SEARCH

SEARCH BY CITATION

Keywords:

  • Embryonic stem cell biology;
  • Rhesus;
  • Real-time reverse transcription-polymerase chain reaction;
  • Lineage analysis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Cell interactions with the extracellular matrix (ECM) play a critical role in their physiology. Here, we sought to determine the role of exogenous and endogenous ECM in the differentiation of nonhuman primate ESCs. We evaluated cell differentiation from expression of lineage gene mRNA and proteins using real-time polymerase chain reaction and immunohistochemistry. We found that ESCs that attached to and spread upon highly adhesive collagen do not differentiate efficiently, whereas on the less adhesive Matrigel, ESCs form aggregates and differentiate along mesoderm and especially endoderm lineages. To further decrease ESC attachment to the substrate, we cultured them either on nonadhesive agarose or in suspension. In both cases, ESCs formed aggregates and efficiently differentiated along endoderm and mesoderm lineages, most strikingly into cardiomyocytes. Aggregates formed by thus-differentiated ESCs started to beat with a frequency of 50–100 beats per minute and continued to beat for approximately a month. In spite of the presence of exogenous ECM, ESCs were dependent on endogenous ECM for their survival and differentiation, as the inhibition of endogenous collagen induced a gradual loss of ESCs and neither a simple matrix, such as type I collagen, nor the complex matrix Matrigel was able to rescue these cells. In conclusion, adhesiveness to various ECM and nonbiological substrates determines the differentiation of ESCs in such a way that efficient cell-cell aggregation, together with less efficient cell attachment and spreading, results in more efficient cell differentiation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

During embryonic development, the extracellular matrix (ECM) plays a critical role in regulating stem cell differentiation into different lineages, as well as in cell migration and proliferation [1, [2], [3], [4], [5]6]. The ECM interacts with cells via cell surface receptors such as integrins; serves as a reservoir for growth factors; and provides a substrate for cell attachment and spreading, contact guidance for cell migration, and a scaffold for building tissues. The morphology of cells determined by their contact with ECM or with nonbiological surfaces may be associated with particular patterns of cell differentiation and proliferation. For example, floating type I collagen gel-induced functional differentiation in the mammary gland is mediated through changes in cell shape [7]; fibronectin and laminin induce extensive spreading and proliferation and enhance survival of granulosa cells [8]; and when keratinocyte spreading is restricted by a palladium-coated island and cells are forced to remain round, involucrin expression is stimulated and DNA synthesis is inhibited [9]. Also, the rigidity of the substrate determines the differentiation of precursor cells into a hepatocyte lineage by modulating the extent of cell spreading [10].

Endogenous ECM is also important for normal and tumor cell attachment [11] and proliferation [12]. When collagen synthesis is enhanced by ascorbic acid, human skin fibroblasts attach more quickly to plastic and cells become more adherent, as evidenced in their slower detachment by trypsin. In contrast, inhibition of collagen synthesis by cis-hydroxyproline decreases the rate of attachment of fibroblasts to the plastic surface and increases the rate of their detachment by trypsin.

With the establishment of cultures of ESCs derived from the inner cell mass of preimplantation mammalian embryos, it became possible to study the differentiation of these pluripotent cells in vitro. The importance of the composition and three-dimensional (3D) environment of synthetic or natural ECM on embryonic stem cell self-renewal and differentiation has been studied [13, [14]15]. Here, we sought to determine the role of the attachment of cells to exogenous ECM and to each other, as well as of endogenous ECM, in cell differentiation. We compared ESC morphology and differentiation on two types of gelled substrates: three-dimensional bioactive ECM and biologically inert agarose. Also, we studied the effect of endogenous ECM production on the differentiation of ESCs. We found that low-adhesive or nonadhesive substrates that promote cell-cell aggregation facilitate differentiation along endoderm and mesoderm lineages. In contrast, ESCs that are attached to and spread on adhesive substrates tend to differentiate inefficiently, without selectivity as to lineage. In spite of the presence of an exogenous ECM, ESC survival is dependent on the deposition of an endogenous ECM. Thus, the pattern of ESC differentiation is determined by attachment of cells to ECM and to each other.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Undifferentiated ESC Culture

Undifferentiated rhesus monkey R366.4 ESCs (WiCell Research Institute, Madison, WI, http://www.wicell.org) were cultured as previously described [16]. Briefly, ESCs were cultured with mitomycin C-treated (8 μg/ml for 2 hours) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) murine embryonic fibroblasts (MEFs) (Cell Essentials Inc., Boston, MA, http://www.cell-essentials.com) in gelatin-coated six-well plates (Nalge Nunc Inc., Naperville, IL, http://www.nalgenunc.com) to prevent the spontaneous differentiation of ESCs. ESCs were maintained in 80% KnockOut Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Rockville, MD, http://www.invitrogen.com) supplemented with 20% defined fetal bovine serum (d-FBS; HyClone, Logan, UT, http://www.hyclone.com), 1 mM l-glutamine, 0.1 mM 2-mercaptoethanol, and 1% nonessential amino acids (Invitrogen). ESCs were detached from the culture plate with 0.8 mg/ml collagenase IV (Invitrogen), split, and seeded onto a new mitomycin C-treated MEF feeder layer in a gelatin-coated six-well plate. Passages 33–38 of undifferentiated ESCs were used for the current experiments.

Culture on Different 3D Substrates

Type I collagen gels, Matrigel (a gelling basement membrane extracellular matrix) (Travigen Inc., Gaithersburg, MD), and agarose were used as substrates. Collagen gels from rat tail type I collagen (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com) were prepared according to the manufacturer's instructions. Collagen at 3.3 mg/ml in 0.2% acetic acid was mixed with 10% 10× DMEM (Invitrogen), 0.2 M HEPES (Invitrogen), 10% FBS (Gemini Bioproducts, Woodland, CA, http://www.gembio.com), 1% penicillin/streptomycin (Invitrogen), and ultrapure water (KD Medical, Columbia, MD, http://www.kdmedical.com) to reach a final concentration of 0.5 mg/ml. The collagen gel was neutralized to a pH of ∼7.4 by addition of 1 N NaOH. Matrigel (BD Biosciences, San Diego, http://www.bdbiosciences.com) was mixed with DMEM (Invitrogen) at 1:1 (volume) to reach a final concentration of ∼5 mg/ml. Agarose (Sigma-Aldrich) was added to phosphate-buffered saline (PBS) and autoclaved to reach 1.5%. Then, the 1.5% agarose was mixed with 10% 10× DMEM, 10% FBS, 1% penicillin/streptomycin, and ultrapure water to reach a final agarose concentration of 0.5%. Each of the 6- or 12-well inserts was coated with 1 or 0.25 ml, respectively, of 0.5 mg/ml type I collagen, 5 mg/ml Matrigel, or 0.5% agarose. After the collagen and Matrigel solidified at 37°C and the agarose solidified at room temperature, culture medium (DMEM supplemented with 10% FBS and 1% penicillin/streptomycin) was added for 24 hours. Then, rhesus monkey R366.4 ESCs were released by 0.8 mg/ml collagenase IV. ESCs from each well of the six-well plate were suspended in 1 ml of the differentiation medium (culture medium supplemented with ascorbic acid at 50 μg/ml) (Sigma-Aldrich) and seeded on two 6-well inserts or eight 12-well inserts coated with different substrates. The differentiation medium was added to the wells and was changed every 2 days. At various times, cells were photographed, collected from the inserts, washed with PBS, and stored at −80°C for RNA extraction. On days 7 and 21, samples were also fixed with 4% formalin at 4°C overnight and washed for paraffin embedding and sectioning. To have a better view of adherent cells on collagen or on Matrigel, we cut each piece of fixed gel vertically into halves along the diameter before paraffin embedding and sectioning. For study of cell proliferation, some samples were pulsed with 5-bromodeoxyuridine (BrdU) at 10 μg/ml for 18 hours before collection.

Culture in Suspension

ESCs were cultured with no addition of ECM in a NASA-designed rotating bioreactor (USA Synthecon, Houston, TX, http://www.synthecon.com) that keeps cells in suspension and facilitates cell-cell aggregation. As a control, ESCs were mixed with Matrigel, which was prepared as described above and added (90 μl) to 16-well Lab-Tek chamber slides (GSS, Reston, VA, http://www.govsci.com). After the gels had solidified and condensed to form pellets, they were maintained in culture medium for 24 hours. Then, the gels were released from the chamber slides and loaded into the bioreactors. Four pellets of each gel and 1 ml of ESC suspension (one well of the six-well plate of ESCs was suspended in 1 ml) were added into each 10-ml bioreactor. After 21 days, cells were collected from the bioreactors, washed with PBS, and stored for RNA extraction. Some samples were also fixed with 4% formalin at 4°C overnight and washed again for paraffin embedding and sectioning. To study cell proliferation, we pulsed samples with 10 μg/ml BrdU for 18 hours before collection.

Culture with cis-Hydroxyproline

To study the effect of endogenous collagen synthesis on ESC attachment, survival, and differentiation, we suspended ESCs released as described above in differentiation media supplemented or not supplemented with cis-hydroxyproline at 100 μg/ml and seeded them onto different substrates in culture. At various times, cells were photographed, collected from the inserts, washed with PBS, and stored for RNA extraction.

Reverse Transcription and Real-Time Polymerase Chain Reaction

We extracted RNA using the RNeasy Fibrous Tissue Kit (Qiagen, Valencia, CA, http://www1.qiagen.com) and quantitated it using a spectrophotometer. We conducted reverse transcription of RNA (300 ng) using the High Capacity Archive Kit (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) and real-time polymerase chain reaction (PCR) using primer probe sets obtained from Assays on Demand (Applied Biosystems) (Table 1) and the TaqMan master mix (Applied Biosystems). All changes in gene expression levels were compared with those of undifferentiated ESCs. The ΔΔCT method [17] was used to calculate the fold differences of gene expression between the differentiating ESCs cultured on different gels and the undifferentiated ESCs. For each sample, the first ΔCT was calculated on the basis of the difference of the CT values between each lineage gene and the housekeeping gene β-actin in the same sample. The ΔΔCT was the difference in the first ΔCT between the differentiating ESCs and the undifferentiated ESCs that were used as a calibrator. On the basis of the mean and standard error (SEΔΔCT) of the ΔΔ CT results, the mRNA expression (fold change) was expressed as the mean ± SEfold = 2−ΔΔCT ± 2−ΔΔCT SEΔΔCT · ln(2).

Table Table 1.. Genes analyzed in the current study
Thumbnail image of

Immunohistochemistry

The primary antibodies and their dilutions used are listed in Table 2. All primary antibodies were diluted in Tris-buffered saline supplemented with 1% BSA. We performed target retrieval on formaldehyde-fixed paraffin sections using 0.01 M citrate buffer (pH 6.0) in a microwave oven according to the instructions from the manufacturers. Unstained sections from ESCs grown on different gel substrates were incubated with primary antibodies. We used the immunoperoxidase system (LSAB+ system; Dako, Carpinteria, CA, http://www.dako.com), which contained biotinylated anti-rabbit, -mouse, and -goat Ig as a secondary antibody; streptavidin conjugated to horseradish peroxidase as a link agent; and 3,3′-diaminobenzidine as a chromogen. After the immunostaining procedures were completed, some sections were lightly counterstained with Mayer's hematoxylin (Sigma-Aldrich). In some cases, positive-stained cells were counted in 9–15 unconnected fields at ×100 magnification, and the percentage of positive staining was calculated on the basis of the total number of cells in each view.

Table Table 2.. Primary antibodies used in this study
Thumbnail image of

Statistical Analysis

We analyzed gene expression using the results of ΔΔCT (n = 3–11). The Dunnett t test was used to compare the gene expression in ESCs cultured under different culture conditions (on 3D substrate or in bioreactors) with that in the undifferentiated ESCs. We compared lineage gene expression and proliferation (percentage of BrdU-positive cells) under different culture conditions and/or durations using one-way analysis of variance and post hoc Tukey tests. The relationship between the lineage gene expression and the endogenous ECM gene expression was analyzed with multiple linear regression using a sequential “backward elimination” procedure to identify the significant variables. We performed statistical analysis using SPSS 13.0 (SPSS, Inc., Chicago, http://www.spss.com).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Lineage Gene and Lineage Protein Expression in Differentiating ESCs

Using culture on various gelled substrates, we analyzed ESC morphology alterations and the expression of the lineage genes. We chose α-fetoprotein (AFP) and albumin as markers for endoderm, NKX2.5 and renin as mesoderm markers, and keratin and neural filament heavy chain (NFH) as ectoderm markers and analyzed the expression of these lineage genes with real-time PCR. The kinetics of lineage gene expression in differentiating ESCs on different gelled ECM substrates was compared with that of the expression of these genes in undifferentiated ESCs (Fig. 1). Also, we studied protein expression using immunostaining. We chose AFP and albumin as markers for endoderm, cardiac troponin I and renal cell carcinoma antigen as mesoderm markers, and cytokeratin and β-III-tubulin as ectoderm markers. We found that under all studied experimental conditions, differentiating ESCs generally started expressing lineage-specific genes by day 3. The efficiency of ESC differentiation along a particular lineage depended on the substrate.

thumbnail image

Figure Figure 1.. The effects of different three-dimensional gel substrates on morphology changes (A–H) and on mRNA expression levels of lineage genes (I–N) on the differentiation of ESCs. The fold change is expressed as mean ± SE (n = 3–11). Scale bar = 100 μm. Abbreviations: AFP, α-fetoprotein; NFH, neural filament heavy chain.

Download figure to PowerPoint

ESCs on Collagen Gel.

Within 2 hours of seeding, ESCs readily attached to a collagen gel and formed colonies with cells at the periphery, spreading their lamellae onto the free substrate (Fig. 1A). After 21 days of culture, cells formed a layer covering the entire surface, with large clusters of cells observed on top of the cell layer (Fig. 1B).

In general, the levels of expression of all lineage genes were not statistically different from those in the undifferentiated ESCs (p was between .17 and 1), except for the mesoderm renin gene (Fig. 1I–1N). We observed a 4 ± 3 × 103-fold (n = 3) increase in renin gene expression on day 21 (p = .001) (Fig. 1L), assuming that its expression in undifferentiated ESCs was just under the detection threshold (45 cycles) (Fig. 1L).

ESCs cultured on collagen gel are organized into spherical glandular-like structures formed by a single cell layer (monolayer) or by multiple cell layers (multilayer). Although immunostaining generally confirmed the lack of efficient differentiation, it allowed the detection of some positively stained cells within particular structures. For example, β-III-tubulin-positive cells were detected at the periphery of the multilayered structures (Fig. 2A). Cytokeratin-positive cells were found either in the stratified mono- or multilayered structures or on the spread cells covering the surface of the collagen gel (Fig. 2B). In spite of the upregulation of renin gene expression, cells positive for anti-renal cell carcinoma antibody were rarely found. AFP-positive cells were detected infrequently as well.

thumbnail image

Figure Figure 2.. Immunohistochemistry demonstrated that differentiating ESCs expressed ectoderm proteins β-III-tubulin (A) and cytokeratin (B) on collagen, endoderm proteins AFP (C) and albumin (D) on Matrigel, and mesoderm proteins cardiac troponin I (E) and renal cell carcinoma (F) on agarose after 21 days of culture. Positive cells are stained brown. Scale bar = 50 μm. Abbreviation: AFP, α-fetoprotein.

Download figure to PowerPoint

In summary, collagen substrate promotes cell attachment but does not facilitate ESC differentiation. ESCs tend to differentiate on collagen along all three lineages, albeit with low efficiency.

ESCs on Matrigel.

ESCs readily attached to Matrigel and formed aggregates within 2 hours after seeding (Fig. 1C). The size of aggregates increased with time, and the cells at the edges of the aggregates were spread. This became even more obvious on day 21 after seeding (Fig. 1D).

On Matrigel, endoderm gene expression was higher than that in undifferentiated ESCs. On day 21, the levels of AFP and albumin gene expression increased 10 ± 5 × 103-fold (n = 11; p = .0001) and 221 ± 112-fold (n = 11; p = .04), respectively, compared with those in the undifferentiated ESCs (Fig. 1I, 1J).

In contrast to endoderm lineage genes, expression of mesoderm genes (except the renin gene) and ectoderm genes was not upregulated: on day 21, the expression levels of NKX2.5, keratin, and NFH were not statistically different from those in undifferentiated ESCs (p = .2, p = .5, and p = .9, respectively) (Fig. 1K, 1M, 1N); renin gene expression in cells on Matrigel was 66 ± 17 × 103-fold (n = 5) higher than in undifferentiated ESCs (p = .001) (Fig. 1L).

Differentiation of ESCs on Matrigel into the endoderm lineage, revealed with real-time PCR, was confirmed by immunostaining of AFP- and albumin-positive cells (Fig. 2C, 2D). Many of these positively stained cells formed clusters that resided within the ECM. The cells were large and polygonal, with distinct round nuclei resembling those of hepatocytes. Thus, Matrigel facilitates cell-cell aggregation and endoderm differentiation.

ESCs on Agarose.

ESCs did not attach to agarose but rather formed floating aggregates that increased in size with time (Fig. 1E, 1F) and at day 11 reached 2–4 mm in diameter. Strikingly, ∼30% of the aggregates began spontaneously beating between days 11 and 14 in culture and continued beating until the end of experiments on day 30 (supplement online Video 1). Nearly one-third of the beating aggregates had more than one beating focus on the aggregate. The frequency of beating was 50–100 beats per minute. Between days 14 and 21, multiple transparent hollow cyst-like structures budded from the edge of the aggregates (Fig. 1F). A similar pattern of differentiation of ESC into beating aggregates of cardiomyocytes was observed in the pilot experiments with human ESC H1 (supplement online Video 2).

On agarose, the levels of expression both of two endoderm and of two mesoderm markers increased with time in culture, whereas the levels of expression of ectoderm markers remained low. In general, the levels of expression of all tested genes in ESCs cultured on agarose were similar to those on Matrigel, except that of the cardiomyocte-specific gene NKX2.5, which on day 21 postseeding was 45 ± 25-fold (n = 9) higher than in undifferentiated ESCs (p = .02). In contrast, keratin and NFH gene expression in ESCs on agarose remained statistically similar to that in the undifferentiated ESCs (p = .2) (Fig. 1M, 1N).

Immunostaining revealed cardiac muscle cell-specific troponin I protein in cells at the edges of the agarose-cultured aggregates, where the most extensive beating was observed (Fig. 2E). Both AFP- and albumin-positive cells were found in groups of cells cultured on agarose (not shown). Also, immunostaining revealed cells positive for anti-renal cell carcinoma antibody located predominantly in the lumen of the multilayered, stratified, glandular-like structures (Fig. 2F).

In summary, the nonadherent culture of ESCs on agarose promoted the formation of cell aggregates with a preferential cardiac-specific differentiation, capable of beating. In general, in these nonadherent cultures, both endoderm differentiation and mesoderm differentiation were facilitated.

ESCs in Suspension.

We further investigated the differentiation of ESCs in nonadherent culture by suspending the cells in a NASA-designed rotating wall vessel bioreactor. To investigate the effect of the ECM on suspended cells, we mixed ESCs with Matrigel in a suspension culture. Aggregates formed both with and without exogenous ECM (Fig. 1G, 1H). Unlike those on agarose, aggregates in the bioreactor were denser, and no budding at the edges was observed. Although in several experiments we observed cell aggregates beating, it was not consistent, in contrast to what we observed in nonadherent cultures on agarose. The spectrum of lineage differentiation genes in suspended cells was in general similar to those in nonadherent cultures on agarose (Fig. 1I–1N). On day 21, the levels of expression of the cardiomyocyte-specific gene NKX2.5 by differentiating ESCs in suspension with and without Matrigel were 15 ± 8-fold (n = 11, p = .08) and 29 ± 8-fold (n = 8, p = .02) higher, respectively, than in undifferentiated ESCs. Thus, like cells cultured on nonadhesive agarose, ESCs in suspension formed cell aggregates and differentiated along endoderm and mesoderm lineages, especially expressing a cardiomyocyte-specific gene, although no consistent beating was observed in this system.

Endogenous ECM in Differentiating ESC

To investigate whether, together with that of exogenous ECM, endogenous ECM production is important for ESC differentiation, we compared the levels of expression of the endogenous ECM genes in ESCs differentiating on various substrates and investigated the consequences of inhibition of endogenous collagen deposition on ESC survival and differentiation.

Expression of Endogenous ECM.

We focused on the expression of the genes for collagen type I α1 (COL1A1), collagen type IV α2 (COL4A2), fibronectin, and laminin-1 (Fig. 3A–3D).

thumbnail image

Figure Figure 3.. Levels of mRNA and protein expression of the endogenous extracellular matrix (ECM) genes (A–D) on ESCs cultured on three-dimensional substrates. The fold change is expressed as mean ± SE (n = 3–11). Endogenous collagen is shown by trichrome Masson's staining (E, F) (blue). The immunostaining (brown) of endogenous ECM protein is shown (G–J). Scale bar = 50 μm.

Download figure to PowerPoint

On collagen gels (which did not promote differentiation as evaluated above from lineage gene expression), levels of expression of the COL1A1, fibronectin, and laminin-1 genes remained similar to those in the undifferentiated ESCs (Fig. 3A, 3C, 3D), with an increase of the expression of the COL4A2 gene on day 21 (to the level of 6.4 ± 0.8-fold higher than that in the undifferentiated ESCs; p = .04; n = 3) (Fig. 3B). On Matrigel, COL1A1 expression was upregulated 5.0 ± 0.6-fold (p = .03; n = 11), a level significantly higher than that on all the other tested substrates (p = .001–.01) (Fig. 3A). A similar upregulation (9.1 ± 2.5-fold; p = .001; n = 9) of COL1A1 was observed when Matrigel was added to the suspension cultures. Moreover, in differentiating ESCs cultured on Matrigel, COL4A2 gene expression was increased 3.0 ± 0.5-fold (p = .0001; n = 3) on day 7 and 6.3 ± 0.2-fold (p = .0001; n = 9) on day 21 postseeding, relative to undifferentiated ESCs. When Matrigel was added to suspension culture, an 11 ± 2-fold (p = .0001; n = 7) increase in COL4A2 gene expression relative to that in undifferentiated ESCs was observed. In nonadherent culture on agarose, COL4A2 expression was upregulated as well: on day 21, it was 7.1 ± 0.5-fold (p = .0001; n = 9) higher than in the undifferentiated ESCs (Fig. 3B). Also, in cultures with agarose, laminin-1 gene expression was upregulated at day 7 (a 4.4 ± 0.2-fold increase compared with the undifferentiated ESCs; n = 5; p = .02) but decreased afterwards (Fig. 3D).

In ESCs cultured with Matrigel, fibronectin mRNA expression was also increased, in particular on day 21, whereas laminin-1 gene expression increased on day 7 but later decreased (Fig. 3D). Exogenous Matrigel in suspension cultures induced a 9.1 ± 1.5-fold (n = 7) increase of fibronectin gene expression relative to undifferentiated ESCs (p = .04) (Fig. 3C). Thus, in cell aggregates formed under adhesive conditions on Matrigel or in suspension, exogenous Matrigel facilitated collagen gene expression and, to a lesser extent, fibronectin expression.

Trichrome staining (Fig. 3E, 3F) showed that in agarose cultures, endogenous collagen was between single cell layers that formed the wall of the transparent hollow cyst-like structure (Fig. 3E) and at the edges of the aggregates (Fig. 3F). Immunostaining could not distinguish endogenous from exogenous collagen in cultures on collagen or Matrigel substrates; the latter also contain collagen. Immunostaining revealed that endogenous fibronectin (Fig. 3G) was abundant in all cultures and was localized predominantly around the basal layer of multicellular structures and also on the scattered cells. Endogenous laminin-1 (Fig. 3H) was expressed uniformly in pericellular areas and was found around cells in the multicellular structures. We stained undifferentiated ESCs for the latter two ECM. Both fibronectin (Fig. 3I) and laminin (Fig. 3J) were expressed on undifferentiated ESCs that formed colonies. Although in agreement with the gene expression for these two ECM, the staining seemed to be less intense, and it was difficult to quantify the difference in immunostainings.

Thus, in spite of the presence of exogenous ECM, differentiating ESCs expressed both ECM mRNA and proteins. Multiple linear regression analysis showed that the levels of lineage gene expression were positively correlated with the levels of expression of the ECM genes: in particular, the expressions of both of the endoderm lineage genes, AFP and albumin, correlated with the expression of the COL1A1 gene (p = .0001); the expressions of the mesoderm genes NKX2.5 and renin correlated with that of the fibronectin gene (p = .005).

Suppression of Endogenous Collagen Synthesis in ESCs.

The data presented above show that endogenous collagen genes and proteins were consistently expressed in differentiating ESCs on various substrates or in suspension, especially in cultures with Matrigel, and that this expression correlated with endoderm differentiation. To investigate the effect of collagen deposition on ESC survival and differentiation, we treated cell cultures with an inhibitor of collagen synthesis, cis-hydroxyproline (100 μg/ml). As observed after 24 hours in culture, cis-hydroxyproline did not affect the attachment of ESCs to collagen gel or to Matrigel, and ESCs remained unattached on agarose. At day 7, however, incubation with cis-hydroxyproline led to an almost complete loss of cells in collagen gels and on agarose cultures and to a decrease in the number of cells attached to Matrigel (Fig. 4). The remaining cells had a spherical morphology and were less spread at the edges of the aggregates than the controls. Lineage gene expression in cells that remained on Matrigel in these experiments was lower than in the controls (data not shown). Therefore, collagen deposition seems to be one of the survival factors for ESCs. The presence of exogenous collagen did not rescue the ESCs when endogenous collagen synthesis was blocked.

thumbnail image

Figure Figure 4.. The effect of cis-hydroxyproline on ESC morphology and attachment to type I collagen gel, agarose, or Matrigel on day 7. cis-Hydroxyproline resulted in an almost complete loss of cells on collagen gels and agarose cultures and in a decrease in the number of cells attached to Matrigel. Scale bar = 100 μm.

Download figure to PowerPoint

ESC Proliferation

Immunostaining with BrdU revealed that both the numbers and the localization of proliferating cells were significantly different in differentiating ESCs cultured under different culture conditions (Fig. 5). The highest proliferation was observed in nonadherent cultures, either on agarose (50% ± 7% of BrdU-positive cells) or in suspension (63% ± 6% of BrdU-positive cells). In suspension culture, addition of Matrigel did not significantly affect the rate of ESC proliferation relative to that of ESCs without an exogenous ECM (59% ± 4% of BrdU-positive cells). In differentiating ESCs cultured on collagen or on Matrigel, proliferation was lower than in nonadherent cultures (23% ± 6%, p = .001, and 22% ± 7%, p = .001, respectively) (Fig. 5F). Microscopic analysis showed that the proliferating cells were localized mainly in the multilayered glandular structures (Fig. 5B, 5D, 5E).

thumbnail image

Figure Figure 5.. Proliferating cells positively stained by anti-BrdU antibody (brown color) in differentiating ESCs cultured on collagen (A), agarose (B), or Matrigel (C) or in the bioreactor without exogenous extracellular matrix (C) or with Matrigel (D). The percentage of positive staining was calculated on the basis of the total number of cells in each view (mean ± SE) (F). Scale bar = 50 μm. Abbreviation: BrdU, 5-bromodeoxyuridine.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Because of their pluripotency, ESCs are a potential source for transplantation therapy, tissue engineering, and drug screening. The development of simple tools for efficient ESC differentiation into a particular lineage is critically important for all these applications. Today, such tools are not yet available. The ECM can be used as one such tool to guide ESC differentiation into a particular lineage in vitro, since the ECM plays pivotal roles in cell differentiation, as well as in cell migration and proliferation in vivo. Unfortunately, neither the patterns of ESC differentiation triggered by different ECM nor the mechanisms mediating this differentiation are well known.

In our current study, we focused on investigating the ways in which the ECM regulates differentiation of rhesus monkey ESCs. We found that the pattern of differentiation is determined by the pattern of cell attachment. In particular, we found that (a) reduced attachment/spreading on the substrate is associated with cell-cell aggregation that promotes mesoderm and endoderm differentiation; (b) suspension of ESCs on nonadhesive substrates or in a bioreactor facilitates cell proliferation and differentiation into beating cardiomyocytes; and (c) in spite of the availability of exogenous ECM, production of endogenous collagen is essential for ESC survival, and the expression of endogenous ECM genes correlates with lineage gene expression.

We compared the ways in which two biological ECM, collagen I and Matrigel, promote ESC attachment and differentiation. The extent of spreading is different on these two types of ECM: ESCs readily attached to and spread on the surface of a collagen gel, which was covered with cells in 21 days, whereas Matrigel was obviously less adhesive, so that cells attached to each other and formed aggregates such that only cells at the edges of these aggregates spread upon the substrate. There were no signs of toxicity on either of these substrates: ESC proliferation in both types of culture was similar.

The two biological ECM differentially promote ESC differentiation. On highly adhesive collagen gels, where the cells readily spread, there was little cell differentiation, as evidenced by the lack of increased expression of the major lineage genes except the renin gene. Moreover, upregulation of this gene did not result in increased protein expression, and the relative upregulation of this gene compared with that in the undifferentiated ESCs may be overestimated because in the latter cells it remained below the detection level. On the less adherent Matrigel, to which ESCs attach less readily and instead form aggregates, ESCs differentiated efficiently, predominantly along the endoderm lineage, as evidenced by strong endoderm gene expression (the two tested endoderm-lineage genes were upregulated 220–10,000 times).

To further investigate whether the attachment is related to ESC differentiation, we cultured cells on the nonadhesive substrate agarose and in suspension in a rotating wall vessel bioreactor. In both cases, ESCs formed aggregates and efficiently differentiated along endoderm and mesoderm lineages. Especially striking in both types of nonadherent cultures was the differentiation into cardiomyocytes. On agarose, the cardiomyocyte-specific NKX2.5 gene was upregulated 45-fold. Also, immunostaining revealed expression of the cardiac muscle-specific troponin I. The expression of both cardiomyocyte-specific mRNA and -protein was associated with the development of functionally differentiated cells that started to beat, and with time, large beating aggregates were formed. Moreover, the peripheries of the cell aggregates were transformed into hollow (spherical) structures that often beat as one entity. These beating aggregates survived for at least a month. In suspension in the rotating wall vessel, cells in aggregates also expressed a cardiomyocyte-specific gene, NKX2.5; however, beating was rare. The morphology of the aggregates in the two suspension cultures differed, as in a true suspension in the bioreactor, cell aggregates appear more dense than those formed in cultures over an agarose substrate. Globally, troponin I expression was strongest in the beating cells on agarose. Also, at the level of individual aggregates, high expression of troponin I seemed to be in the area where beating occurs. However, cell aggregates in suspension expressed troponin I as well but did not beat. Future investigation of colocalization of troponin I expression and beating at the level of individual cells will require a more detailed analysis of vitally stained cells with confocal microscopy. Since the pioneer study on human ESC differentiation into functional cardiomyocytes [18], multiple human ESC lines have been shown to be able to differentiate into cardiomyocytes [19] when EBs are plated on gelatin-coated dishes. Here, our results showed that the beating foci in primate ESC aggregates can grow and possibly mature with time in a simple suspension culture without further attachment of the cell aggregates or addition of growth factors. Thus, our data suggest that the less the ESCs attach to a substrate, the more they aggregate and the more efficiently they differentiate, although it seems there is an optimal aggregation density for such differentiation.

Furthermore, local cell-cell interactions seem to tune ESC differentiation in a finer way: for example, on a collagen gel, where the average expression of almost all lineage genes was as low as in undifferentiated ESCs, immunostaining revealed β-III-tubulin-positive cells located predominantly in multilayered glandular-like structures, whereas cytokeratin-positive cells were localized predominantly in stratified structures, as well as on substrate-attached cells. The very fact that ESCs are capable of selectively expressing specific markers of differentiation in different structures is further evidence of the critical role of complex cell-cell interactions in ESC differentiation.

It seems that cell-cell aggregation is even more critical for ESC differentiation than are the biologically active components of the ECM. For example, differentiation associated with cell-cell aggregation was evident both on Matrigel, which has many biologically active components, and on nonbiological agarose. Also, when Matrigel was added to suspension cultures, it did not significantly promote differentiation of ESCs beyond what was already triggered by cell-cell aggregation. This finding is consistent with the earlier published reports that ECM control of cell differentiation of kidney epithelia and keratinocytes is mediated by changes in cell shape and adhesiveness [20, 21]. Also, it has been reported that cuboidal cell shape determined by basement membrane proteins is important for Matrigel-induced hepatocyte differentiation, including activation of albumin expression [10, 22].

Thus, the efficiency of ESC attachment to a particular substrate resulted in different cell morphologies, and the extent of cell-cell aggregation correlated with cell differentiation. It remains to be understood whether the changes in cell attachment and spreading that are determined by ECM adhesive properties mediate cell differentiation into a particular lineage or whether these changes are the consequence of differentiation. Whatever the mechanism of ESC differentiation in the above-described systems is, our results provide a way to facilitate differentiation into various lineages using different ECM or nonbiological substrates. More studies are needed to understand which of the above-described patterns of ESC differentiation are typical for this particular cell line and which are typical for ESCs in general. Our pilot experiment with human ESC H1 revealed that at least in cultures on agarose, these cells differentiate similarly to monkey ESCs, which form beating aggregates of differentiated cardiomyocytes and express troponin I.

Cell-cell (through cadherins and cell adhesion molecules) and cell-matrix (through integrins) interactions have been proven to play a crucial role during embryogenesis. For example, the compaction of the inner cell mass requires E-cadherin [23], laminin appeared as early as the 2-cell stage, entactin/nidogen appeared at the 16-cell stage [24], and fibronectin and type IV collagen appeared later in the inner cell mass of 3–4-day-old blastocysts [25]. The existence of these various ECM components makes it clear that at a given time and place, the extracellular matrix has the potential to provide specific environmental information to cells. Remarkably, regardless of whether ESCs are cultured on a biological ECM or on a nonbiological substrate, production of an endogenous ECM is required for cell survival. Inhibition of endogenous collagen induced gradual loss of ESCs, starting 24 hours after seeding. Neither a simple matrix, such as type I collagen, nor the complex matrix Matrigel was able to rescue these cells. This result is in agreement with the previously demonstrated importance of endogenous collagen synthesis for fibroblast attachment to the substratum [11, 12]. It has been suggested that cis-hydroxyproline may have an additional effect other than inhibition of collagen synthesis [26]. It is conceivable that the additional effects of cis-hydroxyproline are indirect and secondary to the loss of collagen resulting in the loss of cell adhesion. It is reasonable to suggest that endogenous ECM not only plays a critical role in ESC survival but also affects ESC differentiation. For example, we found that the expression of the cardiomyocyte-specific gene NKX2.5 correlates with an increase in fibronectin gene expression. However, immunostaining revealed fibronectin in undifferentiated ESCs as well. It appears that fibronectin expression in ESCs precedes upregulation of lineage gene expression and may be related to cell aggregation, as fibronectin was earlier implicated in embryonic carcinoma cell aggregation [27] and in an early stage of fetal heart development [28]. However, on the protein level, its relationship with ESC differentiation seems to be more subtle and needs further investigation. We also found that the increase in expression of the endoderm genes AFP and albumin correlates with increased COL1A1 gene expression. This is consistent with an earlier finding that during the differentiation of embryonic carcinoma cells to endoderm-like cells, type I collagen synthesis was increased [29]. However, testing the causal relationship between lineage gene expression and endogenous ECM expression requires more direct experiments. In these future studies, to understand the factors that are critical for the ESCs to maintain their “stemness,” it will be important to evaluate the kinetics of the absolute expression levels of various lineage genes and endogenous ECM genes in the course of ESC differentiation (starting from undifferentiated cells) and to compare these values with the amount of expressed proteins.

In conclusion, adhesiveness to various ECM and nonbiological substrates determines the differentiation of ESCs in such a way that efficient cell-cell aggregation, together with less efficient cell attachment and spreading, results in more efficient cell differentiation. Deposition of endogenous ECM is required for cell survival and is also related to cell differentiation. Although future experiments should determine the relative contribution of exogenous and endogenous ECM to ESC differentiation and its mechanisms, the present study provides a way of designing in vitro systems for directing ESC differentiation into endoderm and mesoderm lineages.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

This work was funded by the NIH/NICHD Intramural Program and, in part, by the National Aeronautics and Space Administration/NIH Center for Three-Dimensional Tissue Culture.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
Margolis_Supp_Legend.pdf13KSupplemental Legends

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.