To facilitate the exploitation of embryonic stem cells (ESCs) and ESC-derived cells, scale-up of cell production and optimization of culture conditions are necessary. Conventional ESC culture methods are impractical for large-scale cell production and lack robust microenvironmental control. We developed two stirred-suspension culture systems for the propagation of undifferentiated ESCs—microcarrier and aggregate cultures—and compared them with tissue-culture flask and Petri dish controls. ESCs cultured on glass microcarriers had population doubling times (∼14–17 hours) comparable to tissue-culture flask controls. ESC growth could be elicited in shear-controlled stirred-suspension culture, with population doubling times ranging between 24 and 39 hours at 100 rpm impeller speed. Upon removal of leukemia inhibitory factor, the size-controlled ESC aggregates developed into embryoid bodies (EBs) capable of multilineage differentiation. A comprehensive analysis of ESC developmental potential, including flow cytometry for Oct-4, SSEA-1, and E-cadherinprotein expression, reverse transcription–polymerase chain reaction for Flk-1, HNF3-β, MHC, and Sox-1 gene expression, and EB differentiation analysis, demonstrated that the suspension-cultured ESCs retained the developmental potential of the starting cell population. Analysis of E-cadherin−/− and E-cadherin+/− cells using both systems provided insight into the mechanisms behind the role of cell aggregation control, which is fundamental to these observations. These cell-culture tools should prove useful for both the production of ESCs and ESC-derived cells and for investigations into adhesion, survival, and differentiation phenomena during ESC propagation and differentiation.
To facilitate the investigation and exploitation of embryonic stem cell (ESC)–derived cells in research, scale-up of cell production and optimization of culture conditions are necessary . To date, however, there have been few advances in the development of scalable culture techniques for the propagation of undifferentiated ESCs . Because the mouse is widely exploited as a model for conducting in vitro and in vivo studies to gain insight into different aspects of developmental biology and regenerative medicine that may eventually benefit humans, it is important to develop a technology for the scalable and controllable production of murine ESCs and ESC-derived cells.
ESCs are currently cultured as a monolayer on tissue-culture plastic and are subject to variations in the cellular microenvironment due the batch-wise passaging conditions. The rapid exhaustion of cultivation area in this system necessitates frequent user intervention and is associated with a heightened risk of contamination. Most importantly, the batch-type approach does not provide steady-state operating conditions and adequate cell physiology control, leading to variations in the cellular microenvironment  (potentially leading to uncontrolled cell-differentiation responses). Thus, new methods for the scale-up and control of ESC cultures are required. Stirred cultures are particularly appealing because of their ability to provide robust spatial and temporal control . Relevant to ESC growth and differentiation, stirred-suspension cultures have been used to control cell aggregation and 3-D tissue development . Adherent cells can also be expanded in stirred suspension using microcarriers as substrate for cell attachment .
One challenge associated with the suspension culture of ESCs and ESC-derived cells is their recognized dependence on cell–cell adhesion and aggregation for propagation . In this report, we examine the use of hydrodynamic shear to control aggregation and agglomeration. Impeller-associated shear effects have been shown to influence the aggregate size of baby hamster kidney  cells and neural stem cells [8, 9]. The control of aggregate size is important in the regulation of cell agglomeration  because the centers of very large aggregates (>300 μm) may experience cell death due to limitations in nutrients and oxygen delivery . It has been reported that the minimum shear needed to remove cells cultured on surfaces is 6.5 dyn per cm2 , whereas shear stress levels in the range of 15–30 dyn per cm2 are shown to cause damage to attached cells [12, 13]. Therefore, a culture system that can control aggregate size effectively must exert a shear stress that falls within the aforementioned ranges.
We report the establishment of two stirred-suspension systems, one using microcarriers as a substratum for ESC growth and a second that used shear to control ESC aggregate size. Both systems could be used to maintain an expanding ESC population; the second system could also be used to generate ESC-derived cells directly upon leukemia inhibitory factor (LIF) removal.
Materials and Methods
ESC Culture Maintenance
CCE , R1 , M8 , and 9J  murine ESCs were maintained with similar conditions as described elsewhere , using ESC maintenance medium containing Dulbecco's modified Eagle's medium (DMEM) (Invitrogen Corporation, Carlsbad, CA, http://www.invitrogen.com) with high glucose and without L-glutamine or sodium pyruvate. The medium was supplemented with 2 mM L-glutamine (Invitrogen Corporation), 2 mM sodium pyruvate solution (Invitrogen Corporation), 0.1 mM MEM non-essential amino acids solution (Invitrogen Corporation), 50 U/ml penicillin and 50 μg/ml streptomycin (Invitrogen Corporation), 15% ESC-qualified fetal bovine serum (FBS) (Gemini Bio-Products, Woodland, CA, http://www.gembio.com), 0.1 mM 2-mercaptoethanol (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), and 500 pM LIF (Chemicon International, Temecula, CA, http://www.chemicon.com).
Glass-coated styrene microcarriers (SoloHill Engineering, Inc., Ann Arbor, MI, http://www.solohill.com) and Cytodex 3 micro-carriers (Amersham Biosciences, Baie d'Urfé, Quebec, Canada, http://www.amersham.com) were prepared and stored according to the manufacturers' instructions.
Fifty-ml test cultures were established with a seeding density of 5.0 × 104 cells per ml (or 2.5 × 106 cells total) and normalized based on the total cultivation area available; 0.075 g Cytodex 3 or 0.56 g of glass-coated styrene beads was used to achieve a cultivation area of approximately 200 cm2. The microcarriers were equilibrated in 10 ml of ESC maintenance medium for 20 minutes at 37°C before inoculation. The microcarrier suspension, the inoculum, and 20 ml of ESC medium were added to siliconized 100-ml spinner flasks (Bellco Glass, Inc., Vineland, NJ, http://www.bellcoglass.com). The impeller height was set prior to sterilization such that the tip of the impeller was 1–2 mm away from the bottom indent. After a 24-hour seeding period with intermittent stirring (2 minutes of stirring for every 30 minutes of rest), 20 ml of ESC medium was added to the culture, and the culture was agitated with bidirectional stirring at 60 rpm for 2 days.
After 3 days of culture, the microcarrier culture was allowed to settle via gravity. The supernatant was aspirated, and the culture was rinsed with phosphate-buffered saline (PBS). The culture was incubated with trypsin-EDTA at 37°C for 3 minutes while being agitated at 120 rpm. The trypsin was neutralized with medium, and the culture was triturated 30 times to achieve single-cell suspension using a 10-ml pipette. The entire contents of the flask were filtered through a 40-μm or 70-μm cell strainer (BD Falcon, BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) to remove all the microcarriers. An inoculum of 5 × 104 cells per ml was reseeded back into the spinner flask with fresh microcarriers after cell count using a hemocytometer. The remaining cells were used for functional and phenotypic assays.
The suspension aggregate culture technique was adopted from the suspension bioreactor system for the culture of mammalian neural stem cells [19, 20]. Aggregates were dissociated periodically to first control their size and thereby prevent the onset of differentiation and to eliminate the differentiated cells through the enzymatic dissociation process. A single-cell suspension of 2.5 × 106 cells was added to a 100-ml spinner vessel containing 50 ml of ESC culture medium. The impeller height was set such that it was half-submerged in the culture medium (the bottom of the impeller was ∼1 cm from the bottom of the spinner flask). The culture was stirred continuously for 3 days at 60 or 100 rpm. At harvest, the aggregates settled via gravity and the supernatant was aspirated. The aggregates were trypsinized for 1.5 minutes at 37°C. Single-cell suspension was achieved by triturating the mixture 30 times using a 1,000-μl micropipette. The trypsin was then neutralized with medium, and the liquid was triturated 30 times using a 10-ml pipette to achieve a single-cell suspension. Upon centrifugation, the cell pellet was resuspended in medium and the cells were enumerated and reseeded back in the spinner vessel at 5 × 104 cells per ml. Ten-ml static–Petri dish aggregate cultures, seeded at a density of 5 × 104 cells per ml and harvested in a similar manner as the suspension system, were maintained as controls.
Induction of EB-Based ESC Differentiation
To form EBs from cells generated in microcarrier and suspension aggregate cultures in bacterial Petri dishes, 4 × 106 cells were inoculated in 10 ml of differentiation medium, which had similar compositions as ESC culture medium but without sodium pyruvate and LIF. Five ml of medium was added on the next day, and the EBs were passaged every 2 days. Day-4 EBs were incubated in trypsin for 2 minutes at 37°C, and the EBs were then triturated with a 1,000-μl micropipette. A single-cell suspension was further achieved with repeated pipetting after the trypsin was neutralized. Day-9 EBs were dissociated by incubating with 0.1% collagenase (Roche Applied Science, Laval, Quebec, Canada, http://www.roche-applied-science.com) (either in PBS supplemented with 20% FBS or in medium) with dispase (Roche Applied Science) for 30 minutes at 37°C. Thirty μl DNAse (Sigma-Aldrich) was added to the collagenase-cell mixture 20 minutes into the incubation period. Cells were triturated using a 1,000-μl micropipette. The collagenase was neutralized upon medium addition and was removed after centrifugation. The cells were then trypsinized (5 minutes at 37°C), triturated, and resuspended for analyses.
To generate EBs in stirred-suspension spinner cultures directly from ESC aggregates, suspension-cultured ESC aggregates were allowed to settle via gravity, the supernatant was removed, and the aggregates were resuspended in 50 ml of differentiation medium. The aggregates developed into EBs over 9 days of culture with frequent medium exchange (half to full medium exchange every day), and samples were taken on days 4 and 9 of differentiation for analysis.
Samples of the cells harvested from microcarrier cultures were suspended in Hanks' balanced saline solution supplemented with 2% v/v FBS at a concentration of 1 × 107 cells per ml, and 100 μl aliquots were analyzed in Eppendorf tubes (Brinkmann Instruments, Inc., Westbury, NY, http://www.brinkmann.com). Each marker was analyzed individually. SSEA-1, E-cadherin, and Oct-4 staining protocols were reported by Viswanathan et al.  Flk-1 and CD34 staining protocol was also described elsewhere . Data acquisition was conducted using the EXPO32 ADC flow cytometry system (Applied Cytometry Systems for Beckman Coulter, Sacramento, CA, http://www.beckmancoulter.com). Data analyses were performed using Summit V3.1 software (DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com). Unless otherwise stated, the cell population was considered to be positively stained if the measured fluorescence intensity exceeded that obtained by >99% of the negative control cells from the same population.
Total RNA was collected from cells using GenEluteTM Mammalian Total RNA Miniprep Kit (Sigma-Aldrich) according to manufacturer's instructions and stored at −80°C. The isolated RNA was quantified by a UV spectrophotometer (Eppendorf, Hamburg, Germany, http://www.eppendorf.com) at a dilution of either 1:50. RT-PCR was carried out using the One-Step RT-PCR kit from Qiagen (Mississauga, Ontario, Canada, http://www.qiagen.com). The oligonucleotide primers used for the amplification for various cDNA are listed in Table 1. PCR amplification conditions were set at 30 cycles of 1 minute of denaturation at 95°C, 1 minute of annealing at the listed temperatures, and 1 minute of extension at 72°C. β-Actin was amplified to verify the presence of RNA in the template. RT-free amplification was performed as a negative control. RT-PCR products were analyzed on a 0.8% Tris-acetate-EDTA (TAE) agarose gel. Bands were visualized under UV light, and images were captured using the ChemiImager 500 system (Alpha Innotech Corporation, San Leandro, CA, http://www.alphainnotech.com).
Table Table 1.. Properties of the oligonucleotide primers used in reverse transcription–polymerase chain reaction
The size is the number of base pairs between the forward and reverse sequences for the primers. Annealing temperature is the temperature at which the primer binds to the RNA template during polymerase chain reaction. The primers were purchased from ACGT Corporation (Toronto).
To measure the potential of our culture system for supporting ESC differentiation to functionally defined cells, suspension-grown ESC aggregates were used to generate EBs directly. These EBs were then plated into cardiogenic and hematopoietic culture conditions, using methods established in our lab [18, 22–24]. Briefly, to generate suspension-grown ESC aggregate–derived cardiomyocytes, EBs formed after 3 days of LIF removal were plated onto 10-cm gelatin-coated tissue-culture dishes with 10 ml of differentiation medium with 15% FBS (lot no. A36202V; NorthBio Inc., Toronto, http://www.northbio.com) and maintained for 5 days. Attached EB cultures were inspected on day 9 of differentiation for the presence of beating areas. For the colony-forming cell (CFC) assays, EBs were dissociated via trypsinization after 7 days of culture and 15,000 single cells in 300 μl of medium were evenly suspended in 2.7 ml of methylcellulose (M3434; StemCell Technologies, Vancouver, British Columbia, Canada, http://www.stemcell.com). One-ml aliquots of the cell suspension were seeded into 35-mm dishes. The methylcellulose cultures were maintained for 6 days, after which erythroid and myelomonocytic colonies were identified.
All data analyses, including graphical representations, were performed using Excel (Microsoft, Redmond, WA, http://www.microsoft.com) or Origin 6.1 (OriginLab, Northampton, MA, http://www.originlab.com). Statistical analyses were performed in Excel. F-test was conducted to determine variance equality between two test populations, and a two-sample t-test assuming equal/unequal variances was used depending on the result. Power calculations were used to determine the sample size needed for each comparison. Statistical analysis was performed regardless of the power calculation results to help describe any observed trend. The mean values were reported in all graphs, with the error bars representing the standard deviation. The sample size was indicated in the caption of each figure.
ESC Growth on Microcarrier Surfaces
Cultivation area restriction is a major concern associated with the culture of ESCs on tissue-culture plastic. Therefore, we explored the feasibility of adapting existing culture techniques to address this problem. The use of microcarriers in stirred suspension was an appealing alternative because the microcarriers have a high surface area–to–volume ratio, and the available cultivation area in each culture vessel can be adjusted easily.
Twelve commercially available microcarriers were used to establish ESC cultures in Petri dish–grade six-well plates. Glass microcarriers (SoloHill Engineering, Inc.) and Cytodex 3 micro-carriers (Amersham Bioscience) were chosen for the stirred-suspension studies based on the extent of ESC growth and/or culture morphology such as cell spreading (data not shown).
Total cell numbers were determined on days of harvest based on cell counts performed using trypan blue exclusion. The net population doubling time was calculated assuming that the viable cells exhibited exponential growth kinetics between seeding and harvest with a negligible lag phase. The extent of cell growth was also assessed via cell fold expansion, which was defined as the ratio of the viable cell number at harvest per 50-ml culture volume to the viable seeding density. The total or cumulative fold expansion, determined at the end of the 15-day culture period, was simply the sum of the cell fold expansions determined on days of harvest. CCE and R1 cells could be cultured on microcarrier surfaces consistently throughout the culture period. In particular, the average net population doubling times recorded for CCE and R1 cells cultured on glass microcarriers (CCE: 13.9 ± 0.7 hours; R1: 17.0 ± 1.9 hours) were not significantly different from their respective tissue-flask controls (CCE: 14.8 ± 1.3 hours; R1: 17.2 ± 2.0 hours) (Fig. 1A). CCE cell cultures could also be established on Cytodex 3 microcarriers, although they exhibited a slower net population doubling time (17.0 ± 2.4 hours). In contrast, R1 cells could not be cultured consistently on Cytodex 3, yielding poor and variable net population doubling times. CCE cells cultured on glass microcarriers achieved cumulatively a 191.8 ± 11.3–fold expansion during the 15-day culture period (compared with the 112.9 ± 10.5–fold and 158.8 ± 27.8–fold expansions achieved by the CCE Cytodex 3 cultures and tissue-flask controls, respectively) (Fig. 1B). R1 cells cultured on glass microcarriers had similar total fold expansions as their corresponding tissue-flask controls (105.3 ± 12.4–fold versus 104.4 ± 10.6–fold, respectively), but R1 Cytodex 3 cultures expanded only approximately 30-fold during the 15-day culture period (Fig. 1B).
To support the above cell-growth analyses, ESC characteristics were examined by monitoring the maintenance of ESCs in their undifferentiated state during the culture period, as well as their differentiation potential using flow cytometry and RT-PCR analysis. SSEA-1, E-cadherin, and Oct-4 expression was measured every 3 days during the 15-day culture period. The cells were then used to generate EBs to analyze ESC differentiation through the detection of markers from the three germ layers: Flk-1, CD34, MHC, HNF3-β and Sox-1. Regardless of the culture method, SSEA-1 and E-cadherin expression measured using flow cytometry remained high (>80%) for both CCE and R1 throughout the culture period (Figs. 1C and 1D, respectively). Oct-4 expression was also maintained (>80%), with a subsequent downregulation upon EB formation (Fig. 1E). Flk-1 and CD34 expression on day 15 of the culture was low; Flk-1 expression was upregulated by day 4–5 of EB formation, followed by the expression of CD34 by day 8–10 of EB formation (Fig. 1F). Flk-1, HNF3-β, MHC, and Sox-1 genes were upregulated as cells differentiated, as detected by RT-PCR (Fig. 1G).
A significant amount of bead-bridging (indicated by arrowheads in Figure 1H), caused by the collision of microcarriers and the adhesion of cells at the points of contact , was observed in both the glass and Cytodex 3 microcarrier cultures (Fig. 1H). Ideally, the surface of each microcarrier should be evenly covered with a monolayer of cells. Bead-bridging created a suboptimal culture environment, and hence the appropriate ratio of beads to cells could not be determined.
ESC Expansion as Spontaneously Formed Aggregates in Stirred Suspension
From the bead-bridging study we noticed that the resulting cell aggregates were not detrimental to ESC growth, indicating that these cellular masses contained live ESCs. Because the maintenance of microcarrier cultures can be laborious and time-consuming, we set out to develop a method to culture ESCs in suspension without a surface onto which cells adhered, allowing the cells to be sampled and harvested with ease.
CCE and R1 cells formed aggregates spontaneously in continuously stirred cultures. Fifteen-day cell-growth data for the stirred-suspension CCE and R1 cell cultures maintained at 60 and 100 rpm were compared with their tissue-flask and static–Petri dish controls. An expanding ESC population could be maintained, in suspension as aggregates, although the extent of cell growth was less than that achieved in tissue-culture flasks and Petri dish cultures (Figs. 2A, 2B). At 60 rpm, aggregates had larger and more variable sizes (CCE: 211.1 ± 89.2 μm; R1: 197.0 ± 98.0 μm) (Fig. 2H). In terms of cell growth, CCE suspension aggregates acquired a longer doubling time and were variable from trial to trial, resulting in the large standard deviation. Consequently, CCE cells expanded only 32.5 ± 22.8 times, while R1 cells expanded 25.2 ± 8.8 times, translating to a population doubling time of 32.1 ± 10.2 hours. To mediate the large variability in cell growth observed for our stirred-suspension culture system at 60 rpm, the agitation rate was increased to 100 rpm. The sizes of the resulting ESC aggregates were smaller and more uniform than those at 60 rpm (CCE: 136.2 ± 40.3 μm; R1: 116.5 ± 38.1 μm) (Fig. 2H). The population doubling times recorded for CCE and R1 cells were 23.5 ± 5.8 hours and 39.4 ± 19.4 hours, respectively. CCE cell aggregates cultured at 100 rpm expanded 53.4 ± 9.6 times in 15 days, as compared with 20.4 ± 11.0 times for R1 cells. Petri dish controls for both cell lines yielded 75.9 ± 17.4– and 49.0 ± 11.5–fold expansions for CCE and R1 cells, respectively.
Importantly, although the population growth of the suspension ESC aggregates was lower than the controls, SSEA-1 and E-cadherin expression was >80% for both cell lines throughout the entire suspension culture period (Figs. 2C and 2D, respectively). Oct-4 also maintained high expression level during the culture period, along with subsequent downregulation upon EB formation (Fig. 2E). Flow cytometric analysis of Flk-1 and CD34 yielded expected expression patterns (Fig. 2F). RT-PCR analysis of Flk-1, HNF3-β, MHC, and Sox-1 gene expression was also consistent with normal ESC development (Fig. 2G).
E-Cadherin Mediated the Extent of Cell–Cell Adhesion in Microcarrier and Suspension Aggregate Cultures
We speculated that this high level of bead-bridging observed in our microcarrier cultures could be due to the expression of the cell adhesion molecule E-cadherin on undifferentiated ESCs . Microcarrier cultures performed using the M8 ESCs heterozygous for E-cadherin behaved similarly to CCE and R1 cells in microcarrier cultures (Fig. 3A). In contrast, similar cultures of the E-cadherin–null 9J ESCs showed that bead-bridging was dramatically reduced (Fig. 3B). A monolayer of 9J cells was able to attach onto the glass microcarrier surface; however, the cells did not spread out and flatten on the substratum, and most of the cells remained unattached, contributing to poor culture viability.
M8E-cadherin+/− ESCs were also cultured as suspension aggregates agitated at 100 rpm for 6 days with dissociation and reseeding on day 3. The average aggregate size was significantly smaller than those of CCE and R1 cells (M8: 69.2 ± 18.1 μm) (Fig. 3C). In contrast, 9J E-cadherin−/− ESCs were unable to form aggregates in suspension at 100 rpm. The culture consisted of mostly singly suspended cells with poor viability. There were also a small number of larger, irregularly shaped cell clumps that were formed (Fig. 3D).
ESC Aggregates Could Be Induced to Form EBs in Differentiation Medium While Being Maintained in Stirred-Suspension Cultures
We asked next if a single-step expansion differentiation bioprocess could be developed based upon the shear-controlled expansion of ESCs. In this study, 50-ml suspension aggregate cultures agitated at 100 rpm were established, as described above, for the expansion of undifferentiated cells. After extended culture, LIF was removed from the media and the ESC aggregates were induced (at 100 rpm) to differentiate into EBs over 9 days. Analyses of cells generated from the EBs formed from the suspension-expanded ESC aggregates revealed that the cells exhibited normal differentiation kinetics. Oct-4, Flk-1, and CD34 had an expression pattern similar to that of the other culture techniques (Figs. 4A, 4B). In addition, phenotypic marker expression was also detected by RT-PCR as previously mentioned (data not shown). The nondissociated aggregates were able to form EBs that increased in size with time (Fig. 4C) in 100-ml suspension without detrimental agglomeration (CCE: day 4 EBs, 6,150 ± 1,485 aggregates; day 8 EBs, 2,600 ± 1,273 aggregates).
To demonstrate that the ESC suspension aggregates were able differentiate into functional cardiac and hematopoietic cells, expanded aggregates were induced to differentiate upon LIF removal and the resultant EBs were tested for their ability to form contractile areas and hematopoietic CFCs, respectively, using standard assays (see Materials and Methods). Beating areas of cells were formed from these EBs, as expected, after attachment (data not shown). Using the CFC assay, we observed that cells dissociated from EBs generated with ESC aggregates generated both erythroid and myleomonocytic colonies (Fig. 4D). In both cases, no detectable differences relative to standard control conditions were observed.
Prolonged ESC culture requires a balance between cell growth and differentiation. We explored the possibility of culturing ESCs in suspension on microcarriers and as spontaneously formed aggregates. We found that both culture systems could be adapted to accommodate the propagation of ESCs, albeit with different efficiencies. CCE cells proliferated well on both glass and Cytodex 3 microcarriers; however, significant bead-bridging occurred. R1 cells could be propagated consistently only on glass microcarriers. Both CCE and R1 ESCs could also form uniform aggregates in stirred suspension without encapsulation. Normal differentiation could be induced in a single-step process from the suspension culture–expanded ESCs.
The nature of the culture substrata provided by the micro-carriers influenced the efficacy of ESC adhesion and spreading. Consistent with our observations, studies involving other cell types also suggested that glass microcarriers were able to support growth better than dextran-cored Cytodex 3 microcarriers . Cell attachment on both substrata also differs; the filopodia on the cells anchor on the surface of the glass microcarriers, while they penetrate into the dextran Cytodex 3 beads . As a result, cells detached with ease when they were trypsinized from the glass surface, but cells cultured on Cytodex 3 could experience cell damage, resulting in variable cell yields. The high cell numbers achieved by our glass microcarrier cultures were convoluted by the significant cell–cell aggregation and bead-bridging observed on the bead surfaces (Fig. 1H). Cell–cell aggregation may create a beneficial microenvironment by releasing autocrine growth factors, leading to the requirement of fewer supplemented growth factors . Also, these masses may be less susceptible to shear because of their elastic properties; the ability to deform under stress may diminish shear impact on individual cells .
Our results from the suspension aggregate culture systems demonstrate that ESCs could be propagated as aggregates with minimal agglomeration, using a high agitation speed at an impeller height of approximately 1 cm from flask base, a setup that provided sufficient agitation for completely suspending the cells while allowing aggregates to form. To control aggregate size, adequate shear must be exerted to remove the outer cells of an aggregate (6.5 dyn/cm2) , while excessive shear stress (15–30 dyn/cm2) may result in cell damage [12, 13]. In a similar system, the shear stress exerted at 60 rpm was reported to be 6.25 dyn per cm2 , which may not be enough to effectively dislodge the outer cells of the aggregates, resulting in large aggregate sizes that may lead to nutrient depletion, causing variable and reduced cell growth. Agglomeration was controlled by increasing the bioreactor agitation rate to 100 rpm exerting a shear stress of approximately 9.86 dyn per cm2  to be exerted on the cells. In addition, the reduction in aggregate size upon an increase in agitation rate (100 rpm) helped improve the quality of the harvested ESCs, likely by facilitating effective nutrient and oxygen transport into the center of the aggregates. This beneficial effect of shear is demonstrated by the ability of the harvested cells to be cultured under high-shear conditions and consistently reaggregate and expand after dissociation.
The performance of our culture systems was influenced by a number of factors. First, E-cadherin expression was found to be critical in the establishment of ESC microcarrier and suspension aggregate cultures. The extent of cell–cell adhesion was likely mediated by the expression of E-cadherin in ESCs, which appeared to support rather than hinder growth. E-cadherin−/− 9J ESCs adhered onto the surface of glass microcarriers without cell spreading, and they could not form aggregates in suspension. In contrast, E-cadherin+/− M8 cells were cultured successfully both on microcarriers and as aggregates in 100-rpm stirred-suspension culture. The reduction in size of M8 ESC aggregates (diameter = 69.2 ± 18.1 μm), compared with those generated by E-cadherin+/+ ESCs (CCE: 136.2 ± 40.3 μm; R1: 116.5 ± 38.1 μm), suggests that cell–cell interaction was compromised due to the reduced expression of E-cadherin. Together, the results indicate that E-cadherin expression is important for ESC adhesion, spreading, and cell–cell interactions, leading to culture robustness.
Second, our data demonstrate that the undifferentiated state of ESC aggregates can be controlled by shear and constant dissociation. During aggregate dissociation, the differentiated cells are separated from the undifferentiated ones, and the undifferentiated cells will reaggregate, likely due to strong E-cadherin expression (Fig. 5A). (The E-cadherin−/− cells failed to reaggregate.) This is consistent with Steinberg's differential adhesion hypothesis, which states that cell adhesion is maximized when weaker cell attachment is displaced by stronger ones . Therefore, the less cohesive (lower E-cadherin–expressing ) differentiated (primitive endoderm) [30, 31] cells will envelope the more cohesive undifferentiated ESCs. The small eddies that form under high agitation rate [25, 32] can shear off the differentiated cells as they collide with the eddies, hence reducing the aggregate size while maintaining an undifferentiated ESC population. We hypothesize that if the marker expression of ESCs is tracked just throughout the aggregation-dissociation-reaggregation process, one should be able to observe an oscillatory expression profile similar to the one shown in Figure 5B.
Our results demonstrate the development of a single-step scalable bioprocess for the generation of ESCs and ESC-derived cells. They can be used to discover cues (e.g., exogenous cytokines and growth factors) that can be added to the suspension ESC culture environment to establish protocols for the directed differentiation of human ESCs into functional tissues.
The authors would like to thank Amersham Biosciences and Dr. William Hillegas from SoloHill Engineering, Inc., for providing advice and reagents associated with the microcarrier culture systems. We also appreciated the provision of the M8 and 9J murine ESC lines from Dr. Lionel Larue (Institut Curie, France). This work was funded by the Natural Sciences and Engineering Research Council (NSERC) of Canada and the National Science Foundation (NSF) through an Engineering Research Center (ERC) grant to the Bioprocess Engineering Research Center (BPEC) at the Massachusetts Institute of Technology. P.W.Z. is the Canada Research Chair in Stem Cell Bioengineering.
The authors indicate no potential conflicts on interest.