Biomedical Microdevices

, Volume 12, Issue 6, pp 1097–1105

Enhanced effects of secreted soluble factor preserve better pluripotent state of embryonic stem cell culture in a membrane-based compartmentalized micro-bioreactor


    • Institute of Industrial ScienceThe University of Tokyo
  • Takeshi Katsuda
    • Institute of Industrial ScienceThe University of Tokyo
  • Kevin Montagne
    • Institute of Industrial ScienceThe University of Tokyo
    • LIMMS/CNRS-IISThe University of Tokyo
  • Hiroshi Kimura
    • Institute of Industrial ScienceThe University of Tokyo
  • Nobuhiko Kojima
    • Institute of Industrial ScienceThe University of Tokyo
  • Hidenori Akutsu
    • Department of Reproductive BiologyNational Research Institute for Child Health and Development
  • Takahiro Ochiya
    • Section for Studies on MetastasisNational Cancer Center Research Institute
  • Teruo Fujii
    • Institute of Industrial ScienceThe University of Tokyo
    • LIMMS/CNRS-IISThe University of Tokyo
  • Yasuyuki Sakai
    • Institute of Industrial ScienceThe University of Tokyo
    • LIMMS/CNRS-IISThe University of Tokyo

DOI: 10.1007/s10544-010-9464-8

Cite this article as:
Chowdhury, M.M., Katsuda, T., Montagne, K. et al. Biomed Microdevices (2010) 12: 1097. doi:10.1007/s10544-010-9464-8


Pluripotent stem cells are under the influence of soluble factors in a diffusion dominant in vivo microenvironment. In order to investigate the effects of secreted soluble factors on embryonic stem cell (ESC) behavior in a diffusion dominant microenvironment, we cultured mouse ESCs (mESCs) in a membrane-based two-chambered micro-bioreactor (MB). To avoid disturbing the cellular environment in the top chamber of the MB, only the culture medium of the bottom chamber was exchanged. Cell growth in the MB after 5 days of culture was similar to that in conventional 6-well plate (6-WP) and membrane-based Transwell insert (TW) cultures, indicating adequate nutrient supply in the MB. However, the cells retained higher expression of pluripotency markers (Oct4, Sox2 and Rex1) and secreted soluble factors (FGF4 and BMP4) in the MB. Inhibition of FGF4 activity in the MB and TW resulted in a similar cellular response. However, in contrast to the TW, inhibition of BMP4 activity revealed that autocrine action of the upregulated BMP4, which acted cooperatively with leukemia inhibitory factor (LIF), upregulated the pluripotency markers expression in the MB culture. We propose that BMP4 accumulated in the diffusion dominant microenvironment of the MB upregulated its own expression by a positive feedback mechanism—in contrast to the macro-scale culture systems—thereby enhancing the pluripotency of mESCs. The micro-scale culture platform developed in this study enables the investigation of the effects of soluble factors on ESCs in a diffusion dominant microenvironment, and is expected to be used to modulate the ESC fate choices.


Embryonic stem cellSoluble factorsDiffusionMicroenvironmentMicro-bioreactor

1 Introduction

The autocrine and paracrine actions of soluble factors have an important role in directing pluripotent stem cell fate choices in vivo (Gadue et al. 2005; Loebel et al. 2003). Pluripotent stem cells and their progenies remain in a diffusion dominant microenvironment enclosed by the trophectoderm and extra-embryonic part until an appreciable amount of mass flow by convection occurs after the onset of blood circulation (Nagy et al. 2003). At the initial stage of embryo development, the fate of pluripotent cells is influenced by the adequate signaling of soluble factors in the microenvironment. In vitro, ESC fate is also modulated by soluble factors (Kunath et al. 2007; Ying et al. 2003a). Although exogenous soluble factors can be added to the in vitro culture systems to control ESC fate, it is necessary to consider the influence of endogenous soluble factors which are secreted by the cells (Wiles and Proetzel 2006). This is highlighted by the fact that the addition of exogenous soluble factors has little influence on the initial differentiation of embryoid bodies (EBs) but influences the successive maturation of differentiated progenies towards the matured cell types (Ogawa et al. 2005; Wiles and Proetzel 2006). Furthermore, neuronal stem cells can be derived efficiently from mouse ESCs (mESCs) without the addition of exogenous factors (Ying et al. 2003b). Therefore, a culture system which mimics the diffusion dominant nature of the in vivo microenvironment is of great importance in order to improve our understanding of stem cell biology and control the stem cell fate decision (Loebel et al. 2003; Murry and Keller 2008).

Microfluidic technology provides advanced tools to develop micro-scale culture systems in an in vivo relevant dimension as well as to control mass transfer modes in the cellular microenvironment (Meyvantsson and Beebe 2008). Various micro-scale culture systems have been developed for ESC culture, but little is known about the effects of secreted soluble factors in these systems. Moreover, before proceeding to the differentiation of ESCs, it is necessary to characterize the differences between the micro and macro-scale cultures, namely regarding the effects of cell secreted soluble factors on ESC behavior. Human ESCs (hESCs) cells were cultured in straight micro-channels in static (Abhyankar et al. 2003) and semi-static (Korin et al. 2009) conditions. Although these cultures facilitated the accumulation of soluble factors around the cells owing to the diffusion dominant nature of static micro-scale culture, their effects on the cells were not investigated. Furthermore, the environment changed abruptly because of the daily replacement of the total culture medium. In micro-fabricated wells, hESCs were found to remain undifferentiated for more than two weeks (Mohr et al. 2006). The reason for that was not identified, but most likely resulted from soluble factors, cell-cell contacts and the extra-cellular matrix (ECM) produced by the cells. Some studies focused on controlling ESC microenvironment using perfusion-based systems (Figallo et al. 2007; Kim et al. 2006). In one of those studies, mESCs were cultured in microfluidic arrays at different flow rates, and the cell colonies showed flow-dependent size variations (Kim et al. 2006). This was attributed to the amount of nutrient delivery as well as the removal of waste and secreted factors. Although perfusion is a way to supply enough nutrients to cells for long-term culture and control the cellular microenvironment by removal of the secreted soluble factors, it disturbs the cellular diffusion-based microenvironment (Walker et al. 2004).

In this context, we developed a membrane-based two-chambered micro-bioreactor (designated as MB hereafter) and culture conditions for ESCs to investigate the influence of secreted soluble factors on cells by mimicking the diffusion-dominant in vivo microenvironment. The culture medium of the top chamber was not replaced during the culture period to avoid disturbance in the cellular microenvironment. In contrast, the culture medium of the bottom chamber was exchanged daily to maintain a sufficient nutrient supply. We cultured mESCs for five days in leukemia inhibitory factor (LIF) supplemented culture medium to study the effects of soluble factors on cellular behavior, such as cell-cell interactions, cell proliferation and differentiation, in which the influence of secreted soluble factors is important (Yu et al. 2005). In the LIF supplemented medium, BMP4 synergistically interacts with LIF to preserve the mESC pluripotency by resisting the differentiation inducing action of FGF4 (Ying et al. 2008c). Therefore, the cell states in the MB, membrane-based macro-scale Transwell Insert (TW) and conventional 6-well plate (6-WP) cultures were compared by the expression of pluripotency markers (Oct4, Sox2, Rex1 and Nanog) and cell secreted soluble factors (FGF4 and BMP4). In addition, we performed cell culture experiments by inhibiting signaling components of FGF4 and BMP4 in the MB and TW. Then, the gene expressions of inhibited and non-inhibited cultures were compared to discern the effects of soluble factors in the micro and macro-scale culture systems.

2 Materials and methods

2.1 Design of the MB

Figure 1 shows the design details of the MB. The reactor had two round chambers (top and bottom) with an area of 2.27 cm2. They were kept separated by a porous membrane. Each of the chambers’ height and volume were 500 μm and 114 μL, respectively. The chambers contained 13 pillars (1 mm in diameter) that kept the membrane horizontal, and enabled a more homogeneous cell seeding on the membrane. Cells were cultured on the top face of the membrane. To avoid culture area other than the membrane, two feeding holes in the top chamber were drilled at the chamber perimeter. On the other hand, feeding holes in the bottom chamber were made approximately 0.7 cm away from the chamber perimeter to get clearance from the membrane perimeter.
Fig. 1

Details of the PDMS chambers and membrane used to fabricate the MB. The membrane is sandwiched between the two PDMS chambers using the common O2 plasma method. Both top and bottom PDMS chambers have two feeding holes to inject cells and exchange culture medium. Cells are cultured only on the top face of membrane

2.2 Fabrication of the MB

Details of the MB fabrication method are presented elsewhere (Kimura et al. 2008). Briefly, negative photoresist SU-8 2100 (Microchem Co.) was used to create the mold masters with the desired pattern. Then PDMS polymer (Silpot 184; Dow Corning Corp.) was mixed with its curing agent at a 10:1 ratio, poured over the mold masters, cured for 2 h at 75°C and peeled off thereafter.

The polyester membranes (pore size 0.4 μm, thickness 10 μm) were removed from Transwell Inserts 3450 (Corning Inc.). To bond the membrane with the PDMS layers, both sides of the membrane were coated with a thin layer of SiO2 by sputtering for 20 s at 150 W and 0.5 Pa. The membrane was sandwiched between the two PDMS chambers following O2 plasma treatment. Flow chips were then attached to the silicon tubing for connecting syringes; medium and reagents were manually introduced using those syringes.

2.3 Pre-treatment of experimental group (6-WP, TW and MB)

Because SiO2 was used to coat the polyester membrane incorporated in the MB, the top side of the TW membrane was also coated with SiO2 by sputtering. The TW and MB were then sterilized for 2 h under UV light. A 0.2% w/v gelatin solution was applied to cover the surface of the 6-WP and the membrane surfaces of the TW and MB, which was followed by 6 h of incubation. Culture medium for the experimental group was added to these culture systems for pre-incubation before seeding mESCs.

2.4 Routine cell culture

mESCs were routinely cultured in 60 mm gelatin coated dishes (Iwaki). Cell inoculation density was 2 × 104 cells/cm2, and the cells were passaged every other day. Culture medium composition for routine culture was high glucose DMEM (DMEM; Gibco) containing 20% ESC qualified Fetal Bovine Serum (FBS; Gibco), 1000 U/ml ESGRO-LIF (Chemicon), 1% MEM non-essential amino acids (Gibco), 2 mM GlutaMax-I (Gibco), 100 U/ml penicillin, 100 U/ml streptomycin (Gibco) and 0.1 mM 2-mercaptoethanol (Gibco). The cells were maintained in a 37°C humidified environment containing 5% CO2.

2.5 Cell culture in the experimental group

mESCs in the experimental group were cultured using the same medium as for routine culture except that DMEM and FBS were replaced with Knockout DMEM (Gibco) and 15% Knockout Serum (KSR; Gibco), respectively. KSR was used because it contains fewer extrinsic proteins. For the experimental group, cell inoculation density was 2 × 104 cells/cm2. Cell culture in the experimental group was continued for 5 days. Culture medium of the 6-WP, the lower chambers of the TW and MB were changed daily. The upper chamber culture medium of the TW and MB were not changed. Morphological examination of the cells under microscope was performed daily. For inhibition experiments, Fibroblast Growth Factor Receptor (FGFR) antagonist SU5402 (Mohammadi 1997) (Calbiochem) at 10 μM and BMP4 antagonist Noggin (Smith and Harland 1992) (R&D Systems) at 100 ng/ml were added to the culture medium.

2.6 Glucose concentration measurement

Culture medium from the 6-WP, the lower chamber of the TW and MB were collected every day during the 5-day culture. On the 5th day, the culture medium from the upper chambers of the TW and MB were also collected. Glucose concentrations were measured with a glucose analyzer (GA05, A&T Corp., Japan).

2.7 Cell collection and qPCR analysis

Isolation of total mRNA was performed using Trizol Reagent (Invitrogen). In all culture systems, cells were dissociated using Trypsin (Gibco), counted and then lysed with Trizol. First-Strand cDNA Synthesis Kit (GE Healthcare) was used to synthesize cDNA from the total mRNA. PCR reactions were carried out with a 7500 Real-Time PCR System (Applied Biosystems) using Quantitect SYBR Green PCR Kit (Qiagen). All steps were performed according to the manufacturers’ instructions. Primers for cDNA amplification are listed in Table 1. qPCR were performed at least in duplicate. Raw data of PCR product amplification curves was analyzed using LinRegPCR v11.4 software (Ruijter et al. 2009) to determine the threshold cycles used in the ∆∆CT method for relative quantification of gene expression. Geometric mean of the threshold cycles of reference genes GAPDH and β-Actin was used to normalize the target gene expressions. mESC culture at day 3 in the 6-WP was used as calibrator.
Table 1

Genes and primers used in qPCR analyses




Sequence (5′-3′)


Housekeeping gene






Housekeeping gene






Pluripotency marker






Pluripotency marker






Pluripotency marker






Pluripotency marker






Cells’ self-secreted soluble factor






Cells’ self-secreted soluble factor





2.8 Statistical analysis

Student’s t-test for comparing two groups and one way ANOVA with Tukey’s post test for comparing more than two groups were performed for statistical evaluation using the demo version of GraphPad software (GraphPad Software, Inc.). Differences with a P < 0.05 (*), P < 0.01 (**), or P < 0.001 (***) were considered to be statistically significant. All data are presented as the mean ± SEM.

3 Results

3.1 Effect of SiO2 coating of the membranes on ESC behavior

SiO2 coating on the membrane was necessary to bond it strongly with the PDMS layers, thereby preventing culture medium leakage in the MB. The coating did not affect permeability of the membranes as the measured glucose permeability of coated and non-coated membranes was the same (3 × 10−12 m2s−1). To examine whether the coating had any effect on cell behavior, we performed mESC culture on SiO2-coated and non-coated membranes of the TW. Spread colonies of mESCs were observed on SiO2-coated membranes (Fig. 2(a)), whereas these colonies remained spherical on the non-coated membranes (Fig. 2(b)). Cells attached weakly on the membranes without SiO2 coating as the PBS wash during cell harvesting caused some cell loss. Consequently, the PBS wash was omitted for the non-coated membranes. However, no difference in cell growth was observed between SiO2-coated and non-coated samples (fold changes in cell number relative to the seeded cells were 35.25 ± 2.58 and 36.72 ± 3.51, respectively). Furthermore, both of the cultures showed similar gene expression profile (P > 0.05) (Fig. 3). Therefore, SiO2 coating on the membrane was used for all the experiments.
Fig. 2

mESC culture on day 5 on SiO2-coated (a) or non-coated (b) polyester membranes of the TW. Cell colonies on the SiO2-coated membrane are spread, whereas colonies on the non-coated membrane are round. Scale bar represents 125 μm
Fig. 3

Relative gene expression (Log2-transformed) of mESCs cultured for 5 days on SiO2-coated (TW_SiO2+) or non-coated (TW_SiO2−) membrane of the TW. Both cultures show similar gene expression profiles (P > 0.05). Zero value represents the gene expression of mESC cultured in 6-WPs for 3 days. Columns and error bars represent mean ± SEM of three independent experiments

3.2 Cell culture condition in the 6-WP, TW and MB

To keep the cellular microenvironment in the upper chamber of the MB and TW minimally disturbed, we only changed the culture medium of the lower chambers. Despite this, nutrients in these culture systems were sufficient as the glucose concentration throughout the 5 days of culture remained over half of the glucose concentration in fresh culture medium (0.02 M). Furthermore, no significant cell death was observed as assessed by the Trypan blue dye-exclusion test (data not shown) on day 5. Cell growth in all culture systems was similar (P > 0.05) (Fig. 4). mESC culture in the TW and MB could be continued for more than 5 days, whereas cells began to die in the 6-WP after that period owing to nutrition depletion.
Fig. 4

Fold increase in cell number relative to the seeded cell number after 5 days of mESC culture in the 6-WP, TW and MB. Cell growths are similar in all culture systems (P > 0.05). Columns and error bars represent mean ± SEM of four independent experiments

mESC colonies were extensively merged in the 6-WP (Fig. 5(a)) but they remained mostly as separated colonies in the TW and MB on day 5 (Fig. 5(b) and (c), respectively). Many differentiated cells showing a different morphology from usual mESCs were observed in the vicinity of cell colonies in the 6-WP and TW cultures (Fig. 5(d) and (e), respectively). In contrast, a few differentiated cells and smooth-bordered mESC colonies were observed in the MB (Fig. 5(f)), thereby indicating homogenous nature of the colonies.
Fig. 5

mESC culture on day 5 in the 6-WP (a, d), TW (b, e) and MB (c, f). Arrows in the higher magnification images (d, e) indicate differentiated cells with a different morphology from the tightly packed colony type morphology of mESCs. Scale bars represents 250 μm (a, b and c), and 62.5 μm (d, e and f)

3.3 Comparison of gene expression profiles among culture systems

In the MB culture, significantly higher expression levels of Oct4, Sox2 and Rex1 were observed compared to the 6-WP (Fig. 6). In addition, Sox2 and Rex1 expression in the MB were considerably higher than in the TW culture. These results showed that mESC pluripotency in the MB culture was higher than that in the macro-scale culture systems (6-WP and TW). Furthermore, both FGF4 and BMP4 were highly expressed in the MB culture compared to the TW (Fig. 6). However, only BMP4 expression in the MB was observed to be higher than in the 6-WP. Only Rex1 expression was different between the TW and 6-WP (Fig. 6).
Fig. 6

Comparison of the gene expression profiles (Log2-transformed) of mESCs cultured for 5 days in the 6-WP, TW and MB. Pluripotency markers (Oct4, Sox2 and Rex1) and soluble factors (FGF4 and BMP4) expression are upregulated in the MB. Zero represents gene expression of mESCs cultured in the 6-WP for 3 days. Columns and error bars represent mean ± SEM of six independent experiments. Statistical significance of the compared pairs are shown using the symbols *, ** and ***, representing P-values below 0.05, 0.01 and 0.001, respectively

3.4 Effects of soluble factors

To investigate whether dissimilarity in the activities of FGF4 and BMP4 between the MB and TW was responsible for the observed differences in the pluripotency markers expression (Sox2 and Rex1; Fig. 6), we performed inhibition experiments of FGF4 and BMP4 activities in the MB and TW cultures. FGF4 activity was inhibited using the small molecule SU5402, an antagonist of FGFR. This resulted in significantly increased expression of Nanog in the MB as well as TW cultures, but the other three pluripotency markers remained essentially unchanged (Fig. 7(a)). Expression of FGF4 increased, whereas that of BMP4 decreased in both MB and TW (Fig. 7(a)). We therefore concluded that the differentiation inducing activity of FGF4 suppressed Nanog expression, but did not affect the expression of Sox2 and Rex1 in the TW or MB. As a result, FGF4 cannot be accounted for the differences in Sox2 and Rex1 expressions between the MB and TW cultures.
Fig. 7

Gene expression profiles (Log2-transformed) of mESCs cultured for 5 days in the TW and MB, (a) with (TW_SU5402 and MB_SU5402) or without (TW and MB) inhibition of FGF4 signaling by SU5402; (b) with (TW_Noggin and MB_Noggin) or without (TW and MB) the BMP4 antagonist Noggin. Nanog expression both in the TW and MB increases following the SU5402 treatment. Sox2 and Rex1 expression decrease only in the MB following the Noggin treatment. Zero represents gene expression of mESCs cultured in 6-WP for 3 days. Columns and error bars represent mean ± SEM of four and three independent experiments for (a) and (b), respectively. Statistical significance of the compared pairs (TW against TW_SU5402; MB against MB_SU5402; TW against TW_Noggin; MB against MB_Noggin) are shown using symbols *, ** and ***, representing P-values below 0.05, 0.01 and 0.001, respectively

We then inhibited BMP4 activity using its antagonist Noggin. Expression of the pluripotency markers Sox2 and Rex1 decreased by the Noggin treatment in the MB, but they remained unchanged in the TW (Fig. 7(b)). In addition, both in the MB and TW culture, FGF4 and BMP4 expression remained unchanged by the same treatment. Sox2 and Rex1, which were upregulated more significantly in the MB as compared to the 6-WP and TW cultures (Fig. 6), decreased significantly by the Noggin treatment (Fig. 7(b)). Therefore, we can conclude that the activity of upregulated BMP4 (Fig. 6) is responsible for the better preservation of the mESC pluripotency in the MB.

4 Discussion

In this study, we developed a micro-scale culture system in which ESCs can be cultured in a diffusion dominant microenvironment without any limitation of nutrient supply for a long period of time. We observed better preservation of the mESC pluripotency in the micro-bioreactor than in the conventional macro-scale 6-WP and TW culture systems. We also demonstrated that autocrine effects of the up-regulated BMP4 cooperated with LIF to preserve the high pluripotency in the MB. Furthermore, the influence of FGF4 was similar in the TW and MB, whereas the influence of BMP4 was observed only in the MB.

A transcription network of Oct4, Sox2, Rex1 and Nanog maintains the pluripotency and proliferation of mESCs by suppressing the gene expression associated with differentiation (Masui et al. 2008; Niwa 2007). Usually, even in undifferentiated culture of mESCs in the presence of LIF, a proportion of the cells can undergo spontaneous differentiation (Smith 2001) which is associated with the decreased expression of those genes. Generally, overgrown differentiating mESC colonies have rough borders compared to the normal colonies. In the MB, mESC colonies were smooth-bordered, had few differentiated cells (Fig. 5(c) and (f)) and retained higher expression of the pluripotency markers (Fig. 6). These results indicated spontaneous differentiation of ESCs occurred less in the MB. Among the pluripotency markers, Sox2 and Rex1 showed prominently higher expression in the MB as compared to the WP and TW (Fig. 6). In fact, downregulation of Sox2 and Rex1 expression has a stronger correlation with loss of pluripotency of mESCs than the downregulation of Oct4 and Nanog expression (Palmqvist et al. 2005).

mESCs produce FGF4 extensively (Niwa et al. 2000) and BMP4 moderately (Johansson and Wiles 1995). FGF4 induces mESCs to differentiate (associated with the decreased expression of pluripotency markers of mESCs), which is counteracted by LIF and BMP4, as shown in Fig. 8 (Ying et al. 2008). In this study, although FGF4 was upregulated, high expression of BMP4 cooperated with LIF to preserve a high expression of pluripotency markers in the MB (Figs. 6, 7(b) and 8). In contrast, downregulated BMP4 in the TW had no observable effect on mESC pluripotecny markers, thereby maintaining a low level of pluripotency markers expression (Figs. 7(b) and 8). Notably, BMP4 expression was significantly upregulated only in the MB culture compared to the 6-WP and TW cultures (Fig. 6). Enclosed micro-scale environment might have facilitated the upregulation of BMP4 in the MB. Because the cell growth behaviors in these cultures were the same (Fig. 4), amount of secreted factors in the culture environment would be approximately the same. However, in the MB, the culture volume was small (114 μL compared to the cell compartment volumes of 1.5 mL and 2 mL in the TW and 6-WP, respectively), and mass transfer was diffusion dominant due to small dimensions as well as the absence of a free interface between the culture medium and air (Yu et al. 2005). On the other hand, surface tension differences at the interface cause rapid convection in the macro-scale culture systems, and that creates a homogenous distribution of secreted soluble factors over the entire culture volume (Yu et al. 2005). Therefore, secreted soluble factors were accumulated and reached higher concentrations in the MB than in the 6-WP and TW. Moreover, they were presumably retained around the cell colonies at high concentrations for a longer time period in the MB owing to the diffusion dominant mass transfer. BMP4 can induce its own expression by a positive feedback mechanism (Vainio et al. 1993). Furthermore, owing to the relatively short half-life of BMP4, it is necessary to retain BMP4 near the cell for its activity (Miljkovic et al. 2008). Therefore, the higher exposure of cells to BMP4 in the MB culture than in the macro-scale cultures (the 6-WP and TW) may facilitate the feedback mechanism. This explains the observed upregulation and downregulation of BMP4 expression in the MB and macro-scale cultures, respectively (Fig. 6).
Fig. 8

A schematic diagram of the influence of soluble factors on the pluripotency markers in the embryonic stem cell culture environment. LIF and BMP4 cooperate to resist FGF4-induced differentiation. Effects of BMP4 on the pluripotency state of mESCs in the micro (MB) and macro-scale (TW) culture systems are also depicted

In spite of FGF4 accumulation, cells in the MB displayed a similar response in gene expression to that observed in the TW following FGF4 inhibition (Fig. 7(a)). However, inhibition of BMP4 activities resulted in significantly different effects in the MB and TW (Fig. 7(b)). Molecular diffusivities (inversely proportional to the cube root of molecular weight, MW) primarily determine the retention behavior of the soluble factors around the cells (Yu et al. 2005). FGF4 has a lower MW (22 kDa) than BMP4 (47 kDa), and may have diffused more quickly out of the cellular milieu. Therefore, it was unable to exert any influence on the cells as BMP4 did in the MB. In addition, extensively secreted FGF4 could have reached the threshold level of its activity equally in the MB and TW. These could be the plausible reason for the similar response observed in the TW and MB.

The average concentrations (total number of molecules divided by volume) of FGF4 and BMP4 in the MB might be the highest among the culture systems owing to the accumulation of these factors in the smallest volume. However, the cellular response to a soluble factor depends on the concentration level of the factor in the vicinity of the cell (local concentration). Both the average and local concentrations are influenced by various parameters of a soluble factor such as secretion, consumption, sequestration and release form the ECM. However, convection and diffusion only influence the local concentration. Owing to the diffusion dominant mass transfer in the MB, a soluble factor could be retained around the cells over time to reach a high concentration—all other parameters being the same in all culture systems. Therefore, in the MB, we could realize the combined effect of accumulation in a small volume and diffusion dominant mass transfer. However, we could not distinguish explicitly which concentration (average or local) reached the threshold level to impart a cellular response. To make the distinction, further investigation (experiments coupled with mathematical simulation) is necessary by taking the various parameters of a soluble factor into account along with diffusion and convection. This study, which characterizes the effects of soluble factors on ESC culture in the MB, provides a basis for the investigation.

mESCs secret FGF5, Nodal and BMP2 at low variable levels besides FGF4 and BMP4 (Wiles and Proetzel 2006). This micro-bioreactor and culture condition will be useful to study their effects in a diffusion dominant cellular environment, and will contribute to the understanding of ESC biology. The heterogeneity of ESCs during differentiation is one obstacle in obtaining lineage-specific cells useful for cell-based transplantation therapies (Singh and Terada 2007). Our micro-bioreactor can be used for obtaining relatively homogenous ESCs. In the absence of LIF, both FGF4 and BMP4 promote the differentiation of ESCs (Ying et al. 2003a). Therefore, the activity of soluble factors observed in the MB will provide an enhanced signaling microenvironment for controlling ESC differentiation process in a monolayer format such as for neuronal (Ying et al. 2003b) or hepatocyte (Teratani et al. 2005) differentiation. By keeping the cellular environment in the upper chamber minimally disturbed, it is also possible to provide other soluble factors or inhibitors through the lower chamber to gain more precise control of the differentiation process.

5 Conclusions

In this study, we developed a membrane-based two-chambered micro-bioreactor for mESC culture to mimic the diffusion dominant mass transfer environment observed in vivo. The influence of soluble factors on cells in the micro-bioreator was compared with a macro-scale culture system. We observed enhanced retention of the pluripotent phenotype of mESCs in the micro-bioreactor owing to the enhanced effect of a soluble factor in a diffusion dominant microenvironment. A similar effect of the soluble factor was not observed in the macro-scale membrane-based Transwell insert culture system, in which soluble factors dissipated away from cell surrounding through inherent convection. This micro-bioreactor offers a suitable platform not only to understand the influence of secreted soluble factors on stem cell biology, but also to address an enhanced signaling environment to direct the ESC fate.


M. M. Chowdhury was supported by Monbukagakusho scholarship from the Japan Ministry of Education, Culture, Sports, Science and Technology (MEXT). This research was supported in part by CREST from Japan Science and Technology Agency and GMSI (Global Center of Excellence for Mechanical Systems Innovation), The University of Tokyo. We would like to thank Dr. Masaki Nishikawa and Dr. Morgan Hamon for their useful suggestions regarding various technical aspects related to this study.

Copyright information

© Springer Science+Business Media, LLC 2010