Biotechnol. J. 2011, 6, 1009–1017 DOI 10.1002/biot.201100293 www.biotechnology-journal.

com

Technical Report

A two-compartment bioreactor system made of commercial

parts for bioprocess scale-down studies: Impact of oscillations
on Bacillus subtilis fed-batch cultivations

Stefan Junne1, Arne Klingner1*, Johannes Kabisch2, Thomas Schweder2 and Peter Neubauer1
1 Chair of Bioprocess Engineering, Department of Biotechnology, Technische Universität Berlin, Berlin, Germany
2 Laboratory of Pharmaceutical Biotechnology, Institute of Pharmacy, Ernst-Moritz-Arndt Universität, Greifswald, Germany

This study describes an advanced version of a two-compartment scale-down bioreactor that sim-
Received 14 June 2011
ulates inhomogeneities present in large-scale industrial bioreactors on the laboratory scale. The Revised 8 July 2011
system is made of commercially available parts and is suitable for sterilization with steam. The Accepted 11 July 2011
scale-down bioreactor consists of a usual stirred tank bioreactor (STR) and a plug flow reactor
(PFR) equipped with static mixer modules. The PFR module with a working volume of 1.2 L is
equipped with five sample ports, and pH and dissolved oxygen (DO) sensors. The concept was
applied using the non-sporulating Bacillus subtilis mutant strain AS3, characterized by a SpoIIGA
gene knockout. In a fed-batch process with a constant feed rate, it is found that oscillating sub-
strate and DO concentration led to diminished glucose uptake, ethanol formation and an altered
amino acid synthesis. Sampling at the PFR module allowed the detection of dynamics at different
concentrations of intermediates, such as pyruvic acid, lactic acid and amino acids. Results indi-
cate that the carbon flux at excess glucose and low DO concentrations is shifted towards ethanol
Supporting information
formation. As a result, the reduced carbon flux entering the tricarboxylic acid cycle is not sufficient
available online
to support amino acid synthesis following the oxaloacetic acid branch point.

Keywords: Bacillus subtilis · Fed-batch cultivation · Inhomogeneities · Scale-down · Two-compartment reactor

1 Introduction zone, are substrate depleted. Consequently, cells

are exposed to starvation conditions. Because the
It has been observed that in industrial-scale biore- cells are moving rapidly in the turbulent fluid flow
actors mixing of the incoming feed solution is not of the bioreactor, they are continuously exposed to
sufficient to prevent the appearance of substrate alternating environmental conditions between
gradients [1, 2]. Hence, concentrations of the ingre- substrate excess and starvation.
dients of the feed solution are higher near the feed- Multiple devices have been introduced to ex-
ing zone, causing an alternation of intracellular pose cells to oscillatory conditions in laboratory
fluxes and supporting the synthesis of (sometimes simulators [4]. Different approaches have consist-
unwanted) byproducts [3]. In contrast, other re- ed of either two stirred tank reactors (STR-STR), or
gions of the bioreactor far away from the feeding of a STR and a plug flow reactor (STR-PFR). Static
mixers were applied in the PFR module to support
the supply of sufficient oxygen to the liquid phase
Correspondence: Dr. Stefan Junne, Chair of Bioprocess Engineering, [5, 6]. A two-compartment reactor was also applied
Department of Biotechnology, Technische Universität Berlin,
to observe effects of pH gradients near the acid/
Ackerstrasse 71–76 ACK24, 13355 Berlin, Germany
E-mail: [email protected]

Abbreviations: DO, dissolved oxygen; PFR, plug flow reactor; STR, stirred * Current address: Institute of Biochemical Engineering, Technische Uni-
tank bioreactor versität Braunschweig, Gaußstrasse 17, 38106 Braunschweig, Germany

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1009

Biotechnology Biotechnol. J. 2011, 6, 1009–1017
Journal

base addition in Escherichia coli and Bacillus sub-

tilis cultivations [7, 8].
Despite the industrial importance of B. subtilis
cultivations, the impacts of oscillating conditions
on the process yield and bacterial physiology in
scale-down approaches are very rare. In this re-
port, we describe the influence of large-scale con-
ditions in a newly designed two-compartment
scale-down reactor made of an STR coupled to a
PFR. Unlike the first systems, the PFR part is
equipped with five pH and dissolved oxygen (DO)
Figure 1. Scheme of the two-com-
sensors as well as sampling ports. All reactor parts partment reactor as described in this
are commercially available. The application is study. (1) STR with a working volume
demonstrated using a non-sporulating B. subtilis of 10 L. (2) Connection between
strain with the focus on the dynamics within the the STR and a PFR module. A hose
main carbon metabolism and the amino acid pool. pump pumps the fluid from the STR
Our results obtained with the new reactor concept into the PFR. The feed solution is
introduced to the system between
are discussed with regard to new information
the pump and the PFR module.
gained, and show that the tool is suitable for com- (3) The PFR module is equipped with
bining systems biology and process optimization static mixer modules. (4) A spacer
approaches. equipped with three ports is mount-
ed between each static mixer mod-
ule. One port is used for the DO
2 Materials and methods sensor, one for the pH electrode
and one for manual sampling
through a membrane.
2.1 The scale-down reactor concept

The scale-down reactor (Fig. 1) consists of a 10-L Switzerland) were applied to monitor the distribu-
stirred tank bioreactor (Biostat E, Sartorius SA, tion of the corresponding parameter measured
Göttingen Germany) equipped with two disk stir- along the height of the PFR. The DO sensor was
rers and a mechanical foam destroyer at the top of mounted so that the membrane reached to the in-
the stirrer shaft. Beside a pt-100 temperature sen- ner wall of the spacer, while the swelling part of the
sor, a pH sensor (65/90 VT) and a DO sensor (both pH sensor reached in the bulk of the fluid stream.
Mettler-Toledo Deutschland GmbH, Gießen, Ger- Prior to utilization, the pH and DO sensors were
many) were used at the STR for control purposes. calibrated using the software package ARC sensor
At the bottom, the STR module was connected to a configurator V2.2 (Hamilton Inc.). Data from all
hose pump, operation range of 50–155 L/h.The tub- sensors were transmitted with a D/O converter
ing inside the pump was made of Ponnprene F DN LabjackTM U12 (Meilhaus Electronic GmbH, Puch-
19 (pump distributed by Lewa GmbH, Leonberg, heim, Germany) and collected with the software
Germany). The tubing between the STR and the LabView® (National Instruments Inc., Austin, TX).
pump was made of silicone rubber with an inner Calibration of the DO sensors was performed in ni-
diameter of 15 mm.The pump was connected to the trogen-sparged water (0% saturation of DO) and air
PFR part with a 32-mm tube coated with EPDM (100% saturation of DO), while the pH sensors were
(Reichelt Chemietechnik GmbH & Co, Heidelberg, calibrated at pH 7.0 and 4.0. The total working vol-
Germany). ume of the PFR part was 1.2 L [1.8 L including the
The PFR part consisted of four static mixer transfer from the STR to the PFR (0.15 L) and back-
modules, each with working volume 0.25 L and di- wards (0.45 L)]. The system was sterilized with hot
ameter 25 mm (Kenics series KM, distributed by steam for 40 min. To avoid backflow at the feed
Lewa GmbH) (Fig. 1). The modules were insulated pump, a back-pressure valve was integrated in the
with a polymer foam jacket to maintain the tem- feed tube. Aeration was possible above a rate of
perature in the PFR module. Between each static 0.1 vvm through a sintered inlet at the bottom, but
mixer module, spacers were mounted that were was not applied in this study. For cultivation exper-
equipped with three sample ports with an inner di- iments, a pump rate of 1.78 L/min was used (corres-
ameter of 12 mm.At each spacer, an optical DO sen- ponding to ~1 min mean retention time in the
sor (Visiferm DO ARC 120) and a pH sensor (Po- whole PFR part).
lilyte Plus ARC 120) (both Hamilton Inc., Bonaduz,

1010 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 1009–1017 www.biotechnology-journal.com

2.2 Determination of the residence time The kLa values are higher at the top part of the
distribution reactor. This increase is not linear but rises more
rapidly along the height of the PFR at the applied
To characterize the PFR part with respect to its re- flow rate: kLa1 = 19/h, kLa2 = 20/h, kLa3 = 28/h,
tention time and plug flow behavior, extrusion ex- kLa4 = 41/h, and kLa5 = 56/h. The static mixers seem
periments with 5 mM H2SO4 were performed at dif- to provide higher oxygen transfer at lower axial
ferent flow and aeration rates. The plug flow char- dispersion. Values are approximately one tenth
acteristics were evaluated with the Bodenstein lower than in usual stirred tank reactors for bio-
number [9].The axial dispersion coefficient Dax can processes. However, when the PFR was directly
be derived based on the obtained Bodenstein num- aerated from the bottom, DO was still measured for
ber Bo, the flow velocity u and the length L of the E. coli cultivations up to OD600 = 80 (results not
observed distance: shown). Oxygen depletion in the connecting tubing
and the hose pump between the STR and PFR can
u ⋅L
Dax = (1) be avoided if the oxygen saturation in the STR is
Bo
maintained at levels at least above 50% of satura-
The characterization of the fluid flow by pulse ex- tion under these cultivation conditions and a flow
periments yielded a narrow residence time (τ) dis- rate above 1.7 mL/min.
tribution. Residence times at the applied flow rate
of 1.78 mL/min for τport1 was 26 s, τport2 35 s, τport3 2.4 Strain and culture conditions
45 s, τport4 56 s, and τport5 62 s. The mean residence
time at the last port was comparably low due to the As stress induction and spore formation are relat-
fact that no other mixing module was located be- ed, it is possible that spore formation is initiated
hind this measurement point. However, the hydro- just when cells are exposed to oscillating conditions
dynamic pressure is the lowest at the upper sample as performed in this study. Therefore, a B. subtilis
port, which likely influences the residence time mutant strain was applied for experiments in which
distribution and the disturbance of the plug flow a central modulator in the sporulation cascade,
characteristics. The deviance between measure- SpoIIGA, was knocked out. The knockout was
ments was below 5%. achieved using a double-crossover procedure, lead-
The Dax values achieved ranged from 0.02 to ing to an insertion that disrupted the target gene
0.005 m2/s. Bodenstein numbers above 7 to 10 in- [11]. The SpoIIGA mutant strain B. subtilis AS3 did
dicate a plug flow behavior [5, 10]. This would re- not show any sporulation, in contrast to the wild-
sult in Dax values below 0.015 m2/s. Hence, resi- type strain, which showed a rate of 98% in Difco
dence times above 30 s (port 2 and above) are char- Sporulation Media (DSM). The growth of the mu-
acterized by a dominating plug flow behavior. tant remained comparable to that of the wild-type.
Backmixing effects at port 1, which is located right In contrast to the commonly applied Spo0A mu-
after the change of the flow direction at the en- tants, the SpoIIGA mutant showed lysis rates simi-
trance of the PFR module, might hinder the devel- lar to those of the wild-type strain.
opment of an ideal plug flow behavior. However, The utilized mineral salt medium was applied as
higher dispersion at the bottom can even be bene- described elsewhere [12]. Two feeding phases were
ficial when the introduced feed solution has to be applied: one used the in situ enzymatic substrate re-
well distributed. lease method EnBase®-Flo [13] until a biomass con-
centration of approximately 4 g/L was achieved, and
2.3 Determination of kLa values the second used mechanical feeding of a concen-
trated feed solution containing 440 g/L glucose as
The dynamic method (gassing out) was applied to the main carbon source at a constant rate thereafter.
determine the volumetric oxygen transfer coeffi- During the mechanical feeding phase, feed was
cient kLa. The kLa value was determined as follows: added at the top of the STR (reference experiment
under non-oscillating conditions) or at the entrance
⎛ c* − c ⎞ of the PFR module in the two-compartment reactor.
ln ⎜ * O2 i,O2 ⎟ A feed rate of 1 mL/min was applied. The pH was
⎝ cO2 − ci +1,O2 ⎠
k La = (2) controlled at 7.0 by adding a 25% NH4OH solution.
ti +1 − ti
As antifoam agent 0.1‰ polypropylene glycol 3000
In Eq. (2), ci,O2 denotes the measured concentration (Fluka AG, Buchs, Switzerland) was used. The tem-
of DO at measurement i, c*O2 is the maximum solu- perature was maintained at 37°C. The aeration rate
bility of O2 under the applied conditions, and t the of the STR was set constant to 1 vvm. All experi-
time point of the measurement. ments were performed in duplicate.

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1011

Biotechnology Biotechnol. J. 2011, 6, 1009–1017
Journal

2.5 Sample preparation the glucose-limited growth mode. An early initia-

tion of glucose-limited conditions (in this case 6 h
For analysis, samples were taken from the STR and before the attachment of the PFR module) is ab-
from the five sampling ports in the PFR. As intra- solutely necessary to gain reliable results that re-
and extracellular analyses of different compounds flect the conditions in industrial scale processes.
should be conducted, samples of the culture super- Otherwise the portion of cells grown under glu-
natant and the suspension were taken. Intracellu- cose-sufficient conditions would be too large in the
lar concentrations were determined by the differ- early samples, leading to misleading results. How-
ential method. To sample the cell suspension, nee- ever, the major importance of the application of an
dles and membrane adapters were applied in situ release system is the provision of glucose
(Monovette®, Sarstedt AG, Nümbrecht, Germany). during the transfer of cells between the STR and
To extract the supernatant syringes equipped with the PFR module.While the STR runs under glucose
hydrophilic nitrocellulose filters (pore size: 0.8 µm, limitation, the substrate supply by external feeding
diameter: 25 mm, Millipore Inc., Schwalbach, Ger- would not be sufficient to provide any substrate in
many) were used. Subsequently, supernatant sam- the connecting tubing.A strong starvation response
ples were immediately frozen at –80°C. in the cells would be triggered, influencing all in-
For analysis of the intracellular concentrations, tracellular measurements. Due to the in situ sub-
two different sampling methods were applied. In strate release (in addition to the mechanic feeding),
the first method for preparing samples for the cells are never exposed to complete glucose starva-
amino acid analysis, cold methanol was added to tion.
stop the metabolism of the cells. For cell disruption, Although high cell density cultivations are gen-
samples were thawed and diluted to an OD600 of erally operated with an exponential feed, in this
0.22 and sonicated for five cycles (each lasting 30 s) study a constant feed rate was applied. This report
with a 30-s break in between each cycle. Samples initially describes the behavior of a B. subtilis
for intracellular nucleotide and volatile fatty sporulation mutant in a two-compartment reactor,
acid/alcohol determination were prepared as de- and different ratios of substrate to cell density (that
scribed elsewhere [14]. is amount of substrate provided per cell) can be ex-
amined. Hence, different conditions can be applied
2.6 Analysis in one experiment, evaluating which conditions
lead to impacts on the central metabolism of B. sub-
The off gas was analyzed with an O2 and CO2 sen- tilis cultivations. At the PFR inlet, a concentration
sor (Bluesense Gas Sensor GmbH, Herten, Ger- of 0.4 g/L glucose is achieved at the beginning of
many). The derivation of the specific oxygen up- the mechanical feeding phase. The excess glucose
take rate (OUR) and carbon dioxide production concentration measured at the first sample port in
(CER) based on a gas balance of the input and out- the PFR module decreases throughout the cultiva-
put flow was performed as described in [15]. Sug- tion to values below 0.1 g/L.
ars and volatile fatty acids/alcohols were analyzed
as published previously [12]. The quantification of 3.2 Comparison of cultivations in the STR and in
amino acids in the culture supernatant and culture the two-compartment reactor
broth followed the methodology described in [16].
In the PFR module, the online pH measurement at
the five sample ports indicated no significant
3 Results change in the pH throughout the cultivation. As-
suming that extracellular pH changes of 0.1 will not
3.1 Inoculation and feeding phase with in situ lead to altered intracellular pH changes in bacteria,
substrate release this parameter seems too unimportant to have any
influence when the PFR module is applied.The on-
The preculture was cultivated to an OD600 of 0.2–0.3 line DO measurement (Fig. 2) indicated low values
before being inoculated into the bioreactor. The in the PFR, but oxygen was not depleted in the PFR
preculture was grown in the same in situ release part throughout the cultivation. Hence, the oxygen
fed-batch cultivation system EnBase-Flo as in the availability might be limited inside the cell, but it is
bioreactor cultivation. After a short lag phase, a still continuously available from outside. These
maximum extracellular glucose concentration of conditions reflect those of the zone near the feed-
0.3 g/L was detected (see Supporting information, ing point in industrial scale fed-batch cultivations,
Fig. S1). At 6 h before the start of the mechanic where the oxygen availability is also low in high cell
feeding phase, the culture turned automatically to density cultivations – due to the large distance to

1012 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 1009–1017 www.biotechnology-journal.com

Figure 2. Comparison of B. subtilis AS3 fed-batch cultivations in an STR (left) and in the two-compartment reactor (TCR, right) at a constant feed rate
(26.4 g/h glucose). (A) Cell dry weight (closed circles), specific growth rate (dashed line) and feed rate (straight line). (B) DO saturation at the STR (upper
black dashed line), DO at sensor 1 (black line), at sensor 2 (light gray line), at sensor 3 (black dotted line), at sensor 4 (dark gray line) and at sensor 5
(gray dotted line). (C) Extracellular glucose polymer (closed circles), glucose (opened squares) and lactic acid (opened triangles) concentration in the STR.
(D) Extracellular ethanol (opened circles) and acetic acid (closed triangles) concentration in the STR. (Max. deviance at concentration measurements of 6%.)

stirrers and the lower portion of O2 in the gas phase accumulation of acetic acid and lactic acid were ob-
at the top of the reactor. served, ethanol was synthesized in larger amounts
As depicted in Fig. 2, bacterial growth was hard- under oscillating conditions. Final values were six-
ly altered when applying the two-compartment re- fold higher than under non-oscillating conditions.
actor concept. In contrast, the specific glucose up- Ethanol is a main product of the fermentative me-
take rate was diminished by 10%.This led to the ac- tabolism of B. subtilis [17]. In accordance, the res-
cumulation of glucose in the first hours of the me- piratory quotient was slightly higher under oscil-
chanical feeding phase. Hence, the culture was no lating conditions (Fig. 3). Surprisingly, no accumu-
longer operating under glucose limitation (also at lation of other fermentative byproducts like acetoin
the STR module). Due to the constant feed rate and and butanediol were detected extra- or intracellu-
the increasing cell concentration, the volumetric larly in any cultivation. Intracellular peak concen-
glucose uptake rate increases. Finally, glucose-lim- trations of ethanol were found in samples taken
ited conditions were achieved after 20 h of cultiva- from the top phase of the PFR module (Fig. 4).
tion. While only small changes in the extracellular Regarding the intracellular concentrations in

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1013

Biotechnology Biotechnol. J. 2011, 6, 1009–1017
Journal

Figure 3. Off gas analysis of B. subtilis AS3 fed-batch cultivations in an STR (left) and in the two-compartment reactor (right) at a constant feed rate
(26.4 g/h glucose). (A) Volumetric portion of O2 (black) and CO2 (gray). (B) Specific oxygen uptake qO2 (black), specific carbon dioxide release qCO2
(light gray) and respiratory quotient RQ (dark gray).

the STR, only pyruvic acid was found to be above

the detection limit (see Supporting information,
Fig. S2). When the extracellular glucose concentra-
tion was high (i.e., when no substrate limitation was
present), the intracellular pyruvic acid concentra-
tion in cells in the STR module of the two-com-
partment reactor was sixfold higher. In between
the first two measurements at the PFR module (3 h
of operation under oscillating conditions), higher
concentrations of pyruvic acid were seen already at
the first ports of the PFR (Fig. 5).

3.3 Impact of oscillatory conditions on the amino

acid synthesis

From the onset of the cultivation, two amino acids

(valine and tryptophan) were found in the super-
natant. Beside the amounts already present in the
inoculum and media, an accumulation of both
amino acids was observed under oscillating condi-
tions (Fig. 5A). If this observation was a result of
stronger cell lysis (which is doubtful due to the
even growth under the two different conditions) or
a stronger release due to an enhanced synthesis
was not clear. A stronger impact on the amino acid
synthesis was observed when measuring the intra-
cellular concentrations (Fig. 5B–G). Aspartate, as-
paragine and glutamine concentrations were di-
Figure 4. Intracellular volatile fatty acid and ethanol concentrations meas-
minished drastically at least during certain process
ured in the PFR module of the two-compartment reactor for B. subtilis AS3 phases under oscillating conditions. In contrast,
fed-batch cultivations. Numbering performed in the order of ports from arginine production was more than doubled. Since
the bottom to the top of the PFR module (max. deviance of 6%). aspartate and asparagine share the same precur-

1014 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 1009–1017 www.biotechnology-journal.com

Figure 5. Concentrations of amino acids of B. subtilis AS3 fed-batch cultivations in an STR (left) and the two-compartment reactor (TCR, right) – samples
were taken from the STR module. (A) Extracellular valine concentrations in the STR (open circles) and in the TCR (closed circles). (B) Extracellular trypto-
phan concentrations in the STR (open squares) and TCR (closed squares). (C) Intracellular aspartate concentrations in the STR (open circles) and TCR
(closed circles). (D) Intracellular glutamic acid concentrations in the STR (open squares) and TCR (closed squares). (E) Intracellular asparagine concentra-
tions in the STR (open triangles) and TCR (closed triangles). (F) Intracellular glutamine concentrations in the STR (open triangles) and TCR (closed trian-
gles). (G) Intracellular arginine concentrations in the STR (open diamonds) and TCR (closed diamonds). (H) Intracellular alanine concentrations in the
STR (open circles) and TCR (closed circles). Max. deviance of 7% among all data.

sor, oxaloacetic acid, it is likely that the carbon flux the precursor glutamic acid. The glutamine syn-
through the tricarboxylic acid cycle is not sufficient thetase (GS) and its expression are known to be
to supply the demand for their production. As men- regulated strongly depending on the availability of
tioned, the redirection of the carbon flux towards nitrogen, and depending on the concentration of
ethanol (and to some extend also to lactic acid) pro- glutamic acid. Small oscillations of concentrations
duction might be one reason for this limitation. of this intermediate and/or the presence of excess
Some carbon flux is redirected from the conversion nitrogen available might lead to a repression of GS
of glutamine to arginine. Both amino acids share activity and, as consequence, a stronger arginine

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1015

Biotechnology Biotechnol. J. 2011, 6, 1009–1017
Journal

production. This assumption is underlined when possible inhibition of the GS by increased gluta-
considering the findings of the intracellular meas- mine concentrations might cause a redirection of
urements of amino acids in the PFR (Fig. 6). While the fluxes towards asparagine synthesis [18]. The
glutamine accumulated initially, an accumulation general transcription factor TnrA activates nitrate
of glutamic acid was observed at higher ports. A and nitrite utilization and transport in B. subtilis
[19, 20]. It is bound tightly to the inactivated form
of GS, mediating nitrogen assimilation.TnrA is also
bound to the non-inhibited form of GS, although
with a lower affinity [21]. This might be problemat-
ic when cells are exposed to different nitrogen
availability at a rate faster than adaptation process-
es (response times) inside the cell (e.g., the intra-
cellular glutamine concentrations might change
more slowly than the oscillating environment).
Surprisingly, asparagine was produced in the PFR
module part, but finally reconverted in the STR
module, since an accumulation was not detected.
The dynamics that lie behind this observation re-
main unclear and have to be further examined.
The oscillating conditions have a profound im-
pact on the main carbon and amino acid synthesis.
Further studies also revealed changes in the nu-
cleotide levels, which are not discussed here. How-
ever, it is clear that the repeated exposure of sub-
strate excess under a reduced availability of DO
changes the amino acid composition of the cell.
As stated in the literature [3, 22], oscillating con-
ditions enhance the substrate uptake rate of E. coli
cultivations. This is differs greatly to observations
in this study with B. subtilis. Many regulatory me-
chanisms for carbon conversion have been identi-
fied in Bacillus sp. [23] Among the most prominent
ones is the catabolite control protein CcpA, which
mediates various regulatory mechanisms effecting
glycolysis [24], fermentative carbon conversion
[23] and ammonia assimilation [25]. To elucidate
the regulatory interactions of the process observed
more thoroughly, a dynamic metabolic flux analysis
based on the stoichiometric matrix of the main car-
bon metabolism and amino acid synthesis has to be
performed in future studies.

4 Concluding remarks
Due to the rapidly varying physiological conditions
(overflow metabolism, sporulation, carbon catabo-
lite repression) and the strong regulation of the
metabolic network, the systems biology of Bacillus
sp. is currently in the focus of many research proj-
ects. The link between common approaches in sys-
tems biology related to fundamental research and
Figure 6. Intracellular amino acid concentrations measured in the PFR
the industrial needs of large-scale bioprocesses
module of the two-compartment reactor at B. subtilis AS3 fed-batch culti- can provide better insights into problems related to
vations. Numbering performed in the order of ports from the bottom to scale-up, strain improvement and development for
the top of the PFR module (max. deviance of 5%). production purposes. However, up to now, no suit-

1016 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 1009–1017 www.biotechnology-journal.com

able tool capable of integrating systems biology ap- [9] Levenspiel, O., Chemical Reaction Engineering, Wiley-VCH,
proaches and scale-up issues in a scale-down reac- New York 1972.
[10] Hagen, J., Chemiereaktoren – Auslegung und Simulation,
tor has been commercially available. Data on dif-
Wiley-VCH, Weinheim, New York 2004.
ferent levels of interaction, combined with pro- [11] Sanchez, H., Cozar, A. C., Martinez-Jimenez, M. I., Targeting
teomics and transcriptomics along the height of the the Bacillus subtilis genome: An efficient and clean method
PFR module, will help us to gain a better under- for gene disruption. J. Microbiol. Methods 2007, 70, 389–394.
standing of the adaptation processes of the cell.The [12] Glazyrina, J., Materne, E. M., Dreher, T., Storm, D. et al., High
impact on the regulatory network of a cell can be cell density cultivation and recombinant protein production
further understood. For observations in a faster with Escherichia coli in a rocking-motion-type bioreactor.
Microb. Cell Fact. 2010, 9, 42.
time range, the described system could be connect-
[13] Krause, M., Ukkonen, K., Haataja, T., Ruottinen, M. et al., A
ed to a rapid sampling unit, e.g., like the Bioscope novel fed-batch based cultivation method provides high
concept [26], closing the time gap between the cell-density and improves yield of soluble recombinant pro-
feeding point and the first sample port. The de- teins in shaken cultures. Microb. Cell Fact. 2010, 9, 11.
scribed methodology, in combination with a strict [14] Buchholz, A., Takors, R., Wandrey, C., Quantification of in-
consideration of findings from computational fluid tracellular metabolites in Escherichia coli K12 using liquid
chromatographic electrospray ionization tandem mass
dynamics in large-scale reactors for the design of
spectrometric techniques. Anal. Biochem. 2001, 295, 129–
the experimental setup, will lead to advanced 137.
process development for the industrial scale. [15] Junne, S., Nicolas Cruz-Bournazou, M., Angersbach, A.,
Gotz, P., Electrooptical monitoring of cell polarizability and
cell size in aerobic Escherichia coli batch cultivations. J. Ind.
We kindly thank Hamilton (Bonaduz, Switzerland) Microbiol. Biotechnol. 2010, 37, 935–942.
for the donation of the pH and DO sensors for the [16] Kroemer, J. O., Systembiologische Studien an Corynebac-
terium glutamicum zur Charakterisierung der Methionin-
PFR module.
synthese. Universität Saarbrücken 2006, p. 182.
[17] Nakano, M. M., Dailly, Y. P., Zuber, P., Clark, D. P., Character-
The authors have declared no conflict of interest. ization of anaerobic fermentative growth of Bacillus subtilis:
Identification of fermentation end products and genes re-
quired for growth. J. Bacteriol. 1997, 179, 6749–6755.
5 References [18] Wray, L. V. Jr., Zalieckas, J. M., Fisher, S. H., Bacillus subtilis
glutamine synthetase controls gene expression through a
[1] Bylund, F., Collet, E., Enfors, S. O., Larsson, G., Substrate gra- protein-protein interaction with transcription factor TnrA.
dient formation in the large-scale bioreactor lowers cell Cell 2001, 107, 427–435.
yield and increases by-product formation. Bioprocess Eng. [19] Nakano, M. M., Hoffmann, T., Zhu, Y., Jahn, D., Nitrogen and
1998, 18, 171–180. oxygen regulation of Bacillus subtilis nasDEF encoding
[2] Enfors, S. O., Jahic, M., Rozkov, A., Xu, B. et al., Physiological NADH-dependent nitrite reductase by TnrA and ResDE. J.
responses to mixing in large scale bioreactors. J. Biotechnol. Bacteriol. 1998, 180, 5344–5350.
2001, 85, 175–185. [20] Wray, L. V., Ferson, A. E., Rohrer, K., Fisher, S. H., TnrA, a
[3] Neubauer, P., Haggstrom, L., Enfors, S. O., Influence of sub- transcription factor required for global nitrogen regulation
strate oscillations on acetate formation and growth-yield in in Bacillus subtilis. Proc. Natl. Acad. Sci. USA 1996, 93, 8841–
Escherichia coli glucose-limited fed-batch cultivations. 8845.
Biotechnol. Bioeng. 1995, 47, 139–146. [21] Kayumov, A., Heinrich, A., Fedorova, K., Ilinskaya, O., Forch-
[4] Neubauer, P., Junne, S., Scale-down simulators for metabol- hammer, K., Interaction of the general transcription factor
ic analysis of large-scale bioprocesses. Curr. Opin. Biotech- TnrA with the PII-like protein GlnK and glutamine syn-
nol. 2010, 21, 114–121. thetase in Bacillus subtilis. FEBS J. 2011, 278, 1779–1789.
[5] George, S., Larsson, G., Enfors, S. O., A scale-down 2-com- [22] Hewitt, C., Nebe-Von Caron, G., An industrial application of
partment reactor with controlled substrate oscillations – multiparameter flow-cytometry: Assessment of cell physio-
Metabolic response of Saccharomyces cerevisiae. Bioprocess logical state and its application to the study of microbial fer-
Eng. 1993, 9, 249–257. mentations. Cytometry 2001, 44, 179–187.
[6] George, S., Larsson, G., Olsson, K., Enfors, S. O., Comparison [23] Stulke, J., Hillen, W., Regulation of carbon catabolism in
of the Baker’s yeast process performance in laboratory and Bacillus species. Annu. Rev. Microbiol. 2000, 54, 849–880.
production scale. Bioprocess Eng. 1998, 18, 135–142. [24] Kim, H. J., Jourlin-Castelli, C., Kim, S. I., Sonenshein, A. L.,
[7] Onyeaka, H., Nienow, A., Hewitt, C., Further studies related Regulation of the Bacillus subtilis ccpC gene by CcpA and
to the scale-up of high cell density Escherichia coli fed- CcpC. Mol. Microbiol. 2002, 43, 399–410.
batch fermentations:The additional effect of a changing mi- [25] Wacker, I., Ludwig, H., Reif, I., Blencke, H. M. et al., The reg-
croenvironment when using aqueous ammonia to control ulatory link between carbon and nitrogen metabolism in
pH. Biotechnol. Bioeng. 2003, 84, 474–484. Bacillus subtilis: Regulation of the gltAB operon by the
[8] Amanullah, A., McFarlane, C. M., Emery, A. N., Nienow, A., catabolite control protein CcpA. Microbiology 2003, 149,
Scale-down model to simulate spatial pH variations in 3001–3009.
large-scale bioreactors. Biotechnol. Bioeng. 2001, 73, 390– [26] De Mey, M., Taymaz-Nikerel, H., Baart, G., Waegeman, H. et
399. al., Catching prompt metabolite dynamics in Escherichia coli
with the BioScope at oxygen rich conditions. Metab. Eng.
2010, 12, 477–487.

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1017