Engineering endostatin-expressing cartilaginous constructs using injectable biopolymer hydrogels

Acta Biomaterialia 8 (2012) 2203–2212

Contents lists available at SciVerse ScienceDirect

Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat

Engineering endostatin-expressing cartilaginous constructs using injectable biopolymer hydrogels Lily Jeng a,b, Bjorn R. Olsen c, Myron Spector a,d,⇑ a

Tissue Engineering, VA Boston Healthcare System, Boston, MA 02130, USA Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA c Department of Developmental Biology, Harvard School of Dental Medicine, Boston, MA 02115, USA d Department of Orthopaedic Surgery, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 02115, USA b

a r t i c l e

i n f o

Article history: Received 2 December 2011 Received in revised form 4 February 2012 Accepted 16 February 2012 Available online 24 February 2012 Keywords: Cartilage tissue engineering Gene therapy Scaffold Mesenchymal stem cell

a b s t r a c t The release of an anti-angiogenic agent, such as type XVIII/endostatin, from an implantable scaffold may be of benefit in the repair of articular cartilage. The objectives of this study are to develop an injectable mesenchymal stem cell (MSC)-incorporating collagen-based hydrogel capable of undergoing covalent cross-linking in vivo and overexpressing endostatin using nonviral transfection, and to investigate methods for the retention of the endostatin protein within the scaffolds. The effects of different cross-linking agents (genipin, transglutaminase-2, and microbial transglutaminase) and different binding molecules for endostatin retention (heparin, heparan sulfate, and chondroitin sulfate) are evaluated. Cartilaginous constructs that overexpress endostatin for 3 weeks are successfully engineered. Most of the endostatin is released into the surrounding media and is not retained within the constructs. The presence of two common basement membrane molecules, laminin and type IV collagen, which have been reported in developing and mature articular cartilage and are generally associated with type XVIII collagen in vivo, is also observed in the engineered cartilaginous constructs. Endostatin-producing cartilaginous constructs can be formulated by growing nonvirally transfected mesenchymal stem cells in collagen gels covalently cross-linked using genipin, transglutaminase-2, and microbial transglutaminase. These constructs warrant further investigation for cartilage repair procedures. The novel finding of laminin and type IV collagen in the engineered cartilage constructs may be of importance for future work toward understanding the role of basement membrane molecules in chondrogenesis and in the physiology and pathology of articular cartilage. Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction Articular cartilage is a naturally avascular tissue that exhibits limited healing potential if damaged. One of the initial features of the reparative response to cartilage repair procedures, including microfracture, and the implantation of cells, tissue grafts, and tissue-engineered constructs, is a vascular response. It has been well documented that vascularity inhibits the embryonic formation of cartilage [1] and that vascular invasion of mature cartilage has been associated with its degradation [2]. Recent studies have investigated angiogenesis inhibitors, such as Flt-1 [3,4], endostatin [5,6], and suramin [7], for the regulation of angiogenesis during the cartilage repair process, to try to improve cartilage regeneration. Two such studies [3,4] have demonstrated improved cartilage formation in repair models in which an anti-angiogenic factor was ⇑ Corresponding author at: Tissue Engineering, VA Boston Healthcare System, Boston, MA 02130, USA. Tel.: +1 857 364 6639; fax: +1 857 364 6791. E-mail address: [email protected] (M. Spector).

delivered. The prolonged release of an anti-angiogenic agent, such as endostatin, from an implantable scaffold may, therefore, be necessary for the regeneration of articular cartilage and the preservation of tissue-engineered cartilage implants. Endostatin, a 20 kDa proteolytic fragment of type XVIII collagen that inhibits endothelial cell proliferation and migration [8], has been found in developing [9] and normal adult articular cartilage [10] and shown to be synthesized by chondrocytes [11]. Besides its anti-angiogenic properties, it is known that endostatin can influence other processes, including promotion of anabolic activity in cartilage [11]. Thus, the application of endostatin in articular cartilage repair is promising for its potential multiple effects. Collagen gels which can be covalently cross-linked in vivo offer benefits in providing control over mechanical properties and degradation rate [12,13]. Commonly used collagen cross-linking agents for biologic tissue fixation, including glutaraldehyde and carbodiimide are, however, cytotoxic and cannot be used for in situ cross-linking of cell-seeded scaffolds [12]. Cross-linking agents that can be used safely with cells, such as transglutaminases

1742-7061/$ - see front matter Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actbio.2012.02.015

2204

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

[14–16] and genipin (GP) [12,17,18], have received increased attention in recent years. The principal objective of this study was to develop an injectable mesenchymal stem cell (MSC)-incorporating collagen-based hydrogel capable of forming cartilaginous material and overexpressing endostatin for several weeks, and to determine the effects of various cross-linking agents (genipin, transglutaminase-2, and microbial transglutaminase) on the development of the construct. This construct could serve as an implant for cartilage defects, providing cartilaginous material for subsequent in vivo maturation and integration, and endostatin-expressing cells to prevent vascular invasion. Previously, our lab has shown that MSCs can be engineered to overexpress endostatin while retaining the ability to differentiate down the chondrogenic lineage in preformed sponge-like scaffolds [6]. Our rationale for the use of a collagen scaffold is based on prior work demonstrating its utility for cartilage applications [19,20]. Another objective of the study was to investigate methods for the retention of the MSC-expressed endostatin protein in the gels and preformed sponge-like scaffolds. Binding of endostatin may protect it from proteolytic degradation and could create a reservoir of the protein for sustained release in a local area, thus potentially increasing the local concentration and amount of time during which the protein is present [21,22]. Endostatin has been shown to have multiple binding partners, including the glycosaminoglycans (GAGs): heparin, heparan sulfate (HS), and chondroitin sulfate (CS) [23–26]. It is hypothesized that endostatin-expressing cartilaginous constructs can be engineered using collagen hydrogels covalently cross-linked using transglutaminase (TG)-2, microbial transglutaminase (mTG), and GP, and that supplementation of the collagen scaffolds with heparin, HS, and CS will result in retention of endostatin within the engineered constructs, compared to non-supplemented collagen scaffolds.

and third experiments. In the second experiment, P2 MSCs were seeded into the sponge-like scaffolds and transfected by the pEndo-containing lipoplex incorporated into the scaffold, for comparison with the hydrogel results. The principal outcome variable for the three experiments was the amount of endostatin released by the cells. Samples of the constructs in all of the experiments were evaluated histologically, and histochemically and immunohistochemically for the presence of Safranin O and type II collagen, cartilage matrix molecules, and for endostatin. In the first experiment immunohistochemical analysis was also performed for the principal basement membrane molecules, laminin and type IV collagen, because type XVIII collagen is most frequently associated with basement membrane, and because laminin and type IV collagen have previously been identified in the pericellular matrix in articular cartilage [29].

2. Materials and methods

1. In the first experiment, monolayer-transfected MSCs were seeded in hydrogels, which were covalently cross-linked using TG-2, mTG, and GP. Control MSCs that were not transfected were also seeded into all gels (TG-2, mTG, and GP). The variable was the cross-linking agent. 2. In the second experiment, MSCs were seeded onto endostatin lipoplex-incorporating collagen sponge-like scaffolds. The variable was GAG supplementation of the sponge. 3. In the third experiment, monolayer-transfected MSCs were seeded in genipin-cross-linked hydrogel scaffolds incorporating heparin (obtained commercially, already bound on the surface of agarose beads), or control agarose beads without heparin (commercially obtained). Non-bead-supplemented hydrogel controls with no additional cross-linking agent and non-beadsupplemented gel controls with genipin were used. Control MSCs transfected using a plasmid encoding for green fluorescent protein (GFP) were also seeded into each non-bead-supplemented gel group. The variable was the incorporation of heparin, using heparin-agarose beads.

2.1. Plasmid propagation and isolation

2.4. Scaffold fabrication

The endostatin plasmid (pEndo) vector pCEP-Pu AC7, which employed the transcriptional control of the cytomegalovirus (CMV) promoter, and the propagation and isolation protocol, have been previously described [5,27]. Briefly, plasmid was obtained by heat shock transformation of Escherichia coli DH5a competent cells (Invitrogen, Carlsbad, CA, USA) and isolation using a Mega QIAfilter™ Plasmid kit (Qiagen, Valencia, CA, USA).

Six groups of hydrogel scaffolds were prepared: (1) soluble rat tail type I collagen (0.2% w/v) (BD Biosciences, San Jose, CA, USA); (2) type I collagen (0.2%) incorporating transglutaminase (TG)-2 (100 lg ml–1) (Sigma); (3) type I collagen (0.2%) incorporating microbial transglutaminase (mTG; 100 lg ml–1) (Ajinomoto Food Ingredients LLC, Chicago, IL); (4) type I collagen (0.2%) incorporating genipin (GP, 0.25 mM) (Wako, Richmond, VA, USA); (5) type I collagen (0.2%) incorporating GP (0.25 mM) and control agarose beads (without heparin) (Sigma); and (6) type I collagen (0.2%) incorporating GP (0.25 mM) and heparin-agarose type I beads (10% weight of heparin/weight of collagen) (Sigma). The volume of agarose beads and heparin-agarose beads added to the hydrogels was determined by the amount of heparin attached to the heparin-agarose beads; for this experiment, 0.16 ml of beads was added per ml of hydrogel mixture. The molecular weight of the heparin was reported by the manufacturer to be 25 kDa. For all hydrogel groups, sodium hydroxide (1 M) (Fisher, Pittsburgh, PA, USA) was added to the hydrogel mixture to obtain a final pH of 7.4, and expansion medium (with no FGF-2) and phosphatebuffered saline (PBS, Sigma) were used as fillers to bring the mixture up to the desired total volume. For the TG-2 group, calcium chloride (5 mM) (Sigma) and dithiothreitol (2 mM) (Sigma) were also added to the mixture. All hydrogel groups were kept on ice until cell seeding and gel casting. Three groups of porous sponge-like scaffolds were prepared: (1) porcine type I/III collagen (CI) (0.5% w/v) (Geistlich Biomaterials, Wolhusen, Switzerland); (2) CI (0.5%) additionally supplemented with CS (7% w/w relative to CI) (Sigma Chemical Co., St Louis,

2.2. Cell isolation and two-dimensional monolayer expansion MSCs were isolated from heparinized bone marrow aspirates from the iliac crests of two adult Spanish goats as previously described [28]. All of the groups in each of the three experiments (described in Section 2.3) used cells from the same animal. Adherent cells were expanded in monolayer using a standard MSC expansion medium consisting of low glucose Dulbecco’s modified Eagle’s medium (DMEM-LG) (Invitrogen), containing fetal bovine serum (FBS, 10% v/v) (Invitrogen), penicillin/streptomycin (PS, 1%) (Invitrogen), and fibroblast growth factor (FGF)-2 (10 ng ml–1) (R&D Systems, Minneapolis, MN, USA). The cells were incubated in a humidified chamber at 37 °C, carbon dioxide (5%), and atmospheric oxygen. MSCs were grown through two subcultures to obtain passage 2 (P2) cells. 2.3. Experimental design P2 caprine MSCs were transfected with pEndo-containing lipoplex in monolayer and then seeded into the hydrogels in the first

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

MO, USA); and (3) CI (0.5%) additionally supplemented with HS (7% w/w relative to CI) (Sigma). Collagen powder, and CS and HS GAGs as appropriate, were blended in hydrochloric acid (0.001 N). The slurry was degassed and poured into molds, and porous spongelike sheets (1.5 mm thick) were fabricated by freeze-drying (VirTis, Gardiner, NY, USA) and sterilized using dehydrothermal treatment as previously described [6]. Disks (8 mm in diameter and 1.5 mm in thickness) were cut from the porous sheets using a dermal biopsy punch. 2.5. Transfection with GenePORTER 2/endostatin plasmid complexes (lipoplexes) and cell seeding and culture Separate plasmids encoding for human endostatin and for GFP were encapsulated in a lipid-mediated transfection reagent, GenePORTER™ 2 (GP2; Gene Therapy Systems, Inc., San Diego, CA, USA) following the manufacturer’s instructions. For the gels (experiments # 1 and 3), P2 MSCs were transfected in monolayer using the pEndo lipoplexes, using a ratio of 5 lg pEndo per 1 million cells. The cells were incubated in DMEM-LG for 4 h to allow for transfection, and then expansion medium (with no FGF-2) was added to the cell flasks for overnight incubation. In the first experiment, the day after transfection, the transfected cells were added to the hydrogel mixtures (cross-linked using TG-2, mTG, or GP) at a seeding density of 0.8 million cells ml–1. No beads were added to any of the hydrogels. Control constructs seeded with non-transfected cells were also prepared. In the third experiment, the day after transfection, the transfected cells were added to the different hydrogel groups (no additional cross-linker control with no beads, GP-cross-linked control with no beads, GP and control beads, and GP and heparin-agarose beads) at a seeding density of 0.8 million cells ml–1. Additional control groups were prepared: (1) constructs with no beads, seeded with pGFP lipoplex-transfected MSCs (5 lg pGFP/1 million cells) and (2) genipin-cross-linked non-cell-seeded constructs. For all gel constructs (experiments # 1 and 3), hydrogels were cast by pipetting 0.5 ml of the cell-seeded mixtures into each well of a 24-well plate and incubating at 37 °C for gelation, to create constructs with a diameter of 16 mm and a thickness of 2 mm. The constructs were incubated overnight in expansion medium (with no FGF-2). In the second experiment, lipoplexes were prepared and incorporated into the porous sponge-like scaffolds, 20 lg of pEndo per scaffold, using two steps. Half of the lipoplex solution was pipetted on one side of the scaffold and incubated for 10 min. The other half of the lipoplex solution was added to the second side of the scaffold and incubated for 10 min. A 1 ml aliquot of an aqueous carbodiimide solution consisting of 0.6 mM 1-ethyl-3-(3dimethylaminopropyl) carbodiimide hydrochloride and 0.6 mM N-hydroxysuccinimide (Sigma) was added to the scaffold, followed by incubation for 30 min. Excess carbodiimide was removed by soaking the scaffolds in PBS for 1 h. The sponge-like scaffolds were placed in agarose-coated wells for cell seeding. P2 MSCs were trypsinized and resuspended in DMEM-LG. Two million MSCs were pipetted onto each side of the scaffold, with an incubation period of 10–30 min for each side, for a total of 4 million cells seeded per scaffold, and using the same ratio of plasmid to cells (5 lg pEndo/1 million cells) as was used for the gels. All constructs were then incubated in DMEM-LG to allow for 4 h of transfection. CI scaffolds with no lipoplex served as controls. All constructs, gels and sponges, were switched to chondrogenic medium (CM), 4 h after cell seeding for sponges and the day after cell seeding for gels, and cultured in CM for the remainder of the experiments. The CM was composed of high glucose Dulbecco’s modified Eagle’s medium (DMEM-HG) (Invitrogen), nonessential amino acids (1% v/v) (Invitrogen), HEPES buffer (1%) (Invitrogen),

2205

penicillin/streptomycin/L-glutamine (1%) (Invitrogen), bovine serum albumin (BSA, 1.25 mg ml–1) (Invitrogen), 1X ITS+1 (Sigma), TGF-b1 (10 ng ml–1) (R&D Systems), dexamethasone (100 nM) (Sigma), and L-ascorbic acid 2-phosphate (0.1 mM) (Wako). Every 1–3 days, expended medium was collected and frozen at –20 °C until analysis, and fresh CM was added. The sponge-like scaffold constructs were cultured in 1.5 ml of medium, and the hydrogel constructs were cultured in 1 ml of medium. Cultures were terminated on select days for histological examination and biochemical analysis. 2.6. Endostatin detection in the constructs and medium Select constructs were collected for homogenization. Homogenizing medium was prepared by dissolving 1 Complete Mini protease inhibitor cocktail tablet (Roche, Indianapolis, IN, USA) per 7 ml of PBS. Constructs were cut into small pieces, as needed, using a safety razor blade. Each construct was placed in a homogenizing vessel containing 500 ll of homogenizing medium and homogenized using the Tissue-Tearor tissue homogenizer (Biospec, Bartlesville, OK, USA) at 30,000 rpm for 1 min on ice. The homogenate was then centrifuged at 10,000–13,000 rcf for 5– 8 min to remove debris and insoluble materials, and the supernatant was stored at 20 °C until analysis. Endostatin protein in the homogenized constructs and in the expended culture medium was measured using a sandwich enzyme-linked immunosorbent assay (ELISA) kit for human endostatin protein (R&D Systems, Minneapolis, MN) following the kit instructions. 2.7. Analysis of DNA and GAG content Constructs were lyophilized and enzymatically digested overnight using proteinase K (Roche Diagnostics, Indianapolis, IN). Determination of the DNA content was carried out using the Picogreen dye assay kit (Molecular Probes, Inc., Eugene, OR, USA) according to the manufacturer’s instructions. The number of cells was estimated from the DNA content using an average value of 5.6 pg DNA per cell for goat MSCs [30]. The sulfated GAG content was determined by the dimethylmethylene blue (DMMB) dye assay, with a standard curve obtained using chondroitin-6-sulfate from shark cartilage (Sigma). 2.8. Histological and immunohistochemical evaluation Constructs allocated for histology were fixed in paraformaldehyde (4% in phosphate buffered saline) for at least 3 h at 4 °C, processed and embedded in paraffin, and sectioned into 6 lm thick sections by microtomy. The sections were mounted on glass slides and stained with Safranin-O using standard histological techniques. Immunohistochemical evaluation was carried out. Endostatin was examined using an anti-endostatin rabbit polyclonal antibody (final concentration 17 lg ml–1, Millipore, Billerica, MA, USA). Type II collagen distribution was examined immunohistochemically using an anti-type II collagen mouse monoclonal antibody (CIIC1, final concentration 4 lg ml–1, Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA, USA). Laminin and type IV collagen, common basement membrane molecules, were also examined immunohistochemically, using an anti-laminin rabbit polyclonal antibody (final concentration 27 lg ml–1, Abcam) and an anti-type IV collagen rabbit polyclonal antibody (final concentration 17 lg ml–1, Abcam), respectively. The immunohistochemical staining was performed using the Dako Autostainer (DakoCytomation, Carpinteria, CA, USA) and the peroxidase-aminoethyl

2206

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

carbazole (AEC)-based Envision + kit (Dako) following the manufacturer’s recommendations. 2.9. Statistical analysis Data are presented as the mean ± standard error of the mean (SEM). Analysis of variance (ANOVA) and Fisher’s protected least significant difference (PLSD) post hoc testing were performed using StatView software (SAS Institute Inc, Cary, NC, USA). Statistical significance was set at P < 0.05. 3. Results Our experience with the seeding methods indicated that 65% of the cells were retained in the porous sponge-like scaffold and 100% of the cells were noted in the hydrogels, 1 day after seeding [6]. 3.1. MSC-seeded hydrogels treated with different cross-linking agents 3.1.1. Endostatin detection in the expended medium The endostatin levels in the medium from the cultures of the three types of cross-linked gels (TG-2, mTG, and GP) seeded with the non-transfected MSCs (in the first experiment), and from non-cross-linked and GP-cross-linked gels seeded with MSCs transfected in monolayer with green fluorescent protein plasmid (pGFP) (in the third experiment) were negligible (data not shown). Endostatin was detected in the expended medium of hydrogel constructs seeded with endostatin plasmid (pEndo) lipoplex-transfected MSCs for all collection periods throughout the 21-day period of the experiment (Fig. 1a). The profile of the release showed a peak in the endostatin content of the medium within the first week, gradually decreasing to a relatively steady level after 14 days (Fig. 1a). The highest endostatin level of the gel experiments, 18 ng ml–1 (a total of 18 ng of endostatin in the expended medium), was measured on the 3-day collection period ending on day 4, for the GP-cross-linked group (Fig. 1a). The cumulative endostatin level at 21 days was highest for the GP-cross-linked group (64 ng ml–1) and lowest for mTG-cross-linked group (28 ng ml–1); the cumulative release for the TG-2-cross-linked group was 51 ng ml–1. Including all pEndo lipoplex-transfected groups, 2-factor ANOVA indicated significant effects of cross-linking agent and collection period (P < 0.0001, power = 1 for both) on endostatin levels. Fisher’s PLSD post hoc testing demonstrated significant differences among all three cross-linking agent groups. 3.1.2. Biochemical analysis of DNA and GAG content Biochemical analyses of samples terminated after 21 days of culture revealed similar cell numbers among the groups, between 1.2 and 1.5 million cells per construct (Fig. 2a). These values represented a threefold increase in cell number over the 21-day period of the study, relative to the 0.4 million cells which were initially incorporated into the gels. There were no notable differences among the different cross-linking agents with respect to cell number. Two-factor ANOVA failed to find significant effects of transfection (P = 0.22, power = 0.21) or cross-linking agent (P = 0.55, power = 0.13) on cell number. For this experiment, the amount of GAG was recorded per volume, as a percentage of that of native goat articular cartilage. The GAG content per volume of native goat articular cartilage has previously been shown to be 15.8 lg mm–3 [31]. One day after seeding, no notable GAG was measured in the constructs (data not shown). All groups displayed substantial amounts of GAG after 3 weeks of culture, with a wide range from 64% to over 200% that of

Fig. 1. Endostatin measured for the transfected cell-seeded constructs (mean ± SEM). (a) Expended medium from collagen hydrogel constructs containing monolayer-transfected MSCs and cross-linked using TG-2, mTG, and GP (n = 6). (b) Endostatin measured in the expended medium of cell-seeded collagen sponge-like scaffolds (CI), and CI constructs supplemented with chondroitin sulfate and heparan sulfate (n = 2–4), which incorporated lipoplex containing 20 lg of pEndo. (c) Endostatin measured in the expended medium of monolayer-transfected cellseeded GP-cross-linked collagen hydrogel constructs incorporating heparin-agarose beads (n = 6). Control groups include the same type of gel containing agarose beads and no beads. An additional control group consisted of cell-seeded gel with no GP cross-linking and no beads.

native cartilage (Fig. 2b). The non-transfected, GP-cross-linked group displayed a remarkably high GAG percentage (more than twofold higher) compared to the other groups (Fig. 2b). Two-factor ANOVA found significant effects of transfection (P < 0.0001, power = 1) and cross-linking agent (P < 0.0001, power = 1) on GAG percentage. Fisher’s post hoc PLSD testing demonstrated a significant difference between the GP-cross-linked group and the other two groups. 3.1.3. Gross and histological evaluation of the constructs All of the 3-week gels displayed cells of chondrocytic (rounded and in lacunae) and fibroblastic (elongated) morphology. Qualitative observation for endostatin in the constructs revealed only a

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

Fig. 2. Biochemical analyses of MSC-seeded collagen gels cross-linked using TG-2, mTG, and GP, after 21 days of culture (mean ± SEM). (a) Cell numbers, estimated from the DNA content (n = 3). (b) Percentage GAG content per volume compared to native articular cartilage (n = 3).

few endostatin-containing cells on select days throughout the culture period, including at the 3-week time period (Fig. 3a), and virtually no endostatin in the extracellular matrix, indicating that no endostatin protein was retained within the constructs. Constructs cultured for 3 weeks demonstrated areas of positive staining for type II collagen (representative construct seen in Fig. 3b) and sulfated GAGs (representative construct seen in Fig. 3c) in the extracellular matrix of the construct, generally in regions occupied by cells of chondrocyte morphology. Both laminin and type IV collagen, common basement membrane molecules, were observed in the constructs during the 3week culture period, with similar staining patterns. One day after seeding, positive intracellular staining for both molecules was noted for many of the cells in the constructs (Fig. 3d and g). After 6 days, diffuse staining was observed throughout much of the extracellular matrix (Fig. 3e and h). Many cells appeared elongated, and intracellular staining of several of the cells was still apparent after 6 days (Fig. 3e and h). At 3 weeks, intense staining was seen in the interior of the constructs, in the interterritorial matrix surrounding rounded cells in lacunae, as well as in the extracellular matrix of the thin layer at the periphery of the constructs (Fig. 3f and i). 3.2. Sponge-like scaffolds and gels supplemented with GAGs 3.2.1. MSC-seeded porous sponges supplemented with HS and CS The MSC-seeded scaffold control group, which did not incorporate lipoplex, displayed no notable endostatin protein in the medium (Fig. 1b). Endostatin levels in the medium from the cell-seeded lipoplex-incorporating scaffold with no GAG supplementation remained elevated at or above 5 ng ml–1 for all of the collection

2207

periods throughout the 20-day experiment. For all of the lipoplex-containing scaffolds, the endostatin levels in the medium increased to a peak in the 3-day collection period ending on day 4, and then gradually decreased (Fig. 1b). The highest endostatin level in the experiment was observed to be 29 ng ml–1 at the three-day collection period ending on day 4, for the lipoplex-incorporating CI control group (Fig. 1b). The endostatin content of the medium from the two groups that received GAG supplementation were comparable, and the levels of both groups were substantially lower (25% or less) than those of the group that did not receive additional GAG supplementation through 15 days of the experiment (Fig. 1b). The cumulative endostatin level at day 20 was much higher for the non-GAG-supplemented group (97 ng ml–1) than for both GAGsupplemented groups (28 ng ml–1 for CI-CS, 9 ng ml–1 for CI-HS). Two-factor ANOVA of the lipoplex-incorporating groups revealed significant effects of collection period (P = 0.0012, power = 0.98) and GAG supplementation (P < 0.0001, power = 1) on endostatin levels in the expended medium. Fisher’s PLSD post hoc testing demonstrated a significant difference between the CI control constructs and the two GAG-supplemented constructs. Four days after seeding, cells could be seen in the pores of the sponge-like scaffold, often clustered together (Fig. 4). Positive staining for endostatin for a few cells was seen in the lipoplexincorporating constructs at day 4 (Fig. 4). Similar sparse staining was seen at days 14, 20, and 28 (data not shown). No staining was observed in the controls that did not receive lipoplex (data not shown). Analysis of homogenized 20-day samples from all of the groups revealed only barely detectable amounts of endostatin retained in the constructs, indicating that the chondroitin and heparan sulfates that were added to the collagen scaffolds did not bind the endostatin. Subsequent analysis was then carried out to determine the amount of GAG that was contained in the collagen sponge-like scaffolds after the DHT and carbodiimide treatment, just prior to cell seeding. The CI scaffolds that were not additionally supplemented with GAGs showed the lowest amount of GAG at 6.97 ± 0.01 lg, near the lower limit of detection for the DMMB assay. The CS-incorporating and HS-incorporating scaffolds had 9.03 ± 0.78 and 11.14 ± 1.23 lg of GAG respectively, compared to the 150 lg of GAG added per scaffold, indicating a lack of retention of the supplemented GAGs in the scaffolds. 3.2.2. Gels incorporating heparin Since the GAG-supplemented sponge-like scaffolds resulted in little GAG retention just prior to cell seeding, a modified protocol using heparin (obtained commercially, already bound to the surface of agarose beads) was developed for endostatin protein retention in hydrogels, to ensure that GAG was retained in the constructs at the time of cell seeding. Agarose beads without heparin (commercially obtained) were also used in the gels as a control group. Immediately after seeding, rounded cells (Fig. 5a–c) were noted in all groups, and beads could be seen dispersed throughout the hydrogels of the bead-incorporating groups (Fig. 5b and c). Beads remained visible in the hydrogels throughout the culture period (Fig. 5). Analysis was carried out to determine the amount of supplemented GAG (from the heparin-agarose beads) retained in the constructs immediately after cell-seeding and during the first week of culture. For the heparin-agarose bead group, 70 ± 4 lg of GAG was measured in the constructs immediately after cell seeding and gel casting, compared to the 100 lg of heparin which was bound to the agarose beads added to the gel. After one day, the non-cell-seeded heparin-agarose control group had 69 ± 5 lg of GAG, compared to less than 6 lg GAG detected in the gels with no beads and in the gels with the agarose beads. But after 1 and 7 days of culture the cell-seeded heparin-agarose groups had only 15 lg of GAG per

2208

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

Fig. 3. Micrographs of TG-2-cross-linked hydrogel constructs. (a) Endostatin immunohistochemistry (arrow, red chromogen) of TG-2-cross-linked gel seeded with transfected MSCs, 21 days after seeding; scale bar, 100 lm. Inset shows higher magnification of positive staining; scale bar, 5 lm. (b) Type II collagen immunohistochemistry (red) of GP-cross-linked gel seeded with nontransfected MSCs, 21 days after seeding. Scale bar, 100 lm. (c) Safranin-O staining (red) of GP-cross-linked gel seeded with nontransfected MSCs, 21 days after seeding. Scale bar, 500 lm. (d) Laminin immunohistochemistry (red) of TG-2-cross-linked gel seeded with transfected MSCs, 1 day after seeding. Scale bar, 100 lm. Inset shows higher magnification of positive staining; scale bar, 10 lm. (e) Laminin immunohistochemistry (red) of TG-2-cross-linked gel seeded with transfected MSCs, 6 days after seeding. Scale bar, 100 lm. (f) Laminin immunohistochemistry (red) of TG-2-cross-linked gel seeded with transfected MSCs, 21 days after seeding. Scale bar, 100 lm. (g) Type IV collagen immunohistochemistry (red) of TG-2-cross-linked gel seeded with transfected MSCs, 1 day after seeding. Scale bar, 100 lm. Inset shows higher magnification of positive staining; scale bar, 10 lm. (h) Type IV collagen immunohistochemistry (red) of TG-2-cross-linked gel seeded with transfected MSCs, 6 days after seeding. Scale bar, 100 lm. (i) Type IV collagen immunohistochemistry (red) of TG-2-cross-linked gel seeded with transfected MSCs, 21 days after seeding. Scale bar, 100 lm.

Fig. 4. Endostatin immunohistochemistry (arrows, red chromogen) of CI control construct (a), CI-CS construct (b), and CI-HS construct (c), 4 days after seeding; scale bar, 100 lm. Inset shows higher magnification of positive staining; scale bar, 5 lm.

construct. After 1 day all of the other cell-seeded gel groups displayed 50% of the GAG content of the cell-seeded heparinagarose bead group. At 7 days the GAG contents of the other cell-seeded gel groups increased to between 50% and 90% of the cell-seeded heparin-agarose group, perhaps reflecting some newly synthesized GAG. The endostatin levels in the cultures of the control non-crosslinked and GP-cross-linked gels seeded with cells transfected with the pGFP were negligible (data not shown). The endostatin levels in the medium of the hydrogel constructs seeded with pEndo lipoplex-transfected MSCs remained elevated through the 21-day experiment (Fig. 1c). The release profile displayed a peak in the first week of culture and then decreased (Fig. 1c), similar to the

endostatin expression profile seen in the first gel experiment using cells from another goat and in the second experiment using the sponge-like scaffolds. The highest endostatin level of the experiment was 11 ng ml–1 (a total of 11 ng of endostatin in the expended medium), measured on the 3-day collection period ending on day 5, for the genipin-cross-linked group with no beads (Fig. 1c). The cumulative endostatin levels at day 21 ranged from 31 ng ml–1 (heparin-agarose bead group) to 36 ng ml–1 (control agarose bead group). Including the four groups with no beads, 3-factor ANOVA revealed significant effects of plasmid (pEndo vs. pGFP) (P < 0.0001, power = 1) and collection period (P < 0.0001, power = 1) on endostatin levels in the expended medium, but did not find an effect

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

2209

Fig. 5. Hydrogel constructs seeded with pEndo lipoplex-transfected MSCs and cross-linked with genipin, with and without beads. Top row, light microscopy images immediately after cell seeding. Each image in row is represented at identical magnification; scale bar, 200 lm. Endostatin immunohistochemistry (arrows, red chromogen) 1 day after seeding (second row), 7 days after seeding (third row), and 21 days after seeding (bottom row). Each immunohistochemistry image is represented at identical magnification; scale bar, 50 lm.

of cross-linking agent (P = 0.74, power = 0.06). Including the three genipin-cross-linked groups seeded with pEndo lipoplex-transfected cells, 2-factor ANOVA revealed significant effects of collection period (P < 0.0001, power = 1) and bead incorporation (P = 0.046, power = 0.59) on endostatin levels. Fisher’s PLSD post hoc testing demonstrated a significant difference between the control agarose bead group and the heparin-agarose bead group. Endostatin amounts within the homogenized constructs were very small (data not shown), ranging from 0.06 to 0.41 ng of protein, when compared to the endostatin amounts in the expended medium. The amount of endostatin for the pGFP controls was negligible. Immunohistochemical analysis for endostatin in the constructs revealed a few cells showing positive staining 1 day after seeding for all pEndo lipoplex-transfected groups (Fig. 5d–f). At day 7 after seeding, sparse positive staining was seen in the two bead-incorporating groups (Fig. 5h and i) but not in the non-bead-supplemented group (Fig. 5g), and by day 21, virtually no positive staining was noted (Fig. 5j–l).

4. Discussion In this study, we engineered endostatin-expressing cartilaginous constructs in vitro using hydrogels capable of being injected

and undergoing covalent cross-linking with transglutaminase-2, microbial transglutaminase, and genipin. Using only a small amount of pEndo (5 lg of plasmid per 1 million cells) for the transfection of MSCs, for safety considerations, we were able to prepare injectable gels that overexpressed endostatin for 3 weeks. General trends in the expression profiles of the expended medium were similar among the different groups, and similar to those seen with the sponge-like scaffolds in this study, as well as in our previous work [6]. Overexpression displayed peak values (ranging from 7 to 18 ng ml–1), shown in prior studies [32,33] to be therapeutic levels, within the first week. Of note is that these injectable gels containing endostatin-expressing MSCs may be of value for other clinical applications, including the treatment of cancerous tumors. Little staining for endostatin was seen in the constructs, and much higher endostatin amounts were detected in the expended media than in the homogenized constructs, suggesting that little protein was retained in the engineered constructs, for both gels and sponges. We report on the distribution of two common basement membrane molecules, laminin and type IV collagen, in engineered cartilaginous constructs. It is important in the context of this study because type XVIII collagen is generally associated with basement membranes in vivo, and the likely reason that the basement membrane serves as a barrier to vascular invasion. The diffuse staining pattern seen in this study is different from the localized pericellular

2210

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

matrix staining that has previously been reported in normal articular cartilage [29], demonstrating that the constructs being engineered in vitro are not the same as normal cartilage. However, the staining pattern seen here is reminiscent of that seen in cartilage development [29,34] and disease [35]. This discovery may provide insight into the development of tissue-engineered constructs. Of note was that the endostatin released by the MSCs did not associate with the basement membrane molecules in the developing cartilaginous constructs, as one might have expected based on the co-localization of type XVIII collagen with type IV collagen and laminin in vivo. It may be that such binding occurs during the maturation of a basement membrane structure that did not occur within the time frame of this study, or that another region of type XVIII collagen is responsible for binding with other basement membrane molecules. It will be interesting to address this issue in long-term investigations. In experiment # 1, the cross-linking agents affected the endostatin levels in the expended medium, with 0.25 mM GP resulting in the highest cumulative endostatin, and 100 lg ml–1 mTG resulting in the lowest. This finding suggests that cross-linking agents may affect the development of endostatin-producing constructs, which is in agreement with other studies that suggest that crosslinking agents can alter cell metabolism [36] and modulate the nature of ECM deposition [17]. The MSC-seeded gels demonstrated the capacity to undergo chondrogenic differentiation, as indicated by the notable GAG densities for all groups at the end of the 3-week culture period in the first experiment. This is in agreement with other studies demonstrating chondrogenesis in gels cross-linked using transglutaminase [16] and GP [17,18]. Qualitatively, there were areas in the constructs that showed signs of chondrogenesis (positive staining for type II collagen and sulfated GAGs and chondrocyte morphology). In this study, GP-cross-linked constructs were found to have a higher percentage of sulfated GAGs compared to the transglutaminase-cross-linked constructs. For experiment # 1, the cell numbers at 3 weeks were comparable for all groups and were increased compared to the initial seeding density, indicating that the cells were undergoing proliferation. No single cross-linking agent stimulated goat MSC proliferation compared to the others. In experiment # 2, low retention of HS and CS was observed in the GAG-supplemented sponges. The small amount of GAG (9 lg per scaffold) measured in the sponges just prior to cell seeding was unexpected, as previous work has demonstrated the ability to fabricate scaffolds with GAG compositions similar to the GAG percentage intended for this study (7 wt.%) [37]. It has been shown that chondroitin sulfates and heparan sulfates vary depending on the species of origin and the manufacturing process used to obtain the GAG, including differences in molecular weights, and evidence from previous work suggests that these differences can affect the amount of GAG that is grafted on the collagen [37–39]. It is possible that only small amounts of the GAGs used in this study formed cross-links with the collagen during DHT treatment, and that subsequent loss of HS and CS GAGs due to elution may have occurred when the sponges were immersed in an aqueous carbodiimide solution. The finding from this experiment suggests that not all GAGs lend themselves to cross-linking well to collagen. The lack of HS and CS retention in the collagen scaffolds may have contributed to the differences in endostatin levels in the expended medium of experiment # 2. It is possible that some of the pEndo lipoplexes may have bound to the HS and CS that were eluted from sponges, resulting in less pEndo lipoplexes for transfection compared to the CI control sponges and contributing to the lower levels observed for the GAG-supplemented constructs than for the non-supplemented constructs at all collection periods.

The addition of HS and CS in the second experiment of this study did not have a beneficial effect on endostatin retention as hypothesized, also likely due to the lack of HS and CS in the scaffolds for endostatin binding. Positive staining for endostatin was seen for only a few cells in the GAG-supplemented sponges, suggesting that the protein amounts in the engineered constructs were negligible, and qualitatively, there were no notable differences among the three sponge groups. Hydrogels incorporating heparin-agarose beads were successfully cast in the third experiment, and analysis of the immobilized heparin content revealed that the majority of the heparin that was added was present in the gels. However, the addition of heparin did not have a beneficial effect on endostatin retention in the hydrogels. Biochemical analysis of the gels indicated that when cells were also added into the hydrogel constructs, the amount of heparin drastically decreased even though the beads were still present. The addition of cells genetically modified to deliver protein, rather than direct delivery of the protein in conventional protein retention studies, added to the complexity of the system and may have contributed to the decrease in GAG levels after cell seeding. It is possible that the cells may have internalized the heparin [40] or released molecules that degraded the heparin [41,42]. While the heparin-agarose bead-incorporating constructs in this study did not aid in endostatin retention using engineered MSCs, the heparin-agarose bead-incorporating gels could be of potential benefit for other retention applications using direct delivery of the proteins. There are strengths and weaknesses of sponge-like scaffolds and hydrogels. The gel is injectable, resulting in less surgical trauma compared to sponge implantation. However, a lipoplex-supplemented sponge-like scaffold may provide the utility of a one-step off-the-shelf transfection construct, compared to the current two-step process of monolayer transfection and subsequent scaffold seeding for the hydrogels. Further research is needed to investigate the possibility of a one-step in situ transfection process for gels. The endostatin levels in the expended medium of monolayertransfected (experiments #1 and 3) vs. scaffold-transfected (experiment # 2) constructs were of the same order of magnitude. However, a much larger number of cells was used with sponges than with gels. On a per cell basis, the data suggest that monolayertransfected cells were producing, on average, more endostatin than scaffold-transfected cells. The cellular environment during DNA uptake – endostatin plasmid lipoplexes were presented in a three-dimensional scaffold environment in the sponge-like scaffold experiment and in a two-dimensional monolayer environment in the hydrogel experiment – may have contributed to the differences in endostatin production per cell, possibly by affecting the presentation/availability of the lipoplexes or by directly affecting cell behavior [5,43]. It is important to note that there were differences in other experimental conditions between the sponge experiments and the gel experiments, and further studies are needed to determine how cell–matrix interactions may influence nonviral gene transfection. It is of interest to compare monolayer-transfected cells in gels (experiments # 1 and 3) vs. sponges (previous work in our lab [5]). In our previous study [5], our lab examined 0.1 million monolayer-transfected MSCs (10–50 lg pEndo per 1 million cells) seeded per sponge-like scaffold and found peak endostatin levels ranging from 3 to 12 ng ml–1. In this work, 0.4 million monolayer-transfected MSCs (5 lg pEndo per 1 million cells) seeded per gel resulted in peak endostatin levels ranging from 6 to 18 ng ml–1. The ranges are similar between the sponges and the gels, and the preliminary data may begin to suggest that scaffold form does not have a significant effect on endostatin production of already transfected cells. However, the differences in number

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212

of cells used and pEndo/cell ratios make it difficult to draw conclusions, and further investigation is needed in this area. 5. Conclusions In this study, we engineered endostatin-producing cartilaginous constructs using covalently cross-linked hydrogels, demonstrating that gels are a feasible scaffold option. Nonviral gene therapy resulted in locally elevated levels of protein compared to nontransfected controls, as evidenced in this study as well as in the literature [6,27]. Most of the endostatin protein was released into the surrounding media and was not retained within the constructs. Of the conditions tested in experiment # 1, 0.25 mM GP yielded the best results; however, TG-2 and mTG should be not ruled out as alternative cross-linking agents, as they also resulted in cartilaginous constructs overexpressing endostatin protein. The ability to engineer novel endostatin-producing constructs, which may be used primarily for their angiogenesis inhibition properties, has important implications not only for articular cartilage tissue engineering, but also for other avascular tissues, for cancer research applications, and for tissues in which angiogenesis inhibition is desired. For example, prominent vascular infiltration has been implicated in the failure of tissue-engineered auricular cartilage [44], and an endostatin-producing construct could be of value for this application. The role of angiogenesis in tumor growth and the importance of angiogenesis regulation have also been recognized for years in the cancer field [45]. Given that endostatin is one of the most widely researched angiogenesis inhibitors in this area, it follows that a novel endostatin-producing construct could be of interest for this field. These gel-based constructs will be of value for future in vivo work investigating the use of endostatin for cartilage repair, as they can be implanted into focal cartilage defects to provide cartilaginous material for subsequent in vivo maturation and endostatin-expressing cells to prevent vascular invasion. 6. Disclosure The authors have nothing to disclose. Acknowledgements This work was supported by the Rehabilitation Research and Development Service of the US Department of Veterans Affairs, the Department of Defense, the National Science Foundation Graduate Research Fellowship (L. Jeng), and the Siebel Scholarship (L. Jeng). M. Spector is a VA Research Career Scientist. The funding sources had no involvement in study design, data collection and analysis, or report writing. The authors are grateful for the assistance of Alix Weaver, and would like to thank Ajinomoto Food Ingredients LLC (Chicago, Illinois) for providing the microbial transglutaminase. Appendix A. Figures with essential colour discrimination Certain figures in this article, particularly Figs. 3–5, are difficult to interpret in black and white. The full colour images can be found in the on-line version, at doi:10.1016/j.actbio.2012.02.015. References [1] Yin M, Pacifici M. Vascular regression is required for mesenchymal condensation and chondrogenesis in the developing limb. Dev Dyn 2001;222:522–33. [2] Fenwick SA, Gregg PJ, Rooney P. Osteoarthritic cartilage loses its ability to remain avascular. Osteoarthritis Cartilage 1999;7:441–52.

2211

[3] Kubo S, Cooper GM, Matsumoto T, Phillippi JA, Corsi KA, Usas A, et al. Blocking vascular endothelial growth factor with soluble Flt-1 improves the chondrogenic potential of mouse skeletal muscle-derived stem cells. Arthritis Rheum 2009;60:155–65. [4] Matsumoto T, Cooper GM, Gharaibeh B, Meszaros LB, Li G, Usas A, et al. Cartilage repair in a rat model of osteoarthritis through intraarticular transplantation of muscle-derived stem cells expressing bone morphogenetic protein 4 and soluble Flt-1. Arthritis Rheum 2009;60:1390–405. [5] Sun X-D, Jeng L, Bolliet C, Olsen BR, Spector M. Non-viral endostatin plasmid transfection of mesenchymal stem cells via collagen scaffolds. Biomaterials 2009;30:1222–31. [6] Jeng L, Olsen BR, Spector M. Engineering endostatin-producing cartilaginous constructs for cartilage repair using nonviral transfection of chondrocyteseeded and mesenchymal-stem-cell-seeded collagen scaffolds. Tissue Eng Part A 2010;16:3011–21. [7] Stevens MM, Marini RP, Schaefer D, Aronson J, Langer R, Shastri VP. In vivo engineering of organs: the bone bioreactor. Proc Natl Acad Sci USA 2005;102:11450–5. [8] O’Reilly MS, Boehm T, Shing Y, Fukai N, Vasios G, Lane WS, et al. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell 1997;88:277–85. [9] Stempel J, Fritsch H, Pfaller K, Blumer MJ. Development of articular cartilage and the metaphyseal growth plate: the localization of TRAP cells, VEGF, and endostatin. J Anat 2011;218:608–18. [10] Pufe T, Petersen WJ, Miosge N, Goldring MB, Mentlein R, Varoga DJ, et al. Endostatin/collagen XVIII—an inhibitor of angiogenesis—is expressed in cartilage and fibrocartilage. Matrix Biol 2004;23:267–76. [11] Feng Y, Wu YP, Zhu XD, Zhang YH, Ma QJ. Endostatin promotes the anabolic program of rabbit chondrocyte. Cell Res 2005;15:201–6. [12] Sundararaghavan HG, Monteiro GA, Lapin NA, Chabal YJ, Miksan JR, Shreiber DI. Genipin-induced changes in collagen gels: correlation of mechanical properties to fluorescence. J Biomed Mater Res Part A 2008;87A:308–20. [13] Pek YS, Spector M, Yannas IV, Gibson LJ. Degradation of a collagenchondroltin-6-sulfate matrix by collagenase and by chondroitinase. Biomaterials 2004;25:473–82. [14] Faye C, Inforzato A, Bignon M, Hartmann DJ, Muller L, Ballut L, et al. Transglutaminase-2: a new endostatin partner in the extracellular matrix of endothelial cells. Biochem J 2010;427:467–75. [15] Kashiwagi T, Yokoyama K, Ishikawa K, Ono K, Ejima D, Matsui H, et al. Crystal structure of microbial transglutaminase from Streptoverticillium mobaraense. J Biol Chem 2002;277:44252–60. [16] McHale MK, Setton LA, Chilkoti A. Synthesis and in vitro evaluation of enzymatically cross-linked elastin-like polypeptide gels for cartilaginous tissue repair. Tissue Eng 2005;11:1768–79. [17] Lima EG, Tan AR, Tai T, Marra KG, DeFail A, Ateshian GA, et al. Genipin enhances the mechanical properties of tissue-engineered cartilage and protects against inflammatory degradation when used as a medium supplement. J Biomed Mater Res Part A 2009;91A:692–700. [18] Dare EV, Griffith M, Poitras P, Kaupp JA, Waldman SD, Carlsson DJ, et al. Genipin cross-linked fibrin hydrogels for in vitro human articular cartilage tissue-engineered regeneration. Cells Tissues Organs 2009;190:313–25. [19] Zheng L, Fan HS, Sun J, Chen XN, Wang G, Zhang L, et al. Chondrogenic differentiation of mesenchymal stem cells induced by collagen-based hydrogel: an in vivo study. J Biomed Mater Res A 2010;93:783–92. [20] Breinan HA, Martin SD, Hsu HP, Spector M. Healing of canine articular cartilage defects treated with microfracture, a type-II collagen matrix, or cultured autologous chondrocytes. J Orthop Res 2000;18:781–9. [21] Sommer A, Rifkin DB. Interaction of heparin with human basic fibroblast growth-factor—protection of the angiogenic protein from proteolytic degradation by a glycosaminoglycan. J Cell Physiol 1989;138:215–20. [22] Pieper JS, Hafmans T, van Wachem PB, van Luyn MJA, Brouwer LA, Veerkamp JH, et al. Loading of collagen-heparan sulfate matrices with bFGF promotes angiogenesis and tissue generation in rats. J Biomed Mater Res 2002;62:185–94. [23] Ricard-Blum S, Feraud O, Lortat-Jacob H, Rencurosi A, Fukai N, Dkhissi F, et al. Characterization of endostatin binding to heparin and heparan sulfate by surface plasmon resonance and molecular modeling—role of divalent cations. J Biol Chem 2004;279:2927–36. [24] Sasaki T, Larsson H, Kreuger J, Salmivirta M, Claesson-Welsh L, Lindahl U, et al. Structural basis and potential role of heparin/heparan sulfate binding to the angiogenesis inhibitor endostatin. Embo J 1999;18:6240–8. [25] Hohenester E, Sasaki T, Olsen BR, Timpl R. Crystal structure of the angiogenesis inhibitor endostatin at 1.5 A resolution. Embo J 1998;17:1656–64. [26] Faye C, Chautard E, Olsen BR, Ricard-Blum S. The first draft of the endostatin interaction network. J Biol Chem 2009;284:22041–7. [27] Capito RM, Spector M. Collagen scaffolds for nonviral IGF-1 gene delivery in articular cartilage tissue engineering. Gene Ther 2007;14:721–32. [28] Vickers SM, Squitieri LS, Spector M. Effects of cross-linking type II collagenGAG scaffolds on chondrogenesis in vitro: dynamic pore reduction promotes cartilage formation. Tissue Eng 2006;12:1345–55. [29] Kvist AJ, Nystrom A, Hultenby K, Sasaki T, Talts JF, Aspberg A. The major basement membrane components localize to the chondrocyte pericellular matrix—a cartilage basement membrane equivalent? Matrix Biol 2008;27:22–33. [30] Vickers SM, Gotterbarm T, Spector M. Cross-linking affects cellular condensation and chondrogenesis in type II collagen-GAG scaffolds seeded

2212

[31]

[32]

[33]

[34]

[35] [36]

[37]

L. Jeng et al. / Acta Biomaterialia 8 (2012) 2203–2212 with bone marrow-derived mesenchymal stem cells. J Orthop Res 2010;28:1184–92. Pfeiffer E, Vickers SM, Frank E, Grodzinsky AJ, Spector M. The effects of glycosaminoglycan content on the compressive modulus of cartilage engineered in type II collagen scaffolds. Osteoarthritis Cartilage 2008;16:1237–44. Kulke MH, Bergsland EK, Ryan DP, Enzinger PC, Lynch TJ, Zhu AX, et al. Phase II study of recombinant human endostatin in patients with advanced neuroendocrine tumors. J Clin Oncol 2006;24:3555–61. Yamaguchi N, Anand-Apte B, Lee M, Sasaki T, Fukai N, Shapiro R, et al. Endostatin inhibits VEGF-induced endothelial cell migration and tumor growth independently of zinc binding. EMBO J 1999;18:4414–23. Melrose J, Smith S, Cake M, Read R, Whitelock J. Perlecan displays variable spatial and temporal immunolocalisation patterns in the articular and growth plate cartilages of the ovine stifle joint. Histochem Cell Biol 2005;123:561–71. Tesche F, Miosge N. Perlecan in late stages of osteoarthritis of the human knee joint. Osteoarthritis Cartilage 2004;12:852–62. Kitano A, Saika S, Yamanaka O, Reinach PS, Ikeda K, Okada Y, et al. Genipin suppression of fibrogenic behaviors of the alpha-TN4 lens epithelial cell line. J Cataract Refract Surg 2006;32:1727–35. Yannas IV, Burke JF, Gordon PL, Huang C, Rubenstein RH. Design of an artificial skin. 2. Control of chemical-composition. J Biomed Mater Res 1980;14:107–32.

[38] Salamone JC. Concise polymeric materials encyclopedia. Boca Raton, FL: CRC Press; 1999. [39] Ellis DL, Yannas IV. Regeneration templates, artifical skin and nerves. Boca Raton, FL: CRC Press; 1999. [40] Harris EN, Weigel JA, Weigel PH. The human hyaluronan receptor for endocytosis (HARE/Stabilin-2) is a systemic clearance receptor for heparin. J Biol Chem 2008;283:17341–50. [41] Vilar RE, Ghael D, Li M, Bhagat DD, Arrigo LM, Cowman MK, et al. Nitric oxide degradation of heparin and heparan sulphate. Biochem J 1997;324:473–9. [42] Raghuvanshi S, Sharma P, Singh S, Van Kaer L, Das G. Mycobacterium tuberculosis evades host immunity by recruiting mesenchymal stem cells. Proc Natl Acad Sci USA 2010;107:21653–8. [43] Kong HJ, Liu J, Riddle K, Matsumoto T, Leach K, Mooney DJ. Non-viral gene delivery regulated by stiffness of cell adhesion substrates. Nat Mater 2005;4:460–4. [44] Britt JC, Park SS. Autogenous tissue-engineered cartilage—evaluation as an implant material. Arch Otolaryngol Head Neck Surg 1998;124:671–7. [45] Abdollahi A, Folkman J. Evading tumor evasion: current concepts and perspectives of anti-angiogenic cancer therapy. Drug Resist Updat 2010;13:16–28.