Long-Term Culturing of Undifferentiated Embryonic Stem Cells in Conditioned Media and Three-Dimensional Fibrous Matrices Without Extracellular Matrix Coating



ESCs have unlimited proliferation potential and capability to differentiate into all tissue types. They are ideal cell sources for tissue engineering and cell therapy, but their supplies are limited. Current in vitro ESC cultures are carried out in tissue flasks with the surface precoated with extracellular matrix (ECM) proteins. T-flask cultures also require frequent subculturing because their limited surface area cannot support long-term growth of ESCs. In this work, ECM coating and frequent subculturing required in two-dimensional (2D) cultures were circumvented by culturing murine ESCs in three-dimensional (3D) polyethylene terephthalate (PET) fibrous matrices. Also, media conditioned with STO fibroblast cells were used to replace leukemia inhibitory factor and to effectively maintain the pluripotency of murine ESCs in a long-term static culture. However, the lactic acid present in the conditioned medium could inhibit ESC growth and induce spontaneous differentiation when its concentration exceeded 1.5 g/l. In addition, the 3D static culture could be limited by oxygen, which was depleted in the long-term culture when cell density in the matrix was high. However, these problems can be alleviated in dynamic culture with improved oxygen transfer and continuous media perfusion. The matrix pore size also had profound effects on ESCs. The smaller-pore (30–60 μm) matrix gave a higher proliferation rate and Oct-4 and stage specific embryonic antigen-1 expressions. Overall, the 3D culturing method is superior to the 2D culture method and can provide an economical way to mass-produce undifferentiated ESCs in uncoated matrices and conditioned media.


ESCs derived from the inner cell mass of blastocyst have unlimited proliferation potential and are totipotent [1, 2]. Therefore, they are the ideal cell sources for tissue engineering and cell therapy. Their vast potential clinical applications include treatments of diabetes, Parkinson disease, spinal cord injury, liver malfunction, heart failure, and skin wounds [3, 4]. ESCs are also invaluable tools for drug discovery and gene therapy. Genetically modified ESCs can be used for high-throughput drug screening and to transmit and express specific genes in target organs [3, 4]. Although the demand for ESCs is high and is expected to grow rapidly once their biomedical applications have been established, there has been little effort aiming at developing an economical method of mass production of ESCs [5]. Currently, the expansion of ESCs is based on common laboratory procedures carried out in two-dimensional (2D) cell culture systems such as T-flasks, which are limited by the available surface area and difficult to scale up. Furthermore, the culture surface needs to be precoated with expensive extracellular matrix proteins, such as gelatin for murine ESCs (mESCs) and Matrigel for human ESCs (hESCs). Frequent subculturing or passaging is also required to maintain the undifferentiated state of ESCs [6, [7], [8], [9]–10]. These expensive, labor-intensive, and time-consuming 2D culturing methods cannot meet the projected market demand for ESCs.

Recently, a three-dimensional (3D) culturing method using fibrous matrices such as polyethylene terephthalate (PET) has been developed for culturing various cell types, including Chinese hamster ovary, hybridoma, osteosarcoma, cytotrophoblast, and mouse embryonic stem cells [11, [12], [13], [14]–15]. In general, 3D fibrous matrices can support high cell densities (3 × 108 cells per milliliter of matrix) and are advantageous for use as tissue scaffolds because of their high porosity, specific surface area, permeability, and mechanical strength. In addition, the unique 3D structure of fibrous matrices can provide cells a biomimetic environment that closely resembles their in vivo conditions. Cells cultured in the porous fibrous matrices are also protected from shear damage, a major concern in large-scale mammalian cell cultures [16]. Such a 3D culturing system is thus considered to be more scaleable and has been shown to be able to support and sustain mESC growth and hematopoietic differentiation [14]. One main objective of this study was thus to evaluate the feasibility of using 3D PET matrices for long-term ESC cultures as an economical method of mass production of undifferentiated ESCs.

Three-dimensional tissue scaffold is an important component in most of the tissue-engineered constructs. The 3D scaffold supports cell population, organization, and function mimicking those of native tissues and plays a vital role in in vitro cell cultures [17]. Many studies have shown that cells cultured in 3D scaffolds behave differently in cell cycle progression, proliferation, and differentiation from those in 2D cultures [11, [12], [13], [14]–15]. Unlike cells in 2D slides, ESCs cultured in 3D collagen gel did not form epithelia cells or large spread cells. Instead, they formed tubular or spherical gland-like structures [18]. When induced to differentiate by certain growth factors, hESCs formed a 3D structure that was not observed in 2D cultures [19]. Although the effect of 3D culturing environment on ESC differentiation has been studied, very little is known about the interaction between ESCs and the 3D scaffolds or the effects of this interaction on ESC expansion and maintenance of ESC pluripotency. The 3D scaffold is a critical component in the development of a scaleable process for mass production of pluripotent ESCs.

In this study, 3D culturing of mESCs in PET matrices and its effect on the expansion of undifferentiated ESCs were studied and compared with those in 2D cultures. The necessity and effects of gelatin coating of the support surface on mESC expansion in 3D cultures were also studied. As an economical alternative to leukemia inhibitory factor (LIF) commonly used in mESC growth media to maintain mESCs at their undifferentiated stage, media preconditioned with STO fibroblast cells were also studied in this work. The effects of lactic acid concentration in the conditioned media and the pore size of the PET matrix on ESC cultures were also investigated, and the results are reported here.

Materials and Methods

Cultures and Media

Murine embryonic stem (ES) D3 cells (CRL-1934; American Type Culture Collection, Manassas, VA, http://www.atcc.org) were maintained on gelatin-precoated T-flasks containing the ES growth medium, which consisted of the knockout Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS), 50 U/ml penicillin, 50 μg/ml streptomycin, 0.1 mM nonessential amino acids, 2 mM l-glutamine, 100 μM β-mercaptoethanol (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), and 100 μM LIF (Chemicon, Temecula, CA, http://www.chemicon.com). STO (CRL-1503; American Type Culture Collection) cells were cultured in DMEM with 10% FBS. Unless otherwise noted, T-flask cultures were incubated in a CO2 incubator at 37°C. The ESCs were passaged every 3 days to maintain their undifferentiated stage. All other cell culture materials were obtained from Gibco (Grand Island, NY, http://www.invitrogen.com), unless otherwise specified.

To replace LIF in the ES growth medium, STO-conditioned media were prepared in either T-flasks or spinner flasks, as follows. After the STO fibroblast cell culture reached 80% confluence in 75-cm2 T-flasks, the media were refreshed, and cells were cultured for an additional 4 days. Supernatant was then collected and used as the conditioned medium after supplementing with glucose, nonessential amino acid, l-glutamine, β-mercaptoethanol, and antibiotics to the same concentrations as those in the ES growth medium described above. The conditioned medium used in the long-term ES culture was prepared in a 250-ml spinner flask by inoculating 3–5 × 106 STO cells to the flask containing the PET matrix and 110 ml of media. After incubation for 6 days, the medium was refreshed, and cells were cultured for another 4 days. The supernatant was then collected and supplemented with additional media components in the ES growth medium as described above.

Nonwoven PET Matrices

Needle-punched nonwoven PET matrices (thickness, 2 mm; porosity, 0.93; pore size: 60–130 μm; fiber diameter, ∼20 μm; fiber density, 1.35 g/cm3) were used as 3D scaffolds. The PET matrices were pretreated with NaOH to increase their surface hydrophilicity. The thickness, porosity, and pore size of the PET matrices were reduced to ∼1 mm, 0.88, and 30–60 μm, respectively, by thermal compression [20]. In this article, the original PET matrices without thermal compression had the higher porosity and are referred to as high porosity (HP) matrices, whereas the matrices with the lower porosity obtained after thermal compression are referred to as low porosity (LP) matrices.

Static Cultures in Multiwells

The effects of gelatin coating, STO-conditioned media, lactic acid, and matrix pore size on ESC proliferation and maintenance of ESC pluripotency were studied in static cultures. Unless otherwise noted, ESCs were inoculated into either gelatin-coated 24-well plates (2D culture) or uncoated 3D PET matrices in 24-well plates. For 3D cultures, each well containing one sterile PET matrix and 100 μl of medium was inoculated with 3–10 × 104 cells. The cell suspension was carefully added from the center of the matrix using a micropipette. Following 3 hours of incubation in a CO2 incubator to allow for cell attachment to the matrix, the seeded matrix was washed with the culture medium to remove unattached and loosely attached cells. Then, the matrix was transferred to a new well, and 1 ml of fresh medium was added to the well.

Effects of Gelatin Coating.

The necessity and effects of gelatin coating of the support surface on ESCs were studied. ESCs were cultured in multiwells each containing either a PET film disk (1.5 cm in diameter) or PET fibrous matrix (∼1 cm square). Both coated and uncoated PET films and matrices were used for direct comparison. For gelatin coating, the PET film or matrix was soaked in 0.3 ml of 0.5% gelatin solution for 2 hours at room temperature, followed by air drying prior to use.

Effects of STO-Conditioned Medium.

The feasibility of using STO-conditioned media to replace LIF for maintaining mESCs in the undifferentiated stage was studied. Each well was inoculated with 2 × 104 cells, which were cultured in either the growth medium (with LIF or without LIF) or the STO-conditioned medium (without LIF). Cells were harvested every 2–3 days for analyses of total cell number and stage specific embryonic antigen-1 (SSEA-1)-positive cells. The harvested cells were subcultured in the same type of medium for six consecutive passages, each time with the same inoculation amount of 2 × 104 cells, to evaluate the long-term stability of ESC culture maintained in different media.

Effects of Lactic Acid.

The effects of lactic acid on ESC proliferation and maintenance were also studied in 2D static cultures. The growth media containing various initial amounts of lactic acid (0.162–3 g/l) were used to culture ESCs (2 × 104) in gelatin-coated 24-well plates. The initial medium pH was adjusted with sodium bicarbonate to ∼7.0. After incubation for 4 days, cells were harvested to quantify the total cell number and SSEA-1-positive cells.

Long-Term 3D Cultures

The conditioned medium was mixed with fresh ES growth medium (without LIF) at different ratios to reduce the amount of lactic acid present in the resulting media, which were used to test their effects on the long-term 3D culture of mESC. Each medium was tested in at least five culture wells, each containing 100 μl of the medium and one sterile PET matrix (1 × 1 cm) seeded with 3.75 × 104 cells. The media in the culture wells were refreshed periodically, and liquid media samples were taken before and after each change of media for analyses of glucose and lactate concentrations. Cells grown in the PET matrices were analyzed for their total number and SSEA-1 and Oct-4 expressions weekly. The long-term culturing study continued for 4 weeks, until all cell-matrices had been sacrificed for the assay purpose.

Long-term cultures in uncompressed (HP, 60- to 130-μm pore size) and compressed (LP, 30- to 60-μm pore size) PET matrices were compared to evaluate the effects of matrix pore size. Approximately 3.5 × 104 ESCs were inoculated into each well containing a piece of the PET matrix (1 × 1 cm) and 100 μl of the ES growth medium. Cells were harvested every 5 days for analyses of total cell number and SSEA-1- and Oct-4-positive cell populations.

EB-Forming Efficiency

To evaluate the pluripotency of the ESCs produced from various culturing systems, their EB-forming efficiency was determined, as follows. Approximately 1,000 ESCs were inoculated into a nonadhesive Petri dish. After 6 days of culture in the ES growth medium without LIF, the resulting EBs were counted. The EB-forming efficiency was defined as the number of EBs divided by the number of inoculated cells [7].

Flow Cytometric Analyses

ESCs were trypsinized from the PET matrices before flow cytometric analyses. For SSEA-1 assay, 2–5 × 105 cells were washed with phosphate-buffered saline (PBS) containing 0.5% bovine serum albumin and 2 mM EDTA (Sigma-Aldrich). Human Fc receptor blocking reagent (BD Pharmingen, San Jose, http://www.bdbiosciences.com/pharmingen) was added to cells in the suspension to prevent nonspecific binding; this was followed by incubation with 20 μl of SSEA-1 antibody (Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww) at 4°C for 60 minutes. After being washed, the samples were incubated with goat anti-mouse IgM-PE secondary antibody (Southern Biotechnology Associates Inc., Birmingham, AL, http://www.southernbiotech.com) for 30 minutes. After being fixed with 2% formaldehyde, the SSEA-1-positive cell population was analyzed using a flow cytometer (FACSCalibur; BD Biosciences, San Diego, http://www.bdbiosciences.com). For quantification of cells with the intracellular protein Oct-4, cells were fixed with 4% formaldehyde at room temperature for 20 minutes, followed by cell membrane perforation with the washing buffer containing 0.5% saponin and 0.1% sodium azide (Sigma-Aldrich) at room temperature for 20 minutes. The samples were later incubated with anti-Oct-4 monoclonal antibody (Chemicon) at room temperature for 30 minutes and then IgG-fluorescein isothiocyanate (FITC) for another 30 minutes. The Oct-4-positive cells were then analyzed and quantified using a flow cytometer (FACSCalibur, BD Biosciences). Unstained cells were used for locating the population, and cells labeled with either IgM-PE or IgG-FITC were used to evaluate the nonspecific binding or background fluorescence reading, respectively.

Scanning Electron Microscopy

Cell morphology and distribution in the PET matrices were observed using a scanning electron microscope (Philips XL 30; Philips Electronics, Eindhoven, The Netherlands, http://www.philips.com). Each matrix sample with cells was washed with PBS solution and incubated in 2.5% (vol/vol) glutaraldehyde at 4°C overnight. The samples were then rinsed with distilled water and progressively dehydrated in 10% (vol/vol) to 100% (vol/vol) ethanol in increments of 10% by soaking the sample for 30 minutes at each ethanol concentration. The dehydrated samples were then dried by soaking in a hexamethyldisilazane (HMDS) (Sigma-Aldrich) and ethanol mixture with ascending HMDS concentrations of 1:3, 1:1, and 3:1 for final dehydration. The dried samples were sputter-coated with gold-palladium at an argon pressure of 14 Pa for 120 seconds and a current of 17 mA to convey electrical conductivity and then observed in the scanning electron microscope (SEM) at 5–25 kV accelerating voltage.

Analytical Methods

Cell Counting.

The PET matrices were placed into the standard nuclei counting solution (0.1 M citric acid, 0.1% [wt/vol] crystal violet) and incubated at 37°C for 24 hours. The matrix was then vigorously vortexed to release cell nuclei, which were counted under a microscope. Cells on 2D surface of multiwells or T-flasks were counted, after trypsinization, using a hemocytometer.

Metabolic Assays.

Lactic acid and glucose concentrations in the culture media were measured with a YSI Biochemistry Select Analyzer (YSI Life Sciences, Yellow Spring, OH, http://www.ysi.com).


Effects of Gelatin Coating

Figure 1 compares the results of ESCs cultured in two and three dimensions with or without surface coating with gelatin. In 2D cultures, cells generally grew better on the gelatin-coated surface than on the uncoated surface, as indicated by the higher expansion fold and fraction of SSEA-1-positive cells. This finding is consistent with common belief that gelatin or extracellular matrix (ECM) protein coating on the support surface is indispensable for cell attachment and growth in ESC cultures [7]. However, 3D cultures with PET scaffolds without gelatin precoating were able to support good cell growth and maintain high SSEA-1 expression (Fig. 1). It is noted that gelatin coating of the PET matrices resulted in poorer cell growth because the uneven coating had blocked and deteriorated the 3D pore structure of the matrices. Eliminating the expensive ECM coating of the support surface would be an advantage of 3D cultures over 2D cultures because it can significantly reduce the cost of the ESC expansion process. This 3D culturing advantage was also demonstrated with human ESCs cultured in PET scaffolds without coating with Matrigel or any ECM proteins as usually required in 2D cultures [21]. In addition, ESCs harvested from the 3D culture showed a comparable or even higher EB-forming efficiency (24.7% ± 7.2% vs. 18.7% ± 7.6% in 2D cultures). The EB-forming efficiency is indicative of the pluripotency of ESCs or their ability to differentiate into various cell types [7]. Therefore, ESCs grown in 3D PET matrices can be maintained in their undifferentiated stage to conserve their pluripotency.

Figure Figure 1..

Comparison of ESC cultures on 2D polyethylene terephthalate (PET) films and 3D PET matrices with and without gelatin precoating. (A): ESC growth expansion (fold). (B): Fraction of SSEA-1-positive cells. Abbreviations: 2D, two-dimensional; 3D, three-dimensional; SSEA-1, stage specific embryonic antigen-1.

Effects of Conditioned Media

STO cells are commonly used as feeder cells to support the undifferentiated mESC growth [8]. These cells secrete several cytokines that are essential for mESCs to maintain their undifferentiated state. The conditioned medium from STO cell culture was thus studied as an economical alternative to the more expensive cytokine LIF commonly used in mESC growth media. As shown in Figure 2, no significant difference in cell growth, as indicated by the total cell number at the end of each passage, was found among the three media studied in 2D cultures. However, SSEA-1 was significantly downregulated in mESCs cultured in the growth medium without LIF, whereas cells cultured in the conditioned medium and the LIF-containing growth medium maintained their high SSEA-1 expression level, ∼85%. The results suggested that the STO-conditioned medium was as good as the more expensive LIF-containing growth media in maintaining mESCs in the undifferentiated stage.

Figure Figure 2..

Two-dimensional cultures of embryonic stem cells with six passages in the growth medium W/ LIF and W/O LIF and the CM. (A): Total cell number. (B): Fraction of SSEA-1-positive cells. Abbreviations: CM, conditioned medium; LIF, leukemia inhibitory factor; SSEA-1, stage specific embryonic antigen-1; W/, with; W/O, without.

Effects of Lactic Acid

However, it was later found that the conditioned media produced in the spinner flask could not sustain ESC growth and maintain their undifferentiated stage in long-term 3D cultures. The fraction of SSEA-1-positive cells dropped from 98% to 58% in the 3D culture after 20 days. Further investigation revealed that the lactic acid production in the 3D STO culture was much higher than that in 2D STO cultures, resulting in the higher lactic acid concentration of the conditioned media from the 3D culture (2.2–2.4 g/l vs. 0.6–0.9 g/l from the 2D culture). The higher density of STO cells in the 3D culture resulted in limited oxygen transfer and thus higher lactic acid production.

Because lactic acid is a known inhibitor to most mammalian cells, it was believed to be the main factor limiting mESC growth in the conditioned media with a high lactic acid content. The effects of lactic acid on mESCs were thus studied and the results are shown in Figure 3. As expected, lactic acid strongly inhibited ESC growth at concentrations of 1.5 g/l and higher. At 3 g/l of lactic acid, it was observed that mESCs could not attach well on the flask surface, and there was no significant cell growth. Moreover, increasing the lactic acid concentration also significantly decreased the cell populations expressing Oct-4 and SSEA-1, implying that mESCs had undergone spontaneous differentiation at elevated lactic acid concentrations. It is clear that ESCs are more sensitive to lactic acid than other animal cells, which can usually tolerate lactic acid up to 3.5 g/l [22, 23]. This result concurred with the finding that frequent media replacement to minimize the accumulation of metabolic wastes was necessary for long-term ESC cultures [24].

Figure Figure 3..

Effects of initial lactic acid concentration in the growth media on ESCs in two-dimensional culture. (A): Final cell number. (B): Fractions of SSEA-1- and Oct-4-positive cells. Abbreviation: SSEA-1, stage specific embryonic antigen-1.

Long-Term 3D Cultures in Conditioned Media

The conditioned medium was mixed with fresh ES growth medium (without LIF) at different ratios to reduce the amount of lactic acid present in the resulting media, which were used to test their effects on the long-term 3D culture of mESC. The results are shown in Figure 4. In general, ESCs grown in these media showed similar growth kinetics. Except for the 50% conditioned medium, ESCs had expanded approximately 80–98 times and reached their maximum numbers after 15 days (Fig. 4A). As expected, the fraction of cells expressing Oct-4 decreased significantly faster in the culture without LIF than with LIF or the conditioned media (Fig. 4B). However, all cultures showed significant decrease in SSEA-1-positive cell population after 15 days (Fig. 4C), which marked the beginning of the stationary phase when cell growth was limited by oxygen and a dramatic drop in SSEA-1 expression occurred. The effect of oxygen starvation in the 3D culture is discussed below.

Figure Figure 4..

Three-dimensional cultures of embryonic stem cells in the growth medium with and without LIF and the CM at 10, 25, and 50% concentrations. (A): Cell number. (B): Fraction of Oct-4-positive cells. (C): Fraction of SSEA-1-positive cells. Abbreviations: CM, conditioned medium; LIF, leukemia inhibitory factor; SSEA-1, stage specific embryonic antigen-1.

SSEA-1 is a surface marker and thus more sensitive to the culture environment than the intracellular protein Oct-4. It is noted, however, that SSEA-1 expression decreased much faster in 25% and 50% conditioned media, in which the lactic acid concentrations increased above the inhibiting level of 1.5 g/l (data not shown). It is thus clear that a high lactic acid concentration not only inhibited cell growth but also induced spontaneous differentiation. In general, the 10% conditioned medium was able to replace the LIF-containing medium for the long-term 3D murine ES culture. As can be seen in Table 1, for the first 15 days, ESC growth in the 10% conditioned medium was 16% higher and Oct-4 and SSEA-1 levels were within the range compared with the LIF-containing ES growth medium. Therefore, undifferentiated ESCs can be expanded and maintained in STO-conditioned media, provided that the lactic acid concentration is properly controlled at a low level. This conclusion was confirmed with a dynamic ES culture carried out with a perfusion bioreactor, where oxygen and lactate concentrations were controlled at proper levels [21].

Table Table 1.. Murine ESC culture performance after 15 days of culturing in various media
original image

Effects of Matrix Pore Size

ESCs were cultured in PET matrices with two different pore sizes. As shown in Figure 5, ESCs cultured in the smaller pore matrix (LP) grew faster, with 60% more cells after 20 days as compared with the HP culture. There were also more SSEA-1- and Oct-4-positive cells in the LP culture. With smaller pores, the contact probability between the cells and the fibers is higher. Accordingly, more attachment and bridging, which are beneficial for cell proliferation and maintenance of undifferentiated state, are likely to occur. This pore size effect was confirmed by SEM images showing that ESCs were more likely to form large aggregates in large-pore matrices (HP) and, on the other hand, spread more evenly in the entire fibrous matrix when cultured in small-pore matrices (LP) (Fig. 6). Aggregation of cells, which was mostly found in large-pore matrices, is not conducive to proliferation and can induce spontaneous differentiation caused by contact inhibition. These findings are consistent with the results previously reported [13] and can explain why cells grown in the larger-pore matrices were more likely to differentiate.

Figure Figure 5..

Effects of matrix pore size on ESC cultures. (A): Cell growth. (B): SSEA-1 expression. (C): Oct-4 expression. Abbreviations: HP, high porosity; LP, low porosity; SSEA-1, stage specific embryonic antigen-1.

Figure Figure 6..

Scanning electron micrographs of ESCs in polyethylene terephthalate (PET) matrices. In general, large-cell aggregates were formed in the large-pore matrix, whereas cells were more evenly distributed in the small-pore matrix. (A, B): Small-pore PET matrix. (C, D): Large-pore PET matrix. Arrows point to the large-cell aggregates. Abbreviations: Acc. V., accelerating voltage; Magn, magnification; WD, working distance.

As also can be seen in the SEM pictures shown in Figure 6, 3D PET matrices provided plenty of fiber surface and void space for cell anchorage and growth. There were still tremendous amounts of unoccupied surface and space in the PET scaffolds, even in the long-term cultures that reached a relatively high cell density of ∼108 cells per milliliter of matrix volume. It is therefore unlikely for the 3D culture to have the contact inhibition that commonly occurs in 2D cultures.

Effects of Oxygen Transfer in 3D Fibrous Matrix

Nutrient limitation in the media, another common cause for cell growth inhibition, was believed to be the main reason for cell growth stagnation in the long-term 3D cultures observed in this study. In a 3D static culture, the main mass transfer mechanism is diffusion, which is inefficient for oxygen with a low solubility in water. Oxygen must diffuse through the liquid that occupied the pores of the fibrous matrix to reach the cells in the scaffold. When the cell density in the matrix increased as a result of cell growth, the oxygen consumption rate also increased, whereas the permeability of the matrix decreased, resulting in poor oxygen transfer in the scaffold and oxygen starvation.

To verify this oxygen starvation hypothesis, a diffusion model, shown below, was used to simulate the oxygen concentration changes during the period of long-term 3D culturing in the PET matrices.

For diffusion in the liquid phase (z ≤ 0.415 cm),

equation image

For diffusion in the PET matrix (0.415 cm < z ≤ 0.515 cm),

equation image

The boundary and initial conditions are as follows: C = 0.212 mM at z = 0 (oxygen saturation at the surface of the medium); δCt = 0 at z = 0.515 cm (no diffusion at the well bottom); C = 0.212 mM at t = 72n hours (n = 0, 1, 2, 3, 4, 5, since media were refreshed every 72 hours).

The dissolved oxygen concentrations can be found by solving the above transient partial differential equations using Matlab (The MathWorks, Inc., Natick, MA, http://www.mathworks.com/) and the parameter values as follows: oxygen diffusivity in the liquid, D1 = 2.5 × 10−5 cm2/second; oxygen diffusivity in the matrix, D2 = 1.8 × 10−5 cm2/second; specific oxygen consumption rate, Q = 1 × 10−10 mmol/cell · hour [25]; initial cell number, X0 = 3.75 × 104; specific growth rate, μ = 0.012 hours−1. The model-predicted oxygen concentration profiles in the z-direction or the depth from the medium surface during the culturing period were calculated (data not shown), which showed that the oxygen concentration at the bottom of the culture well reduced to 0 at 387 hours, which is consistent with the experimental observation that cell growth stopped at ∼15 days of culturing (Fig. 4A). It is thus concluded that oxygen depletion was the dictating factor in limiting the long-term 3D static culture. Better long-term culture performance was obtained when a dynamic culture with better oxygenation was used [21].


Spontaneous differentiation usually occurs when embryonic stem cells are cultured in vitro [4]. Upon the first isolation, it was found that murine embryonic fibroblasts (MEFs) were necessary as the feeder-layer cells in maintaining ESC pluripotency [5]. However, this coculture method generated a complication in their application because of the cell source contamination from the feeder layer cells, which cannot be easily separated when harvesting the ESCs. Further investigation later demonstrated that LIF could be used as a substitute for the feeder cells to sustain undifferentiated mESC growth [6]. The minimum concentration of LIF necessary to maintain the undifferentiated mESCs was reported to be 500 pM [9], which adds a substantial cost to the ES culture medium. Likewise, human ESCs also require either a feeder layer or expensive growth factors, mainly basic fibroblast growth factor (bFGF), in their culturing media to maintain their undifferentiated state and pluripotency [10, 26, [27], [28]–29]. In this work, STO-conditioned media were demonstrated to be an effective alternative to LIF in maintaining mESC pluripotency in the 3D culture, which has a better scale-up potential for mass production purposes than 2D cultures. Similarly, hESCs can be maintained in MEF-conditioned media supplemented with only 4 ng/ml of bFGF [10] when cultured in uncoated PET matrices [21].

In general, 3D culturing environment can closely mimic in vivo growth conditions. In this work, PET matrices without ECM coating were shown to support mESC growth well and maintain mESC pluripotency, another advantage over 2D ESC cultures. It is noted that mESCs grown in the 3D PET scaffold had an average doubling time of 38 hours. Fok and Zandstra [30] also reported that mESCs in suspension culture at 60 rpm had a doubling time of ∼40 hours and expanded ∼75-fold in 15 days, which was improved to ∼105-fold when mESCs were cultured with microcarriers as the support substrate in the suspension culture. Microcarrier cultures have the advantage of easier scaling up for a suspension culture but also may suffer from high shear damage. Our static cultures in 3D PET matrices showed a similar expansion fold. The 3D culturing process also has a good scale-up potential and can be further improved in a dynamic culturing environment [21].

It is noted that mESCs grown in 2D cultures on a gelatin-coated surface had a significantly shorter doubling time of 14 hours. The longer doubling time in the 3D cultures might be partially attributed to the oxygen limitation in the static culture and possibly to cells having a requirement for some adaptation from 2D to 3D environment. It has also been reported that cells grown on a 2D surface can stretch better, which stimulates proliferation [31]. However, the apparently shorter doubling time in the 2D cultures required frequent subculturing, which not only is labor-intensive but also can only reach a relatively low cell number or density at each passage, limiting its potential for large-scale applications.

In addition to growth factors, other culture conditions, such as mechanical strain, can also inhibit ESC spontaneous differentiation [32]. In addition, lactic acid has long been recognized as a potential toxic metabolite to many mammalian cells [22, 23]. However, the effects of lactic acid accumulation in the culture medium on ESC growth and differentiation have not been well studied before. The present study shows that mESCs have a lower tolerance to lactic acid than other known mammalian cell lines [22, 23]. Therefore, reducing or avoiding lactic acid accumulation in the culture media would be critical in the development of a process for mass production of ESCs. Lactic acid metabolism is known to be greatly influenced by the dissolved oxygen level in the media, which has also been shown to have significant effects on stem cell growth and differentiation [33, [34], [35], [36], [37], [38]–39].

Lennon et al. found that rat mesenchymal stem cells could produce more bone cells at 5% oxygen than at 20% oxygen [33]. Ouyang et al. also concluded that low oxygen could prolong the life span of epidermal stem cells [35]. On the other hand, Kallos and Behie claimed that oxygen should be kept above 20% in neuron stem cell suspension culture to obtain more cells [36]. Wang et al. showed that 5% oxygen tension would inhibit adipose-derived stem cell growth, but would increase chondrogenesis process [37]. Bauwens et al. observed that low oxygen can improve ESCs differentiation into cardiomyocytes [38]. Kurosawa et al. demonstrated that a high oxygen tension (40%) inhibited ESC spontaneous differentiation [39]. Based on these studies, it can be concluded that a high oxygen concentration is conducive to proliferation and low oxygen stimulates differentiation. Therefore, a high dissolved oxygen concentration should be better for ESC expansion and maintenance in the undifferentiated stage. In this study, we found that ESC spontaneous differentiation accelerated and cell growth was inhibited after 15 days of culturing because of oxygen depletion in the static culture. The oxygen effect is closely related to lactic acid, which also influenced ESC proliferation and differentiation. Because the effects of oxygen vary with cell lines, the influence of oxygen on ESC expansion needs to be further investigated in a well-controlled bioreactor, which can provide a dynamic environment and accommodate media perfusion to alleviate the oxygen starvation problem found in static cultures [21].


The 3D culturing method with conditioned media provides an economical way to mass-produce undifferentiated ESCs. ECM coating and frequent subculturing required in 2D cultures are circumvented in 3D cultures. STO-conditioned media can replace LIF in long-term mESC cultures to effectively maintain the pluripotency of ESCs. These two aspects should pave the way for developing an economical process for mass production of ESCs. Lactic acid at a moderate concentration of 1.5 g/l has a negative effect on ESC growth and can induce spontaneous differentiation. The smaller-pore PET fibrous matrix is favorable for growing and maintaining undifferentiated ESCs. The growth of ESCs is inhibited and more spontaneous differentiation takes place when the oxygen is limited in the long-term, static, 3D cultures. A dynamic culture with better process control and media perfusion is thus recommended to overcome this oxygen limitation problem in the further development of 3D cultures for ESC expansion.


The authors indicate no potential conflicts of interest.