The Importance of Using Small Scale Bioreactor Mimics to Scale up Human Embryonic
Stem Cell Culture
Iwan Roberts1, Nathalie Moens1, Fabien Moncaubeig2, Mahieu Egloff2 , Pete Coffey3,4, Chris Mason1
Advanced Centre for Biochemical Engineering, University College London, 2ATMI LifeSciences, 3Instute of Ophthalmology, University
College London, 4Neuroscience Instute, University of California, Santa Barbara
The Promise of Bioreactors
• The commercialisaon of allogeneic human embryonic stem cell therapies for large
paent populaons is reliant on both minimising manual handling and realising
economies of scale
• Bioreactors are a proven technology for the producon of biopharmaceucals
Some Problems of Tradional Bioreactors
• Moving from well understood adherent 2D cell culture in flasks to a suspension or
micro carrier based culture at commercial scale will require significant me and
investment.
• Requires new analycal methods to track cell growth and phenotype?
• Are the cells produced sll the same? (aggregates create microenvironments)
– Need to repeat clinical/pre clinical work?
The Integrity Xpansion mulplate bioreactor (ATMI)
• Easiest way to scale up 2D flask culture is to ‘stack’ the layers and remove the air
space between the layers.
• Minimises the process development required, e.g. cells can sll be imaged
Integrity Xpansion
Mulplate bioreactor
‘Stack’
Remove Air Space
Using the Xpansion One to Predict Large Scale Performance
• Large scale experimentaon is me consuming and expensive
• The Xpansion One (128cm2) is used to explore key variables
before a move to the large scale bioreactors is made.
2) Xpansion Bioreactor Design
Integrity Xpansion Mulplate Bioreactor
• Available Scales: 6,147cm2, 30,735 cm2,
110,646 cm2.
• Equivalent surface area of to 18 multray
10 layer stacks
• 35cm diameter, 60cm height
• Liquid volume 20L
• Gas exchange is provided via a gas
permeable silicon tube in the central
column.
• Mixing/circulaon is iniated when
needed to control the pH and Dissolved
Oxygen (DO) to setpoints.
Protocol
• hESC were seeded into triplicate T25 flasks to act as controls for each Xpansion One
experiment using the protocol below.
Small Differences in Environment make a Significant Difference in Cell Growth
• Xpansion one performance improved through an iterave process
• pH set point, Flow rate used, O2 and CO2 levels in the incubator were varied
• Best performance yielded 0.12 million cells cm 2
• 2.9mm/s linear flow rate, pH Set point 7.32, CO2 reduced to 3% on day 3
• However this cell growth was 38% lower then flask controls
• Interesngly pH (Fig. 1), glucose and lactate (Fig. 2) on spent media show very
similar relaonship between the flasks and the Xpansion One, with pH levels
on the Xpansion One becoming slightly lower by harvest.
• It is believed that these small deviaons in pH (seen on the online plot in the laer
days with the pH falling below the set point) account for the reduced growth.
• It would be expected that beer control of pH at higher densies would be obtained
through automac control of CO2 levels – this is possible on the larger scale
bioreactors
Day 0
• Mytomycin C inacvated mouse embryonic
fibroblasts seeded at 10,000 cells/cm2
Day 1
• hESC seeded at 20,000 cells/cm2
• 6 hours post seeding regulaon started
Day 2
• Media exchanged at a rao of 0.18ml/cm2
Day 3
• Media exchanged at a rao of 0.18ml/cm2
Gas IN/OUT
Gas exchange
occurs in central
column
Fixed bed
made up of
plates
Liquid OUT
Liquid IN
Integrated mixing
system
Media circulaon through Xpansion plates
Day 4
• Cells Harvested
Regulaon System
• Dissolved gasses in the media are
equilibrated with the gas in the
incubator by pumping through a
silicon tube
• When pH is between the setpoint
and alarm values there is no
pump acon.
• When the pH is below the alarm
value the pump is turned on.
• As cell densies increased around
day 4 the pH fell below the set
point. To try and combat this the
impact of manually reducing the
CO2 levels in the incubator was
explored.
Media Exchange
Media Exchange Media Exchange
pH Setpoint: 7.32
Alarm: 0.05
Pump Acon
Small Scale Mimic
Xpansion One Set up
Flow Rate Increases did not impact Cell Growth
or hESC Phenotype
• Cell growth was maintained over a 25 fold
range of linear flowrates (0.2 5mm/s)
• The pluripotent phenotype was also
maintained, with no significant difference in
marker expression, as determined by flow
cytometry
• Bars represent averages and standard
deviaons from 5 different Xpansion One
experiments and accompanying flask
control cultures.
100
Posive Cells (%)
Xpansion One
• Surface area: 128cm2 (23ml)
• Mimics culture condions of the Integrity
Xpansion
• Media flows across the plate and
through an external gas permeable
silicon tube to mimic the hydrodynamic
environment and control of the large
scale bioreactor
4,00
1,60
7,8
3,50
1,40
3,00
1,20
2,50
1,00
2,00
0,80
1,50
0,60
7,2
1,00
0,40
7
0,50
0,20
0,00
0,00
7,4
4) Impact of pH Regulation and Flow on Cell Phenotype
Connectors/opcal
fibre (pH, DO)
8
7,6
Analysis
• Cell Number, Size and Viability: ViCell (Beckman Coulter)
• Spent media analysis: glucose and lactate YSI 2700 (YSI Life Sciences), pH
• Flow Cytometry: Epics XL MCL (Beckman Coulter)
• Xpansion One: Online pH and DO
Objecve
Large Scale Reactor
5) Importance of Online Control
Glucose (g/L)
Bioreactor Background
3) Materials and Methods
pH
1) Objective
80
60
40
20
0
SSEA 4
Xpansion
One
SSEA 4
Flask
TRA 1 60 TRA 1 60
Xpansion
Flask
One
6,83
6,8
6,66
6,6
0
20
40
60
Time (h)
pH: Xpansion One, 2.9mm/s
80
Flask
100
0
Lactate (g/L)
1The
20
40
60
80
Time (h)
Glucose: Xpansion One, 2.9mm/s
Flask Glucose
Lactate: Xpansion One, 2.9mm/s
Flask Lactate
6) Conclusions
• Cell growth was not impacted by increasing the regulaon flow rate 25 fold.
• The hESC cell phenotype was also maintained
• At high cell densies towards the end of the culture the pH could no longer be
maintained at the set point which resulted lower pH levels at harvest compared to
flasks.
• However Glucose and lactate levels remained largely the same.
• CO2 levels in the incubator were manually reduced down from 5% to try and migate
against this.
• It is likely that these small differences in pH accounted for the decreased cell growth
in the Xpansion One system when compared to flask culture (38% fewer cells per
cm2)
• It is believed that if the CO2 and O2 composion of the gas could be automacally
altered that beer environmental control would produce comparable cell yields
between the Xpansion One and flask culture.
• This is possible on the larger scale reactors, and could potenally be retrofied
to the Xpansion One.
• In addion if the media was provided by perfusion addional control could be
obtained by increasing the amount of media supplied.
In summary
Through use of the Xpansion One small scale mimic we were able to explore key
variables quickly and cost effecvely before moving to large scale operaon. pH and
control of CO2 levels in the gas mix were idenfied as being crical to further improve
cell yields per cm2. With a disparity of only 38% fewer cells in the Xpansion One in our
semi opmized set up comparable performance to flask culture at large scale is very
likely with minimal changes. This highlights the importance of environmental control
systems and use of small scale mimics when moving cell culture away from tradional
methods to large scale novel bioreactors for commercial producon.