SEARCH

SEARCH BY CITATION

Keywords:

  • Human embryonic stem cells;
  • Chondrogenesis;
  • Cartilage;
  • Tissue engineering

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

This study describes the development and application of a novel strategy to tissue engineer musculoskeletal cartilages with human embryonic stem cells (hESCs). This work expands the presently limited understanding of how to chondrogenically differentiate hESCs through the use of chondrogenic medium alone (CM) or CM with two growth factor regimens: transforming growth factor (TGF)-β3 followed by TGF-β1 plus insulin-like growth factor (IGF)-I or TGF-β3 followed by bone morphogenic protein (BMP)-2. It also extends the use of the resulting chondrogenically differentiated cells for cartilage tissue engineering through a scaffoldless approach called self-assembly, which was conducted in two modes: with (a) embryoid bodies (EBs) or (b) a suspension of cells enzymatically dissociated from the EBs. Cells from two of the differentiation conditions (CM alone and TGF-β3 followed by BMP-2) produced fibrocartilage-like constructs with high collagen I content, low collagen II content, relatively high total collagen content (up to 24% by dry weight), low sulfated glycosaminoglycan content (∼4% by dry weight), and tensile properties on the order of megapascals. In contrast, hESCs treated with TGF-β3 followed by TGF-β1 + IGF-I produced constructs with no collagen I. Results demonstrated significant differences among the differentiation conditions in terms of other biochemical and biomechanical properties of the self-assembled constructs, suggesting that distinct growth factor regimens differentially modulate the potential of the cells to produce cartilage. Furthermore, this work shows that self-assembly of cells obtained by enzymatic dissociation of EBs is superior to self-assembly of EBs. Overall, the results of this study raise the possibility of manipulating the characteristics of hESC-generated tissue toward specific musculoskeletal cartilage applications.

Disclosure of potential conflicts of interest is found at the end of this article.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

There is a need for musculoskeletal cartilage replacements due to the inability of native cartilages to heal effectively. In particular, fibrocartilages like the knee meniscus and temporomandibular joint (TMJ) disc, as well as hyaline articular cartilage, are the focus of tissue engineering research efforts due to the wide prevalence of their disease and injury. The diverse strategies to engineer these cartilages share the major obstacle of identifying a readily available cell source that can be used in patients. Although there are challenges with the use of human embryonic stem cells (hESCs), such as teratoma formation, these cells have tremendous potential for regenerative medicine efforts [1]. For example, the pluripotency of hESCs makes it conceivable that the naïve cells can be differentially coaxed in vitro and used to functionally produce different musculoskeletal cartilages. Accomplishing this feat requires significant investigation in two areas: chondrogenic differentiation and tissue engineering.

Studies with adult stem cells [2, [3], [4], [5]6], as well as recent work with hESCs [7, [8], [9], [10], [11], [12]13], provide important knowledge in formulating new strategies for using hESCs in cartilage tissue engineering. Chondrogenic differentiation of mesenchymal and embryonic stem (ES) cells has been performed most commonly with growth factors such as transforming growth factor (TGF)-β1, TGF-β3, bone morphogenic protein (BMP)-2, BMP-4, and insulin-like growth factor (IGF)-I, singly [7, [8], [9], [10], [11], [12], [13], [14], [15], [16], [17]18], or in combination [19]. Also widely used are biochemical agents that have roles in chondrogenesis and collagen production, such as dexamethasone; insulin, transferrin, and selenious acid (ITS); ascorbic acid; and l-proline [5, [6]7, 20]. Additionally, the growth environment of the ES cells may be important [7, 8, 13]. For the most part, these studies have demonstrated chondrogenic differentiation through analysis of gene expression with polymerase chain reaction [7, 9, 11, 13, 21], characterization of cartilage matrix with histology [7, 9, 11, 13], or identification of cell surface and cell associated matrix markers with flow cytometry [13, 22, [23], [24], [25], [26], [27]28]. Although these characterizations help to determine whether cells exhibit a chondrocytic (or fibrochondrocytic) phenotype, a functional approach to using any cell source for engineering purposes should incorporate quantitative evaluations of the biochemical and biomechanical properties of the generated tissue [29].

Quantitative evaluations should be used in engineering studies because the amounts of specific collagens and glycosaminoglycans (GAGs) can vary considerably among different cartilages, and the structural arrangement of these matrix molecules largely defines their biomechanical functions [30]. Thus, cell applicability toward cartilage applications can be determined with a complement of quantitative and qualitative assessments of the engineered tissues. Few studies have applied this functional approach to cartilage tissue engineering with any stem cell source, and, to our knowledge, no study has demonstrated the ability to differentially modulate the functional chondrogenic potentials of hESCs (i.e., engineer tissues with different biochemical and biomechanical properties). This study offers this functional approach to assess how distinct differentiation conditions affect hESCs.

In addition to the chondrogenic differentiation of stem cells, using the differentiated cells for cartilage tissue engineering remains a challenge since there are many potentially fruitful strategies, such as the use of hydrogels and scaffolds [2, 13, 31, [32], [33], [34], [35]36]. As an alternative, we have recently developed a scaffoldless strategy called self-assembly [37, 38]. Originally, this process utilized native chondrocytes that were sedimented at a high density into an agarose-coated well. No scaffold material was necessary for the cells to form tissue with hyaline-like qualities in terms of morphology, protein composition, and biomechanics [37, 38]. The principles derived from these self-assembly studies provide a promising starting point for an emerging technology like hESCs.

Considering the progress with stem cells and cartilage tissue engineering, the aims of this study were (a) to analyze the differentiation of hESCs after applying specific combinations of cartilage-relevant growth factors (i.e., TGF-β3, TGF-β1, IGF-I, and BMP-2) and (b) to tissue engineer cartilage with these differentiated cells using self-assembly. We sought to address whether cells with different chondrogenic potentials would be generated when hESC embryoid bodies (EBs) were exposed to different combinations of growth factors. To determine how to tissue engineer cartilages using self-assembly, two groups were sedimented into agarose wells: EBs or enzymatically dissociated cells (DCs) from the EBs. The expectation was that DCs would result in improved tissue properties compared with EBs. A full-factorial design, using three differentiation conditions and two modes of self-assembly (EB or DC), was employed to address the aims and overarching questions of this study.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

Culture Conditions

hESC Expansion.

The NIH-approved hESC line BG01V [39, 40] (American Type Culture Collection [ATCC], Manassas, VA, http://www.atcc.org) was cultured according to standard protocols. Briefly, a feeder layer of γ-irradiated CF-1 (Charles River Laboratories, Wilmington, MA, http://www.criver.com) mouse embryonic fibroblasts (MEFs) at a density of 5 × 105 MEFs per well of a Nunc six-well dish (Fisher Scientific, Hampton, NH, http://www.fishersci.com) was used in the expansion of the hESCs. Frozen hESCs at passage (p)16 were thawed according to standard protocol and subcultured. A growth medium consisting of Dulbecco's modified Eagle's medium (DMEM)/F-12 (Gibco, Gaithersburg, MD, http://www.invitrogen.com), ES-qualified fetal bovine serum (FBS) (ATCC), l-glutamine (Gibco), knockout serum replacer (Gibco), and nonessential amino acids (Gibco) was used. The hESCs were passaged with collagenase IV (Gibco) every 4–5 days and were used for the experiment at p21.

Embryoid Body Formation and Differentiation Conditions.

Dispase solution (0.1% wt/vol in DMEM/F-12) was applied for 10–15 minutes to colonies of undifferentiated hESCs in monolayer when the colonies reached 70%–80% confluence. This enzymatic treatment predominantly lifts the hESC colonies from the culture dish, leaving MEFs behind and forming EBs from the hESC colonies [41]. After two washes and centrifugations with DMEM/F-12, the EBs were suspended in a chondrogenic medium (CM) consisting of high-glucose DMEM, 10−7 M dexamethasone, ITS+ Premix (6.25 ng/ml insulin, 6.25 mg transferrin, 6.25 ng/ml selenious acid, 1.25 mg/ml bovine serum albumin, and 5.35 mg/ml linoleic acid; Collaborative Biomedical, San Jose, CA, http://www.bdbiosciences.com), 40 μg/ml l-proline, 50 μg/ml ascorbic acid, 100 μg/ml sodium pyruvate, and 1% FBS (Gemini Bio-Products, West Sacramento, CA, http://www.gembio.com). The EBs were distributed into bacteriological Petri dishes (Fisher) by placing EBs from two six-well culture plates into each Petri dish and using 18 ml of medium per dish. Three differentiation conditions were applied to the EBs in this experiment: (a) CM alone for 28 days (designated CM), (b) CM with TGF-β3 (10 ng/ml) for 7 days followed by the combination of TGF-β1 (10 ng/ml) and IGF-I (100 ng/ml) for 21 days (designated D1), and (c) CM with TGF-β3 (10 ng/ml) for 7 days followed by BMP-2 (10 ng/ml) for 21 days (designated D2). For the entire experiment, medium and, when applicable, growth factors were completely changed every 48 hours. EBs were used for self-assembly or for histological analysis at t = 4 weeks.

Self-Assembly of Chondrogenically Differentiated hESCs.

After 28 days of differentiation (t = 4 weeks), EBs in each of the three differentiation groups were separated into two equal subgroups. One subgroup of EBs from each differentiation condition was digested in 0.05% trypsin-EDTA (Gibco) for 1 hour. Cells from each digest were counted with a hemocytometer, washed with DMEM containing 1% FBS, centrifuged at 200g, and resuspended at a concentration of 5.0 × 105 cells per 20 μl in CM. Constructs were made by seeding the DC suspension into 3-mm wells of 2% agarose (5.0 × 105 cells per well).

The other subgroup comprised the undigested EBs, which were centrifuged at 200g and resuspended in CM. EBs were seeded into 5-mm wells of 2% agarose using an equivalent of 1 × 106 cells per construct (based on the hemocytometer count). The two self-assembly modes (EB and DC) were carried out over the ensuing 4 weeks, culturing all constructs made from the three differentiation conditions in CM alone (i.e., without any exogenous growth factors).

Assessments

Analysis of Differentiated EBs.

At t = 4 weeks, a small number of EBs from each differentiation condition were collected for analysis. For visualization of Sox-9, some of the cells obtained from the trypsin digestion at 4 weeks of differentiation were plated at a density of 4.0 × 105 per milliliter onto a glass slide and allowed to attach overnight. The cells were then fixed with 3.7% paraformaldehyde for 20 minutes, incubated with Triton X-100 for 20 minutes at room temperature, blocked with 3% bovine serum albumin for 30 minutes, incubated with Sox-9 primary antibody (AnaSpec Inc., San Jose, CA, http://www.anaspec.com) for 2 hours, and then incubated with Alexa Fluor 546 conjugated goat anti-rabbit IgG1 secondary antibody (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) for 1 hour. Phosphate-buffered saline (PBS) washes were performed between each of these steps.

EBs were also cryosectioned and stained for collagens using picrosirius red, GAGs using Alcian Blue, and collagen I and collagen II using immunohistochemistry (IHC) as previously described [37]. Stains for mesodermal tissue markers were used to detect unwanted differentiation, including von Kossa (calcified tissues such as bone), Masson's trichrome (muscle), and oil red O (adipose). Standard protocols were followed for each of these stains.

Analysis of Self-Assembled Constructs.

At the t = 8 weeks time point (after 4 weeks of self-assembly), each construct was measured for wet weight after carefully blotting excess water. Diameter and thickness measurements were made using digital calipers with an accuracy of 0.01 mm (Mitutoyo, Aurora, IL, http://www.mitutoyo.com). Constructs were either used for histology, biochemical assays, or biomechanical testing. Histological assessments for self-assembled constructs included picrosirius red, Alcian Blue, IHC for collagen I and collagen II, von Kossa, Masson's trichrome, and oil red O. Additionally, picrosirius red samples were analyzed with a polarized microscope (Nikon, Melville, NY, http://www.nikonusa.com) to visualize collagen alignment. Biomechanical testing included tensile testing using an Instron 5565 (Instron, Norwood, MA, http://www.instron.us) and unconfined compression using a modified creep indentation apparatus [42]. For tensile testing, specimens were cut from the cylindrical constructs into dog-bone shapes and pulled at a strain rate of 1% per second until failure. Gauge length, thickness, and width of the specimens were measured with digital calipers so that load and extension measurements could be converted to stress and strain. Similar to the whole constructs, collagen alignment of the tensile specimens was analyzed with picrosirius red staining and polarized light. For unconfined compression testing, constructs were allowed to equilibrate in PBS for 10 minutes and then subjected to an instantaneous 1.96 mN test load. The creep test was allowed to run for at least 1 hour, which was long enough to achieve deformation equilibrium. With the unconfined compression creep data, intrinsic material properties of the constructs were obtained using a previously developed viscoelastic model [43].

Biochemical assays included dimethylmethylene blue (DMMB), hydroxyproline, PicoGreen, and enzyme-linked immunosorbent assays (ELISAs) for collagens I and II. Samples were lyophilized for 48 hours, and dry weights were measured. Previously described protocols were used for DMMB and hydroxyproline tests, and one set of samples was used for these two assays [37]. For collagens I and II, Chondrex reagents and protocols were used (Chondrex, Redmond, WA, http://www.chondrex.com) with the exception that constructs were digested with papain (rather than pepsin) at 4°C for 4 days followed by a 1-day elastase digest. The PicoGreen assay was performed with this set of samples using a multiple of 7.7 pg DNA per cell.

Statistics

Data were analyzed with a two factor analysis of variance using Tukey's post hoc test when applicable and a significance value of p < .05. At least four samples were analyzed for biochemical assays and biomechanical tests for all groups. All data are reported as mean ± standard deviation.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

Chondrogenic Differentiation of EBs at t = 4 Weeks

During the 4 weeks of differentiation in EB form, EBs noticeably grew in size with the CM (chondrogenic medium without growth factors) and D2 (CM with additives of TGF-β3 followed by BMP-2) groups, whereas D1 (CM with additives of TGF-β3 followed by TGF-β1 and IGF-I) EBs did not appear to change in size. The morphology and histology of the EBs at t = 4 weeks is shown in Figure 1A. The collagen I and collagen II IHC illustrate that the cartilaginous matrix in the EBs was loosely connected and unorganized, with all three differentiation conditions exhibiting collagen I most prominently. Alcian Blue staining for all groups at this time point was minimal (data not shown). Dissociation of the EBs with trypsin resulted in a cell suspension, although some cells were still connected with extracellular matrix (ECM) after the 1-hour digestion. Most of the cell suspension was used to make constructs, with at least 8 DC constructs being self-assembled from each differentiation regimen. Similarly, at least 8 EB constructs were self-assembled from each group. A small portion of the cell suspension was used to analyze Sox-9 expression and cell morphology (Fig. 1B). While the cells generated from each differentiation regimen at t = 4 weeks exhibited Sox-9 protein expression, they demonstrated distinct cell morphologies. CM and D1 cells were rounded and approximately the same size as native articular chondrocytes. D2 cells appeared larger and fibroblastic. Histological analyses for calcified tissue (von Kossa), muscle (Masson's trichrome), and adipose (oil red O) were negative at this time point (data not shown).

thumbnail image

Figure Figure 1.. Chondrogenic differentiation of human ESCs (t = 4 weeks). (A): Collagens I and II were detected in all three differentiation conditions with immunohistochemistry at t = 4 weeks (10×). The embryoid bodies in all groups appeared highly hydrated and cellular with a loosely organized extracellular matrix. Due to this, obtaining good frozen sections for these structures was challenging. Calcified tissue (i.e., bone), muscle, and adipose were not detected (data not shown). (B): The SOX-9 transcription factor was detected in all three differentiation regimens at t = 4 weeks (green). The blue fluorescence is a Hoechst stain for the nucleus. Whereas chondrogenic medium alone and D1 cells were approximately the same size and had a similar rounded shape as the positive control of native articular chondrocytes (bottom row, left), D2 cells were larger and appeared fibroblastic. The negative control of MEFs (bottom row, right) did not exhibit SOX-9. The white bar is 10 μm (40×). Abbreviations: CM, chondrogenic medium; Col, collagen; D1, differentiation condition 1; D2, differentiation condition 2; MEFs, mouse embryonic fibroblasts.

Download figure to PowerPoint

Self-Assembled Constructs

Morphology.

After the initial seeding of the dissociated cells into the 3-mm agarose wells, cells coalesced within 24 hours into constructs that were slightly smaller than the well. Over the following weeks, the spacing between cells in each construct increased as they produced ECM, causing the constructs to appear smooth and cartilaginous (Fig. 2A). The amount of EBs for each group seeded into the 5-mm wells was enough to cover the entire bottom surface initially. Over the ensuing weeks, CM and D2 constructs filled the well, whereas D1 constructs appeared to shrink away from the outer edges. EB constructs never achieved homogeneity during the experiment. A clear matrix connected EBs in a construct, and the constructs appeared highly hydrated (Fig. 2A).

thumbnail image

Figure Figure 2.. Gross morphology and histology of self-assembled constructs (t = 8 weeks). (A): Dissociated cell (DC) constructs appeared more uniform than embryoid body (EB) constructs. The DC group also held their shape when manipulated, whereas the EB group did not. Differentiation condition 1 constructs were generally smaller than constructs from the other two groups, as shown in the pictures and the morphological measurements. EB constructs were engineered larger (5-mm molds vs 3-mm molds for DC constructs) because the EBs at t = 4 weeks were too large for the 3-mm wells. (B): Collagens I and II (top two rows) were detected in the chondrogenic medium alone and differentiation condition 2 groups with immunohistochemistry at t = 8 weeks, regardless of self-assembly mode (EB or DC) (10×). D1 constructs had collagen II but did not demonstrate much collagen I staining. Intense picrosirius red and spotty Alcian Blue stains (4×) are shown in the bottom row for each differentiation condition. Calcified tissue (i.e., bone), muscle, and adipose were not detected at t = 8 weeks (data not shown). Abbreviations: CM, chondrogenic medium; Col, collagen; D1, differentiation condition 1; D2, differentiation condition 2.

Download figure to PowerPoint

Construct morphological measurements are shown in Figure 2 below the gross morphological pictures. D1 constructs had significantly lower thickness and wet weight compared with CM constructs for both EB and DC groups (p < .05), whereas D2 constructs were not different from either of the differentiation conditions. At t = 8 weeks, CM and D2 constructs demonstrated uniform staining for collagens I and II, regardless of self-assembly mode (EB or DC, Fig. 2B). D1 constructs also demonstrated uniform staining for collagen II but no significant staining for collagen I (Fig. 2B) for both EB and DC self-assembled constructs. Intense picrosirius red staining in all self-assembled constructs illustrated the matrix-producing capacity of the differentiated cells (Fig. 2B). Conversely, Alcian Blue staining was minimal (Fig. 2B). An interesting finding with histology was that a central pocket of fluid had formed within the DC constructs (Fig. 2B). This was noted primarily in the CM and D2 constructs. At the end of the 8-week experiment, other mesodermal tissues (bone, muscle, adipose) were not detected by histology (data not shown).

Biochemical Analysis.

When comparing EB and DC self-assembled groups for biochemical content, normalized by dry weight (dw), DC constructs demonstrated greater matrix production (both collagen and GAG) (p < .05), as shown in Figure 3. The measurements for hydroxyproline showed that the D1 DC group did not produce as much collagen (5.2% by dw) as the other two groups, with CM and D2 DC constructs producing 17.9% and 24.1% by dw, respectively (Fig. 3A). Although Alcian Blue staining was not substantial, the DMMB assay demonstrated the presence of sulfated GAGs in all constructs (Fig. 3B). The water content for engineered constructs in all groups was approximately 90% (91.1% ± 2.7% for CM DC, 85.5% ± 5.8% for D1 DC, 89.7% ± 5.1% for D2 DC, 92.8% ± 3.3% for CM EB, 94.2% ± 2.6% for D1 EB, and 91.7% ± 2.3% for D2 EB).

thumbnail image

Figure Figure 3.. Biochemical analysis of total collagen and sulfated GAGs (t = 8 weeks). (A): Self-assembly with DCs caused an increase in total collagen content compared with self-assembly with EBs (p = .002). Significant differences were also detected due to differentiation agent, with chondrogenic medium alone and differentiation condition 2 constructs being higher than differentiation condition 1 constructs (p = .0007). Note: The conventions used to show statistically different results are upper or lower case letters (one set for each experimental factor). Groups not connected by the same letter are significantly different (p < .05). (B): Sulfated GAG content was higher in DC constructs compared with EB constructs (p = .038). Differentiation condition was not a significant factor for GAG production. Abbreviations: CM, chondrogenic medium; D1, differentiation condition 1; D2, differentiation condition 2; DC, dissociated cell; DW, dry weight; EB, embryoid body; GAG, glycosaminoglycan.

Download figure to PowerPoint

PicoGreen demonstrated that the number of cells per construct was significantly different between CM and D1 groups (p < .05), whereas D2 constructs were not different from the other two groups (Fig. 4A). ELISAs for collagens I and II demonstrated that the production of collagens I and II varied between each differentiation regimen and between DC and EB constructs (Fig. 4B, 4C). Specifically, collagen I production per cell was significantly higher in CM constructs compared with the other two differentiation agents (for example, in μg × 10−2 per cell, 4.8 ± 1.2 for CM DC, −0.5 ± 0.5 for D1 DC, and 3.8 ± 0.9 for D2 DC, p < .05). D1 constructs demonstrated undetectable collagen I, which echoed the IHC results for this group. The ELISA data also demonstrated that DC constructs had higher collagen I and lower collagen II production per cell than EB constructs (p < .05). Differentiation condition was a significant factor when analyzing the collagen II ELISA, with CM constructs having higher collagen II content compared with D2 constructs. For example, CM DC samples had over twofold higher collagen II content per cell than D2 DC samples (0.8 ± 0.4 vs. 0.3 ± 0.1 μg × 10−5 per cell, p < .05). D1 constructs were not significantly different compared with the other two differentiation agents in terms of collagen II content per cell.

thumbnail image

Figure Figure 4.. Enzyme-linked immunosorbent assays (ELISAs) for collagens I and II (t = 8 weeks). (A): PicoGreen results from the ELISA digest showed that chondrogenic medium alone constructs had higher cell numbers than differentiation condition 1 constructs at t = 8 weeks. CM dissociated cell constructs had almost twice as many cells as the other two DC groups. All constructs were initially seeded with the same amount of cells. Additionally, D1 EB constructs exhibited lower cell numbers than the other EB constructs. These results generally mirror the gross morphology of the constructs. (B): Collagen I per cell was undetectable in D1 constructs, whereas CM and differentiation condition 2 constructs exhibited relatively high amounts of collagen I per cell. Overall, CM constructs had higher collagen I content (p < .0001). Also, DC constructs had more collagen I per cell than EB constructs (p < .0001). (C): Collagen II per cell demonstrated differences between EB and DC constructs (p = .008). CM constructs had more collagen II per cell than D2 constructs (p = .001). Groups not connected by the same letter are significantly different (p < .05). Abbreviations: CM, chondrogenic medium; D1, differentiation condition 1; D2, differentiation condition 2; DC, dissociated cell; EB, embryoid body.

Download figure to PowerPoint

Biomechanical Analysis.

Unconfined compression testing of the self-assembled constructs demonstrated that DC constructs had a significantly higher instantaneous modulus compared with EB constructs (p < .05), whereas there was no significant difference among CM, D1, and D2 constructs (Fig. 5). There was no statistical difference among any treatments in terms of their relaxed modulus (2.2 ± 1.5 kPa for CM DC, 1.7 ± 0.8 kPa for D1 DC, 1.3 ± 0.3 kPa for D2 DC, 0.7 ± 0.1 for CM EB, 1.8 ± 0.7 kPa for D1 EB, and 0.8 ± 0.2 kPa for D2 EB). The CM and D2 DC constructs exhibited a higher apparent viscosity than all other treatments (2,778 ± 817 kPa-s for CM DC, 1,489 ± 857 kPa-s for D1 DC, 2,487 ± 980 kPa-s for D2 DC, 539 ± 208 kPa-s for CM EB, 1,445 ± 572 kPa-s for D1 EB, and 693 ± 356 kPa-s for D2 EB).

thumbnail image

Figure Figure 5.. Compressive properties (t = 8 weeks). Dissociated cell constructs had a higher instantaneous modulus than embryoid body constructs (p = .005). Differentiation condition had no effect. Groups not connected by the same letter are significantly different (p < .05). Abbreviations: CM, chondrogenic medium; D1, differentiation condition 1; D2, differentiation condition 2; DC, dissociated cell; EB, embryoid body.

Download figure to PowerPoint

Tensile testing (Fig. 6) showed that D2 DC constructs had an over 5.5-fold higher tensile modulus (3.3 ± 0.7 vs. 0.6 ± 0.5 MPa) and 2.8-fold higher ultimate tensile strength compared with D1 constructs (1.1 ± 0.1 vs. 0.4 ± 0.3 MPa). Comparing these tensile properties of D2 with CM constructs yielded similar increases (6.6-fold and 2.8-fold, respectively). Polarized light microscopy performed directly on tensile tested specimens demonstrated collagen alignment in the direction of tensile testing for D2 DC tensile specimens, whereas CM and D1 tensile specimens did not (Fig. 6). Moreover, D2 constructs exhibited a higher degree of collagen alignment than CM and D1 constructs in the untested DC samples. EB constructs were not testable under tension.

thumbnail image

Figure Figure 6.. Tensile properties of dissociated cell (DC) constructs (t = 8 weeks). (A): DC constructs had enough mechanical integrity to be tested under tension, whereas embryoid body constructs did not have this degree of mechanical integrity and could not be tested. In terms of both tensile modulus and ultimate tensile strength, differentiation condition 2 constructs were significantly higher than chondrogenic medium alone and differentiation condition 1 constructs. Also notable was the fact that the values for these properties were on the order of megapascals. Groups not connected by the same letter are significantly different (p < .05). (B): Collagen alignment (demonstrated by picrosirius red and polarized light) in the specimens along the axis of tensile testing (double-headed arrow) was seen best in the D2 group, whereas the CM and D1 specimens demonstrated no preferred direction (top row). Pictured on the top row is one half of the tensile specimens, with the broken end (where failure occurred) being on the left of each picture (white arrow, 10×). Analyzing the untested whole constructs (bottom row) also demonstrated a higher degree of collagen alignment in D2 constructs compared with the other groups (10×). Abbreviations: CM, chondrogenic medium; D1, differentiation condition 1; D2, differentiation condition 2.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

This study establishes a new methodology to study cartilage tissue engineering with hESCs. The use of self-assembly as a tissue engineering strategy resulted in quantitative data that addressed two questions. First, we investigated whether cells with different chondrogenic potentials would be generated when hESCs were exposed to distinct growth factor regimens for 4 weeks. We assessed this after the cells had formed neocartilage (at t = 8 weeks), showing differences in the chondrogenic potential of CM, D1, and D2 cartilage constructs in terms of morphology, biochemistry, and biomechanics. These properties also addressed the second major question of the study, illustrating that DC constructs outperform EB constructs and thereby highlighting the importance of enzymatic dissociation of EBs prior to self-assembly. These findings represent incremental steps toward functional engineering of different types of musculoskeletal cartilages with hESCs.

Functional goals for engineering a cartilage replacement with any cell source, including hESCs, logically begin by studying the healthy native tissue. A number of characterization studies have been carried out to understand the relationship between cartilage structure and function (reviewed in [30]). The constructs engineered in this study generally exhibited properties most similar to the fibrocartilages, particularly the TMJ disc and the outer portion of the knee meniscus. The constructs had relatively high total collagen contents (up to 24% by dw in this study vs. ∼80% by dw for native TMJ and outer meniscus), low sulfated GAG contents (approximately 4% by dw in this study vs. 0.6%–10% for native TMJ and outer meniscus), and relatively high tensile properties (order of 1 MPa in this study vs. order of 10–100 MPa for the native fibrocartilages). These fibrocartilages are also notable for their high collagen I content and low to absent collagen II content. Both CM and D2 constructs demonstrated this pattern, whereas D1 constructs did not contain detectable collagen I.

Although the absolute values for certain tissue design parameters may not seem substantial when the hESC-generated cartilages are compared with native tissues, the results are notable in the context of previous work in the cartilage and stem cell communities. Compared with studies using biomaterials as scaffolds, as well as our original work describing self-assembly, the constructs produced by chondrogenically differentiated hESCs have comparable collagen content (approximately 1%–2% by wet weight) but lower sulfated GAG [37, 44, 45]. Even though the current study produced mostly fibrocartilage and these previous tissue-engineering studies [37, 44, 45] produced hyaline-like cartilage with native chondrocytes, this comparison demonstrates the matrix-producing capacity of the differentiated hESCs. We have also measured tensile properties on the order of 1 MPa with native chondrocyte self-assembled constructs (unpublished results). Pioneering work by Elisseeff and associates [13] with hESCs that were chondrogenically differentiated with members of the TGF-β superfamily in hydrogels resulted in a sulfated GAG content around 7% by dw (vs. approximately 4% in this study). The engineered tissue in this study builds upon this original work and is at least on par with the prior art in several important design parameters.

This study also contributes evidence that the chondrogenic potential of hESCs can be altered with soluble differentiation agents. We constructed the differentiation regimens such that TGF-β3 would be administered during the critical early period of ES cell differentiation when the specification of mesodermal cells into precursors of different lineages may occur [9]. After this initial stage, we tested whether the functional capacity of these cells to produce cartilage could be altered by either the combination of TGF-β1 with IGF-I or BMP-2 alone, considering the success of these growth factors with other systems [2, 5, [6], [7], [8], [9], [10], [11], [12]13, 23, 46, 47] and taking into account their varying roles in the differentiation and biological actions of cells [47, [48], [49]50]. Differences were observed at t = 4 weeks in terms of cell morphology and at t = 8 weeks in terms of construct morphology (Fig. 2), biochemistry (Figs. 3, 4), and tensile properties (Fig. 6). Since cells from each differentiation condition were cultured in the basal chondrogenic medium without exogenous growth factors during self-assembly, these data collectively indicate that the cells generated after 4 weeks of EB differentiation had varying capacities to produce cartilage.

Tensile testing revealed the most dramatic difference among differentiation conditions. D2 tensile specimens exhibited the highest degree of collagen alignment, and this finding appears to account for the higher tensile modulus and ultimate tensile strength of this group (Fig. 6). Whether this is a true functional difference needs further investigation. One explanation for the apparent differences in degree of alignment and tensile properties is that the D2 cells, which had a more fibroblastic morphology (Fig. 1B), had a better ability to organize the collagen network. The link between cell shape and function has been well-established in various types of cartilage [30]. Additionally, in native cartilages, the resident cells, such as chondrocytes, remodel the matrix on a regular basis [51].

Another curious finding was the pocket of fluid inside of the CM and D2 constructs. We have previously encountered fluid-filled interiors in other self-assembled constructs [52]. In that instance, it was postulated that the phenomenon was a result of the use of a xenogenic serum, which was a medium component used in this study. Others have used human chondrocytes with bovine serum and noted fluid-filled spheroids [53]. Our initial self-assembly study used bovine cells and bovine serum and encountered no fluid-filled region [37]. Another possibility for the fluid-filled interior encountered in this study is that a different cell population (chondrogenic or nonchondrogenic) accumulated in this space, but the histological evidence did not offer support of this idea. In future studies, we will seek to understand the underlying cause of this phenomenon and eventually eliminate this undesirable feature.

Although characterization of the differentiation process was one major goal of this study, we also determined how the differentiated hESCs responded to the transition from differentiation in EB form to tissue engineering. Although constructs made with both self-assembly modes, EB and DC, expressed cartilage proteins, the gross appearance (Fig. 2), total collagen and sulfated GAG contents (Fig. 3), and biomechanical properties (compressive, Fig. 5, and tensile, Fig. 6) of the DC constructs were better. Additionally, the ELISA results (Fig. 4) suggested that the process of digesting the EBs after 4 weeks of differentiation and subsequently placing the cells into agarose wells for self-assembly increases collagen I content and decreases collagen II content. In comparing EB and DC constructs, it is important to note that the difference in initial construct size (3-mm wells for DC constructs and 5-mm wells for EB constructs) was necessary due to difficulty with seeding the EBs into 3-mm wells. This difference in construct size between EB and DC groups necessitated comparisons normalized by cell number and dry weight. Given the marked differences found between these two groups with this analysis, we postulate that the ECM produced by the EBs during the first 4 weeks hindered cell-cell contacts and lowered the concentration of cells when they were placed in agarose molds for self-assembly. On the other hand, enzymatic dissociation of the EBs and subsequent seeding of the cells into agarose molds promoted direct cell contacts and a higher cell density. Even in normal development of cartilaginous tissues, such as articular cartilage, mesenchymal precursors aggregate at high density with direct cell contacts as an early step of chondrogenesis [54].

The findings of this study offer proof of concept for the developed system, although certain challenges remain. The mouse-derived feeder layer creates the possibility of having a small amount of irradiated xenogenic contaminants in the hESC constructs. Current progress with hESCs in feeder-free and human feeder systems indicates that this hurdle can be overcome [55, [56], [57]58]. Additionally, although we stained for other mesodermal tissues, we cannot completely rule out the presence of unwanted or incomplete differentiation. Any future therapy using hESCs must eliminate the possibility of teratoma formation.

Though the system cannot address all of these concerns currently, it offers a simple and robust design that is amenable to improvement. Chondrogenically-differentiated hESCs, chondrogenic medium, and an agarose mold are the major components of this system, with EB differentiation and self-assembly representing two important phases of the experiment. The modular design of this tissue engineering methodology accommodates perturbations to each of the key components during each phase to study how hESCs differentiate and how these differentiated cells can be used to engineer cartilage. With this system, a number of hypothesis-driven investigations into the effects of different seeding densities, different growth environments, and other biochemical and biomechanical differentiation agents can be conducted in the future. Beyond the goal of functional cartilage engineering, a major challenge will be to understand how these factors affect specific biochemical differentiation pathways. The developed methodology can also be used as a model system for this fundamental research.

In summary, a system for studying cartilage tissue engineering with hESCs has been developed that can discern functional differences among engineered cartilages made from chondrogenically differentiated hESCs that were exposed to distinct differentiation conditions. We have also shown the importance of enzymatic dissociation of EBs during the transition from EB differentiation to self-assembly. For our laboratory, this work sets the stage for an exciting new line of investigations into the engineering of specific musculoskeletal cartilages with hESCs, including the knee meniscus, TMJ disc, and hyaline articular cartilage.

Disclosure of Potential Conflicts of Interest

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

The authors indicate no potential conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References

We gratefully acknowledge support from the NSF (traineeship for E.J.K.), the Hertz Foundation (fellowship for G.M.B.H.), and an unrestricted fund from Rice University.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References