Scale-up of Human Mesenchymal Stem Cells on Microcarriers in Suspension in a Single-use Bioreactor

Published on: 
, , , ,
BioPharm International, BioPharm International-03-01-2012, Volume 25, Issue 3

Demonstration of large-scale stem-cell scale-up.


Clinical demand for mesenchymal stem cells (MSCs) drives the need for development of robust large-scale production. This study demonstrates the utility of a 3-L single-use bioreactor and collagen-coated microcarriers for the expansion of human bone marrow derived MSCs. This proof of principle study is a demonstration of the potential for large-scale stem-cell scale-up using stirred bioreactors.

Mesenchymal stem cells (MSCs) are multipotent cells with the ability to differentiate into a variety of cell types including osteoblasts, chondrocytes, and adipocytes. These cells have been explored for the repair and regeneration of connective tissues such as cartilage and bone, and for transfusion therapy in patients following bone marrow or peripheral blood stem cell transplants to reduce complications from life-threatening graft-versus-host disease (1, 2).

As demand for stem cells for both drug discovery and clinical applications grows, effectively translating the promise of stem cells into therapeutic reality will require large-scale industrialized production under tightly controlled conditions. Achieving this level of production while meeting rigorous quality and regulatory standards will depend on further progress in the areas of cell culture and scale-up, characterization, enrichment, purification, and process control to deliver a consistent and reproducible supply of cells in a safe and cost-effective manner.

To meet the market needs and clinical demand for MSCs, rapid, robust expansion methods are required. To date, large-scale production is typically achieved using two dimensional (2D) tissue culture vessels—an expensive and time-consuming process. The research presented here examines the utility of a single-use, stirred-tank bioreactor in combination with microcarriers for mesenchymal stem-cell expansion, and comprehensively compares the characteristics of the product cells with those grown in standard 2D cultures.


MSCs (EMD Millipore SCR108) were cultured under static conditions with low glucose DMEM (Invitrogen 11054), 10% FBS (HyClone SH30070.03), Pen/Strep (EMD Millipore TMS-AB2-C), L-Glutamine (EMD Millipore TMS-002-C), and 8 ng/mL human recombinant β fibroblast growth factor (EMD Millipore GF003AF-MG) in T-150 flasks coated with gelatin (EMD Millipore ES-006B). Low oxygen conditions were used during 2D propagation as well as during the attachment phase in which the MSCs were attached to the collagen-coated microcarriers (Solohill C102-1521) in Petri dishes.

For agitated culture, the growth medium was supplemented with pluronic acid (Sigma P5556) and antifoam C emulsion (Sigma A8011-500ML). Spinner flasks (Corning 3152) were pre-coated with Sigmacote (Sigma SL2-100ML) and operated at 30 RPM. The impeller speed in the the Mobius CellReady 3-L bioreactor (EMD Millipore CR0003L200) was set to 25 RPM at low volume (1 L) and then increased to 40 RPM at the larger volume (2 L). The cell concentrations were measured daily using a NucleoCounter (Eppendorf M1293-0000) after lysing the cells off the microcarriers. Supernatant was analyzed daily on a BioProfile Flex (Nova Biomedical) to generate the profiles of metabolite accumulation and nutrient consumption. Cells on microcarriers were fixed with 4% paraformaldehyde (USB 19943 1 LT) and stained with DAPI (Invitrogen D1306) to fluorescence the nuclei for images.

RNA was analyzed by reverse transcriptase–polymerase chain reaction (RT–PCR, Invitrogen 10928-042 kit) following isolation on glass fiber filters post guanidinium thiocyanate treatment (Ambion AM1912 kit). Custom DNA oligonucleotide primers (Invitrogen) were used for the PCR at 200 nM. Expression of cell surface markers on the MSCs (e.g. positive markers CD44, CD105, and CD90, and negative marker CD19), were measured using corresponding antibodies (EMD Millipore SCR067). An adipocyte differentiation kit (EMD Millipore SCR020) was used to identify cells containing lipid vacuoles that stained positive with oil red stain, indicative of cells that have undergone adipogenic differentiation.


To better understand the growth of bone marrow-derived stem cells, initial experiments were designed to optimize growth rates while the cells were grown two-dimensionally. Various seeding densities and oxygen levels were investigated to understand which parameters led to the highest growth rate or lowest doubling time so that these seeding densities and oxygen levels could later be used in the three dimensional (3D) culture (see Figure 1). To accomplish this, MSCs were grown over multiple passages in tissue culture dishes at varying cell densities. A lower cell density of 5,000 cells per cm2 led to a shorter average doubling time of 49 h compared with cells seeded at a density of 20,000 cells/cm2, which resulted in a longer doubling time of 63 h. Differences in doubling times at different seeding densities could be due to a number of factors, including inhibitory signals from neighboring cells, competition for growth factors and nutrients, or waste product build up at higher density cultures. The seeding density of 5,000 cells per cm2 was selected as a seeding density for experiments designed to translate culture conditions from 2D to 3D culture.


Figure 1: (A) Determination of optimal seeding density on mesenchymal stem cell doubling time; (B) Determination of oxygen levels on mesenchymal stem cell doubling time. (ALL FIGURES ARE COURTESY OF EMD MILLIPORE)

To determine the best oxygen level to support MSC growth, cells were grown for multiple passages at different ambient oxygen levels. Hypoxic conditions were preferable for cell growth, with a shorter average doubling time of 45 h for cells grown at 5% oxygen versus cells grown under normoxic conditions (i.e., 21% oxygen; doubling time of 54 h). The hypoxic conditions that led to superior growth are likely closer to the physiological oxygen levels within bone marrow than are normoxic conditions (3, 4). The depth of liquid in 2D culture is measured in millimeters, while 3D or deep cultures have vertical liquid levels that can be measured in centimeters for spinner flasks and smaller stirred tank bioreactors like the Mobius CellReady 3L, or meters for larger scale reactors. This height difference could cause lower dissolved oxygen levels in the spinner flask than in 2D culture even when they are both exposed to normal atmospheric oxygen levels. Spinner flask cultures maintained at different oxygen levels (5%, and 21%) showed similar growth (data not shown). For subsequent Mobius CellReady 3L cultures, dissolved oxygen levels were monitored and shown to remain above 5% oxygen using headspace aeration. In addition to better defining the desired operational window, these experiments led to the observation that a 48 hour doubling time was adequate for 2D growth and this benchmark was used later to gauge 3D growth.

Different microcarrier types were evaluated for their ability to support MSC attachment, growth, and viable detachment (see Figure 2). The attachment and propagation of MSCs on various microcarriers was first investigated. Four microcarriers showed robust growth over the four day culture and were then further evaluated in the follow-up comparison. Top performers were compared to reveal how efficiently viable cells could be harvested from the microcarriers. After five days of growth, cells were removed from the microcarriers using three different dissociation reagents (e.g., trypsin, acutase, and collagenase) for 10 min at 37 °C and the results were averaged.

Figure 2: (A) Attachment and propagation of mesenchymal stem cells on various microcarriers; (B) Recovery of viable cells.

The collagen-coated microcarriers led to the highest viable cell recovery and were subsequently used in 3D studies. A high percentage of viable cells were also recovered from Hilex microcarriers, and these microcrocarriers gave the best recovery using an animal product-free microcarrier. The material of the microcarrier may have played a crucial role in promoting viable recovery. Both the Hillex and collagen microcarriers are made of polystyrene, while the Cytodex beads are constructed from dextran matrices. Even though the Cytodex beads showed a growth rate advantage over collagen beads, the number of viable cells recovered was much higher from the collagen microcarrier culture. Because the stem cells are the product, the remainder of the experiments were performed on the collagen-coated polystyrene microcarriers.

Following the attachment study, MSCs were cultured under stirred agitation conditions in spinner flasks. The MSCs were seeded onto the collagen-coated microcarriers in ultra low adherent petri dishes under static conditions for two days, then transferred to a spinner flask and agitated at thirty revolutions per minute. The MSCs were then sequentially passaged from one spinner flask to another using 20% of the cells on microcarriers from the first spinner after five days and seeding them with 80% of media with fresh microcarriers. This procedure was then repeated and a four- to six-fold increase in cell number was observed over the first five days of the culture for each passage.

Figure 3 shows the cell concentration profile versus time and cell nuclei stained with DAPI after 5, 10, and 20 days of culture in stirred agitation spinner flasks. The cells readily propagated and after 3 days cells were found attached to nearly every microcarrier. This was a very important result because it shows that the MSCs can jump from one microcarrier to another under stirred agitation. Because only around 4% of microcariers that were seeded under static conditions were still in the culture after 10 days, and yet the day 20 image shows more than 4% of the microcarriers populated with cells, the cells must have traversed from a confluent surface to a fresh surface to populate the fresh microcarriers.

Figure 3: (A) Cell nuclei and (B) cell culture concentration profile after 5, 10, and 20 days of culture in stirred agitation spinner flasks.

After it was shown that MSCs could propagate under stirred agitation, studies relating to the scale-up from the 125-mL spinner flask to the Mobius CellReady 3L bioreactor were undertaken. MSCs on microcarriers that were still in the exponential growth phase were transferred from spinner flasks (200 mL at 100,000 cells/mL) to the single-use Mobius CellReady 3L bioreactor which contained 800 mL of media with fresh microcarriers that were at temperature and agitating at 25 RPM.

On day three, one liter of media with fresh beads was added to the one liter culture and the impeller rate was increased to 40 RPM to keep the microcarriers suspended. The cell concentration profile of cells on microcarriers growing in the Mobius CellReady 3L and images of cell nuclei on microcarriers at various time points are depicted in Figure 4. Cell number increased by 5.2-fold over the five day culture period.

Figure 4: (A) Increase in cell number in single-use bioreactor. (B) By the seventh day of culture, greater than 90% of the microcarriers had mesenchymal stem cells growing on them.

This growth rate is similar to the 48 h doubling time exponential curve for the first five days; the cell concentration then drops off this pace slightly by the seventh day. As shown in the images, after one day of growth in the Mobius CellReady 3L bioreactor there are still fresh beads that have not been populated and microcarriers that are near confluence. The latter are probably beads that were transferred from the spinner flask, but there are also several microcarriers that have nuclei, indicating new growth. The day four image shows many microcarriers with many MSCs, but there are also some naked microcarriers, most likely introduced when the number of microcarriers was diluted 1:1 with new micorcariers with the day three media addition.

In the day seven image, propagation of cells from bead to bead is most evident; a culture that begins with a minority of confluent microcarriers (10%) has MSCs on greater than 90% of the microcarriers a week later. Some small aggregates of microcarriers (2–5 beads) were observed, but because the static 3D petri dish cultures also contained small aggregates which was the model for 3D growth, these small clumps were not discouraged. The cell concentration measured by counting nuclei lysed off the microcarriers and the images of nuclei on the microcarriers give a strong indication that the cell number was increasing and that the cells were propagating to fresh microcarriers. These analyses, however informative, do not supply information regarding the health of the cells, nor do they indicate if these cells have the typical characteristics of MSCs. Accordingly, further characterization studies were performed.

To better understand the health of the cells and to determine if the cells are active, several additional parameters were measured. Daily supernatant samples from the single-use bioreactor were collected and analyzed and the profiles of nutrients and metabolites in the media during the culture were generated (see Figure 5). Glucose levels decreased from 0.79 to 0.48 g/L between day one and three; after the media addition, levels fell again from 0.57 to 0.02 g/L between days four and seven. The average specific glucose consumption rate over the culture was 2.7E-09 g/cell-day. Lactate levels increased steadily from 0.34 to 0.68 g/L between days one and three, and again after the media addition, from 0.51 to 1.0 g/L over the last three days. The average specific generation rate of lactate was 1.9 E-09 g/cell-day during the culture. Because of the decreasing glucose levels, it was decided to perform characterization, viable recovery, and differentiation studies after five days in the Mobius CellReady 3L bioreactor. Cultures longer than five days may require more media additions or perhaps higher levels of glucose in the growth media (low glucose was used for this study) to maintain suitable levels of glucose for exponential cell growth. Showing that the MSCs are consuming glucose and producing lactate over the first five days of the culture indicates that the culture is healthy, but it does not indicate how the cells might perform after being removed from the microcarriers after the 3-L culture.

Figure 5: (A) Analysis of glucose and (B) lactate levels from the 3-L single-use bioreactor.

To show that the MSCs were still viable after five days of growth on collagen-coated microcarriers in the Mobius CellReady 3L bioreactor, the growth capability of MSCs harvested from the suspension culture was investigated. MSCs were withdrawn from the bioreactor and trypsinized for 5 minutes at 37°C to remove the cells from the microcarriers. The suspension was passed through a 100 μm sieve, and the cells were seeded on gelatin-coated tissue cultureware for propagation. The cell number versus time and bright field images of these cells are depicted in Figure 6. Cells retained their ability to propagate for multiple passages and there was no observable lag in growth after the cells were transitioned from the collagen microcarriers back to 2D gelatin-coated flasks. The MSCs show the morphology that is classically indicative of MSCs and this can be observed during various points of this culture in the bright field images. Another signal of healthy stem cells is the capacity of sustained growth after microcarrier culture; this was observed, as expected. Our next studies aimed to use additional genotypic and phenotypic analyses to confirm that these cells were definitively MSCs.

Figure 6: (A) Cells removed from the microcarriers retained the ability to propagate when seeded onto gelatin coated flasks; (B) Cells removed from the microcarriers had consistent morphology.

In addition to growth and morphology, MSCs typically express several markers at high levels, while not expressing others. To determine whether the MSCs remain MSCs after the 3-L bioreactor culture, mRNA from the cells was isolated and the expression levels of several genes were probed using RT-PCR. The MSC characterization genes that were investigated were positive markers CD44, CD105, and CD90, and negative marker CD19. The relative expression of these genes was compared with samples of cells derived from the 3-L single-use bioreactor, spinner flasks, static cells grown on 3D microcarriers, and gelatin-coated tissue culture flasks (2D gel). Cells expanded in the 3-L bioreactor showed similar levels of gene expression of the MSC characterization genes when compared with cells grown on gelatin-coated tissue culture flasks.

High RNA levels of the positive markers were observed for all culture conditions, while expression of markers chosen as negative controls was not observed. The housekeeping gene GAPDH was present in similar levels between the samples. Differentiation of the MSCs in the 3-L bioreactor would have caused the gene expression of the characteristic markers to change; that these genes were expressing similar amounts of messenger RNA across the samples indicated that the cells remained undifferentiated.

A second method to characterize MSCs is to examine protein expression levels of MSC characterization surface proteins. Cells taken from the 3-L bioreactor and dissociated from the beads were seeded onto gelatin-coated glass slides, along with control MSCs; these cells were then exposed to antibodies specific for these surface proteins. Labeled secondary antibodies (red) were incubated with these samples and then counterstained with nuclei marker DAPI (blue) so that immunofluorescent images could be collected. CD44 and CD90 protein levels of cells taken from the bioreactor were comparable to the levels observed in cells that were grown on gelatin in tissue culture flasks. The expression of negative control markers CD14 and CD19 was not observed. Results from both MSC characterization experiments are shown in Figure 7. In subsequent studies, a broader array of positive and negative surface markers and quantification of positive cells using flow cytometry can be used.

Figure 7: Mesenchymal stem cell characterization. (A) Similar RNA levels were found from all culture configurations; (B) Similar levels of protein were also observed in mesenchymal stem cells grown in the 3-L single-use bioreactor compared with gelatin-coated culture flasks.

To show that the MSCs retained their ability to differentiate, a study was performed to coax MSCs towards adipocytes. MSCs removed from microcarriers from the 3-L bioreactor culture and 2D control cultures were compared to evaluate their ability to differentiate. The adipocyte differentiation was conducted over three weeks; cells were fixed and the differentiation was assessed via oil red staining of lipid vacuoles present in adult adipose cells (see Figure 8). Cells from the Mobius CellReady 3L and gelatin control cultures both contained the characteristic vacuoles indicative of adipocytes, indicating that the cells had some of the same differentiation capabilities as the control MSCs. Future experiments will include additional differentiation protocols to study the differentiation of the cells to additional lineages.

Figure 8: : Lipid vacuoles of apidocytes were stained red after differentiating cells removed from the 3-L bioreactor and cells grown on gelatin following a two-week differentiation.

Results indicate that MSCs are capable of being expanded in the 3-L bioreactor on collagen-coated microcarriers. Cells grown in the bioreactor showed similar growth rates to control cells that are grown on tissue cultureware. The MSCs produced from these cultures also express similar levels of characteristic stem cell genes as observed in the RNA and protein expression. They also retain the ability to grow for multiple passages and to be differentiated to adipocytes at similar degrees as control MSCs grown in 2D gelatin cultures.


Clinical demand for MSCs drives the need for development of robust large-scale production. This study demonstrates the utility of a 3-L single-use bioreactor and collagen-coated microcarriers for the expansion of human bone marrow derived MSCs. This proof of principle study is a demonstration of the potential for large-scale stem-cell scale-up using stirred bioreactors.

Optimal seeding density and oxygen levels were first established for MSC cultures. A seeding density of 5,000 MSCs per cm2 propagated more effectively than other seeding densities that were tested as did cells grown at low oxygen when compared with normoxic conditions. Collagen-coated microcarriers led to the highest recovery of viable cells after static culture. MSCs grown on microcarriers were able to propagate from one microcarrier to another while under stirred agitation in spinner flasks and were able to retain their proliferation rates after two sequential passages.

MSCs propagated in the 3-L single-use bioreactor for five days, while doubling the working volume, with a 5.2-fold increase in total cell number from 30 to 150 million cells. MSCs were capable of growing for multiple passages after being removed from the bioreactor and showed similar levels of gene and protein expression of MSC characterization genes. After differentiating to adipocytes, both the cells from the 3-L bioreactor and cells grown on gelatin contained lipid vacuoles that stained positively red, confirming successful differentiation.

The next step to augment the results presented here will be to use multiple MSC lines from different donors to better characterize the robustness and reproducibility of this process. Additionally, upcoming studies will use a wider and more quantitative array of stem cell markers at both the RNA and protein levels to more fully characterize the cells produced using this method.

Daniel Kehoe, PhD,* is a research scientist in Process Solutions, Aletta Schnitzler, PhD, is a research scientist in Pharmaceutical Chemical Solutions, Janice Simler, PhD, is a senior research scientist in Process Solutions, Anthony DiLeo, PhD, is director, Corporate Technology Office, and Andrew Ball, PhD, is R&D manager, Systems Pharmacology, all at EMD Millipore, Billerica MA. *To whom correspondance should be addressed,


1. P-HG Chao, W. Grayson, and G. Vunjak-Novakovic, J. Orthop. Sci. 12 (4), 398–404 (2007).

2. M. von Bonin et al., Bone Marrow Transplant 43 (3), 245–251 (2009).

3. K. Parmar et al., Proc Natl Acad Sci. 104 (13), 5431–5436 (2007).

4. C. Holzwarth et al., BMC Cell Biol. 11, 11 (2010).