The effects of glycosaminoglycan content on the compressive modulus of cartilage engineered in type II collagen scaffolds

Osteoarthritis and Cartilage (2008) 16, 1237e1244 ª 2008 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.joca.2008.02.014

International Cartilage Repair Society

The effects of glycosaminoglycan content on the compressive modulus of cartilage engineered in type II collagen scaffolds

E. Pfeiffer B.S.yz, S. M. Vickers Ph.D.yza, E. Frank Ph.D.x, A. J. Grodzinsky Ph.D.x and M. Spector Ph.D.yzk*

y Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA z Tissue Engineering, VA Boston Healthcare System, Boston, MA 02130, USA x Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA k Department of Orthopaedic Surgery, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 02115, USA Summary Objective: The current study determined the unconfined compressive modulus of tissue-engineered constructs with varying sulfated glycosaminoglycan (GAG) density produced by goat articular chondrocytes in type II collagen scaffolds prepared with a range of cross-link densities and various times in culture. The purpose of this work is to establish a basis for future studies employing constructs of selected maturity (e.g., 25%, 50%, or 75% normal GAG content) for cartilage repair in vivo. Methods: Porous scaffolds (8 mm diameter by 2 mm thick) were fabricated from porcine type II collagen by freeze-drying, followed by dehydrothermal treatment and carbodiimide cross-linking. In a pilot study, passage 3 adult caprine articular chondrocytes isolated from one goat were grown in scaffolds with six cross-link densities for 2, 3, 4, and 6 weeks (n ¼ 3). The goal was to select scaffold cross-link densities and times in culture that would produce constructs with approximately 25%, 50% and 75% the GAG density of native articular cartilage. Based on the results of the pilot study, chondrocytes from three goats were grown in scaffolds with two cross-link densities for three time periods: 3, 5, and 9 weeks (n ¼ 6; one of the cross-link groups was run in quadruplicate). The equilibrium modulus from unconfined compression testing of these samples was correlated with GAG content. Results: There was a notable increase in GAG density with decreasing cross-link density. Histological analysis verified a chondrogenic phenotype and revealed various amounts of GAG and type II collagen-containing cartilage. The correlation between modulus and GAG density had a linear coefficient of determination of 0.60. One group with a mean GAG density of 22 mg/mm3, which was 140% the GAG density of normal caprine articular cartilage, averaged a compressive modulus of 31.5 kPa, which was 10% of caprine articular cartilage tested in this study. Conclusions: The GAG density and modulus of tissue-engineered constructs can be controlled by the degree of cross-linking of type II collagen scaffolds and time in culture. ª 2008 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. Key words: Cartilage, Tissue engineering, Mechanical properties, Modulus, Glycosaminoglycan, Scaffold, Collagen.

the articular surface of the stifle joint1 resulted in more reparative tissue and a greater percentage of hyaline cartilage after 15 weeks, compared to a chondrocyte-seeded scaffold cultured for only 12 h prior to implantation2. While the constructs cultured for 12 h and 4 weeks had different compositions and mechanical behaviors those parameters were not quantitatively evaluated. Therefore, it was not possible to correlate certain features of the make-up of the constructs with the in vivo findings. Several approaches have been successfully employed for the production of cartilaginous constructs with varying degrees of maturity (e.g., glycosaminoglycan, GAG, content) in vitro. Numerous studies have demonstrated the effects of growth factors individually3e5 and in combination6e8 on chondrocytes in monolayer and three-dimensional (3-D) cultures. One prior investigation showed a substantial increase in the GAG content of chondrocyte-seeded type II collagen scaffolds grown in serum-free medium supplemented with fibroblast growth factor (FGF)-2, compared to cultures with no FGF-2 supplementation, after only 2 weeks in culture9. Another recent investigation of chondrocyte-seeded type II

Introduction One of the critical issues related to the implementation of tissue engineering strategies for the treatment of a wide array of problems relates to the degree to which the tissueengineered construct should be allowed to develop (i.e., mature) in vitro prior to implantation. For example, how mature should a cartilaginous construct, produced in culture, be before it is implanted into a defect in articular cartilage? In our prior investigations using a canine model, an autologous chondrocyte-seeded type II collagen scaffold cultured for 4 weeks prior to implantation into a defect in a

Current address: Medtronic Spinal and Biologics, 2600 Sofamor Danek Dr., Memphis, TN 38132, USA. *Address correspondence and reprint requests to: Dr Myron Spector, Ph.D., VA Boston Healthcare System, 150 S. Huntington Avenue, Mail Stop 151, Boston, MA 02130, USA. Tel: 1-857-3646639; Fax: 1-857-364-6791; E-mail: [email protected]. harvard.edu Received 2 July 2007; revision accepted 17 February 2008.

1237

1238

E. Pfeiffer et al.: Compressive modulus of cartilage engineered in type II collagen scaffolds

collagen scaffolds, however, took an alternative approach to regulate the rate of chondrogenesis in vitro by altering the cross-link density of the scaffolds10. The study demonstrated that the cross-link density of the scaffold significantly affected chondrogenesis in vitro. Scaffolds with low crosslink densities, which underwent cell-mediated contraction, displayed a substantially greater GAG density than scaffolds of higher cross-link density, which did not contract10. In order to prepare constructs of a particular diameter for implantation, allowance would have to be made for the amount a construct would contract, in selecting the original diameter of the scaffold to be seeded with cells. One of the challenging aspects of the investigation of the degree of maturity that a tissue-engineered construct should have prior to implantation relates to the parameters to employ for the evaluation of the maturity of the tissueengineered construct: morphological/histological, biochemical, and/or mechanical. To begin to address this issue we selected the sulfated GAG density of tissue-engineered cartilaginous constructs as one measure of their degree of maturity, relative to the GAG density of native articular cartilage. The goal is to eventually implant constructs with 25%, 50%, and 75% the GAG density of native cartilage into defects in an animal model. The objective of the current study was to evaluate the GAG density and compressive modulus of chondrocyte-seeded type II collagen scaffolds using various combinations of cross-link density and time in culture; caprine cells were used because a goat model will be used for the in vivo evaluation. Materials and methods TYPE II COLLAGENeGAG SCAFFOLD FABRICATION Sheets of the type II collagen scaffold, 2.5 mm in thickness, were prepared from a 1% slurry of porcine type II collagen (Geistlich Biomaterials, Wolhusen, Switzerland) in 0.001 N HCl using a freeze-drying protocol found to generate scaffolds with porosity of about 90% and pores with a diameter of approximately 200 mm10. Disks, 8 mm in diameter, punched from the sheets were sterilized (and lightly cross-linked) by dehydrothermal treatment (DHT)10. Additional cross-linking was produced with a water-soluble carbodiimide11,12. Matrices were immersed in a solution containing 1-ethyl-3-(3dimethylaminopropyl) carbodiimide hydrochloride (EDAC, Sigma Chemical Co., St. Louis, MO, USA) and N-hydroxysuccinimide (NHS, Sigma Chemical Co., St. Louis, MO, USA) for 2 h10. Various EDAC:NHS molar ratios were employed, relative to the carboxyl contained in the collagen scaffold (1.2 mM COOH per mole of collagen)12. Five groups received EDAC treatments and a sixth group was only cross-linked by DHT (Table I). Relative cross-link density of the scaffolds (n ¼ 2e4)10 was determined indirectly by their swelling ratios, SRs13,14; the cross-link density varies with the inverse of SR. CHONDROCYTE CULTURE, SEEDING OF THE SCAFFOLDS, AND VOLUME MEASUREMENTS Articular cartilage was harvested from the stifle joints of adult Spanish goats immediately postmortem. Chondrocytes were isolated using

Table I Cross-linking agent molar ratios employing EDAC and NHS. Ratios are shown relative to the carboxyl contained in the collagen scaffold Mean  S.D. Cross-link group DHT only EDAC1 EDAC1.5 EDAC2 EDAC2.5 EDAC3

Molar ratios

a previously published protocol10, and cultured separately. In a pilot study, passage (P) 2 chondrocytes isolated from one goat were grown in scaffolds with six cross-link densities for 2, 3, 4, and 6 weeks (n ¼ 3). The goal was to select scaffold cross-link densities and times in culture that would produce constructs with approximately 25%, 50% and 75% the GAG density of native articular cartilage. Considering the results of the pilot study, P3 chondrocyte-seeded scaffolds with two cross-link densities (EDAC1.5 and EDAC2) were cultured for three time periods: 3, 5, and 9 weeks. Based on the available P3 cells (Table II), a sample size of six was achieved using Goat 1 cells to seed three groups, cells from Goat 2 to seed two groups, and cells from Goat 3 to seed the sixth group of scaffolds. For the EDAC1.5 groups, the experiment was conducted in quadruplicate (n ¼ 24), because the pilot data indicated that this treatment might be the most promising to yield the widest range of GAG density. The scaffolds were seeded with 4  106 cells and grown in chondrogenic medium as previously described10. The diameters of the cell-seeded matrices and unseeded controls were measured at the end of the culture period using circular templates ranging from 1 mm to 10 mm diameter in 0.5 mm increments. Thickness was measured using a micrometer with a resolution of 0.01 mm. The volume of the constructs was calculated from these dimensions.

MECHANICAL TESTING Unconfined compression testing was conducted to determine the equilibrium modulus of elasticity, E, of cell-seeded and non-cell-seeded control scaffolds (n ¼ 6) and full-thickness (mean, 0.8 mm) samples of caprine articular cartilage tissue (n ¼ 3 plugs from flat surfaces from one goat knee) of about the same diameter as the scaffolds. A cylindrical polymethyl methacrylate loading platen, 9.5 mm in diameter, was used in the test apparatus. Stress relaxation tests were conducted to steps of 2%, 4%, 6%, 8%, 10%, 15% and 20% engineering strain, with each step applied at a constant rate over 5 s and held for a subsequent 300 s, based on pilot data which showed that values at 300 s were within w5% of the values at 100 s. The compressive modulus was taken to be the slope of a linear regression line fit to the equilibrium stress and strain calculated at the completion of each step. Tests were accepted if the linear regression coefficient of determination was at least 0.9, as a check for test error such as sample slip or tearing, which would affect the stressestrain relationship (modulus). The multiple measurements indicated that the stressestrain relationship was linear (free of strain-stiffening or softening) in most samples, validating the calculation of modulus. Test elimination could usually be traced to a visible tear or unusual slipping of a rounded sample in the test chamber.

BIOCHEMICAL ANALYSIS AND HISTOCHEMISTRY/ IMMUNOHISTOCHEMISTRY At the termination of the cultures and the mechanical tests the constructs were cut into half, for allocation of samples for determination of GAG content using the dimethylmethylene blue dye assay15 and for Safranin-O histochemical staining and type II immunohistochemistry10. Unseeded control scaffolds and samples of goat articular cartilage were also analyzed for GAG content. Cartilage was removed (down to the tidemark) from the two sloping sides of the trochlea from one of the stifle joints of two goats (for a total of four samples). Disks (4 mm in diameter) were punched from the samples which ranged in thickness from 0.6 mm to 0.9 mm. As in prior immunohistochemical studies10, the scaffold material did not stain with the antibody, likely due to the breakdown of the epitope in the course of processing the material.

STATISTICS A sample size of n ¼ 6 was used for most of the analyses using a power calculation based on detecting a significant difference between two groups of 30%, with a ¼ 0.05 and b ¼ 0.05 and a standard deviation (SD) of 15%. P < 0.05 was set as the criterion of significance for statistical tests.

SR (n)

EDAC

NHS

COOH

0 1 2.5 5 10 10

0 1 1 2 4 4

1 5 5 5 5 1

8.18  2.49 7.36  0.67 5.73  0.29 5.30  0.12 5.06  0.44 4.36  0.33

(4) (3) (3) (4) (2) (4)

Table II Number of cells (millions) Goat # 1 2 3

Thawed cells

P1

P2

P3

10.3 9.5 4.6

30 26.1 9.6

86.5 82 40

171 138 74.6

1239

Osteoarthritis and Cartilage Vol. 16, No. 10

Results The various cross-link protocols yielded a range of crosslink densities reflected in the SRs (Table I). As has been observed in previous work10, cell-seeded constructs which were only lightly DHT cross-linked contracted to less than 30% of their original diameter after only 2 weeks. This contraction of the scaffolds was visible after as little as 1 week in culture. The small size of some scaffolds resulted in loss of a few samples in the pilot

experiment. Other lightly cross-linked samples were lost owing to their disintegration in culture. HISTOLOGY AND IMMUNOHISTOCHEMISTRY

There was a notable reduction in size of the 3-week constructs related to their cross-link density [Fig. 1(aed)]. DHTtreated scaffolds contracted to irregular spherical structures only about 2 mm in diameter [Fig. 1(a)]. The DHT- and EDAC1-treated constructs displayed folding of the scaffolds

Fig. 1. Micrographs of chondrocyte-seeded type II collagen scaffolds after 3 weeks in culture: (a) DHT cross-linked; (b) EDAC1.5 group; (c) EDAC2 group; and (d) EDAC3 group. The micrographs in (a)e(d) were stained with Safranin-O and taken at the same magnification. (e) and (f) Micrographs of EDAC1.5 samples after 3 and 9 weeks, respectively, taken at the same magnification. (b)e(d) Planar sections made through the middle part of the disk-shaped scaffold. (e) Safranin-O and (f) type II collagen immunohistochemical stain (brown chromogen).

E. Pfeiffer et al.: Compressive modulus of cartilage engineered in type II collagen scaffolds

during the process of contraction in the first 3-week period of culture; all of the EDAC2, EDAC2.5, and EDAC3 specimens and many EDAC1.5 scaffolds remained in the disk shape throughout the culture periods. Scaffolds with higher cross-link densities retained more of the initial sponge-like architecture of the scaffold [comparing Fig. 1(aed)]. Chondrogenesis was evident based on the characteristic rounded morphology of cells located in lacunae [Fig. 1(e)], histochemical staining of GAG using Safranin-O stain [Fig. 1(aec), and (e)], and type II immunohistochemistry [Fig. 1(f)]. Of note was the decrease in chondrogenesis with increasing cross-link density [comparing Fig. 1(aed)]. For the EDAC1.5, EDAC2, and EDAC3 cross-link groups there was an increase in the amount of tissue containing GAG and type II collagen with time in culture. This was less evident in the DHT group, because of the small size of the constructs and substantial amount of cartilage found by 3 weeks. Also of note was the inverse relationship between the degree of chondrogenesis and the presence of residual scaffold, which could be distinguished in histological section by its morphology. Little residual scaffold was found in the DHT only samples [Fig. 1(a)] which displayed intense GAG staining and a majority of cells of chondrocytic morphology, whereas much of the initial type II collagen scaffold appeared to remain in the EDAC3 group [Fig. 1(d)] which displayed substantially less cartilage formation. In individual samples in the EDAC1.5 [Fig. 1(b)] and EDAC2 [Fig. 1(c)] groups, regions of cartilage formation with little remaining scaffold could generally be found in the interior portions of the samples with remaining scaffold found in the peripheral regions which displayed no chondrogenesis. While chondrogenesis was clearly evident in the scaffolds in the less cross-linked groups, cells of a non-chondrocytic phenotype were also present in the scaffolds of lower cross-link density. Even though immunohistochemistry for type I collagen was not performed, it was evident that some must be present because the type II antibody did not label all of the newly synthesized matrix. The fibrous appearance of the matrix in cartilaginous regions, including some that stained positive for type II collagen [Fig. 1(f)], indicated the presence of fibrocartilage. The mixed make-up of tissues (hyaline cartilage, fibrocartilage and fibrous tissue) within individual constructs was a common finding. Also of interest was the high cell number density in many of the cartilaginous regions [Fig. 1(e)]. Some cells in these hypercellular areas displayed pyknotic nuclei, and cell debris could also be seen, suggesting the eventual reduction in cell number. GAG DENSITY

The GAG/volume values for the four goat articular cartilage samples were: 3.3 mg/mm3 and 18.9 mg/mm3 for one goat and 8.6 mg/mm3 and 12.7 mg/mm3 for the samples from the second goat; compared to 27 mg/mm3 for native adult bovine cartilage16. Interestingly, there was a strong inverse correlation between GAG density and sample thickness (linear regression analysis; R2 ¼ 0.99). Our cartilage samples included the superficial zone which may have had a decreased GAG content. The inverse correlation might be explained, then, if the GAG-depleted superficial zone increased with increasing thickness, but this will require additional work to resolve. For the purposes of comparative analysis, the GAG content from the pilot experiment was expressed per volume of the constructs and presented relative to the average of the highest GAG

density values for the two goats, 15.8 mg/mm3. Thus we chose a conservative approach in estimating the GAG density of the native goat cartilage. The GAG density of the tissue-engineered constructs generally increased with decreasing cross-link density (Fig. 2). There was no systematic increase in GAG density with time in culture across all cross-link groups (Fig. 2). Of note was the finding that selected constructs achieved GAG densities substantially higher than native cartilage after only 2 weeks in culture. Based on the goal of producing constructs with approximately 25%, 50%, and 75% the GAG density of native cartilage, it was decided to employ the EDAC1.5 and EDAC2 cross-link protocols for 3, 5, and 9 weeks. For these two cross-link densities, there was a significant five-fold increase in total GAG content with time from about 150 mg at 3 weeks to 750 mg at 9 weeks [Fig. 3(a)]. The total GAG content was comparable for the two cross-link groups at each of the three time periods [Fig. 3(a)]. Three-factor analysis of variance (ANOVA) revealed a significant effect of time on GAG content (P < 0.0001, power ¼ 1), but no effect of cross-link treatment (P ¼ 0.27) or of animal source of cells (P ¼ 0.07, power ¼ 0.51). Fisher’s protected least squares differences (PLSD) post hoc testing showed that the differences in GAG content between time periods were significant. Two-factor ANOVA performed on the EDAC1.5 group demonstrated that there was no effect of quadruplicate set on GAG content, as would be expected. Of note, however, was the effect of animal source of cells on the GAG content for the EDAC1.5 data, with cells from one of the animals resulting in 30e50% higher values of GAG content after 5 and 9 weeks; two-factor ANOVA (time and animal) revealed a significant effect of animal source of cells on GAG content (P ¼ 0.0003, power ¼ 0.98). There was a dramatic difference in the volumes of the constructs prepared with these two cross-link protocols [Table III; Fig. 3(b)]. The EDAC1.5 group was about 20% of its original volume at week 3, while the EDAC2 group remained at about 80% of its original value through the 9-week culture period. Three-factor ANOVA revealed significant effects of cross-link treatment and animal source of cells on volume (both P < 0.0001, power ¼ 1); there was also a significant effect of time on volume, but with a low power (P ¼ 0.04, power ¼ 0.6). In analyzing the data for

250

% Normal GAG per Volume

1240

Week 2 Week 3 Week 4 Week 6

225 200 175 150 125 100 75 50 25 0

DHT

EDAC1 EDAC1.5 EDAC2 EDAC2.5 EDAC3

Cross-Link Group Fig. 2. Pilot study comparing the effect of cross-link group and time in culture on GAG content. GAG content (GAG/volume) is shown as percent of that in native caprine articular cartilage (15.8 mg/mm3).

1241

Osteoarthritis and Cartilage Vol. 16, No. 10

a

1000 900

GAG Content, µg

800

Week 3 Week 5 Week 9

700 600 500 400 300 200 100 0

EDAC2

EDAC1.5

Cross-Link Group

b

120 Week 3

Volume, mm3

100

Week 5 Week 9

80

COMPRESSIVE MODULUS

60 40 20 0

EDAC2

EDAC1.5

Cross-Link Group

GAG per Volume, µg/mm3

c

30 25

Week 3 Week 5 Week 9

20 15 10 5 0

Non-cell-seeded control scaffolds were found to have a GAG density of 0.14  0.09 mg/mm3 (mean  SD; n ¼ 4). Owing to the large difference in the volumes of the constructs there was a large difference in the GAG density between the two cross-link groups [Fig. 3(c)], with the EDAC1.5 constructs displaying a three-fold greater GAG density than the EDAC2 group at 3 weeks, two-fold greater at 5 weeks, and 2.5-fold greater at 9 weeks [at 9 weeks, 21.5  2.5, mean  standard error of the mean (S.E.M.) vs 8.6  1.3 mg/mm3]. For both cross-link groups there was an increase in GAG density with time [Fig. 3(c)]: the mean value for the 3-week EDAC1.5 group was about 60% of the native value of 15.8 mg/mm3, and the 5-week EDAC1.5 samples exceeded that value. Three-factor ANOVA revealed significant effects of crosslink treatment (P < 0.0001, power ¼ 1), time (P ¼ 0.0002, power ¼ 0.99), and animal source of cells (P ¼ 0.035, but with a low power ¼ 0.63) on GAG density. Fisher’s PLSD post hoc testing showed significant differences in GAG density between weeks 3 and 5 (P ¼ 0.005) and 3 and 9 (P < 0.0001), but no difference in GAG density between weeks 5 and 9 (P ¼ 0.15). In analyzing the data for the EDAC1.5 group only, two-factor ANOVA (time and quadruplicate set) showed no significant difference in the GAG values in the quadruplicate samples at each of the three time periods.

EDAC1.5

EDAC2

Cross-Link Group Fig. 3. Graphs showing the (a) total GAG content per specimen, (b) the volume of the constructs, and (c) the GAG per volume (GAG density), for each time point and cross-link combination. Mean  S.E.M. (n ¼ 24 for the EDAC1.5 groups, and n ¼ 6 for the EDAC2 groups).

the EDAC1.5 group only, two-factor ANOVA (time and quadruplicate set) showed no significant difference in the volumes of the constructs in the quadruplicate sets of data at each of the three time periods.

Non-cell-seeded, hydrated scaffolds in the EDAC1.5 crosslinking group had a modulus of 0.88  0.38 kPa (mean  SD; n ¼ 4) while scaffolds in the EDAC2 group had a modulus of 1.45  0.84 kPa (n ¼ 3). The modulus of native caprine articular cartilage was 314  221 kPa (mean  SD; n ¼ 3). The values for compressive moduli for the less crosslinked group (EDAC1.5) were higher than the values for the EDAC2 group at each of the three time periods of culture [Fig. 4(a)]. After 9 weeks the EDAC1.5 modulus was three-fold higher than that of the EDAC2 group (w31.5  3 kPa, mean  S.E.M., vs 10.7  5 kPa). The moduli of each cross-link group increased with time in culture with the EDAC1.5 group increasing five-fold (6.1  2 kPa to 31.5  5 kPa) from 3 to 9 weeks [Fig. 4(a)]. Even the highest mean value of 31.5 kPa was only 10% of the native modulus, despite the fact that the GAG density for this group was 140% of the native value. Two-factor ANOVA revealed significant effects of crosslink treatment (P ¼ 0.001, power ¼ 0.94) and time (P ¼ 0.0002, power ¼ 0.991) on modulus. Fisher’s PLSD post hoc testing showed that the differences between individual cross-link groups and among all time points were significant. For the EDAC1.5 group, two-factor ANOVA (time and quadruplicate set) showed a significant effect of time (P < 0.0001, power ¼ 1) but not of quadruplicate samples (P ¼ 0.06, power ¼ 0.69) on modulus . Two-factor ANOVA (time and animal) revealed a significant effect of time on modulus (P < 0.0001, power ¼ 1), but not of animal source of cells (P ¼ 0.35). The relationship between GAG density (mg/mm3), rGAG, and modulus (kPa), EComp, from linear regression analysis [Fig. 4(b)], was: EComp ¼ 1  rGAG P < 0:0001; R 2 ¼ 0:6 where the S.E.M. for the coefficient was 0.1.

1242

E. Pfeiffer et al.: Compressive modulus of cartilage engineered in type II collagen scaffolds Table III Dimensions of the constructs cross-linked with the EDAC1.5 and EDAC2 carbodiimide protocols (mm); mean  S.E.M. (n) EDAC1.5 Thickness 1.8  0.2 2  0.1 2.5  0.1 2.7  0.1

Initial 3 Weeks 5 Weeks 9 Weeks

EDAC2 Diameter 8.3  0.2 3.6  0.3 4.1  0.3 4.2  0.2

(4) (24) (20) (24)

Discussion This study showed that the cross-link density of the type II collagen scaffolds was a principal determinant of the GAG density of chondrocyte-seeded constructs, thus supporting earlier findings10. A notable finding was that the GAG density of some of the constructs substantially exceeded the value for goat articular cartilage after only a few weeks of culture. Recent work17 growing primary 1-week old calf chondrocytes in scaffold free-culture also found GAG values which exceeded that of native (bovine) cartilage: two-fold at 4 weeks, and then decreasing to two-thirds at 12 weeks. In the pilot study there was no systematic change of GAG density with time across all groups, while in the

a Compressive Modulus [kPa]

40 Week 3 Week 5 Week 9

35 30 25 20 15 10 5 0

EDAC1.5

EDAC2

Cross-Link Group 80

b Modulus, kPa

70 60

Week 3, E1.5

Week 5, E2

Week 3, E2

Week 9, E1.5

Week 5, E1.5

Week 9, E2

50 40 30 20 10 0 0

10

20

30

40

50

60

GAG Density, µg/mm3 Fig. 4. (a) Compressive modulus from each time point and crosslink combination. Mean  S.E.M. (n ¼ 24 for the EDAC1.5 groups, and n ¼ 6 for the EDAC2 groups). Native caprine articular cartilage had a modulus of 314  221 kPa (mean  SD; n ¼ 3). (b) Scaffold compressive modulus plotted against GAG density, including all samples. The line shows the linear regression fit. E1.5, EDAC1.5; E2, EDAC2.

(4) (24) (20) (24)

Thickness 2.1  0.2 1.8  0.2 1.9  0.1 2.2  0.1

(3) (6) (6) (6)

Diameter 8.3  0.8 7.2  0.4 7.1  0.4 7.0  0.4

(3) (6) (6) (6)

follow-up work with the EDAC1.5 and EDAC2 groups there was an increase with time. This might be explained in the groups with lower cross-link density (viz., DHT and EDAC1) by the substantial amount of cartilage that had already formed in the constructs after only 2 weeks. The increased GAG density in the less cross-linked, more contracted scaffolds was not due to a densification of the GAG that had already formed, because the majority of the contraction occurred before most of the GAG production. Also of importance is that the difference in the GAG density among the constructs was associated with a phenotypic difference in the cells. Cells in the constructs which contracted and had a high GAG density displayed the chondrocytic phenotype of rounded cells in lacunae, whereas cells in constructs with a low GAG density were not chondrocytic in appearance. It has been well-documented that isolated chondrocytes de-differentiate during monolayer culture18 associated with their adoption of a flattened morphology, particularly in the presence of FGF-26. These cells re-differentiate to chondrocytes in vitro under conditions which allow (or perhaps force) them to assume a spherical shape19e21. The fact that individual constructs can be populated by both re-differentiated and de-differentiated chondrocytes, which may be synthesizing various types of GAG at different rates, complicates investigations of the effects of crosslink density and contracture on the rate of cartilage-specific GAG production. Contraction of the scaffolds has previously been attributed to a-smooth muscle actin expression22. The contracture of the more compliant EDAC1.5 scaffolds resulted in an increase in cell number density in selected regions of the construct10. This finding recalls the ‘‘condensation’’ of mesenchymal progenitor cells necessary for the development of cartilage in vivo23e25 and in ‘‘micro-mass’’ cultures26. Contracture-facilitated aggregation of cells in the scaffolds, increasing the packing density, may have induced chondrocyte expression by increasing cellecell contacts27e31 and/or by favoring a rounded morphology32. Once cells re-differentiated in the contracted scaffolds, contracture may have favored chondrogenesis by facilitating the retention of newly synthesized matrix molecules within the collapsed pores of the constructs10. The scaffolds of higher cross-link density and stiffness, resisting cell-mediated contraction, may have inhibited/delayed the chondrocytic re-differentiation of the cells by providing an environment permissive of the maintenance of a flattened/elongated morphology of the de-differentiated chondrocytes. Also of interest was the finding that the greater the degree of chondrogenesis the lesser the amount of residual scaffold10. The increased rate of scaffold degradation associated with cartilage formation, i.e., perhaps to be considered a remodeling process, may have been due to enzymes synthesized by the re-differentiated chondrocytes acting on the scaffolds of lower cross-link density.

1243

Osteoarthritis and Cartilage Vol. 16, No. 10 Cross-link density and culture-time combinations could be found to engineer constructs with mean GAG densities from 20% to 100% of normal caprine GAG density, for constructs ranging in diameter from 3 mm to 7 mm. The results provide a basis for determining the cross-link density and time in culture required to produce constructs of a particular size and GAG density for implantation. While cartilaginous constructs may be formed in scaffold free-cultures33,34, the use of the type II collagen scaffold may provide a wider range of dimensions of the construct and shorter culture times. The compressive modulus of natural caprine articular cartilage reported in this study (314 kPa) compares with values from 300 kPa to 800 kPa for bovine articular cartilage tested under unconfined compression35. The compressive modulus of selected constructs reached about 25% of normal. The highest mean modulus was, however, 10% of the normal value, even though the GAG density for the group was about 140% of the native value. Unconfined compression moduli reported for other chondrocyte-seeded scaffolds include: 12 kPa after 28 days in culture for a biodegradable elastomer, poly(1,8-octanediol citrate), construct36; and 100 kPa for a hyaluronan scaffold37 with selected cellseeding densities and times in culture. There was only a modest correlation between compressive modulus and GAG density, reflected in the R2 value of 0.60. This is understandable in the light of the known contributions of type II collagen, water content, and other matrix molecules, and microstructure, to the mechanical behavior of articular cartilage38. Prior work39 demonstrated a coefficient of determination of about 0.40 for the correlation between GAG content and equilibrium modulus for human articular cartilage samples obtained from several joints; a multivariate analysis of the correlation of the GAG, water, and hydroxyproline contents with modulus revealed R2 values greater than 0.80, with GAG being the most significant factor. Other studies correlating GAG content and unconfined compressive modulus for chondrocyte-seeded scaffolds found: R2 ¼ 0.61 for calf chondrocyte-seeded agarose hydrogels40; and R2 ¼ 0.36 for adult bovine chondrocyte-seeded non-woven hyaluronan fleeces37. The moduluseGAG relationship for tissue-engineered constructs can be confounded by the non-uniformity of the tissue in the scaffolds. Well-developed cartilage could be found in the interior regions of constructs which displayed residual scaffold in the periphery of the samples. Moreover, time in culture and maturation/remodeling of the constructs can also affect the relative modulus and GAG values. The confined compression modulus of calf chondrocyte-seeded non-woven polyglycolic acid fiber meshes41 increased from 20% of normal for 3-month constructs of native GAG content to the normal values (about 1000 kPa) for 7-month constructs with a GAG content 120% of the native value. Results of the current study provide a basis for future work investigating the effects of the degree of maturation of tissue-engineered cartilaginous constructs on cartilage repair in vivo. It will be useful in future studies to investigate other measures of construct maturity, including collagen content, to relate to the mechanical behavior of the constructs and their contribution to cartilage repair in vivo.

Conflict of interest None.

Acknowledgements This material is based upon work supported by the Office of Research and Development, Rehabilitation Research and Development Service, Department of Veterans Affairs. MS is a VA Research Career Scientist Awardee. The type II collagen antibody developed by Holmdahl and Rubin was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. The authors would like to thank Alix Weaver for her technical support.

References 1. Lee CR, Grodzinsky AJ, Hsu H-P, Spector M. Effects of a cultured autologous chondrocyte-seeded type II collagen scaffold on the healing of a chondral defect in a canine model. J Orthop Res 2003;21: 272e81. 2. Breinan HA, Martin SD, Hsu H-P, Spector M. Healing of canine articular cartilage defects treated with microfracture, a type II collagen matrix, or cultured autologous chondrocytes. J Orthop Res 2000;18:781e9. 3. Toolan BC, Frenkel SR, Pachence JM, Yalowitz L, Alexander H. Effects of growth-factor-enhanced culture on a chondrocyteecollagen implant for cartilage repair. J Biomed Mater Res 1996;31:273e80. 4. Martin I, Vunjak-Novakovic G, Yang J, Langer R, Freed LE. Mammalian chondrocytes expanded in the presence of fibroblast growth factor 2 maintain the ability to differentiate and regenerate three-dimensional cartilaginous tissue. Exp Cell Res 1999;253:681e8. 5. Fortier LA, Balkman CE, Sandell LJ, Ratcliffe A, Nixon AJ. Insulin-like growth factor-I gene expression patterns during spontaneous repair of acute articular cartilage injury. J Orthop Res 2001;19:720e8. 6. Jakob M, Demarteau O, Schafer D, Hintermann B, Dick W, Heberer M, et al. Specific growth factors during the expansion and redifferentiation of adult human articular chondrocytes enhance chondrogenesis and cartilaginous tissue formation in vitro. J Cell Biochem 2001;81: 368e77. 7. Blunk T, Sieminski AL, Gooch KJ, Courter DL, Hollander AP, Nahir AM, et al. Differential effects of growth factors on tissue-engineered cartilage. Tissue Eng 2002;8:73e84. 8. Chiyokura Y. Effects of IGF-1, IGF-II, and PDGF on the proliferation of rabbit articular chondrocytes in culture. J Orthop Sci 1996;1:59e63. 9. Veilleux N, Spector M. Effects of FGF-2 and IGF-1 on adult canine articular chondrocytes in type II collageneglycosaminoglycan scaffolds in vitro. Osteoarthritis Cartilage 2005;13:278e86. 10. Vickers SM, Squitieri LS, Spector M. Effects of cross-linking type II collageneGAG scaffolds on chondrogenesis in vitro: dynamic pore reduction promotes cartilage formation. Tissue Eng 2006;12:1345e55. 11. Lee CR, Grodzinsky AJ, Spector M. The effects of cross-linking of collageneglycosaminoglycan scaffolds on compressive stiffness, chondrocyte-mediated contraction, proliferation and biosynthesis. Biomaterials 2001;22:3145e54. 12. Olde Damink LH, Dijkstra PJ, van Luyn MJ, van Wachem PB, Nieuwenhuis P, Feijen J. Cross-linking of dermal sheep collagen using a water-soluble carbodiimide. Biomaterials 1996;17:765e73. 13. Weadock K, Olson RM, Silver FH. Evaluation of collagen crosslinking techniques. Biomater Med Devices Artif Organs 1983;11: 293e318. 14. Yannas IV, Burke JF, Gordon PL, Huang C, Rubenstein RH. Design of an artificial skin. II. Control of chemical composition. J Biomed Mater Res 1980;14:107e31. 15. Farndale RW, Buttle DJ, Barrett AJ. Improved quantitation and discrimination of sulphated glycosaminoglycans by use of dimethylmethylene blue. Biochim Biophys Acta 1986;883:173e7. 16. Sah RL, Chen AC, Grodzinsky AJ, Trippel SB. Differential effects of bFGF and IGF-1 on matrix metabolism in calf and adult bovine cartilage explants. Arch Biochem Biophys 1994;308:137e47. 17. Hu JC, Athanasiou KA. A self-assembling process in articular cartilage tissue engineering. Tissue Eng 2006;12:969e79. 18. Benya PD, Padilla SR, Nimni ME. Independent regulation of collagen types by chondrocytes during loss of differentiated function in culture. Cell 1978;15:1313e21. 19. Benya PO, Schaffer JD. Dedifferentiated chondrocytes reexpress the differentiated collagen phenotype when cultured in agarose gels. Cell 1982;30:215e24. 20. Brown PD, Benya PD. Alterations in chondrocyte cytokeletal architecture during phenotypic modulation by retinoic acid and dihydrocytochalasin B-induced reexpression. J Cell Biol 1988;106:171e9.

1244

E. Pfeiffer et al.: Compressive modulus of cartilage engineered in type II collagen scaffolds

21. Lee CR, Grodzinsky AJ, Spector M. Modulation of the contractile and biosynthetic activity of chondrocytes seeded in collageneGAG matrices. Tissue Eng 2003;9:27e36. 22. Kinner B, Spector M. Smooth muscle actin expression by human articular chondrocytes and their contraction of a collageneglycosaminoglycan matrix in vitro. J Orthop Res 2001;19:233e41. 23. Fell HB. The histogenesis of cartilage and bone in the long bones of the embryonic fowl. J Morphol 1925;40:417e59. 24. Hall BK, Miyake T. Divide, accumulate, differentiate: cell condensation in skeletal development revisited. Int J Dev Biol 1995;39:881e93. 25. Hall BK, Miyake T. All for one and one for all: condensations and the initiation of skeletal development. Bioessays 2000;22:138e47. 26. Ahrens PB, Solursh M, Reiter RS. Stage-related capacity for limb chondrogenesis in cell culture. Dev Biol 1977;60:69e82. 27. Ahrens PB, Solursh M, Reiter RS, Singley CT. Position-related capacity for differentiation of limb mesenchyme in cell culture. Dev Biol 1979; 69:436e50. 28. Goldring MB, Tsuchimochi K, Ijiri K. The control of chondrogenesis. J Cell Biochem 2006;97:33e44. 29. Woods A, Khan S, Beier F. C-type natriuretic peptide regulates cellular condensation and glycosaminoglycan synthesis during chondrogenesis. Endocrinology 2007;148:5030e41. 30. Delise AM, Tuan RS. Analysis of N-cadherin function in limb mesenchymal chondrogenesis in vitro. Dev Dyn 2002;225:195e204. 31. Merker HJ, Zimmermann B, Barrach HJ. The significance of cell contacts for the differentiation of the skeletal blastema. Acta Biol Hung 1984;35:195e203. 32. Swalla BJ, Solursh M. Inhibition of limb chondrogenesis by fibronectin. Differentiation 1984;26:42e8. 33. Mainil-Varlet P, Rieser F, Grogan S, Mueller W, Saager C, Jakob RP. Articular cartilage repair using a tissue-engineered cartilage-like

34. 35.

36. 37.

38.

39.

40.

41.

implant: an animal study. Osteoarthritis Cartilage 2001;9(Suppl A): S6e15. Grogan SP, Barbero A, Winkelmann V, Rieser F, Fitzsimmons JS, O’Driscoll S, et al. Visual histological grading system for the evaluation of in vitro-generated neocartilage. Tissue Eng 2006. Korhonen RK, Laasanen MS, Toyras J, Rieppo J, Hirvonen J, Helminen HJ, et al. Comparison of the equilibrium response of articular cartilage in unconfined compression, confined compression and indentation. J Biomech 2002;35:903e9. Kang Y, Yang J, Khan S, Anissian L, Ameer GA. A new biodegradable polyester elastomer for cartilage tissue engineering. J Biomed Mater Res A 2006;77:331e9. Kelly DJ, Crawford A, Dickinson SC, Sims TJ, Mundy J, Hollander AP, et al. Biochemical markers of the mechanical quality of engineered hyaline cartilage. J Mater Sci Mater Med 2007;18: 273e81. Mow V, Rosenwasser M. Articular cartilage: biomechanics. In: Woo SL-Y, Buckwalter JA, Eds. Injury and Repair of the Musculoskeletal Soft Tissue. Park Ridge, IL: Amer Academy of Orthopaedic Surgeons 1988:427e63. Treppo S, Koepp H, Quan EC, Cole AA, Kuettner KE, Grodzinsky AJ. Comparison of biomechanical and biochemical properties of cartilage from human knee and ankle pairs. J Orthop Res 2000;18: 739e48. Mauck RL, Seyhan SL, Ateshian GA, Hung CT. Influence of seeding density and dynamic deformational loading on the developing structure/function relationships of chondrocyte-seeded agarose hydrogels. Ann Biomed Eng 2002;30:1046e56. Martin I, Obradovic B, Treppo S, Grodzinsky AJ, Langer R, Freed LE, et al. Modulation of the mechanical properties of tissue engineered cartilage. Biorheology 2000;37:141e7.