Differentiation of embryonic stem (ES) cells generally occurs after formation of three-dimensional cell aggregates, known as embryoid bodies (EBs). This differentiation occurs following suspension culturing of EBs in media containing a high (25 mM) glucose concentration. Although high-glucose-containing media is used for maintenance and proliferation of ES cells, it has not been demonstrated whether this is a necessary requirement for EB development. To address this, we examined the growth and differentiation of EBs established in 0-mM, 5.5-mM (physiological), and 25-mM (high) glucose concentrations, through morphometric analysis and examination of gene and protein expression. The effect on EB development of supplementation with basic fibroblast growth factor (FGF2) was also studied. We report that the greatest rate of EB growth occurs in 5.5 mM glucose media. A morphological study of EBs over 104 days duration under glucose-containing conditions demonstrated the development of all three major embryonic cell types. The difference from normal human development was obvious in the lack of rostrocaudal control by the notochord. In the latest stages of development, the main tissue observed appeared to be cartilage and cells of a mesodermal lineage. We conclude that physiological glucose concentrations are suitable for the culturing of EBs, that the addition of FGF2 enhances the temporal expression of genes including POU5F1, nestin, FOXA2, ONECUT1, NEUROD1, PAX6, and insulin, and that EBs can be cultured in vitro for long periods, allowing for further examination of developmental processes.
The establishment and maintenance of human embryonic stem (hES) cell lines derived from the inner cell mass of blastocysts has been delineated only in recent years [1, 2]. These studies demonstrated that hES cells are capable of extensive proliferation and multilineage differentiation, introducing the prospect of using hES-derived cells in cell replacement therapies. Even so, to date there have been few reported studies of embryoid body (EB) differentiation over an extended period in culture. This study attempts to investigate the development of EBs at selected time periods up to 104 days, using three distinct culture media.
To obtain physiologically functional cells from embryonic stem (ES) cells that are suitable for transplantation, controlled differentiation procedures must be implemented. Induction of ES cell differentiation generally occurs through an intermediate step of formation of EBs, which are complex three-dimensional cell aggregates . The formation and early differentiation of EBs occurs in two phases: within the first 2–4 days of suspension culture endoderm forms on the surface of EBs, giving rise to structures termed “simple embryoid bodies”; subsequently, around Day 4, “cystic” EBs develop with the formation of a central cavity and differentiation of a columnar epithelium with a basal lamina . In vitro culturing of EBs commences with the development of cells indicative of the ectoderm, mesoderm, and endoderm germ lineages . Continued culturing can result in more differentiated cell types, including insulin-producing cells [6–8], neuronal cells , and hematopoietic cells . Specialized cells derived through differentiation procedures may thus be useful in the future treatment of diseases involving cell loss, such as Parkinson disease, ischemic heart disease, and type 1 diabetes.
Currently, despite numerous research efforts examining ES cell differentiation, limited knowledge exists concerning the optimum basic medium requirements for culturing EBs—in particular, the glucose concentration that is best for sustaining EB growth and differentiation. The significance of this can be seen in the results of a study on insulin production by ES-derived cells, which was improved when media with a physiological glucose concentration of 5.6 mM was used in the final differentiation stage, following growth at the high glucose concentration of 25 mM . It was hypothesized that the physiological glucose concentration in the final differentiation stage was instrumental in developing improved cell function . This is supported by research reporting the loss of differentiation  and impairment of insulin secretion  in adult pancreatic β cells exposed to sustained high glucose concentrations.
Additionally, wider implications exist regarding the effect of glucose concentration on the differentiation of ES cells into other cell types, because early human and mouse embryo development in vitro was shown to be enhanced in medium lacking glucose [12, 13]. However, previously described protocols for ES cell differentiation, including those using adherent or suspension cultures of EBs, predominantly use media containing high glucose concentrations, presumably based on the use of high glucose media for ES cell maintenance. Therefore, the necessity of using media with high glucose content for ES cell culture and differentiation is brought into question.
Media used in ES cell culture and differentiation are often supplemented with basic fibroblast growth factor (FGF2; also known as Fibroblast Growth Factor 2). FGF2 has a role in stimulating cell proliferation , and in the maintenance of ES cells in an undifferentiated state, by repressing bone morphogenetic protein (BMP) signaling . Additionally, FGF2 has been suggested to have a role in the differentiation of ES cells, with reports suggesting a role in inducing neural differentiation , as well as insulin production by EBs . Furthermore, a study using an EB model of early development reported that FGF signaling contributes to the regulation of basement membrane formation during epithelial morphogenesis .
These studies led us to investigate the effect of glucose and FGF2 on EB development and differentiation, particularly with respect to selected developmental periods at 28, 60, and 104 days. This was achieved through the examination of EB gross morphology, gene expression by RT-PCR, EB histomorphology, and protein expression by immunofluorescent staining.
Materials and Methods
Unless otherwise stated all cell culture reagents were obtained from Invitrogen/Gibco.
hES Cell Line
The hES cell line, ESI-hES3, was obtained from Embryonic Stem Cell International Pte Ltd. and maintained in gelatin-coated organ culture dishes (Becton Dickinson) on gamma-irradiated (45 Gy) primary human fetal fibroblast (HFF; passage 6) feeder layers. Informed consent and approval from the Human Research Ethics Committee (HREC) of The University of New South Wales (UNSW) (HREC 02247) was given for use of human fetal material; UNSW Institutional Ethics Committee approval (HREC 01270) and South Eastern Sydney Area Health Services HREC (02/202) approval was given for use of hES cells. Subculturing of hES cells was performed every 7 days by mechanical dissection. The hES cells were maintained at 37°C, 5% CO2 in Dulbecco modified Eagle medium (DMEM) high glucose, supplemented with 20% defined fetal bovine serum (HyClone, Perbio Science Co.), 2 mM L-glutamine, 0.1 mM nonessential amino acids, 0.01 mM 2-mercaptoethanol, 1× insulin-transferrin-selenium, 25 U/ml penicillin, and 25 μg/ml streptomycin.
EB Formation and Differentiation In Vitro
Formation of EBs was induced by mechanically dissecting undifferentiated hES colonies into pieces less than 200 μm in size using a 1-μm microdissection chisel (Eppendorf). Pieces were transferred to 90-mm sterile petri culture dishes (Sarstedt Australia Pty. Ltd.) and grown in suspension culture at approximately 30 EBs per dish.
EBs were cultured at 37°C, 5% CO2, under three culture conditions: DMEM high glucose (25 mM), DMEM physiological glucose (5.5 mM), and DMEM no glucose (0 mM). Culture media consisted of DMEM supplemented with 20% Knockout serum replacement, 0.1 mM nonessential amino acids, 0.01 mM 2-mercaptoethanol, 4 mM L-glutamine, 1 mM sodium pyruvate, 25 U/ml penicillin, and 25 μg/ml streptomycin. Additionally, EBs were cultured in the presence and absence of 4 ng/ml human recombinant FGF2 (reconstituted in 0.1% albumin Fraction V). Culture media was changed every 2–3 days with media equilibrated for 24 h at 37°C, 5% CO2 before use. For FGF2-supplemented cultures, FGF2 was added immediately after changing media.
To determine whether glucose concentration affected EB growth and development, gross morphological analysis was performed. Phase contrast images were captured of 10 EBs in each glucose concentration every 3–4 days for measurement. Because of the three-dimensionality of EBs, the measurement recorded for an individual EB was the average of the smallest and largest diameter of the EB. The mean diameters of EBs at specific time points was calculated from the average diameters of 10 EBs at each time point. To maintain consistency across measurements, EBs were only included in the analysis if the morphology met the following criteria: rounded, cystic, and with well-defined borders.
Data presented are mean ± SEM. Statistical analysis was performed using analysis of variance and a Duncan multiple range test (NCSS statistical package). Statistical significance was determined as P < 0.05.
RNA Extraction and RT-PCR Analysis
Total RNA was extracted from three or four EBs collected at days 2, 7, 16, 21, 57, and 104 from each glucose concentration of the non-FGF2-supplemented condition. From the FGF2-supplemented set, seven EBs were harvested at days 7, 15, 22, and 28. EBs collected were representative of the morphologically heterogeneous population in the petri dishes. Total RNA was also extracted from hES colonies (1 colony per extraction) and from human fetal tissue (pancreas, kidney, brain, liver and cartilage) for use as controls. Total RNA was extracted using the RNeasy Mini kit (Qiagen) and treated with DNase I (Qiagen) following the supplier’s recommended protocol. Reverse transcription was performed using SuperScript II First Strand Synthesis RT kit and oligo(dT) primers (Invitrogen) following the manufacturer’s protocol, and cDNA obtained was diluted 1:2.
A Platinum Taq polymerase kit (Invitrogen) was used for detection of REN (renin), FOXA2 (Forkhead Box A2), IPF1 (insulin promoter factor 1, also known as PDX1), SLC2A2 (solute carrier family 2; facilitated glucose transporter; previously known as GLUT2), ONECUT1 (one cut domain, family member 1; also known as HNF6), NEUROD1 (neurogenic differentiation 1), glucagon, and insulin mRNA, and RedHot Taq polymerase (ABgene; Epson) was used for detection of remaining genes. The amount of diluted (1:2) cDNA used in each reaction was 2 μl for IPF1, ONECUT1, NEUROG3 (neurogenin 3), NEUROD1, PAX6 (paired box gene 6), and glucagon, 8 μl diluted cDNA for insulin, or 1 μl diluted cDNA for the remaining 10 genes. PCR reactions were performed 2–4 times and the results obtained were consistent. Negative controls were performed in the absence of cDNA template.
The cycling parameters for PCR using Platinum Taq were as follows: initial denaturation step at 94°C for all reactions except NEUROD1 (95°C) for 2 min, followed by 35 cycles of denaturation (94°C for 30 sec or 96°C for 30 sec for NEUROD1), primer annealing (30 sec or 45 sec for glucagon at annealing temperatures shown in Table 1) and extension (72°C for 1 min). A final extension step was carried out at 72°C for 5 min.
|Gene||GenBank accession no.||Sequenc e (5′—3′)||Product size (bp)||Annealing temp. (°C)||Positive control|
|Germ lineage markers|
|Pancreatic endocrine markers|
|Glucagon||NM 002054||tctctcttcacctgctctgtt||acaatggcgacctcttctg||466||64||Fetal pancreas|
|Glucose transport marker|
|Gene||GenBank accession no.||Sequenc e (5′—3′)||Product size (bp)||Annealing temp. (°C)||Positive control|
|Germ lineage markers|
|Pancreatic endocrine markers|
|Glucagon||NM 002054||tctctcttcacctgctctgtt||acaatggcgacctcttctg||466||64||Fetal pancreas|
|Glucose transport marker|
Primer sequence obtained from .
Sense primer from ;antisense primer designed by Melissa Khoo.
The PCR performed using RedHot Taq began with initial denaturation at 96°C for 2 min (POU5F1 (POU domain, class 5, transcription factor 1; also known as OCT3/4), ENO2 (enolase 2, gamma, neuronal), MATN1 (matrilin 1, cartilage matrix protein) or 97°C (others). PCR cycling was carried out for 35 cycles, with the denaturation steps conducted at 96°C (β-actin, nestin, ENO2, MATN1) or 97°C (others) for 30 sec, or 94°C (POU5F1) for 1 min. Annealing steps were carried out at the temperatures shown in Table 1 for 30 sec, except for AFP (α-fetoprotein, 45 sec), and were followed by extension at 72°C for 1 min. A final extension step was carried out at 72°C for 5 min. PCR products were visualized following separation by agarose gel electrophoresis, and size of products was approximated using 100-bp or 1-Kb Plus DNA ladders (Invitrogen). The POU5F1 primers were kindly supplied by Dr. Tomas Stojinov (Sydney IVF). Primer data used for PCR reactions are shown in Table 1.
Histological and Immunofluorescence Analysis
To examine the general histomorphology of EBs, hematoxylin-eosin staining was performed on paraffin-embedded EBs that had been fixed overnight at 4°C in 10% neutral buffered formalin. Paraffin blocks were sectioned at 5 μm. Periodic acid-Schiff (PAS) and Alcian Blue staining were performed using standard techniques to detect neutral and acidic mucopolysaccharides as an indication of putative endodermal differentiation.
For immunofluorescence detection, slides were dewaxed in HISTOSOLV (Amber Scientific; 2 × 5 min) and rehydrated though an EtOH/H20 gradient (2 × 5 min 100%, 1 × 5 min 95%, 1 × 3 min 90%, 1 × 3 min 85%, and 1 × 3 min 70%) followed by 3 × 1 min washes in H2O. Slides were then treated with REVEAL (Biocare Medical) antigen retrieval solution for 20 min at 90°C, or for Nanog and Oct4 staining with antigen retrieval citrate buffer for 10 min at >100°C. Slides were cooled and washed in H2O before Image iT FX signal enhancer (Molecular Probes; Invitrogen) was applied for 30 min at room temperature to block nonspecific binding. Slides were washed with TRIS buffer, pH 7.4, and primary antibodies were applied for 60 min at room temperature. Slides were then washed with TRIS buffer pH 7.4 and secondary antibodies were applied in the dark for 30 min at room temperature. Finally, slides were washed with TRIS buffer, pH 7.4, and mounted with Prolong Gold Antifade containing DAPI (Molecular Probes) and a coverslip. Primary anti-human antibodies used were mouse anti-Oct4 (1:20), mouse anti-nestin (1:20), rabbit anti-laminin (1:100), and goat anti-collagen type IV (1:20), all from Chemicon; goat anti-nanog (1:50) from Santa Cruz; and guinea pig anti-insulin (1:2000) and rabbit anti-AFP (1:20), both from DakoCytomation. Secondary antibodies, all at 1:500 dilution, were Alexa Fluor 594 donkey anti-mouse IgG, Alexa Fluor 488 donkey anti-rabbit IgG, and Alexa Fluor 568 donkey anti-goat IgG (Molecular Probes), and FITC rabbit anti-guinea pig IgG (DakoCytomation).
Morphology and Histomorphology of EBs
A representative set of images of EBs cultured in 5.5 mM glucose media is depicted in Figure 1. Typically, dissected hES colony pieces formed rounded discrete structures and acquired an EB appearance within 12 h (Fig. 1, A and D). The external morphology of EBs was in some cases highly variable, even between EBs growing under the same conditions. Continued growth of EBs in suspension culture resulted in the acquisition of a cystic appearance (Fig. 1C) and increased diameter.
Differences were observed between EBs cultured in the absence (Fig. 1, A–C) and presence of FGF2 (Fig. 1, D–F). Histological analysis after hematoxylin-eosin staining showed EBs cultured in FGF2-supplemented media to have a higher cell density than EBs cultured in media lacking FGF2 (Fig. 2, A and B). Additionally, EBs supplemented with FGF2 had a reduced cystic appearance.
Histological examination of EBs using hematoxylin-eosin staining in FGF2-supplemented and nonsupplemented EBs at Days 28, 60, and 104 showed the presence of all three embryonic germ layers at a basic morphological level. At Day 28, there was evidence of marked proliferation within EBs cultured in 5.5 mM and 25 mM glucose media, accompanied by apoptosis observed as dark rounded apoptotic profiles (condensed rounded dark nuclear material). Furthermore, at the periphery of EBs, particularly the larger EBs, there was development of tissue resembling layers of ectodermal neuroblasts and neuroepithelium (Fig. 2A, arrows). Neural rosettes were also observed in EBs cultured in both 5.5 mM and 25 mM glucose concentrations (Fig. 3, A and B). However, there appeared to be no evidence of a basement membrane for this layer and some evidence of apoptotic profiles in the neuroepithelium (Fig. 2B insert). Additionally, there was a considerable size range of EBs and occasional presence of small cystic bodies (Fig. 3B).
Cultures grown for 60 days using 5.5 mM and 25 mM glucose media contained large EBs with cystic bodies lined by cuboidal epithelium that occasionally became stratified (Fig. 2E, arrow and arrowhead). Areas of amorphous matrix as well as regions resembling mesenchyme developing into cartilage were present (Fig. 2, D and E). Some EBs at Day 60 contained an epithelium lining at the periphery, which developed into endodermal columnar and pseudostratified columnar epithelia with goblet cells (see Fig. 6A). When stained with PAS, these inclusions stained positively for mucopolysaccharides (see Fig. 6B). Although there was evidence of apoptotic profiles in the lumina of some cysts, there did not appear to be cell death in the body of the EBs at Day 60.
At Day 104, EBs cultured in both 5.5 mM and 25 mM glucose media had a noticeably solid texture on touch. The bodies of the EBs resembled developing mesodermal hyaline cartilage (Fig. 3, C and D). In a few areas, the surface columnar epithelium formed cysts that occasionally contained inclusions of intensely staining eosinophilic material (Fig. 3D) or mucopolysaccharides. There was no apparent major cellular development from Day 60 to Day 104.
At Days 28, 60, and 104, culturing of EBs without glucose appeared to severely retard the cellular differentiation and development of EBs (Fig. 2F).
Effect of Glucose Concentration and FGF2 on EB Size
The mean diameter of non-FGF2-supplemented EBs cultured in the different glucose concentrations over a period of 28 days was variable (Fig. 4); however, EBs cultured in 5.5 mM glucose were found to have a significantly greater diameter (P < 0.05) in comparison to EBs cultured in 0 mM and 25 mM glucose at 7 of the 11 time points examined. The absence of glucose in culture medium resulted in EBs of the smallest size, with the difference being statistically significant at Day 1 and at Days 15–20. Similar gross differences were observed in the morphology of FGF2-supplemented EBs cultured at the different glucose concentrations; however, this could not be quantified, because FGF2-supplemented EBs tended to have an aggregated or budding morphology. In addition, EBs cultured in FGF2-supplemented media, but in the absence of glucose, appeared much smaller and less developed.
From Figure 4, it appears that the growth of EBs peaks at Day 18. Beyond this time point, EBs tended to display budding and/or aggregated morphologies, making them unsuitable for measurement according to the criteria selected. This factor was taken into account when determining the percentage increase in size of EBs by using the mean diameter obtained at Day 18 as the final diameter. EBs cultured in 5.5 mM glucose increased 2.5-fold in size over the 18-day time course, whereas EBs grown in 25 mM and 0 mM glucose both had lesser increases of approximately 2-fold.
Pluripotency and Germ Lineage Marker Expression by EBs
To determine whether the differences observed in gross morphology were reflected in the pattern of gene expression, we analyzed pluripotency and germ lineage marker expression in developing EBs by RT-PCR and immunofluorescent staining. Markers of pluripotency (POU5F1, NANOG (Nanog homeobox), and the germ lineages, ectoderm (nestin, ENO2), endoderm (AFP) and mesoderm (REN, MATN1) were included in this analysis.
In the non-FGF2-supplemented EBs, the expression of POU5F1, NANOG, and nestin progressively decreased with time; however, low-level expression of POU5F1 and nestin persisted until Day 21 (Fig. 5A). Immunofluorescent staining confirmed nestin expression within the neuroepithelial-like structures and the neural rosettes (Fig. 6C). In contrast, the expression of REN and, in particular, AFP increased with time (Fig. 5A). Minor differences were observed in the expression of lineage markers cultured in the different glucose concentrations, suggesting that glucose concentration did not greatly affect gene expression.
When EBs were cultured in the presence of FGF2, the most marked difference observed was an increase in nestin expression with time and the earlier expression of REN (Fig. 5B). Immunofluorescent staining for nestin revealed an increased expression within the neuroepithelial-like structures and the neural rosettes (results not shown). Furthermore, expression of nestin and REN was seen to be dependent on the presence of glucose in the culture medium, with lack of glucose resulting in lack of expression. In addition, an increase in expression was observed in a second ectodermal marker, ENO2, and mesodermal marker, MATN1, which both showed a dependence on glucose. Both pluripotency markers POU5F1 and NANOG had declining expression through to Day 28 (Fig. 5B), contrary to the declining expression observed by Day 16 in the non-FGF2-supplemented EBs (Fig. 5A). POU5F1 and NANOG were also observed to be weakly, randomly expressed in cells in the core of the EBs by immunofluorescent staining (results not shown).
Immunofluorescent staining revealed that cells that stained positively for PAS (Fig. 6B) and were classified as endodermal columnar and pseudostratified columnar epithelial cells were also observed to be positive for AFP, an endodermal marker (Fig. 6D).
Dual immunofluorescent staining for laminin and collagen type IV confirmed the presence of cells of the mesodermal lineage. There was an increase in staining for both markers at Days 60 (Fig. 6E) and 104 (Fig. 6F), over that observed in Day 28 EBs. The strongest immunofluorescent staining for both laminin and collagen type IV was revealed in the majority of EBs at Day 104, which were mainly developing cartilage (Fig. 3, C and D).
Pancreatic and Neural Marker Expression by EBs
To determine whether EBs exhibited a pattern of expression indicative of more advanced development, markers of pancreatic (endodermal) and neural (ectodermal) differentiation were examined by RT-PCR. Genes studied in this analysis included NEUROG3 and NEUROD, which are expressed in both pancreatic and neural lineages. Also included were the definitive endoderm marker FOXA2, pancreatic transcription factors IPF1, ONECUT1, and PAX6, pancreas/liver specific glucose transporter SLC2A2, insulin, and glucagon.
EBs cultured in the absence of FGF2 showed decreasing expression of FOXA2 over the 21-day culture period, whereas most markers indicative of precursor pancreatic or neural cells, including IPF1, ONECUT1, PAX6, NEUROG3, and NEUROD1, as well as the β cell marker insulin, were absent from most samples (Fig. 7A). The only markers that showed increasing levels of expression were SLC2A2 and the islet α cell marker glucagon (Fig. 7A). In addition, higher SLC2A2 and glucagon mRNA levels were observed in EBs cultured in glucose-containing media.
Because insulin mRNA expression was not detected at any time point or glucose concentration up to 21 days, the non-FGF2-supplemented EBs were maintained in culture for extended periods. At Day 57 and 104 strong glucagon expression and very weak expression of insulin became apparent in EBs cultured in 5.5 mM and 25 mM glucose. EBs cultured in 0 mM glucose displayed only very weak expression of glucagon (Fig. 7C).
EBs cultured in the presence of FGF2 were also examined using the same markers. The most notable differences observed were the expression of NEUROD1, PAX6, ONECUT1, and insulin by EBs in both glucose-containing media at earlier time points and at greater levels of expression (Fig. 7B), unlike the absent/inconsistent low-level expression in non-FGF2-supplemented EBs (Fig. 7A). Glucagon expression was also present in a glucose-dependent pattern in FGF2-supplemented EBs similar to that seen in unsupplemented EBs, with expression absent or low in EBs grown in 0 mM glucose and highest in EBs cultured in 5.5 mM glucose. Expression of FOXA2 in the FGF2-supplemented EBs differed from its expression in the non-FGF2 EBs, in that expression was observed to increase with time and was greater when glucose was present in the culture medium (Fig. 7B). As with the non-FGF2 condition, NEUROG3 and IPF1 expression were not detected in any sample regardless of glucose concentration (Fig. 7). Additionally, SLC2A2 mRNA expression was similar to that seen in the non-FGF2-supplemented EBs and was correlated with glucose concentration (Fig. 7).
The simplest interpretation of these results is that FGF2 supplementation of media induces expression of some neural and pancreatic endocrine markers in suspension cultured EBs. Additionally, a 5.5-mM glucose concentration was found to enhance the expression of certain genes, including SLC2A2, glucagon, and ONECUT1, with the greatest effect observed in EBs cultured with FGF2 supplementation.
The aim of the present study was to examine the importance of glucose and FGF2 on EB growth and development, particularly through examination of gross morphology, histomorphology, and gene expression. The key findings were that the greatest rate of EB growth occurred in 5.5 mM glucose rather than in a high glucose concentration (25 mM); FGF2 supplementation induced higher neural gene expression (nestin and NEUROD1) in EBs at an earlier stage and in a glucose-dependent manner; supplementation with FGF2 also induced higher expression of pancreatic endocrine genes, including insulin, ONECUT1, and PAX6. Additionally, EBs were still viable after extended periods in culture, and displayed aspects of embryonic differentiation.
In the morphological studies of EB development over 104 days’ duration, ectoderm-, mesoderm-, and endoderm-derived tissues were observed in a considerably disorganized pattern. Neuroepithelial-like peripheral tissue observed in EBs at Day 28 was similar to that found in the neural tube of the 28-day human embryo, albeit appearing to lack a basement membrane and without precise delineation into the neural tube . The disorganized development in the EBs is to be expected because there is no rostrocaudal induction of the EB in comparison to the implanted embryo. In the early embryo, the notochord functions as the primary inductor, instigating a series of signal-calling events, which ultimately transform unspecialized embryonic cells into definitive tissues . In this study, neuroepithelial development had occurred in EBs by Day 28, but without the presence of the basement membrane and the resulting lack of inductive messages, further development was unable to proceed. Instead, the neuroepithelial-like tissue appeared to revert to a more primitive lineage, probably by dedifferentiation into a mesenchymal cell lineage.
The apoptotic profiles observed among the proliferating cells of Day 28 EBs resembled those observed in the developing brains of Day 21 guinea pig embryos subjected to hyperthermic stress . Additionally, the neural rosettes seen in the EBs were similar to those produced at Embryonic Day 13 (equivalent to human Embryonic Day 21–23) in the guinea pig embryo following maternal hyperthermia , reflecting abnormal development.
Our observations show that cell growth is not detrimentally affected by reduced, but not zero, glucose concentrations. In addition, the results indicate that the greatest rate of EB growth and differentiation occurs at a physiologically normal glucose concentration rather than at the high glucose concentrations currently used in ES cell differentiation procedures. Furthermore, we found no advantages in the gene expression of EBs cultured in 25 mM glucose media as opposed to 5.5 mM glucose media.
Substrate utilization studies with human and mouse preimplantation blastocysts have demonstrated enhanced early embryo development in medium lacking glucose, as well as a greater dependency on pyruvate  or EDTA and glutamine  instead of glucose. Consistent with this, our results suggest that cells within EBs cultured in 0 mM glucose are capable of adapting to other substrates present in the medium (e.g., pyruvate and glutamine) for metabolism. The types of substrate used in media for ES cell differentiation may need to be reassessed, particularly if substrate preferences are as complex as those in utero, where a switch from pyruvate to glucose requirement has been reported during mouse and human embryonic development [20, 21].
Supplementation of the culture media with FGF2 resulted in the observation of striking differences in gene expression when compared to the unsupplemented condition. The effects of glucose concentration were also more apparent in the presence of FGF2, with EBs grown in the absence of glucose appearing restricted in growth and gene expression. However, histomorphological examination revealed no major difference between FGF2-supplemented and unsupplemented EBs, apart from slightly greater neuroepithelial development in FGF2-supplemented EBs.
It has been reported that FGF2 belongs to a group of factors that promotes differentiation to the ectodermal and mesodermal lineages . Additionally, FGF2 has been shown to have numerous roles in the developing and adult nervous system, including the stimulation of proliferation and differentiation of multipotent progenitors in the mammalian brain . Our observations of increased nestin (neuroectoderm), ENO2, and NEUROD1 (neuronal) expression with time in FGF2-supplemented EBs is consistent with this. The temporal pattern of nestin and NEUROD1 expression observed in the present study is similar to that described in the human embryonal carcinoma cell line NTERA-2 model of retinoic acid-induced neurogenesis . Recently, NEUROD1 has been identified as an upstream activator of the PAX6 gene . It is therefore likely that the increased expression of PAX6 seen in EBs cultured in the presence of FGF2 is caused by increased neural progenitor proliferation and differentiation in EBs, with concomitant upregulation of NEUROD1 expression.
The induction of pancreatic endocrine gene expression observed in this study is consistent with studies reporting a role for FGF2 [26, 27] and fibroblast growth factor receptor (FGFR) signaling [28, 29] in pancreatic development. In the present study, strong insulin expression was detected only from EBs cultured in media containing FGF2, a high-affinity ligand of FGFR1. Therefore, it appears that FGF2, through FGFR1 signaling, may have induced proliferation and differentiation of cells with the potential to produce insulin. However, expression of IPF1 or NEUROG3 in EBs cultured in either FGF2-supplemented or unsupplemented media was not detected, with the exception of one sample showing low-level IPF1 expression (Day 16, non-FGF2). IPF1 is considered to be one of the most significant transcription factors in pancreatic development and function, as well as a key regulator of glucose-dependent insulin transcription [30, 31, 32], and NEUROG3 is a key transcription factor in the switch between pancreatic endocrine and exocrine cell fates . Lack of expression of both these genes suggests an absence of progenitor β cell development. Conversely, the expression in neural cells of both PAX6  and insulin  has been reported. Therefore, it is feasible that the expression of insulin mRNA in EBs supplemented with FGF2 reflects expression from neural rather than pancreatic progenitor cells. Furthermore, this is not contradicted by the expression of glucagon, also seen in this study, which has been detected in the brain stem and hypothalamus of humans and rats [36, 37].
At Day 21 and Day 28, EBs are relatively undeveloped, and in terms of normal human development at this period, it is not surprising that pluripotent cells are present, as observed by POU5F1 and NANOG expression at time points earlier than Day 28. This is also consistent with a previous study, which observed the presence of pockets of undifferentiated embryonic progenitor cells within cystic EBs , albeit at earlier time points than the present study. Additionally, at concentrations of 40 ng/ml or 100 ng/ml, FGF2 has been found to support ES cells in an undifferentiated state . Furthermore, a lower FGF2 concentration (4 ng/ml), similar to that in the present study, was still capable of reducing the level of BMP signaling and hence differentiation, although not to the same extent as the higher concentrations . This is consistent with our results, and suggests that addition of FGF2 to the culture allowed the survival of a small number of ES cells expressing pluripotency markers.
Immunofluorescent examination of the EBs at Days 28, 60 and 104 was used to confirm the presence of both pluripotent and differentiating cells as determined by histomorphology and/or RT-PCR analysis. The presence of nestin in Day 28 EBs confirmed the presence of neuroectodermal cells isolated to neuroepithelial-like structures and the neural rosettes of the EBs. With the addition of FGF2 to the culture media, the presence of nestin-positive cells was observed to increase, supporting evidence that FGF2 can enhance differentiation toward an ectodermal lineage . AFP was detected in Day 60 EBs in the lining epithelium described as endoderm by PAS staining. AFP staining in Day 104 EBs was not detected. There was little positive staining for collagen type IV in Day 28 EBs. By Day 60, there was increased detection in the cores of the EBs, and by Day 104, the majority of the cores of the EBs were positive for this marker, with the tissue resembling developing cartilage. From this evidence, it appears that EBs in long-term culture proceed along a mesodermal lineage, particularly producing hyaline cartilage.
This study appears to be the first report describing the long-term culture and development of EBs. Previously, the first two stages of EB development were identified to be the formation of simple EBs within the first 2–4 days of culture, followed by the formation of cystic EBs at approximately Day 4. Observing long-term growth of EBs in the present study allowed further characterization of the stages of EB development. The stages identified in this study include a third stage, at approximately 1 mo of in vitro culture, in which the main form of development observed in EBs was structures resembling neuroectodermal tissue; a fourth stage, at approximately 2 mo, in which EBs contained endodermal- and mesodermal-like structures; and a final stage, in which EBs were composed predominantly of mesodermal tissue, with the lack of neuroectodermal and endodermal tissue suggesting dedifferentiation of EBs to the mesodermal lineage.
To achieve efficient directed differentiation of ES cells into functional mature cell types, optimization of culture conditions and differentiation strategies is required. In the present study, we demonstrated that growth of EBs at the physiological glucose concentration of 5.5 mM yields a greater growth rate than the current practice of using media with a high glucose concentration (25 mM). The use of 5.5 mM glucose produced a similar gene expression pattern to that of EBs cultured at 25 mM glucose, therefore allowing the desired cell differentiation to occur with the advantage of avoiding potential impairment of cell function. In addition, long-term culturing of EBs was achieved in this study, indicating that examination of EB development may have more extensive potential for studying early developmental processes than previously believed.
The authors thank the following for their assistance: Patricia Palladinetti, Mark Lutherborrow, Appavoo Mathiyalagan, and Bartlomiej Getta (molecular analyses and technical assistance), Dr. Lily Bai and Pauline Khoury (statistical analysis), Lindy Williams (immunohistochemical analyses), Gavin McKenzie (histological analyses), and Prince of Wales Hospital SEALS Anatomical Pathology Department (histological analyses).