Biotechnol. J. 2011, 6, 959–967 DOI 10.1002/biot.201000407 www.biotechnology-journal.

com

Research Article

Simulation of dissolved CO2 gradients in a scale-down system:

A metabolic and transcriptional study of recombinant
Escherichia coli

Antonino Baez1, Noemí Flores2, Francisco Bolívar2 and Octavio T. Ramírez1

1 Departamento de Medicina Molecular y Bioprocesos, Instituto de Biotecnología, Universidad Nacional Autónoma de México,
Cuernavaca, Morelos, México
2 Departamento de Ingeniería Celular y Biocatálisis, Instituto de Biotecnología, Universidad Nacional Autónoma de México,

Cuernavaca, Morelos, México

Deficient mixing in industrial-scale bioreactors is an important concern as it results in a hetero-

Received 5 March 2011
geneous environment that may affect microbial cell physiology. Dissolved carbon dioxide (dCO2) Revised 20 May 2011
fluctuations, which can occur in large-scale bioreactors, were simulated for the first time and their Accepted 10 June 2011
effects were evaluated on Escherichia coli expressing recombinant green fluorescent protein (GFP).
The dCO2 gradients were simulated by continuously circulating the medium between two vessels
of a scale-down system to mimic mean circulation times (tc) of 50, 170, and 375 s. Specific growth
rate (μ) decreased by 11% and acetate concentration increased by 23% at the highest tc compared
to reference cultures. An increase in the time needed for attaining maximum GFP concentration
was also observed. The effect of dCO2 fluctuations on the transcriptional levels of genes involved
in the glutamate decarboxylase (gadA and gadC) and α-ketoglutarate dehydrogenase (sucA and
sucB) were analyzed by quantitative real time PCR. Such genes are known to be highly over- or
underexpressed at elevated constant dCO2. Expression levels of gadA and gadC increased up to
60% and 72%, and sucA decreased 1.8-fold in the culture performed at the highest tc. Only a mi-
nor decrease of sucB expression was observed at tc of 170 and 375 s. Although exposure to con-
tinuous high dCO2 can affect culture performance, in this work it was shown that E. coli is able to
rescue its metabolism in very short times when cells are intermittently returned to low dCO2 con-
ditions after being exposed to high dCO2.

Keywords: Acetate accumulation · Circulation time · Dissolved carbon dioxide · Scale-down · Transcription

1 Introduction

Accumulation of dissolved carbon dioxide (dCO2)

Correspondence: Dr. Octavio T. Ramírez, Departamento de Medicina
Molecular y Bioprocesos, Instituto de Biotecnología, Universidad
Nacional Autónoma de México, Ave. Universidad 2001 Col. Chamilpa,
Cuernavaca, Morelos 62210, México
E-mail: tonatiuh@ibt.unam.mx
Abbreviations: Bhigh dCO2, bioreactor maintained at high dCO2 concentration
(high dCO2 compartment); Blow dCO2, bioreactor maintained at low dCO2
concentration (low dCO2 compartment); dCO2, dissolved carbon dioxide
(mbar); GFP, green fluorescent protein; μ, specific growth rate (h–1); qRT-
PCR, quantitative real-time polymerase chain reaction; tc, mean circulation
time (s); vvm, gas volume/medium volume minute
in culture media has been recognized as an impor-
tant problem in large-scale animal cell culture [1,
2], as well as in high-pressure fermentations [3, 4].
High levels of dCO2 can also occur during high cell
density E. coli cultures performed at large scale,
particularly in poorly ventilated regions or bioreac-
tor sections exposed to a high total pressure, such
as in the bottom of the vessel [5]. Carbon dioxide is
a toxic byproduct of aerobic fermentations that can
be detrimental to growth, metabolism, and product
formation [5–12]. Elevated CO2 levels are known to

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 959

Biotechnology
Journal
affect both recombinant protein production [5] and
the specific growth rate (μ), and to increase acetate
accumulation in E. coli cultures [5, 7, 9]. Recently,
Baez et al. [5] developed a culture system for con-
trolling dCO2 at constant and predetermined levels,
and showed that recombinant E. coli can respond at
the transcriptional level upon exposure to increas-
ing dCO2. In particular, selected genes of the acid
resistance systems (gadA and gadC, which code for
glutamate decarboxylase and glutamate/γ-amino-
butyric acid antiporters, respectively) were up-reg-
ulated at constant dCO2 concentrations as low as
70 mbar, whereas TCA cycle genes involved in de-
carboxylation reactions (sucA and sucB, which code
for the components E1(0) and E2(0) of 2-α-keto-
glutarate dehydrogenase complex, the rate-deter-
mining enzyme of TCA cycle [13]) were down-reg-
ulated [5].
Limitations in the design and operation of
large-scale bioreactors can lead to deficient mix-
ing, resulting in heterogeneous environmental
conditions that may affect cell metabolism and
physiology [14–16].The effects of environment het-
erogeneities that can occur in large-scale bioreac-
tors have been studied through the scale-down
methodology, which allows the estimation of the
performance of large-scale fermenters through ex-
perimentation at the small scale [17]. Using such an
approach, the overall effects of gradients in dis-
solved oxygen tension, substrate concentration,
and pH have been thoroughly studied in the past
for a variety of microorganisms and cells [15,
18–25]. Although no data have been reported to
date on the presence of dCO2 concentration gradi-
ents, it is reasonable to expect them in large-scale
bioreactors as dCO2 can accumulate to high values
and spatial and temporal variations in the intensi-
ties of mixing and mass transfer are prevalent in
such systems [16, 26, 27]. However, to our knowl-
edge no scale-down studies exist where the effects
of dCO2 gradients have been determined for micro-
bial fermentations. Consequently, the magnitude
and effects of such putative dCO2 gradients on the
cellular physiology is unknown.
Mixing and circulation times are among the
most useful parameters for characterizing the de-
gree of homogeneity of the liquid phase in fer-
menters [28]. Such parameters are commonly used
to compare the characteristic times of physical
phenomena with those of microbial physiology to
predict whether concentration gradients (e.g., sub-
strate, oxygen) and even limitations occur [17, 18].
To simulate environmental gradients through tra-
ditional scale-down studies, the variable of interest
is commonly fluctuated at a period equivalent to
the mean circulating time (tc) found under typical
960
Biotechnol. J. 2011, 6, 959–967
full-scale operating conditions. In turn, the ampli-
tude of the environmental fluctuations (experi-
mentally determined through mixing time studies)
encountered by the microbial cells can also be eas-
ily established for the particular scale-down sys-
tem selected [16].The range of tc prevalent in large-
scale stirred tank reactors (STR) of microbial fer-
mentations is generally between 3 to 60 s [15, 25,
28] and can be as high as 250 s or higher in other
reactor configurations (for instance, in a 10-m3 air-
lift bioreactor, 150-m3 bubble column, or in 50-m3
fluidized-bed bioreactors [16, 17, 29]). In addition
to the tc, scale-down studies performed in two-
compartment systems must also define the volume
ratio between the two compartments, each repre-
senting a characteristic volume of two extreme re-
gions present in large-scale bioreactors. For in-
stance, to simulate substrate gradients that occur at
the feeding zone of large-scale fed-batch cultures,
a combination of plug-flow reactor (PFR) and STR,
with a volume ratio of 0.5 or 1.1 to 10, is commonly
employed [20]. For the case of dissolved oxygen
gradients, the ratio of the volume employed to sim-
ulate the well-to-poorly aerated zones is between
0.25 and 0.50 [18, 21, 29].
In the present work, a two compartment scale-
down system for simulating dCO2 gradients was
designed, constructed, and employed for charac-
terizing the effects of fluctuating dCO2 concentra-
tion on recombinant E. coli cultures expressing re-
combinant GFP. The effect on macroscopic kinetic
and stoichiometric variables, and the transcription-
al response of four selected genes (gadA, gadC,
sucA, and sucB) were also determined. This study
represents the first effort to address the effects of
dCO2 gradients on microbial fermentations.
2 Materials and methods
2.1 Bacterial strain, inoculum preparation,
and culture medium
Escherichia coli strain W3110 (ATCC 27325) carry-
ing the plasmid pV21 was used to express the re-
combinant SupergloTM (Qbiogene) green fluores-
cent protein (GFP). Plasmid pV21 contained the
GFP gene (gfp) placed under control of the lacZ
promoter and a spectinomycin resistance gene [5,
15]. A working cell bank was prepared from a con-
struction similar to that described by Baez et al. [5]
and maintained in 2-mL cryo-preservation tubes at
–70°C. For each experiment, inoculum was pre-
pared by transferring the total content of a defrost-
ed working cell bank vial to a shake flask contain-
ing a volume of culture medium equal to 10% of the
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 959–967
total fermentation volume. Shake flasks were incu-
bated at 37°C and 280 rpm for approximately 21 h.
Culture medium prepared for inoculum propaga-
tion and bioreactor cultures contained: 5.0 g/L glu-
cose, 2.0 g/L Na2SO4, 2.7 g/L (NH4)2SO4, 0.5 g/L
NH4Cl, 14.6 g/L K2HPO4, 4 g/L NaH2PO4 · H2O,
1.0 g/L sodium citrate, 0.01 g/L thiamine, 0.1 g/L
spectinomycin, and 0.5 g/L MgSO4 · 7H2O in dis-
tilled water. Solutions of glucose, trace metals, thi-
amine, spectinomycin, and MgSO4 · 7H2O were sep-
arately sterilized, cooled down, and added asepti-
cally to the medium, as described by Baez et al. [5].
Isopropyl-β-D- thio-galactoside (IPTG) was added
as a chemical inducer of the gfp gene during inoc-
ulation time at a final concentration of 0.5 mM.
2.2 Two-compartment scale-down bioreactor
system
dCO2 gradients were mimicked in a two-compart-
ment scale-down system based on a design previ-
ously described by Sandoval-Basurto et al. [29].
Some modifications were implemented to monitor
and control the dCO2. Briefly, the system consisted
of two interconnected 1.5-L baffled stirred-tank
bioreactors (Virtis, Gardiner NY) equipped with
two six-blade Rushton turbines. Culture medium
was circulated between the two vessels through sil-
icone tubing (inside diameter 12.7 mm) using two
peristaltic pumps. The mean circulation time (tc)
was established as the total culture volume divided
by the circulation flow rate. Working volumes were
0.8 L for the low dCO2 vessel and 0.4 L for the high
dCO2 vessel when using tc of 50 and 170 s. For tc of
375 s, volumes of 1.0 and 1.5 L were used in the low
and high dCO2 vessels, respectively. Cultures were
performed in batch mode and both vessels were
controlled at the same constant values of dissolved
oxygen tension (DOT) (20%), temperature (37°C)
and pH (7.0, by automatic addition of a 2 M NaOH
solution) through the BioFlo 110 control modules
(New Brunswick Scientific, Co., Inc., New Bruns-
wick, NJ). At the same time, dCO2 was continuous-
ly measured in situ in each vessel with potentio-
metric probes (B&C Electronics, Milano, Italy) and
controlled by continuous step changes in the inlet
gas composition. The dCO2 gradients were initiated
3.5 h after inoculation by maintaining a low dCO2
concentration in one vessel (Blow dCO2) and a high
dCO2 concentration in the other vessel (Bhigh dCO2).
In this form, from 3.5 h to the end of the culture, the
cells were continuously exposed to dCO2 fluctua-
tions when circulating between both vessels. The
dCO2 levels were equal in both compartments and
close to zero from inoculation to 3.5 h of culture. For
comparison, an additional set of cultures at a tc of
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
375 s with generation of dCO2 gradients from the
moment of inoculation were also performed by du-
plicate. For all tc tested, the dCO2 in the Blow dCO2
compartment was kept around 33 mbar by strip-
ping dCO2 with a high flow rate [1–3 gas volume/
medium volume minute (vvm)] of an air/nitrogen
mixture. In contrast, high dCO2 concentrations (55,
95, and 213 mbar) were achieved in Bhigh dCO2 for tc
of 50, 170, and 375 s, respectively, by sparging a
CO2/air/nitrogen mixture. The high dCO2 levels,
which usually accumulate under high cell density
cultures, were modeled here as low biomass con-
centration batch cultures with exogenous CO2 sup-
plementation. With such a model, it is possible to
dissect the effect of dCO2 from other changes that
can occur at higher cell densities, including accu-
mulation of toxic by-product, such as acetate, and
quorum sensing phenomena.
2.3 Analytical methods
Biomass concentration was determined from OD
readings at 600 nm measured in a Beckman DU 610
spectrophotometer (Palo Alto, CA) and converted to
dry cell weight using calibration plots of samples
dried to constant weight in an oven maintained at
80°C. Glucose and organic acids were quantified
from culture supernatants by high-pressure liquid
chromatography using an Aminex HPX-87H col-
umn (Bio-Rad, Hercules, CA) maintained at 50°C.
A 5 mM H2SO4 solution at a constant flow of
0.5 mL/min was used as the mobile phase. Organic
acids were detected at 210 nm in a photodiode ar-
ray detector and glucose by refractive index. Re-
combinant GFP production was followed by fluo-
rescence readings in a Perkin Elmer LS55 Lumi-
nescence spectrometer (Wellesley, MA) and GFP
concentration was determined by fluorescent read-
ings using a standard curve as described by Baez et
al. [5].
2.4 Samples for RNA extraction, cDNA synthesis
and qRT-PCR
During the first 3.5 h of culture the cells circulated
in the scale-down system but without CO2 addition
(therefore without dCO2 gradients). The first sam-
ple for RNA extraction was taken at 3.5 h of culture;
afterwards, dCO2 gradients were initiated (Fig. 1A).
At approximately 5 h of culture, samples from the
high and low dCO2 compartments were simultane-
ously taken for RNA purification. The transcription
levels of samples taken at 5 h were compared with
those from samples collected at 3.5 h (reference
condition). Samples for RNA extractions were im-
mediately poured on an ice-cooled tube containing
961
Biotechnology Biotechnol. J. 2011, 6, 959–967
Journal

A 250 mentas, Hanover, MD). This cDNA was used as

200 template for quantitative real-time RT-PCR (qRT-
dCO2, mbar

PCR) assays performed in an ABI Prism 7000 Se-

150 B high dCO2 quence Detection system (Perkin-Elmer/Applied
100 B low dCO2 Biosystems, Foster City, CA) using the SYBR Green
50
PCR Master Mix (Perkin-Elmer/Applied Biosys-
tems, Foster City, CA). Amplification conditions
0 were described by Flores et al. [30], with the fol-
0 2 4 6 8 10 12 lowing slight modifications: the total volume of re-
B 60 7.5 action mixture was 12 μL and 0.42 μM was the final
50 concentrations for both primer and target cDNA
7 concentrations. Mathematical treatment of qRT-
DOT, %

40
pH
PCR was performed using the 2–ΔΔCT method as de-
30 scribed by Livak and Schmittgen [31]. Amount of
6.5
20 target cDNA was normalized to an endogenous ref-
erence (in this case ihfB gene, which did not change
10 6
in the range of dCO2 tested [30]) and relative to a
0 2 4 6 8 10 12
C 2.5 6 calibrator (sample collected at 3.5 h, with an as-
signed transcription level value of one) through
Glucose conc., g/L

5
Biomass conc., g/L

2
the 2–ΔΔCT equation. The specific oligonucleotide
4
1.5 primers used for gene amplification have been de-
3 scribed previously [5]. Data shown in Fig. 3 are
1
2 plotted as base 2 logarithms.
0.5 1
0 0
0 2 4 6 8 10 12 3 Results and discussion
D 1.25 0.4
3.1 Control experiments
Acetate conc., g/L

GFP conc., g/L

1 0.3
0.75
0.5
0.25
0 0
0 2 4 6 8 10
Time (h)
Figure 1. Typical dissolved oxygen tension (DOT), pH and dCO2 control
profiles of a culture maintained at tc of 375 s in the two-compartment
scale-down system. (A) Dissolved CO2 profiles in the low and high dCO2
compartments. (B) DOT and pH controls in the low and high dCO2 com-
partments. (C) Biomass and glucose concentration; the same values were
obtained in either compartment. (D) Acetate and GFP concentration; the
same values were obtained in either compartment. For clarity, only one
set of data for cultures performed at tc of 375 s has been plotted. The axis
corresponding to each variable in panels B, C and D is indicated by hori-
zontal arrows.
RNAlater® (Ambion, Austin,TX).Total RNA extrac-
tion, performed using hot-phenol equilibrated with
water, and cDNA synthesis were performed as de-
scribed by Flores et al. [30]. Total RNA was treated
with DNase kit (DNA-freeTM, Ambion, Foster City,
CA) and its concentration was determined using
absorption of light at 260 and 280 nm. Agarose (2%)
gels were used to confirm the integrity of isolated
RNA. The cDNA was synthesized using Rever-
tAidTM H First strand cDNA Synthesis Kit (Fer-
To determine any possible hydrodynamic effects
0.2
on the cells due to the continuous pumping and cir-
0.1 culation between the two vessels of the scale down
system, replicates of two types of control experi-
ments were performed. First, cultures were per-
12
formed without medium circulation in a single
bioreactor and compared with cultures with medi-
um circulation (tc of 50 s) performed in the two-
compartment scale-down system but without gen-
erating a dCO2 gradient. In both types of cultures,
the CO2 generated by cells was continuously meas-
ured but not controlled. No differences on the ki-
netic parameters were observed for these types of
control cultures (Table 1, experiments A). This re-
sult indicated that hydrodynamic stress generated
in the scale-down system without dCO2 control was
not detrimental to E. coli cells and confirms previ-
ous observations by Hewitt et al. [32] and Cham-
sartra et al. [33]. However, it has been reported that
elevated CO2 levels can modify the membrane
composition and its permeability [34, 35]. There-
fore, to test if high dCO2 concentrations could in-
crease the susceptibility of cells to shear stress, a
second set of controls was performed at 300 mbar
of dCO2 with and without medium circulation
(Table 1, experiments B). These results indicated
that the culture performance with medium circula-

962 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 959–967 www.biotechnology-journal.com

Table 1. Kinetic and stoichiometric parameters of control cultures performed with and without medium circulationa)

Culture Specific growth rate Yx/s Maximum acetate YGFP/X YGFP/S Maximum GFP
(h–1) (gbiomass/ggluc) conc. (g/L) (gproduct/g cell) (gproduct/ggluc) conc. (g/L)
Without circulationb) 0.40 ± 0.037 0.39 ± 0.03 1.07 ± 0.100 0.15 ± 0.015 0.06 ± 0.006 0.32±0.04
With circulationb) 0.38 ± 0.011 0.39 ± 0.014 0.86 ± 0.030 0.16 ± 0.004 0.07 ± 0.004 0.35±0.03
Without circulationc) 0.23 ± 0.001 0.247 ± 0.003 2.04 ± 0.001 0.14 ± 0.002 0.03 ± 0.001 0.19±0.004
With circulationc) 0.27 ± 0.010 0.244 ± 0.002 1.73 ± 0.034 0.14 ± 0.018 0.03 ± 0.004 0.20±0.01
a) The ± represents the error between duplicates.
b) Culture performed without CO2 addition.
c) Culture performed at 300 mbar of dCO2.

tion was slightly better than without circulation of those typically found in large-scale bioreactors,
[maximum specific growth rate (µ) increased by and was sufficiently large to facilitate the experi-
18% and acetate accumulation decreased by 15%]. mental handling. Under these conditions, the max-
Accordingly, it can be concluded that the shear imum dCO2 gradient generated (maximum differ-
stresses generated during medium circulation ence in dCO2 between the high and low dCO2 ves-
through the tubing loops and pump heads of the sels) was only of 22 mbar. Such a dCO2 gradient did
scale-down system do not negatively affect the be- not affect culture performance in comparison to
havior of E. coli cultures even in presence of ele- the reference culture, as inferred from the main
vated dCO2 concentration. Thus, any observed ef- stoichiometric and kinetic parameters (Fig. 2). As
fect can be related to the dCO2 gradient. no deleterious effect of such dCO2 gradient was ob-
served, it was decided to explore the effect of larg-
3.2 Simulation of dCO2 gradients er values of the dCO2 gradient. Increasing the max-
imum agitation (600 to 1100 rpm) and aeration (1 to
The design of a two-compartment scale-down sys- 3 vvm) rates did not increase the dCO2 gradient
tem to generate dCO2 gradients represented an en- generated under a tc of 50 s. It was possible to at-
gineering challenge as a very high CO2 desorption tain larger dCO2 gradients at the higher circulation
rate was needed in Blow dCO2 to rapidly eliminate the times tested (170 and 375 s). Thus, a gradient of
dissolved carbon dioxide from the incoming broth 62 mbar (33 and 95 mbar were kept in the low and
of Bhigh dCO2. In contrast, a high CO2 absorption rate high dCO2 compartments) was achieved at tc 170 s.
was needed in Bhigh dCO2 for rapidly increasing the For a tc of 375 s, a gradient of 180 mbar (constant 33
dCO2 from the incoming broth of Blow dCO2.The low- and 213 mbar were maintained in the low and high
est tc selected in this study (50 s) is within the range dCO2 compartments) was accomplished (Fig. 1A). It

E
Maximum biomass conc., g/L

2.5 A 0.4 C 0.45

Maximum GFP conc., g/L

0.4
2 0.35
0.3
0.3
Y x/s, g/g

1.5
0.25
0.2
1 0.2
0.15
0.1 0.1
0.5
0.05 Figure 2. Effects of dCO2 gradients on the main kinetic and
stoichiometric parameters of E. coli cultures. (A) Maximum
0 0 0
Maximum acetate conc., g/L

0.4 B D F biomass concentration, (B) maximum specific growth rate (μ)

.

.
Specific growth rate, h-1

1 calculated during exponential growth phase, (C) maximum

0.15
GFP concentration, (D) maximum GFP yield on biomass
Y GFP/x, g/g

0.3 0.8 (YGFP/X), (E) maximum biomass yield on glucose (Yx/s), (F)
0.1 0.6 maximum acetate concentration. Error bars represent the dif-
0.2 ference between duplicates. YGFP/X and Yx/s values were calcu-
0.4 lated at the time of maximum GFP and maximum biomass
0.1 0.05 concentration, respectively. The dCO2 was fluctuated between
0.2 33 and 55 mbar, 33 and 95 mbar, and 33 and 213 mbar in cul-
tures performed at tc of 50, 170, and 375 s, respectively, where-
0 0 0
as for the reference culture it progressively accumulated until
.

Reference culture tc 50 s tc 170 s tc 375 s maximum concentration of 12 mbar was reached.

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 963

Biotechnology
Journal
should be noticed that even at this high dCO2 gra-
dient, it was possible to control the pH in the two
vessels at a constant and similar value, as the max-
imum difference was no larger than 0.1 pH units
(Fig. 1B). Typical exponential growth profile and
acetate production were observed, while glucose
was present in the medium (Figs. 1C and D). After
glucose depletion (around 8 h), acetate was com-
pletely re-assimilated. GFP was produced from the
beginning of the culture, closely following the bio-
mass concentration profile (Fig. 1D). GFP produc-
tion ceased 1 h after growth stopped.
3.3 Performance of recombinant E. coli cultures
under dCO2 gradients
A summary of the most relevant stoichiometric and
kinetic parameters, as determined from the kinetic
profiles of E. coli cultures subjected to the various
dCO2 gradients (Figs. 1C and D), is shown in Fig. 2.
Maximum biomass concentration remained con-
stant and similar to the reference culture
(2.18 ± 0.025 g/g) for all tc evaluated (Fig. 2A). In
contrast, as tc increased, and thus the dCO2 gradi-
ent was larger, μ decreased. Whereas the reference
culture had a µ of 0.382 ± 0.011 h–1 it decreased by
9% and 11% for the cultures at tc of 170 and 375 s,
respectively (Fig. 2B). Compared to the reference
culture, maximum GFP concentrations only de-
creased slightly for cultures maintained at tc of 170
and 375 s (5% and 4%, respectively) (Fig. 2C),
whereas the maximum GFP yield on biomass
(YGFP/X) remained relatively constant for all condi-
tions tested (Fig. 2D). The maximum biomass yield
on glucose (YX/S) was not affected by any of the
dCO2 gradients tested (Fig. 2E), as the values found
in all cultures were similar to reference the cul-
tures (0.38 ± 0.014 g/g). Acetate and formate were
the main mixed-acid fermentation metabolites
found in the cultures, although formate only accu-
mulated to concentrations below 0.09 g/L (data not
shown). As the dCO2 gradients increased, acetate
accumulation also increased, reaching maximum
values of 0.857 ± 0.032, 0.982 ± 0.001, and 1.06 ±
0.01 g/L for the reference culture, and the cultures
at a tc of 170 s and tc of 375 s, respectively (Fig. 2F).
Table 2. Comparison of cultures exposed to dCO2 gradients (33–213 mbar) from the time of inoculation and after 3.5 h of inoculation at a tc of 375 sa)
dCO2 gradients at inoculation time
Specific growth rate (h–1) 0.364 ± 0.005
Max. GFP conc. (g/L) 0.347 ± 0.004
YGFP/X (gGFP/gcell)b) 0.174 ± 0.015
a) The ± represents the error between duplicates.
b) Maximum GFP yield on biomass calculated at the time of maximum GFP concentration.
964
Biotechnol. J. 2011, 6, 959–967
Even when carbon from glucose was wasted into
acetate production as tc increased, recombinant
GFP yield on glucose (YGFP/S) was not affected
(data no shown). It should be noted that no changes
in GFP fluorescence at pH values from 5 to 9 were
observed (data not shown) and that previous re-
ports for other GFP variants have shown charac-
teristic times for chromophore maturation above
the 1600 s range [36]. Accordingly, changes in GFP
fluorescence should not be expected as a response
to dCO2 fluctuations.
A volume ratio of 0.667 was set for the low to
high dCO2 compartments of the cultures performed
at a tc of 375 s. Accordingly, bacterial cells intermit-
tently spent on average 225 s under a dCO2 of
213 mbar and 150 s under a dCO2 of 33 mbar. Un-
der such condition µ only decreased by 11% with
respect to the reference cultures. In comparison,
we have previously observed a much higher detri-
mental effect of dCO2 if it is maintained at a con-
stant level throughout the culture [5]. For instance,
at a constant dCO2 of 150 mbar (i.e., 30% lower than
that controlled in Bhigh dCO2 at a tc of 375 s), µ de-
creased as much as 33% for the same E. coli strain
harboring the same plasmid as in the present study
[5]. In the work done by Baez et al. [5], cells were
exposed to constant high dCO2 concentration from
the inoculation time (lag growth phase), whereas
dCO2 gradients were initiated 3.5 h after inocula-
tion (early exponential growth phase) for the data
shown in Figs. 1–3. To rule out any possible effects
due to the period that cells were exposed to dCO2
gradients, cultures were performed in the scaled
down system initiating the dCO2 gradients at the
time of inoculation. However, as shown in Table 2,
no differences were observed between cultures re-
gardless of whether dCO2 gradients were initiated
at the time of inoculation or 3.5 h later. Therefore,
the results of the present study show that transient
exposure to high dCO2 concentrations is less detri-
mental for E. coli cells than constant exposure to
high dCO2 concentrations. Accordingly, the time
spent in the low dCO2 compartment was enough for
the cells to recover, at least partially, their home-
ostasis and return to a more efficient metabolism.
This can be possible as the characteristic times for
dCO2 gradients 3.5 h after inoculation time
0.340 ± 0.008
0.342 ± 0.001
0.163 ± 0.014
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 959–967 www.biotechnology-journal.com

A B
Transcription levels, Log2 fold change

Transcription levels, Log2 fold change

1 gadC gene 1 gadA gene

0.5 0.5

0 0

-0.5 tc 50s tc 170s tc 375s -0.5 tc 50s tc 170s tc 375s

Figure 3. Transcriptional response of E. coli cultures to dCO2

Transcription levels, Log2 fold change
Transcription levels, Log2 fold change

sucA gene sucB gene gradients performed in a scale-down system. Expression levels
C D
0.5 0.5 of mRNA from high (shaded bar) and low (white bar) dCO2
Low dCO2
compartments were normalized relative to those obtained from
High dCO2 the reference condition (mRNA sample taken just before the
0 0 onset of dCO2 gradients) and are reported as the Log2 values.
Transcription levels of the reference condition are denoted as 0
in the graphs. Error bars indicate the difference between dupli-
-0.5 -0.5 cates, except for tc of 50 s, where only single data were ob-
tained. The low dCO2 compartment was maintained at a dCO2
of 33 mbar for all tc tested, whereas the high dCO2 compart-
-1 ment was kept at 55, 95, and 213 mbar for tc of 50, 170, and
-1
tc 50s tc 170s tc 375s tc 50s tc 170s tc 375s 375 s, respectively.

enzymatic allosteric control are shorter than the and high dCO2 compartments, respectively. Such
residence times found in the low dCO2 compart- increments in the expression of gadA and gadC, al-
ment of our scale-down system [17]. though relatively modest, are in agreement with
the hypothesis that a high dCO2 concentration pro-
3.4 Transcriptional profile of selected genes motes cytoplasm acidification, which in turn acti-
vates the acid resistance systems [5]. As shown in
We have recently shown that E. coli cells can re- Fig. 3C, the transcription levels of sucA decreased
spond at the transcriptional level upon exposure to 1.8- and 1.6-fold in the compartment with high
increasing dCO2 concentrations [5]. In particular, dCO2 for cultures maintained at tc of 170 and 375 s
genes from the acid resistance systems, gadA and with respect to the reference condition, respective-
gadC, showed the highest transcription levels, ly. In the case of sucB, the expression levels de-
whereas the genes sucA and sucB, involved in de- creased 1.25- and 1.39-fold in cultures performed
carboxylation reactions of the TCA cycle, were the at tc of 170 and 375 s with respect to the reference
most highly repressed [5]. Accordingly, the tran- condition, respectively (Fig. 3D). Such minor
scriptional response of only these four genes upon changes of sucA and sucB expression were consis-
exposure to dCO2 gradients was determined in this tent with those found by Baez et al. [5], but they
study (Fig. 3). Transcription levels of samples taken should be taken with caution because they are
from the low and high dCO2 concentration com- close to the experimental error (within ± 0.3-fold)
partments were compared with those of the refer- of the qRT-PCR technique employed [30]. Yet, the
ence condition (RNA samples taken from the same down-regulation of these two genes involved in de-
culture just before the onset of dCO2 gradients). carboxylation reactions indicate a reduction in the
The expression of gadC and gadA genes was not TCA cycle activity, concomitant with acetyl-CoA ac-
significantly affected in cultures exposed to dCO2 cumulation as the main cause of acetate accumula-
gradients at tc of 50 and 170 s (Figs. 3A and B). Such tion, although not to the levels previously observed
results are consistent with the negligible changes at high constant dCO2 [5].
observed on growth and yields of these cultures
compared to the reference cultures (Fig. 2).
Nonetheless, at a tc of 375 s, expression levels of 4 Concluding remarks
gadC increased by 66% and 72% in the low and high
dCO2 compartments, respectively. Similarly, gadA The scale-down methodology was established
expression increased by 60% and 54% in the low more than two decades ago; yet, to our knowledge

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 965

Biotechnology
Journal
this is the first time that dissolved CO2 gradients
like those that can be found in large-scale bioreac-
tors have been simulated. The scale-down system
designed in this work was able to mimic, in a sim-
plified way, dCO2 fluctuations that E. coli cells can
experience when circulating through the different
regions of a large fermenter. The designed system
was based on two instrumented stirred tank biore-
actors, which allowed the control of the main
process variables. Accordingly, only the effects at-
tributed to dCO2 gradients were determined. For
cultures performed at the highest circulation time
(375 s), the dCO2 gradient achieved was also the
highest; consequently the detrimental effects of
dCO2 on growth and yields were more pronounced.
The dCO2 gradients were less toxic to the cells com-
pared to previous reports [5] when E. coli were ex-
posed to constant high dCO2 concentrations. Such
results show that E. coli cultures can rapidly adapt
their metabolism when cells are intermittently
changed from a high to low dCO2 environment.The
time (150 s) that the bacterial cells spent in the ves-
sel at low dCO2 concentrations was enough to
reestablish the metabolism and homeostasis. Fi-
nally, the results presented here can be of interest
for industrial applications, as dCO2 gradients that
can occur due to poor mixing (tc of 170 and 375 s),
will not significantly affect growth and recombi-
nant protein production of E. coli cultures.
This work was supported by grants 183288 and
126793 from Consejo Nacional de Ciencia y Tec-
nología, México. Technical support by V. Hernandez
is gratefully acknowledged.
The authors have declared no conflict of interest.
5 References
[1] Zhu, M. M., Goyal, A., Rank, D. L., Gupta, S. K. et al., Effects
of elevated pCO2 and osmolality on growth of CHO cells and
production of antibody-fusion protein B1: A case study.
Biotechnol. Prog. 2005, 21, 70–77.
[2] Matsunaga, N., Kano, K., Maki,Y., Dobashi, T., Culture scale-
up studies as seen from the viewpoint of oxygen supply and
dissolved carbon dioxide stripping. J. Biosci. Bioeng. 2009,
107, 412–418.
[3] Knoll, A., Maier, B., Tscherring, H., Buchs, J., The oxygen
mass transfer, carbon dioxide inhibition, heat removal and
energy and cost efficiencies of high pressure fermentation.
Adv. Biochem. Eng. Biotechnol. 2005, 92, 77–99.
[4] Knoll, A., Bartsch, S., Husemann, B., Engel, P. et al., High cell
density cultivation of recombinant yeasts and bacteria un-
der non-pressurized and pressurized conditions in stirred
tank bioreactors. J. Biotechnol. 2007, 132, 167–179.
[5] Baez, A., Flores, N., Bolivar, F., Ramírez, O. T., Metabolic and
transcriptional response of recombinant Escherichia coli to
966
Biotechnol. J. 2011, 6, 959–967
elevated dissolved carbon dioxide concentrations. Biotech-
nol. Bioeng. 2009, 104, 102–110.
[6] Jones, R. P., Greenfield, P. F., Effect of carbon dioxide on
yeast growth and fermentation. Enzyme Microb. Technol.
1982, 4, 210–223.
[7] Pan, G., Rhee, J., Lebeault, J., Physiological constraints in in-
creasing biomass concentration of Escherichia coli B in fed
bath culture. Biotechnol. Lett. 1987, 9, 89–94.
[8] Dixon, N. M., Kell, D. B., The inhibition by CO2 of the growth
and metabolism of micro-organisms. J. Appl. Bacteriol. 1989,
67, 109–136.
[9] Castan, A., Näsman, A., Enfors, S., Oxygen enriched air
supply in Escherichia coli processes: Production of biomass
and recombinant human growth hormone. Enzyme Microb.
Technol. 2002, 30, 847–854.
[10] El-Sabbagh, N., McNeil, B., Harvey, L. M., Dissolved carbon
dioxide effects on growth, nutrient consumption, penicillin
synthesis and morphology in batch cultures of Penicillium
chrysogenum. Enyme Microb. Technol. 2006, 39,185–190.
[11] Bäumchen, C., Knoll, A., Husemann, B., Seletzky J. et al., Ef-
fect of dissolved carbon dioxide concentrations on growth
of Corynebacterium glutamicum on D-glucose and L-lactate.
J. Biotechnol. 2007, 128, 868–874.
[12] Amoabediny, G., Büchs, J., Determination of CO2 sensitivity
of microorganisms in shake bioreactors. I. Novel method
based on the resistance of sterile closure. Biotechnol. Appl.
Biochem. 2010, 57, 157–166.
[13] Li, M., Ho, P. Y., Yao, S., Shimizu, K., Effect of lpdA gene
knockout on the metabolism in Escherichia coli based on
enzyme activities, intracellular metabolite concentrations
and metabolic flux analysis by 13C-labeling experiments. J.
Biotechnol. 2006, 122, 254–266.
[14] Patnaik, P. R., Can imperfections help to improve bioreactor
performance? Trends Biotechnol. 2002, 20, 135–137.
[15] Lara, A. R., Leal, L., Flores, N., Gosset, G. et al., Transcrip-
tional and metabolic response of recombinant Escherichia
coli to spatial dissolved oxygen tension gradients simulated
in a scale-down system. Biotechnol. Bioeng. 2006a, 93, 372–
385.
[16] Lara, A. R., Galindo, E., Ramirez, O. T., Palomares, L. A., Liv-
ing with heterogeneities in bioreactors: Understanding the
effects of environmental gradients on cells. Mol. Biotechnol.
2006, 34, 355–381.
[17] Sweere, A. P. J., Luyben, K. C. A. M., Kossen, N. W. F., Regime
analysis and scale-down: Tools to investigate the perform-
ance of bioreactors. Enzyme Microbiol. Technol. 1987, 9, 386–
397.
[18] Oosterhuis, N. M. G., Kossen, N.W. F., Olivier, A. P. C., Schenk,
E. S., Scale-down and optimization studies of the gluconic
acid fermentation by Gluconobacter oxidans. Biotechnol.
Bioeng. 1985, 27, 711–720.
[19] Bylund, F., Collet, E., Enfors, S. O., Larsson, G., Substrate gra-
dient formation in the large scale bioreactor lowers cell
yield and increases by-product formation. Bioprocess Eng.
1998, 18, 171–178.
[20] Bylund, F., Guillard, F., Enfors, S., Tragardh, C. et al., Scale-
down of recombinant protein production: A comparative
study of scaling performance. Bioprocess Eng. 1999, 20, 377–
389.
[21] Amanullah, A., McFarlane, C., Emery, A. N., Nienow, A. W.,
Scale down model to simulate spatial pH variations in large-
scale bioreactors. Biotechnol. Bioeng. 2001, 73, 390–399.
[22] Trujillo-Roldán, M. A., Peña, C., Ramírez, O. T., Galindo, E.,
Effects of oscillating dissolved oxygen tensions in the pro-
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Biotechnol. J. 2011, 6, 959–967
duction of alginate by Azotobacter vinelandii. Biotechnol.
Prog. 2001, 17, 1042–1048.
[23] Onyeaka, H., Nienow, A. N., Hewitt, C. J., Further studies re-
lated to the scale-up of high cell density Escherichia coli
fed-batch fermentations: The additional effect of a chang-
ing microenvironment when using aqueous ammonia to
control pH. Biotechnol. Bioeng. 2003, 84, 474–484.
[24] Serrato, J. A., Palomares, L. A., Meneses, A., Ramírez, O. T.,
Heterogeneous conditions in dissolved oxygen affect N-gly-
cosylation but not productivity of a monoclonal antibody in
hybridoma cultures. Biotechnol. Bioeng. 2004, 88, 176–188.
[25] Lara, A. R., Vazquez-Limón, C., Gosset, G., Bolívar, F. et al.,
Engineering Escherichia coli to improve culture perform-
ance and reduce formation of by-products during recombi-
nant protein production under transient intermittent
anaerobic conditions. Biotechnol. Bioeng. 2006, 94, 1164–1175.
[26] Oosterhuis, N. M. G., Kossen, N., Dissolved oxygen concen-
tration profiles in a production scale bioreactor. Biotechnol.
Bioeng. 1984, 26, 546–550.
[27] Enfors, S., Jahic, M., Rozkov,A., Xu, B. et al., Physiological re-
sponses to mixing in large-scale bioreactors. J. Biotechnol.
2001, 85, 175–185.
[28] Amanullah, A., Buckland, B. C., Nienow, A., Mixing in the
fermentation and cell culture industries, In: Paul E. L.,
Atiemo-Obeng V. A., Kresta S. M. (Eds.), Handbook of In-
dustrial Mixing: Science and Practice, John Wiley & Sons,
New Jersey 2004, pp. 1071–1170.
[29] Sandoval-Basurto, E.A., Gosset, G., Bolívar, F., Ramírez, O.T.,
Culture of Escherichia coli under dissolved oxygen gradi-
ents simulated in a two-compartment scale-down system:
Metabolic responses and production of recombinant pro-
tein. Biotechnol. Bioeng. 2005, 89, 453–463.
© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.biotechnology-journal.com
[30] Flores, N., Flores, S., Escalante, A., de Anda, R. et al., Adap-
tation for fast growth on glucose by differential expression
of central carbon metabolism and gal regulon genes in an
Escherichia coli strain lacking the phosphoenolpyruvate:
carbohydrate phosphotransferase system. Metab. Eng. 2005,
7, 70–87.
[31] Livak, K. J., Schmittgen, T. D., Analysis of relative gene ex-
pression data using real-time quantitative PCR and the
2[-Delta Delta C(T)] method. Methods 2001, 25, 402–408.
[32] Hewitt, C. J., Boon, L. T., McFarlane, C. M., Nienow, A. W., The
use of flow cytometry to study the impact of fluid mechani-
cal stress on Escherichia coli W3110 during continuous cul-
tivation in an agitated bioreactor. Biotechnol. Bioeng. 1998,
59, 612–620.
[33] Chamsartra, S., Hewitt, C. J., Nienow,A.W.,The impact of flu-
id mechanical stress on Corynebacterium glutamicum dur-
ing continuous cultivation in an agitated bioreactor. Biotech-
nol. Lett. 2005, 27, 693–700.
[34] Garcia-Gonzalez, L., Geeraerd, A. H., Spilimbergo, S., Elst,
K. et al., High pressure carbon dioxide inactivation of mi-
croorganisms in food: The past, the present and the future.
Int. J. Food Microbiol. 2007, 117, 1–28.
[35] Garcia-Gonzalez, L., Geeraerd, A. H., Mast, J., Briers,Y. et al.,
Membrane permeabilization and cellular death of Es-
cherichia coli, Listeria monocytogenes, and Saccharomyces
cerevisiae as induced by high pressure carbon dioxide treat-
ment. Food Microbiol. 2010, 27, 541–549.
[36]. Iizuka, R., Yamagishi-Shirasaki, M., Funatsu, T., Kinetic
study of de novo chromophore maturation of fluorescent
proteins. Anal. Biochem. 2011, 414, 173–178.
967