Professional Documents
Culture Documents
www.elsevier.com/locate/ces
Abstract
Pressures on pharmaceutical companies to speed bioprocess development have led to signicant interest in small scale, parallel experimentation.
A particular focus is cell cultivation and the optimisation of protein synthesis because of the number of biological and engineering variables
involved. In this work, we briey review the current understanding of mixing and mass transfer phenomena in shaken bioreactors with a view
to dening criteria for the scale-up of results obtained in shaken microwell systems to conventional laboratory scale. Scale-up approaches are
illustrated for two different cell cultures. The rst involves an automated microscale process (1000 l) for the aerobic fermentation of E. coli
JM107:pQR706 overexpressing transketolase (TK) which is subsequently used for asymmetric carboncarbon bond formation. The kinetics of
both the fermentation and bioconversion stages are rst quantied as a function of fermentation medium composition (LB or LB-glycerol) and
shaking frequency with oxygen transfer rates being identied as rate limiting in certain cases. Successful scale-up of the microwell process (in
terms of maximum cell growth rate, biomass yield and specic TK activity) to a 1.4 l scale mechanically stirred bioreactor is then demonstrated
based on experiments performed at constant kL a values. The second process investigated involved antibody production in suspension cultures
of VPM8 hybridoma cells. Initial results suggest that experiments performed at constant mean energy dissipation rates provide a satisfactory
basis for scale translation from shaken microwells (800 l) to conical asks (100 ml) and are indicative of results obtained in a mechanically
stirred bioreactor (3.5 l). Overall this work provides an initial insight into the engineering characterisation of shaken bioreactors and how key
parameters may be used to dene suitable scale-up criteria for different cell cultures.
2005 Elsevier Ltd. All rights reserved.
Keywords: Fermentation; Cell culture; Biocatalysis; Scale-up; Automated microscale processing
1. Introduction
The pressure on pharmaceutical companies to cut bioprocess
development times and costs has led to signicant interest in
small scale, parallel experimentation. A particular focus is cell
cultivation and the optimisation of protein synthesis because
of the number of biological and engineering variables involved
and their high degree of interaction. One approach has been
to miniaturise (5100 ml) and re-design conventional mechanically stirred laboratory bioreactors (Gill et al., 2005; Kostov
et al., 2001; Lamping et al., 2003; Weuster-Botz, 2005). While
maintaining information content per fermentation the degree
of parallelisation and hence throughput is limited. The second
Corresponding author. Tel.: +44 0207 679 7942; fax: +44 0207 679 0703.
2940
(1)
ds
(2N )df
Ph =
1 + 3 log10
df
4
1/3
1/3 2 2
4 VL
1
1
.
df
(2)
sulphite-oxidation method for a wide range of shaking frequencies (01000 rpm), shaking diameters and ll volumes. They
observed that, for shaking speeds higher than a critical value,
OTRmax increased exponentially. Studies on the well geometry also revealed that OTRmax values in wells with a square
cross-section are approximately twice those measured in round
wells for the same shaking frequency and ll volume. The
latter observation may be explained by the bafing effect of
the corners of the square well. The liquid hydrodynamics was
also monitored using a CCD camera and a critical frequency
(Ncrit ) was identied at which the liquid height and surface
area started to increase, explaining the higher OTR measured
previously. The variation of the liquid height with N was
measured at different shaking diameters and the experimental
results showed good agreement with Eq. (3):
dw
Ncrit =
.
(3)
4VL ds
Eq. (3) was obtained for a single ll volume (VL = 200 l)
in a 96-SRW plate. Due to the different ow hydrodynamics
observed in 96-deep square well plates (96-DSW, Duetz and
Witholt, 2001) a different relationship between Ncrit and the
operating conditions can be expected. Gasliquid mass transfer coefcients (kL a) values for various round-well diameters
have also been reported by Doig et al. (2005), based on both
dynamic gassing out measurements and on the growth kinetics
of an obligate aerobe. The following correlation was proposed
to enable prediction of kL a values:
kL a = c1
DO2
ai Re0.68 Sc0.36 Frx Boy ,
dw
(4)
2941
7
Microwell (LB)
Microwell (LB-glycerol)
10 mm
17 mm
40 mm
4
3
2
1
0
0
8 mm
7 mm
17 mm
96-DSW
96-SRW
24-SRW
previous studies have shown that LB-glycerol cultures obtain higher cell yields (Losen et al., 2004) and higher TK
expression levels (Hobbs, 1994) than other LB modications.
The overnight culture was prepared as previously reported by
Ferreira-Torres et al. (2005) and was used to inoculate the
respective fermentations at 10% of the nal working volume.
2.1.2. Microwell and shake ask bioreactor operation
Microwell fermentations were carried out in polypropylene
96-DSW in a contained environment (Fig. 1). On each plate
48 wells were lled with either LB or LB-glycerol medium
and incubated at 37 C at different agitation speeds. The nal
working volume in each well was VL = 1 ml. The plates were
left unsealed to allow efcient oxygen transfer recognising that
there would be some liquid losses due to evaporation. Control
experiments, using pure water, suggested a maximum evaporation rate of 24 l h1 per well but this had no signicant effect
over the time ranges for which max values were calculated. E.
coli growth kinetics were determined using a sacricial well
approach as described previously (Ferreira-Torres et al., 2005).
Shake asks fermentations were performed in 1 l shake asks
inoculated with the same overnight cultures as used in microwell experiments. The asks were incubated at 300 rpm and
T = 37 C for 812 h. All shake asks fermentations were performed in duplicate with a typical standard deviation in the
calculation of the maximum specic growth rate of 3%.
2.1.3. Stirred bioreactor operation
For laboratory scale fermentations, a 2 l mechanically stirred
bioreactor was used, having an internal diameter Tv = 120 mm
and lled with a total liquid volume VL = 1.4 l (liquid height
HL
Tv ) of either LB or LB-glycerol medium. The vessel was
equipped with four equally spaced vertical bafes of width
B = Tv /10 and thickness tb = Tv /100 and was stirred by two
6
Time [hr]
10
12
2942
60
Concentration [mM]
50
40
HPA
30
L-erythrulose
y = f(x)
20
y = abx
10
0
0
40
20
60
80
100
Time [min]
(a)
50
L-erythrulose concentration [mM]
40
30
20
10
0
0
(b)
20
40
60
Time [min]
80
100
120
2943
Table 1
Summary of linked E. coli JM107:pQR706 fermentation and bioconversion process kinetics in microwell, shake ask and mechanically stirred bioreactors
Bioreactor geometry
(a) LB medium
Microwell (1000 l)
OTRmax (mmol l1 h1 )
N (rpm)
Fermentation
Bioconversion
(h1 )
Biomass
(g l1 )
Initial rate
(U ml1 )
Specic activity
(U g1
DCW )
12.9
28.1
48.7
0.10
0.25
0.37
0.9
1.6
2.0
0.83
1.09
0.64
9491
6740
3631
300
nd
0.44
1.8
0.50
2383
700
1000
52.8
116.7
0.37
0.45
1.7
1.8
0.54
nd
2916
nd
500
750
1000
12.9
28.1
48.7
0.31
0.45
0.55
0.7
4.0
5.8
0.95
3.06
0.71
13 540
7674
1550
300
nd
0.60
6.2
0.83
1445
700
1000
52.8
116.7
0.43
0.39
5.8
6.4
1.06
nd
1820
nd
500
750
1000
N
(rpm)
Rea
500
750
1000
700
1060
1400
300
106 800
700
1000
24 720
35 320
120
730
Ph
PG /VL
(kW m3 )
kL a
(s1 )
nd
nd
nd
nd
nd
nd
0.018
0.041
0.072
2.04
na
na
2.6b
3.8c
11.6c
nd
0.077c
0.17c
8.2
0.037d
0.067
nd
0.017
120
18 400
2.68
0.043b
150
17 000
na
0.00364
calculations. The Reynolds number in shaken systems is commonly calculated using the microwell or maximum shake ask
diameter (df = 127 mm) as the characteristic length. It is
noteworthy, however, that mixing characteristics in microwells have yet to be well established and very few parameters
characterising ow dynamics in such geometries have been
suggested to date. For example, while Re is calculated and reported in Table 2(a), no study has been carried out relating Re
to ow regimes in microwells. Calculation of the Ph number
2944
7
N = 500 rpm
PG
= 0.155N 3.141 Vs0.365 + gv s ,
VL
PG 0.699 0.58
vs .
kL a = 0.0108
VL
6
N = 750 rpm
N = 1000 rpm
5
4
0
0
10
12
Time [hr]
(a)
7
N = 500 rpm
N = 750 rpm
(7)
(8)
5
N = 1000 rpm
4
3
2
1
0
0
(b)
2945
10
12
Time [hr]
OTR = KL a(CL CL ),
(9)
120
100
60
DOT [%]
40
0
4
Time [hr]
(a)
0
0
100
60
40
20
10
12
50
80
6
Time [hr]
(a)
120
20
80
40
30
20
10
0
0
(b)
DOT [%]
2946
6
Time [hr]
10
12
0
0
(b)
20
40
60
80
Time [min]
100
120
140
stirred (N = 700 rpm) bioreactors, fermentations were performed at matched kL a values as described previously. For the
shaken ask (N =300 rpm) no estimate of kL a is available however it is noted from Table 2(a) that the energy dissipation rate
is similar. Apart from some differences in the duration of the lag
phase in microwell cultures, the three biomass curves (Fig. 6(a))
exhibited similar values of max and Xnal as summarised in
Table 1(b). Likewise, the rates of L-erythrulose formation when
the cells from the respective fermentations are used for the subsequent HPA/GA bioconversion are also similar as conrmed
by the specic activity values in Table 1(b). The corresponding
fermentations and bioconversion data, when the process was
repeated with LB medium, conrm the good agreement observed between microwell and mechanically stirred bioreactor
experiments under matched kL a conditions.
3.5. Scale-translation of hybridoma cultures based on power
consumption
The oxygen uptake rate of mammalian cells has been shown
to be in the range 2.101012 g cell1 h1 whereas for bacteria
1.4E+06
Shake flask (N = 120 rpm)
1.2E+06
1.0E+06
8.0E+05
6.0E+05
4.0E+05
2.0E+05
0.0E+00
20
(a)
40
60
Time [hr]
80
100
80
100
45
Microwell (N = 120 rpm)
40
35
30
lgG1 [mg I-1]
2947
25
20
15
10
5
0
20
40
60
Time [hr]
(b)
4. Concluding remarks
In this study, the potential for microscale process sequences
to quantitatively distinguish the performance of cultures grown
in different media was demonstrated. In addition, the fermentation and bioconversion kinetics of the whole cell biocatalyst
E. coli JM107:pQR706 were determined using different shaking frequencies. The experiments showed that oxygen transfer is a critical parameter in microwell cultures and therefore
a critical parameter for scale-up predictions to mechanically
stirred bioreactors. By dening experimental conditions that
gave matched kL a values, it was shown that microscale results could be predictive of those obtained in shaken asks and
conventional bioreactor congurations at the laboratory scale.
Similarly, constant power consumption has been identied for
the initial scale comparison of mammalian cell cultures. Our
current work is applying these methods to the analysis of larger
libraries of evolved biocatalysts and mammalian cells.
Notation
a
ai
b
2948
B
Bo
c1
C
CL
CL
CO2
df
ds
dw
D
DO2
Fr
g
HL
kL
KL
N
OTR
P
PG
Po
ri
Re
Sc
t
tb
Tv
T
vs
VL
W
x
X
y
yo
bafe width, m
bond number (=gd 2w /W ), dimensionless
constant in Eq. (4), m1
off-bottom clearance, m; constant in Eq. (1), dimensionless
oxygen concentration in the liquid phase, mol l1
saturation concentration of oxygen in the liquid
phase, mol l1
oxygen concentration uptake by the cells, mol l1
shaken ask diameter, m
shaken diameter, m
shaken microwell diameter, m
impeller diameter, m
oxygen diffusion coefcient, m2 s1
froude number (=ds (2N )2 /2g), dimensionless
gravitational acceleration, m s2
liquid height, m
liquid-lm mass transfer coefcient, m s1
overall mass transfer coefcient, m s1
impeller rotational speed, shaking frequency, s1
oxygen transfer rate, mol l1 s1
power consumption, W
gassed power consumption, W
modied power number, dimensionless
initial rate of product formation, mol l1 s1
Reynolds number (=N d 2 /), dimensionless
Schmidt number (=/DO2 ), dimensionless
time, s
bafe thickness, m
vessel internal diameter, m
temperature, C
supercial gas velocity, m s1
liquid volume, m3
wetting tension, N m1
constant in Eq. (4), dimensionless
biomass concentration, mol l1
constant in Eq. (4), dimensionless
constant in Eq. (6), mol l1
Greek letters
C
max
Acknowledgements
The authors would like to thank the UK Engineering and
Physical Sciences Research Council (EPSRC) for support of
the multidisciplinary Biocatalysis Integrated with Chemistry
and Engineering (BiCE) programme (GR/S62505/01). Financial support from the 12 industrial partners supporting the BiCE
programme is also acknowledged. The authors would also like
to thank the UK Joint Infrastructure Fund (JIF), the Science
Research Investment Fund (SRIF) and the Gatsby Charitable
References
Barrett, T.A., Zhang, H., Levy, M.S., Lye, G.J., 2005. Engineering
characterisation of hybridoma cells in shaken microwell plates: application
to high-throughput process development. Biotechnology Bioengineering,
submitted for publication.
Bchs, J., Maier, U., Milbradt, C., Zoels, B., 2000. Power consumption
in shaking asks on rotary shaking machines: I. Power consumption
measurement in unbafed asks at low liquid viscosity. Biotechnology
Bioengineering 68, 589593.
Bchs, J., Lotter, S., Milbradt, C., 2001. Out-of-phase conditions, a hitherto
unknown phenomenon in shaking bioreactors. Biochemical Engineering
Journal 7, 135141.
Chisti, Y., 1993. Animal cell culture in stirred bioreactors: observations on
scale-up. Bioprocess Engineering 9, 191196.
Dinnis, D.M., James, D.C., 2005. Engineering mammalian cell factories
for improved recombinant monoclonal antibody production: lessons from
nature? Biotechnology Bioengineering 91, 180189.
Doig, S.D., Pickering, S.C.R., Lye, G.J., Woodley, J.M., 2002. The use
of microscale processing technologies for quantication of biocatalytic
Baeyer-Villiger oxidation kinetics. Biotechnology Bioengineering 80,
4249.
Doig, S.D., Pickering, S.C.R., Lye, G.J., Baganz, F., 2005. Modelling surface
aeration rates in shaken microtitre plates using dimensionless groups.
Chemical Engineering Science 60, 27412750.
Duetz, W.A., Witholt, B., 2001. Effectiveness of orbital shaking for the
aeration of suspended cultures in square-deepwell microtiter plates.
Biochemical Engineering Journal 7, 113115.
Duetz, W.A., Redi, L., Hermann, R., OConnor, K., Bchs, J., Witholt,
B., 2000. Methods for intense aeration, growth, storage and replication
of bacterial strains in microtiter plates. Applied and Environmental
Microbiology 66, 26412646.
Elmahdi, I., Baganz, F., Dixon, K., Harrop, T., Sugden, D., Lye, G.J., 2003.
pH control in microwell fermentations of S. erythraea CA340: inuence
on biomass growth kinetics and erythromycin biosynthesis. Biochemical
Engineering Journal 16, 299310.
Ferreira-Torres, C., Micheletti, M., Lye, G.J., 2005. Microscale process
evaluation of recombinant biocatalyst libraries: application to BayerVilliger monooxygenase catalysed lactone synthesis. Biosystems
Bioengineering 28, 8393.
Gill, N.K., Appleton, M., Baganz, F., Peacock, M., Lye, G.J., 2005. Design and
instrumentation of a novel miniature bioreactor system for high throughput
fermentation optimisation. Biochemical Engineering Journal, submitted for
publication.
Girard, P., Jordan, M., Tsao, M., Wurm, F.M., 2001. Small-scale
bioreactor system for process development and optimization. Biochemical
Engineering Journal 7, 117119.
Hashimura, Y., Li, F., Lee, B., 2005. Cell culture scale-down model
development through impeller agitation rate optimization. Biochemical
Engineering XIV. Harrison Hot Spring, BC, Canada, 1014 July 2005.
Hermann, R., Lehmann, M., Bchs, J., 2003. Characterization of
gasliquid mass transfer phenomena in microtiter plates. Biotechnology
Bioengineering 81, 178186.
Hobbs, G.R., 1994. The product and use of transketolase as a catalyst for
carboncarbon bond formation. Ph.D. Thesis, University College London,
London, UK.
Hobbs, G.R., Mitra, R.K., Chauhan, R.P., Woodley, J.M., Lilly, M.D., 1996.
Enzyme-catalysed carboncarbon bond formation: large-scale production
of Escherichia coli transketolase. Journal of Biotechnology 45, 173179.
2949