of articular cartilage have varying degrees of success because they are aimed at diminishing the pain or joint swelling but cannot restore the original hyaline cartilage [1]. The last 10 years have seen the emergence of a novel approach that liberates chondrocytes from their matrix [2,3]. The chondrocytes are cultured and expanded prior to being transplanted into cartilaginous defects. This however requires a method of delivery and stabilising the cells in the defects. Chondrocytes when cultured in a *Corresponding author. Tel.: +44-115-951-3486; fax: +44-115-9513058. E-mail address: [email protected] (S.M. Howdle). 0142-9612/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2003.10.023

two-dimensional (2-D) monolayer culture lose their rounded morphology and assume a more fibroblastic appearance [4,5]. This change in morphology is indicative of the loss of chondrocyte phenotype and is evidenced by changes in the production of extracellular matrix (ECM) proteins. The cells switch from production of cartilage specific type II collagen to the production of type I collagen and switch production of aggrecan to low molecular weight proteoglycan [2,6]. For this purpose, a 3-D scaffold structure is used to aid chondrocyte delivery, attachment and proliferation. Many 3-D scaffolds are being evaluated for a possible role in cartilage repair. These include: (a) biological scaffolds such as types I and II collagen, proteoglycans, hyaluronan and fibrin clots, (b) degradable materials such as poly(glycolic acid) and poly(lactic acid), and (c) non-degradable materials made from cellophane, silicone rubber, carbon fibres, PTFE (Teflon), polyester (Dacron), poly(HEMA) sponges, poly(vinyl alcohol) sponges. Many of these have shown success, with the repair tissue resembling normal hyaline cartilage [7,8].

ARTICLE IN PRESS 3560

J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

This paper describes the fabrication of porous 3-D scaffolds using a non-degradable polymer system consisting of poly(ethyl methacrylate)/tetrahydrofurfuryl methacrylate (PEMA/THFMA). Although non-biodegradable, the advantage of using this material is that it has been demonstrated to support the repair of fullthickness defects in vivo. Rabbits with osteochondral defects had the material implanted just below the level of the subchondral bone and after 8 months the defect had completely filled with dense cartilaginous tissue that was integrated into the surrounding normal cartilage [9–11]. In vitro studies, also using the non-porous PEMA/ THFMA, have demonstrated that the material supported bovine chondrocyte growth and differentiation [8,12,13]. The ability of PEMA/THFMA to support such growth has been attributed to both surface and bulk properties of the polymer system. McFarland et al. [14] have demonstrated that the material appears to present adsorbed fibronectin in a more favourable conformation to support cell adhesion when compared to other polymers. Whilst such adsorption facilitated cell attachment, chondrocyte phenotype and morphology was maintained over extended periods [13]. The surface wettability of the polymer system may play a role here [15]. It has also been suggested that the water uptake properties of the material may contribute to the cartilage repair observed in vivo, by localising soluble factors from the surrounding environment to stimulate repair [9]. Many novel technologies have been applied to the manufacture of 3-D scaffolds. These include: mechanical stretching, fibre extrusion and bonding, template synthesis, phase separation and the use of gases and solvents as porogens (for review see Hutmacher [16]). Carbon dioxide gas above a critical temperature (Tc ¼ 31:1 C) and pressure (Pc ¼ 73:8 bar) functions as a porogen. In this state, the carbon dioxide is said to be supercritical (scCO2), and it is a unique processing medium with the properties of both gas and liquid. It is capable of diffusing into materials easily and extracting scCO2 soluble residues [17,18]. Importantly some polymers when treated with scCO2 swell, creating porous foams examples include poly(methyl methacrylate) (PMMA), polystyrene, polycarbonate and poly(ethelyeneterephthalate). More recently the biomedical polymers poly(d,l-lactide) and poly(d,l-lactide-co-glycolide) have been foamed using scCO2 [19–23]. The foaming of these polymers relies on the following principles: (a) the polymer is saturated with carbon dioxide at high pressure, (b) the polymer/gas mixture is quenched into a supersaturated state by reducing the pressure, and (c) nucleation and growth of gas bubbles dispersed throughout the sample evolves until all thermodynamic forces driving mass transport vanish and we are left with a porous structure [24].

This paper reports the preparation of porous PEMA/ THFMA foams for use as scaffolds, their characterisation and the response of chondrocytes to these novel materials in vitro.

2. Materials and methods 2.1. Preparation of PEMA/THFMA discs PEMA/THFMA polymer discs were made by mixing 5 g of PEMA powder (Bonar Polymers Ltd., Newton Aycliffe, UK) and 3 ml of THFMA (Rohm Chemie, Darmstadt, Germany) monomer liquid. N,N-dimethylp-toluidine (DMPT) was added, (2.5% v/v) to effect polymerisation. This mixture was placed in a custom fabricated PTFE mould to cure, producing discs 10 mm in diameter and 10 mm in thickness. The discs were left overnight to polymerise. Foaming was achieved as follows. Discs were placed into a 10 ml Thar extraction vessel and saturated with CO2 at 1470 psi (100 bar) at 40 C for periods of time ranging from 1 to 48 h. Following this exposure, the scCO2 was vented over 30 s. Control of the venting rate was maintained by a backpressure regulator (Jasco UK, Model BPR-1580–81). The swollen discs were removed from the vessel and freeze fractured using liquid nitrogen to produce discs of 0.5 cm thickness. Pore size, percentage porous material and interconnectivity of pores were determined using scanning electron microscopy (SEM), mercury intrusion porosimetry and helium pycnometry, respectively. For cell culture, the non-porous skin was trimmed producing discs 13 mm in diameter. Unfoamed PEMA/THFMA discs (13 mm diameter) were made by casting in 24-well plates. The unfoamed discs were removed and trimmed. All the discs were sterilised in a solution of PBS with penicillin/streptomycin (pen/strep; 500 units ml1) for 24 h. The pen/strep solution was washed out over 7 days using several changes of sterile water. The sterile discs were mounted in 24-well plates and 13 mm diameter silicone ‘O’ rings (RS supplies) were then placed over each sample to hold them in place. Gaps between the well wall and sample were filled using sterile low-temperature gelling agarose (4% solution, Sigma). Tissue culture plastic (Thermanox) discs were used as control surfaces. 2.2. Characterisation of PEMA/THFMA foams The pore size distribution and the average pore size were calculated using both mercury porosimetry (Micromeritics Accupore IV) and SEM (Philips XL-30 SEM). For porosimetry, pressures ranging from 0 to 25 psi were used. For SEM, the fractured material was

ARTICLE IN PRESS J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

sputter coated with gold and the diameter (across the mouth of the pores) was determined using standard XL30 software. Pores were counted randomly beginning at the centre and travelling set distances from this point. A minimum sample size of 100 was counted. Histograms of the frequency of each diameter were constructed to determine the pore size distribution and the mean pore diameter. The porosity of the material, both open and closed, was calculated by modifying formulae published by Tampieri et al. [25] and Harris et al. [26]. These formulae relate the porosity (P) to the densities (r) of the foamed material:

3561

2.4. Electron microscopy of chondrocyte cultures At given time points the samples were fixed in 1.5% glutaraldehyde in 0.1 m phosphate buffer for 30 min. The samples were washed and post-fixed in 1% osmium tetroxide in phosphate buffer for 30 min. After a further two washes in phosphate buffer, the sample discs were left to air dry in a desiccator overnight, sputter coated with gold and viewed in a Philips 501B SEM. Conventional SEM preparation using dehydration in alcohol could not be used, as it has a detrimental effect on the polymer architecture of the foams. 2.5. DNA content of chondrocyte cultures

Ptotal ¼ ð1  rgeometric =runfoamed Þ  100; Pclosed ¼ ð1  ðrpycnometry =runfoamed Þ  100: Gross measurements and weight were used to calculate the geometric density rgeometric  rgeometric ¼ m=v; where m is the mass and v is the volume of the foamed PEMA/ THFMA disc. The volume of the discs was calculated using the equation v ¼ pr2 h; where r is the radius of the disc and h is the height. The mean disc height and diameter were calculated using standard engineering calipers (Mitutoyo UK Ltd.). Helium pycnometry (Micromeritics AccuPyc 1330 Pycnometer) was used to calculate the density of the foams (rpycnometry ) and the absolute density of the unfoamed PEMA/THFMA (runfoamed ). The method employs Archimedes’ principle of fluid displacement to calculate density.

The DNA content of each sample was measured using the DNA Hoechst assay [28]. This is a fluorimetric method based on the binding of the bis-benzimide Hoechst stain (Hoechst 33258, Sigma) to DNA. After appropriate incubation periods, the samples were washed twice with PBS and 1 ml of double-distilled water was placed in each well. The cells were lysed using a three-cycle freeze thaw process between 80 C and 37 C. The Hoechst stain was prepared at a concentration of 0.5 mg ml1 in TNE buffer. Test samples and standard solutions (100 ml) were pipetted into a 96-well plate. To each of these wells, 100 ml of the Hoechst stain was added. The plate was then gently shaken and read on a cytofluor 2300 fluorimeter (Perseptive Biosystems, UK) at l ¼ 360 nm excitation and l ¼ 460 nm emission. The concentrations of DNA in test samples were calculated using a standard curve generated from the known concentrations of DNA (Calf Thymus DNA, Sigma).

2.3. Chondrocyte cell culture 2.6. Assessment of cell activity on the sample surfaces Full-thickness articular chondrocytes were obtained from bovine cartilage (aged 30 months) using a method adapted from Archer et al. [27]. Cartilage pieces were removed from the proximal side of an open metacarpalphalangeal joint using a scalpel blade. The cartilage pieces were washed in PBS and incubated overnight in Dulbeccos modified eagles medium (DMEM) containing 20% foetal calf serum, 2% HEPES buffer, 1% glutamine, 100 units ml1 penicillin/streptomycin, 0.85 mm ascorbic acid. The following day, the cartilage was finely chopped and incubated with pronase type E (700 units ml1) (BDH Ltd., Poole, UK) in medium for 1 h followed by collagenase type la (300 units ml1) (Sigma-Aldrich Co. Ltd., Poole, UK) in medium for 2 h. The cell suspension was filtered using 70 mm cell strainers (Falcon, Becton Dickinson, Oxford, UK) and centrifuged at 1500 rpm for 5 min to pellet the cells. The cells were washed twice in serum free (sf) medium and then counted using a haemocytometer and seeded onto the foams at 5  105 cells per well.

Metabolic activity was assessed using the Alamar Blue assay (Serotec, Oxford, UK). After the incubation period, the culture medium was removed from wells and the cells on the samples were rinsed twice with Hanks’s Balanced Salt Solution (HBSS). A 1 ml volume of 1:10 Alamar Blue:HBSS was added to each well which was then incubated at 37 C for 90 min. A 100 ml aliquot of the Alamar Blue solution from each well was placed into a 96-well plate, which was read on a Cytofluor 2300 fluorimeter (Perseptive Biosystems, UK) at l ¼ 530 nm excitation, l ¼ 590 nm emission. 2.7. Proteoglycan production Total sulphated glycosaminoglycan (GAG) content of the cell lysate and medium was measured using a 1,9dimethylmethylene blue (DMB) (Serva, Heidelberg, Germany) method [29] with shark chondroitin sulphate C (Sigma, Poole, UK) as a standard. DMB solution

ARTICLE IN PRESS 3562

J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

(250 ml) was added to the wells on a 96-well plate containing 40 ml of sample. The DMB solution was made by mixing 16 mg DMB, 5 ml ethanol, 2 g sodium formate and 2 ml formic acid in 1 l of distilled water. The absorbance was measured at l ¼ 620 nm on an Anthos 2001 plate reader. Cell lysates were prepared as in Section 2.5. Media samples were collected when the culture medium was changed. 2.8. Statistical analysis One-way analysis of variance followed by the Tukey post hoc test was carried out to determine significant differences in metabolic activity and GAG production at each time point with significance determined at the 95% level.

3. Results and discussion 3.1. Characterisation of the discs On removal from the mould, the polymerised PEMA/ THFMA discs had a glassy opaque appearance. After foaming with scCO2 there were two notable changes in the discs. First, the volume of the discs was increased by

ca. 4.5 times. The second change that could be seen was the discs had lost their glassy opaque appearance and were now white. Closer inspection of the foamed discs revealed three distinct regions: an outer skin, a porous region under this and at the centre of the discs was a glassy region (Fig. 1). We speculated that the presence of a glassy core was due to insufficient time for the carbon dioxide to diffuse through the PEMA/THFMA sample. This diffusion front effect has been investigated in many plasticiser/glassy polymer systems [30]. The effect of increasing the exposure of the discs to scCO2, at constant pressure and temperature can be seen in Fig. 1a. As the exposure time increases, there is a greater diffusion of the gas into the material and a decrease in the volume of the glassy core. Following exposure of the discs to scCO2 for 8 h the glassy core was no longer present (Figs. 1a and b). Based on this observation, all foams for tissue engineering scaffolds were exposed to scCO2 for 8 h. SEM analysis of the foam revealed that the outer skin was non-porous (Fig. 2a). Increasing the saturation pressures tended to reduce the thickness of the skin, and lower the average pore diameter a trend also witnessed by Goel and Beckman [31] in PMMA and Arora et al. [32] in polystyrene. Arora et al. attributed this finding to the higher saturation pressure increasing the nucleation

Fig. 1. (a) Differences in swelling sizes that occurred with increasing scCO2 exposure. Left to right, foams that were saturated for 18, 8, 4, 2 and 1 h. (b) Decrease in the volume of the glassy centre with increased diffusion of scCO2. The arrow denotes the point at which the volume of the glassy centre is zero.

Fig. 2. (a) SEM of the foamed PEMA/THFMA disc showing the non-porous outer skin. (b) SEM of one of the foams following removal of the outer skin.

ARTICLE IN PRESS J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

3563

Previously, the supercritical method for producing scaffolds has been attempted by other authors and they have concluded that the gas foaming technique mostly results in a closed cellular structure within the scaffold. Mooney et al. [19] foamed poly(lactide-co-glycolide) and generated sponges with pores of 100 mm diameter and porosities up to 93%. However, they were only able to attain, on average 10–30% of interconnection between pores. These authors have found that a combination technique of gas foaming and salt leaching had to be

density. Reducing the thickness of the skin was not viewed as being important because in our studies the skin is removed prior to cell culture. Analysis of the porous structure that remains upon removal of the skin reveals a distribution up to and including pores of the order of 200 mm (Fig. 2b). A mean pore diameter of 99760 mm was determined by SEM. Mercury intrusion porosimetry returned a mean pore diameter of 3572.6 mm and a similar range of pores (Fig. 3a). One explanation for the differences between the two methods of assessing mean pore diameter is that the porosimetry may be measuring the diameter of the pore opening and this may be smaller than the pore cavity diameter [33]. The presence of a large number of very small pore openings (o20 mm) detected by porosimetry (Fig. 3b) would dramatically underestimate the mean pore diameter. The foamed PEMA/THFMA discs were highly porous, containing a large percentage of open pores (Table 1).

Table 1 Porosity of the PEMA/THFMA discs treated with scCO2

Average Standard error

Open porosity (%)

Closed porosity (%)

Total porosity (%)

57.4 5.5

24.3 5.5

81.7 0.7

Note: The values shown are the means7standard error.

0.050 0.045

Pore Number Fraction

0.040

3.5

0.035 0.030 0.025 0.020 0.015 0.010 0.005

3

0.000 0

5

10

15

20

25

30

35

40

45

50

Pore Radius (µm)

dV/dlogD

2.5

2

1.5

1

0.5

0 0

50

100

150

200

250

Pore Diameter (mm)

Fig. 3. Pore distribution obtained by mercury intrusion porosimetry of the PEMA/THFMA foams. Solid line shows the average for all points (n ¼ 4). Inset shows the fraction of pores of a given radius and demonstrates the high incidence of smaller pores in the sample.

ARTICLE IN PRESS 3564

J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

used in order to increase the interconnectivity between the pores [26]. Our study demonstrates that it is possible to generate interconnectivity between pores without resorting to combination methods such as particulate leaching.

4. Biological assessment of chondrocyte growth and proliferation 4.1. Microscopy Chondrocytes grown on the Thermanox discs initially had a rounded morphology. Within 2 days of culture, the cells had started to spread and become fibroblastic in appearance. By day 3 the proportion of cells that were spreading out had increased and on day 4, multilayering of the cells had occurred. It was observed that in the confluent lower layer the cells were fibroblastic in appearance (data not shown). This is typical of chondrocytes grown in monoculture and has been used to indicate de-differentiation [2,34]. The morphology of cells seeded on the foamed and unfoamed polymer was analysed at days 2, 4 and 8 using SEM. By contrast, chondrocytes grown on the flat unfoamed PEMA/THFMA polymer and on the foamed polymer appeared to behave quite similarly during the first 4 days after cell seeding. The cells on both surfaces have the same round morphology indicating retention of

their chondrocytic phenotype for longer than the Thermanox discs. There were signs however that the chondrocytes were beginning to proliferate on the unfoamed surface losing their rounded morphology and becoming spindle shaped (Figs. 4a and b). Sawtell [35] suggested one of the reasons why the morphology was retained longer on the PEMA/THFMA could be due to differences in the ability of the chondrocytes to adhere to the substratum. The cells adhere strongly on Thermanox and so spreading occurs, but less strongly on PEMA/THFMA, allowing the cells to remain round. By day 8 the cells on the unfoamed polymer had become more fibroblastic in appearance, having lost their rounded shape and almost covering the complete surface (Fig. 4c). On the foamed material the chondrocytes within the pores retained their rounded morphology within the pore (Fig. 4d). Wyre and Downes [13] have previously shown that the non-porous PEMA/THFMA helped chondrocytes to retain their differentiated state for longer than Thermanox controls. Interestingly, on the flat strut connecting the pores, the chondrocytes had lost their rounded morphology in just the same way as those on the unfoamed material (Fig. 4d). These results suggest that creating pores provides a more favourable environment for the retention of the rounded morphology and the prevention of chondrocyte spreading. It has been suggested that the differences in cell responses can be explained by the presence of curvature in the pores,

Fig. 4. Chondrocytes growing on the unfoamed (left) and foamed (right) PEMA/THFMA substrates. a: Chondrocytes on the unfoamed polymer discs at day 4, b: Chondrocytes on the foamed polymer discs, at day 4. c: Chondrocytes on the unfoamed polymer discs at day 8. Spread cells on flat surface (white arrows). d: Chondrocytes on foamed polymer discs at day 8. Spread cells on flat surface above pore (black arrows). Round cells in pore (white arrow).

ARTICLE IN PRESS J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

providing optimum compression and tension of cell mechanoreceptors [36,37]. The observation of cell clustering within pores may provide an alternative explanation for the retention of round cell morphology for extended periods. Watt [38] demonstrated that seeding chondrocytes at higher densities led to the stabilisation of the chondrocyte phenotype for longer. This was attributed to a reduction in available area for spreading and an increase in intercellular communication, either by cell–cell contact or secreted factors. Therefore, the pores may be inducing localised high cell density regimes, thereby aiding in the retention of the chondrocyte phenotype. 4.2. Cell metabolic activity Total chondrocyte metabolic activity was greatest on the Thermanox control surfaces throughout the 14-day study (Fig. 5a). Initially, the activity increased rapidly (day 2–6) and a plateau was reached (day 6–14). Chondrocytes seeded on the unfoamed polymer showed higher metabolic activity than the foams but less than the Thermanox. This activity increased rapidly initially but after day 6 the activity again reached a plateau. By

3565

contrast, the metabolic activity of chondrocytes seeded on the foams was very slow to increase initially, but after day 6, this became more rapid. The activity of chondrocytes on the foams was significantly lower than the unfoamed polymer up until day 14 when there were no significant differences in activity (p > 0:05). When these results were normalised to the amount of DNA in the sample, chondrocytes on Thermanox still display the greatest metabolic activity (Fig. 5b). The graph also shows that the activity/DNA is constant throughout the 14-day period. The normalised activity for the chondrocytes on the unfoamed material declined over time until, at day 14, it reached a level that was not significantly different to the chondrocytes grown on the two foams (P > 0:05). The activity/DNA of the chondrocytes grown on the foams declined from day 2 to 6 but there was no significant difference between day 6 and 14 (P > 0:05). The increase in metabolic activity of the cells with time (Fig. 5) is in agreement with earlier studies, reflecting differences in proliferation [35]. When these results were normalised to DNA content, the reductions in activity per unit of DNA at the longer time points, seen for both polymer surfaces, may be associated with

150

Fluorescence

(Ex. 530nm, Em. 590 nm)

Thermanox U nfoamed PEMA/THFMA Foamed PEMA/THFMA

100

50

0 2

6

(a)

Thermanox U nfoamed PEMA/THFMA Foamed PEMA/THFMA

25

Fluorescence per µg/DNA

14

Days in Culture

20

15

10

5

0 2

(b)

6

14

Days in Culture

Fig. 5. (a) Change in metabolic activities of chondrocytes over a 14-day period using the Alamar Blue assay. (b) Change in the metabolic activity of chondrocytes, normalised to the amount of DNA present. Each point shows the mean7standard error n ¼ 3: Data demonstrate the increased metabolic activity of the de-differentiated chondrocytes.

ARTICLE IN PRESS 3566

J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

differences in activity between fibroblastic and chondrocytic phenotypes. The fibroblasts exhibit the greater metabolic activity of the two cell types, and hence the consistently higher metabolic activity of cells on Thermanox surfaces compared to the PEMA/THFMA surfaces, with and without normalisation to DNA, may be associated with adoption of a fibroblastic phenotype. These are supported by the observed differences in cell morphology between these surfaces. The activity of chondrocytes on the foams remained the lowest until day 14 when there were no significant differences between the unfoamed and the foamed PEMA/THFMA surfaces. 4.3. Glycosaminoglycan production The cell culture medium was changed every 2 days and the amount of sulphated GAG present in the media measured using the DMB assay. Fig. 6a shows the

cumulative GAG release (mg ml1 GAG) to the medium for the different materials. The Thermanox media contained significantly greater amounts of GAG than those of the unfoamed and foamed materials at all time points. Similar amounts of GAG were secreted to the media by the cells grown on the PEMA/THFMA, although by day 12 lower levels of sulphated GAG were detected in the media from the unfoamed discs. When these values were normalised to DNA concentrations a similar trend was observed. No evidence of GAG secretion was detected in the PEMA/THFMA system in the first 2 days, while the GAG secreted by the chondrocytes cultured on the Thermanox was clearly apparent. On days 4–6 sulphated GAG could be measured for all the samples but there was no significant difference between values for foamed and unfoamed materials (p > 0:05). By day 14, there was no significant difference between all samples (Fig. 6a).

Fig. 6. (a) Cumulative GAG release into the medium over a 14-day period. (b) Total GAG produced per unit of DNA and relative contributions from release to the medium (broken bars) and cellular (solid bars) associated GAG normalised per unit of DNA. n ¼ 3 for the graph and each point shows the mean7standard error.

ARTICLE IN PRESS J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

The levels of cellular associated total sulphated GAG normalised to DNA concentrations show a very different trend. On day 2, cellular GAG was highest for those cells grown on the foams and was lowest for those cells grown on the Thermanox. On days 4–6 and 12–14 there was no significant difference between GAG production on any of the materials (Fig. 6b). The increased release of GAG into the medium by cells on Thermanox compared to the unfoamed PEMA/ THFMA surfaces is consistent with earlier studies [39]. Sawtell [39] demonstrated that, whilst cumulative release of GAG to the medium was always lower on PEMA/ THFMA, when GAG production was normalised to DNA concentrations, values were similar, with production on PEMA/THFMA exceeding Thermanox after 25 days. When comparing the foamed and unfoamed polymer, it is clear that GAG release to the medium is greater on the foamed surfaces after 14 days. This may be explained by a loss of chondrocytic phenotype by cells on the unfoamed material relative to cells on the foamed surfaces, as suggested by the greater proportion of cells demonstrating a fibroblastic morphology. Examination of the GAG concentrations associated with cell and media samples when normalised to DNA concentrations indicates that, following initial fluctuations, cells cultured on the foamed surfaces show the least reduction in GAG production between 4 and 14 days culture, again indicative of relatively greater retention of a chondrocytic phenotype by cells on these surfaces compared to the unfoamed samples.

5. Conclusions Carbon dioxide is both a rapid and clean method of processing polymer scaffolds. It is non-toxic and leaves no solvent residues in the polymer matrix. We have demonstrated that scCO2 may be used to produce foamed scaffolds of PEMA/THFMA. This study has demonstrated that the non-porous PEMA/THFMA supported the chondrogenic phenotype of bovine chondrocytes for longer than Thermanox controls, an observation in agreement with previous work [8,12,13]. Our results demonstrate that the change in the nature of the substrate, i.e. from flat discs to foamed PEMA/ THFMA appeared to further enhance this phenotypic retention as indicated by cell morphology and sulphated GAG production. This is important in manipulating the cells to produce the correct extracellular matrix components, which would allow the formation of hyaline cartilage repair tissue. Scaffold fabrication methods should allow precise control of structural features relating to porosity and internal pore architecture (e.g. pore geometry, size, interconnectivity, orientation and branching) so as to maximise the diffusion of nutrients as well as conferring

3567

appropriate mechanical properties [40]. Whilst foams presented in this study demonstrated favourable chondrocyte responses, further modifications in the supercritical processing [22], are required to determine the optimum porosity and pore size for cell migration and cartilage tissue formation.

Acknowledgements The authors would like to acknowledge the assistance of the following persons in undertaking this work, Ms. R. Butler, Dr. A.I. Cooper (University of Liverpool) and Professor K.M. Shakesheff (University of Nottingham). In addition Ms. B. Sim, Mr. P. Fields, Mr. R. Wilson, and Mr. J.M. Whalley for their technical help and advice. Funding for this project was provided by grants from the BBSRC (JJAB) and EPSRC (HSG).

References [1] Fuss M, Ehlers EM, Russlies M, Rohwedel J, Behrens P. Characteristics of human chondrocytes, osteoblasts and fibroblasts seeded onto a type I/III collagen sponge under different culture conditions. A light, scanning and transmission electron microscopy study. Ann Anat 2000;182:303–10. [2] Benya PD, Shaffer JD. Dedifferentiated chondrocytes reexpress the differentiated collagen phenotype when cultured in agarose gels. Cell Biol Int 1982;30:215–24. [3] Brittberg M, Lindahl A, Nilsson A, Ohlsson C, Isaksson O, Peterson L. Treatment of deep cartilage defects in the knee with autologous chondrocyte transplantation. N Engl J Med 1994;331:889–95. [4] Temenoff JS, Mikos AG. Review: tissue engineering for the regeneration of articular cartilage. Biomaterials 2000;21:431–40. [5] Freed LE, Hollander AP, Martin I, Barry JR, Langer R, VunjakNovakovic G. Chondrogenesis in a cell-polymer-bioreactor system. Exp Cell Res 1998;240:58–65. [6] Glowacki J, Trepman E, Folkman J. Cell shape and phenotypic expression in chondrocytes. Proc Soc Exp Biol Med 1983;172: 93–8. [7] Ghivizzani SC, Oligino TJ, Robbins PD, Evans CH. Cartilage injury and repair. Phys Med Rehabilitation Clin North Am 2000;11:289–307. [8] Sawtell RM, Downes S, Kayser MV. An in vitro investigation of the PEMA/THFMA system using chondrocyte culture. J Mater Sci: Mater Med 1995;6:676–9. [9] Downes S, Archer RS, Kayser MV, Patel MP, Braden M. The regeneration of cartilage using a new polymer system. J Mater Sci: Mater Med 1994;5:88–95. [10] Reissis N, Downes S, Kayser M, Lee D, Benntley G. A simple method of cartilage regeneration using a new polymerising system: ultrastructural characteristics of the repair tissue. J Mater Sci: Mater Med 1994;5:793–7. [11] Reissis N, Downes S, Kayser M, Lee D, Benntley G. Characterization of the repair tissue in articular cartilage defects using silver-enhanced colloidal gold immunostaining. J Mater Sci: Mater Med 1994;5:402–6. [12] Sawtell RM, Kayser MV, Downes S. A morphological assessment of bovine chondrocytes cultured on poly(ethyl methacrylate)/ tetrahydrofurfuryl methacrylate. Cells Mater 1995;5:63–71.

ARTICLE IN PRESS 3568

J.J.A. Barry et al. / Biomaterials 25 (2004) 3559–3568

[13] Wyre RM, Downes S. An in vitro investigation of the PEMA/ THFMA polymer system as a biomaterial for cartilage repair. Biomaterials 2000;21:335–43. [14] McFarland CD, Mayer S, Scotchford C, Dalton BA, Steele JG, Downes S. Attachment of cultured human bone cells to novel polymers. J Biomed Mater Res 1999;44:1–11. [15] Hutcheon GA, Downes S, Davies MC. Interactions of chondrocytes with methacrylate polymers. J Mater Sci: Mater Med 1998;9:815–8. [16] Hutmacher DW. Scaffolds in tissue engineering bone and cartilage. Biomaterials 2000;21:2529–43. [17] Cooper AI, Howdle SM, Hughes C, Jobling M, Kazarian SG, Poliakoff M, Shephard LA. Spectroscopic probes for hydrogen bonding, extraction impregnation and reaction in supercritical fluids. Analyst 1993;118:111–6. [18] McHugh MA, Krukonis VJ. Supercritical fluid extraction: principles and practice, 2nd ed. Boston: Butterworth-Heinemann; 1994. [19] Mooney DJ, Baldwin DF, Suh NP, Vacanti JP, Langer R. Novel approach to fabricate porous sponges of poly(d,l-lactic-coglycolic acid without the use of organic solvents. Biomaterials 1996;17:1417–22. [20] Chen G, Ushida T, Tateishi T. Scaffold design for tissue engineering. Macromolecular Biosciences 2002;2(2):67–77. [21] Cooper AI. Recent developments in materials synthesis and processing using supercritical CO2. Adv Mater 2001;13:1–4. [22] Howdle SM, Watson MS, Whitaker MJ, Popov VK, Davies MC, Mandel FS, Wang JD, Shakesheff KM. Supercritical fluid mixing: preparation of thermally sensitive polymer composites containing bioactive materials. Chem Commun 2001;1:109–11. [23] Kazarian SG. Polymer processing with supercritical fluids. Polym Sci 2000;42:78–101. [24] Nealey PF, Cohen RE, Argon AS. Limited-supply non-Fickian diffusion in glassy polymers. Polymer 1995;36:3687–95. [25] Tampieri A, Celotti G, Sprio S, Delcogliano A, Franzese S. Porosity-graded hydroxyapatite ceramics to replace natural bone. Biomaterials 2001;22:1365–70. [26] Harris LD, Kim B-S, Mooney DJ. Open pore biodegradable matrices formed with gas foaming. J Biomed Mater Res 1998;42:396–402. [27] Archer CW, McDowell J, Bayliss MT, Stephens MD, Benntley G. Phenotypic modulation in sub-populations of human articular chondrocytes in vitro. J Cell Sci 1990;97:361–71.

[28] Rago R, Mitchen J, Wilding G. DNA fluorometric assay in 96well tissue culture plates using Hoechst 33258 after cell lysis by freezing in distilled water. Anal Biochem 1990;191:31–4. [29] Farndale RW, Sayers CA, Barrett AJ. A direct spectroscopic microassay for sulphated glycosaminoglycans in cartilage cultures. Connect Tissue Res 1982;9(4):247–8. [30] Argon AS, Cohen RE, Patel AC. A mechanistic model of case-II diffusion of a diluent into a glassy polymer. Polymer 1999;40:6991–7012. [31] Goel SK, Beckman EJ. Generation of microcellular polymeric foams using supercritical carbon dioxide. II: cell growth and skin formation. Polym Eng Sci 1994;34:1148–56. [32] Arora KA, Lesser AJ, McCarthy TJ. Preparation and characterization of microcellular polystyrene foams processed in supercritical carbon dioxide. Macromolecules 1998;31:4614–20. ! y Leon ! CA. New perspectives in mercury porosimetry. Adv [33] Leon Colloid Interface Sci 1998;76–77:341–72. [34] Marijnissen WJCM, Van Osch G JVM, Aigner J, Verwoerdverhoef HL, Verhaar JAN. Tissue-engineered cartilage using serially passaged articular chondrocytes. Chondrocytes in alginate, combined in vivo with a synthetic (E210) or biological biodegradable carrier (DBM). Biomaterials 2000;21: 571–80. [35] Sawtell RM, Downes S, Patel MP, Clarke RL, Braden M. Heterocyclic methacrylates for clinical applications—further studies of water sorption. J Mater Sci: Mater Med 1997;8: 667–74. [36] Boyan BD, Humbert TW, Dean DD, Schwartz Z. Role of material surfaces in regulating bone and cartilage cell response. Biomaterials 1996;17:137–46. [37] Ingber DE. Cellular tensegrity: defining new rules of biological design that govern the cytoskeleton. J Cell Sci 1993;104: 613–27. [38] Watt F. Effect of seeding density on stability of the differentiated phenotype of pig articular chondrocytes in culture. J Cell Sci 1988;89:373–8. [39] Sawtell RM. An in vitro investigation of a heterocyclic methacrylate polymer system for cartilage repair. PhD Thesis. Nottingham: University of Nottingham; 1996. [40] Taboas JM, Maddox RD, Krebsbach PH, Hollister SJ. Indirect solid free form fabrication of local and global porous, biomimetic and composite 3D polymer-ceramic scaffolds. Biomaterials 2003;24:181–94.