Download Advanced continuous cultivation methods for systems microbiology

Microbiology (2015), 161, 1707–1719
Review
DOI 10.1099/mic.0.000146
Advanced continuous cultivation methods for
systems microbiology
Kaarel Adamberg,1,23 Kaspar Valgepea134 and Raivo Vilu1,3
Correspondence
1
Competence Center of Food and Fermentation Technologies, Akadeemia tee 15a,
12618 Tallinn, Estonia
2
Tallinn University of Technology, Department of Food Processing, Ehitajate tee 5,
19086 Tallinn, Estonia
3
Tallinn University of Technology, Department of Chemistry, Akadeemia tee 15,
12618 Tallinn, Estonia
Kaarel Adamberg
[email protected]
Increasing the throughput of systems biology-based experimental characterization of in silicodesigned strains has great potential for accelerating the development of cell factories. For this,
analysis of metabolism in the steady state is essential as only this enables the unequivocal
definition of the physiological state of cells, which is needed for the complete description and
in silico reconstruction of their phenotypes. In this review, we show that for a systems
microbiology approach, high-resolution characterization of metabolism in the steady state –
growth space analysis (GSA) – can be achieved by using advanced continuous cultivation
methods termed changestats. In changestats, an environmental parameter is continuously
changed at a constant rate within one experiment whilst maintaining cells in the physiological
steady state similar to chemostats. This increases the resolution and throughput of GSA
compared with chemostats, and, moreover, enables following of the dynamics of metabolism
and detection of metabolic switch-points and optimal growth conditions. We also describe the
concept, challenge and necessary criteria of the systematic analysis of steady-state metabolism.
Finally, we propose that such systematic characterization of the steady-state growth space of
cells using changestats has value not only for fundamental studies of metabolism, but also for
systems biology-based metabolic engineering of cell factories.
Introduction
An increasing number of genetically engineered recombinant cell factories with superior characteristics or totally
novel functions has led to the successful industrial-scale
production of food supplements, bulk chemicals and pharmaceuticals (Becker et al., 2011; Huang et al., 2012; Jantama
et al., 2008; Lee et al., 2012; Nakamura & Whited, 2003;
Paddon et al., 2013; Sun & Alper, 2015; Yim et al., 2011).
The development of such bioprocesses mainly involves
metabolic engineering of cells based on the knowledge
of their metabolism, use of genome engineering technologies and optimization of bioprocess conditions (e.g. pH,
aerobicity, cultivation mode). Additionally, various metabolic modelling techniques, mainly constraint-based
3These authors contributed equally to this paper.
4Present address: Australian Institute for Bioengineering and
Nanotechnology (AIBN), The University of Queensland, St Lucia, QLD
4072, Australia.
Abbreviations: ACS, acetyl-CoA synthetase; GAS, growth space
analysis; PTA, phosphotransacetylase; TCA, tricarboxylic acid.
000146 G 2015 The Authors
stoichiometric models, are increasingly and successfully
contributing to the creation of superior recombinant cells
(Bordbar et al., 2014; McCloskey et al., 2013). However,
even for the bacterium whose metabolism is best described,
Escherichia coli, the number of potentially superior producer
cells designed in silico exceeds the experimental throughput
for comprehensive characterization of their metabolism,
which is the best proof for validating in silico designs.
Reduction of the gap between throughput of in silico cell
design and experimental validation is important for accelerating systems microbiology-based bioprocess optimization.
This could potentially be achieved by a more accurate and
quantitative understanding of metabolism at the whole-cell
level through high-resolution experimental characterization
of steady-state metabolism of the in silico-designed cells.
Analysis of metabolism in the steady state is essential as
only this enables the unequivocal definition of the physiological state of cells (see below; Hoskisson & Hobbs, 2005),
which is needed for the complete description and in silico
reconstruction of their phenotypes. Steady-state cell culture
can be defined as the state of unchanging concentrations of
different molecules inside and outside of the cells, which
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Printed in Great Britain
1707
K. Adamberg, K. Valgepea and R. Vilu
Steady-state growth space analysis (GSA)
An accurate definition of the physiological state of cells is essential for a complete description and successful reproduction of
their phenotypes in silico. The physiological state of cells is
essentially determined by the concerted effect of numerous
different environmental parameters, e.g. temperature, pH,
oxygen concentration and nutrient availability, and the nearinfinite combinations of their values (Málek, 1958; Stephanopoulos et al., 2002). The effects of several environmental parameters on cell growth have been conceptually organized
into growth and non-growth spaces (Koseki, 2009; Le Marc
et al., 2005). In this review, we define the set of aforementioned
combinations, i.e. physiological states, where cells show growth
[specific growth rate (m) w0 h21] as the growth space. Importantly, such growth space can accurately be described by
steady-state analysis as only this enables the unequivocal definition of the physiological state of cells by providing a oneto-one correspondence between biochemical processes and
environmental conditions (Hoskisson & Hobbs, 2005).
We propose that a more complete and quantitative understanding of cell metabolism can be achieved by steady-state
GSA: systematic high-resolution characterization of the
steady-state growth space of cells through collecting systemslevel data in different physiological states and in silico analysis
of quantitative relationships between the combinations of
environmental parameters and characteristics of cell metabolism. In our view, GSA should form an important part of systems microbiology research.
Ideally, GSA should determine all the necessary cell characteristics (e.g. m, substrate and product flows, cellular composition) to understand the metabolism behind the
physiological states forming the surface of the growth
space. Fig. 1 illustrates an example of a 3D growth space
related to m as a function of pH and residual substrate
1708
co
nc
en
tra
Su
bs
tr
Substrate limitation
pH
ate
Substrate excess
tio
n
Chemostat
Turbidostat
A-stat
D-stat
Auxo-accelerostat
Specific growth rate (µ)
results from a one-to-one correspondence between biochemical processes and environmental conditions. Steady-state
metabolism could, in theory, be studied using batch cultures,
but in practice the rapid transient changes of substrate and
product concentrations do not allow an accurate definition
of the physiological state of cells (Hoskisson & Hobbs,
2005), and thus cannot be used to validate in silico designs.
Therefore, continuous culture methods chemostat and turbidostat are commonly used for steady-state cell physiology
studies (Bryson & Szybalski, 1952; Monod, 1950; Novick &
Szilard, 1950). However, these methods are too resourceexhaustive for high-resolution systems microbiology. In this
review, we describe the systematic analysis of steady-state
metabolism and summarize the research highlighting that
high-resolution characterization of steady-state metabolism
can be achieved by using advanced continuous cultivation
methods termed changestats, such as the accelerostat
(A-stat; Paalme et al., 1995) and dilution rate stat (D-stat;
Kasemets et al., 2003). We will not focus on the specific details
of control algorithms, which were presented earlier in
Kasemets et al. (2003).
Fig. 1. High-resolution steady-state GSA using different changestats. The surface was generated based on the Monod equation
(see main text; Monod, 1950) and near-to-real effects of pH and
substrate concentration on m. Note: residual substrate concentration starts inhibiting mmax above a certain concentration due to
osmotic stress. Blue and green points represent chemostat and
turbidostat experiments, respectively, which can be carried out
during one A-stat or auxo-accelerostat experiment, respectively.
Refer to text for details of the methods.
concentration, simply generated based on the Monod
equation (see below; Monod, 1950) and near-to-real effects
of pH and substrate concentration on m. The illustration is
just one possible case out of many and certainly more complex growth space surfaces based on more than three parameters can be envisaged. Although their visualization is
very challenging, these surfaces can be analysed in silico.
Ultimately, systematic GSA should contribute to the acceleration of systems biology-based metabolic engineering of
an organism into a cell factory.
Current challenge of steady-state GSA
GSA under defined steady-state conditions can be achieved
by application of continuous cultivation methods (e.g. chemostat and changestats), which enables the study of cells at
strictly defined physiological steady states with unchanging
concentrations of intra- and extracellular molecules. Whilst
continuous cultivation methods have been around for decades, analytical methods for high-throughput measurement
of the molecules relevant for quantitatively defining steady
states of metabolism have emerged quite recently. These
-omics methods enable the measurement, for instance, of
mRNAs (transcriptomics), proteins (proteomics), and
intracellular metabolites and extracellular compounds
(metabolomics). Whilst these methods are generally used
for relative comparison between two growth conditions,
absolute quantification of intracellular molecule abundances/concentrations is now also possible (Arike et al.,
2012; Esquerré et al., 2014; Ishii et al., 2007), thus enabling
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Microbiology 161
Advanced continuous cultivation methods
Three criteria for steady-state GSA
Ideally, continuous cultivation methods should meet three
basic criteria for the desired quality of steady-state GSA
described above: (i) be able to conduct experiments in
strictly defined physiological steady states through controlling
environmental conditions (e.g. temperature, pH, limiting substrate concentration), (ii) be able to change environmental
conditions within one experiment to cover a region, instead
of a single point, of the growth space, and (iii) be able to conduct GSA within a reasonable time frame and with reasonable
resources.
The first criterion is a priori achieved if a cultivation system
enabling the control of environmental parameters (e.g. temperature, pH) and substrate feeding rates (for both fluids
and gases) together with culture outflow is available. The
first application of such a system was the development of
the chemostat continuous cultivation method 65 years ago
allowing the study of cell physiology under steady-state conditions (Monod, 1950; Novick & Szilard, 1950). Whilst
Novick & Szilard (1950) developed the chemostat for following the emergence of spontaneous mutations under constant
conditions over a long period of time (weeks and months),
Monod (1950) designed his chemostat mainly for studying
growth kinetics of bacterial populations. Over decades, chemostats have been widely used for various applications,
which have been excellently reviewed previously (Bull, 2010;
Ferenci, 2007; Hoskisson & Hobbs, 2005) and thus will not
be discussed further in this review. Briefly, steady-state metabolism is achieved in continuous cultures by adding fresh
medium and removing cultivation broth from the cultivation
vessel at a constant rate, historically defined as the dilution rate
(D, h21): D5F/V, where F is the feed rate (l h21) and V the
volume of cultivation broth (l). Most importantly, one can
control m of the cells through D as under steady-state conditions the concentration of the growth-limiting substrate is
in a unique relationship with m. This can be described by the
Monod equation: m5mmax|s /(s +KS), where s is the residual
substrate concentration in the cultivation broth and KS is the
affinity constant to the particular substrate (Monod, 1950).
The second criterion for analysing steady-state metabolism at
various environmental conditions within one experiment
could be fulfilled by automated stepwise control of environmental parameters in a chemostat culture (Fig. 2). However,
such stepwise chemostats are very time-consuming due to
the need to stabilize the culture after each step change,
making them also prone to the emergence of unwanted
mutations (Ferenci, 2007; Gresham & Hong, 2015; Harder
et al., 1977; Helling et al., 1987; Novick & Szilard, 1950).
Both of these issues can be circumvented by using changestat
cultivation methods (e.g. the A-stat and D-stat), which enable
http://mic.sgmjournals.org
0.6
Chemostat
Stepwise chemostat
A-stat
0.5
Dilution rate (h–1)
GSA when coupled with continuous cultures. However, the
current challenge of steady-state GSA is increasing
the throughput of experimental characterization of the
growth space to match that of in silico cell design, strain
construction and -omics analyses.
0.4
0.3
0.2
Batch
0.1
Chemostat
0
0
50
100
150
Time (h)
200
250
Fig. 2. Comparison of time and resolution of the A-stat, stepwise
chemostat and single chemostats to scan through the specific
growth rate region 0.1–0.5 h21. A-stat, acceleration 0.01 h22.
Five working volumes assumed to reach steady state. Note: single
chemostats at similar dilution rates with stepwise chemostats
take a longer time as each separate chemostat experiment also
includes the batch growth phase (dotted line), whilst only one is
needed for the whole stepwise chemostat experiment.
the continuous change of one or several environmental parameters within a single experiment without needing the
long stabilization phases after each change (Fig. 1). Importantly, cells grown in changestats are formally in a quasi
steady state, defined here as a state where the cell culture is
moving continuously from one steady state to another one.
However, if the experiment is carried out properly, changestats describe steady-state physiology equally to chemostats
(see below). Hence, changestats are more suitable than chemostats for covering a region of the growth space of interest.
The third criterion for conducting GSA within a reasonable
time frame can be fulfilled by either the use of changestats
or automation of small-scale continuous cultivation systems. As outlined above, changestats save time through
not needing stabilization phases. The same can be achieved
by conducting chemostats in a multiplexed micro-scale
continuous culture platform, the development of which,
especially for micro-organisms, has recently seen considerable progress (Dénervaud et al., 2013; Long et al., 2013;
Moffitt et al., 2012). However, as all the three criteria for
realizing the desired level of steady-state GSA can be
achieved by using changestat methods, we conclude that
changestats are currently the most suitable for GSA.
Advanced continuous cultivation methods –
changestats
The unifying concept behind all different changestats,
which also differentiates them from a classical chemostat
or turbidostat, is the continuous change of one or several
environmental parameters within a single experiment,
whilst maintaining the physiological state comparable
with a chemostat or turbidostat. The ‘family’ of changestat
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
1709
K. Adamberg, K. Valgepea and R. Vilu
methods can be divided into two groups: chemostat-based
and turbidostat-based methods. Whilst the former enables
the study of cells at various m under nutrient limitation,
cells are grown at mmax and in substrate excess conditions
in the latter. Thus, steady-state growth both under nutrient
limitation and excess can be studied using changestat continuous cultivation methods, making them superior to
batch cultures for studying metabolism.
The most important benefit of using changestats for GSA is
following dynamic changes of steady-state metabolism with
the accurate detection of metabolic switch-points and optimal
growth conditions. For instance, in the shift of metabolism
from respiratory to respiro-fermentative growth, environmental conditions providing the best target product formation
characteristics and inhibition effects on substrate consumption can be determined reliably (see below for examples).
It was stated above that changestats describe steady-state
cell physiology equally well to chemostats if the experiments are carried out properly. In this particular context,
this means that the rate of change (acceleration/deceleration) of the environmental parameter of interest has to
maintain the cells in a state comparable with the steady
state (quasi steady state), meaning that cells are able to
adapt to changing environmental conditions. This can be
validated during a changestat experiment most easily by
stopping the acceleration/deceleration and setting the culture into chemostat mode: if growth characteristics do
not change then changestat data are equal to that of the
chemostat. The best evidence for changestat and chemostat
equally describing steady-state metabolism is provided by
the comparison of data from independent experiments of various organisms, including transcriptome and proteome data
(Albergaria et al., 2000, 2003; Barbosa et al., 2003, 2005;
Girbal et al., 2000; Nahku, 2012; Valgepea et al., 2010, 2011;
van der Sluis et al., 2001; van Dijken et al., 2000). It is also
important to note that the A-stat and D-stat have been
shown to be very reproducible (Barbosa et al., 2005; Lahtvee
et al., 2011; Valgepea et al., 2010, 2011; van der Sluis et al.,
2001; van Dijken et al., 2000).
Two factors have to be considered when planning a changestat
experiment because stress responses might be induced with
the change of environmental conditions. If the rate of
change of an environmental parameter is too fast, metabolism
will be disrupted and the quasi steady state lost. However, if
the rate of change is too slow, cells can specifically adapt to
changing conditions leading to phenotypes not seen in chemostats under identical environmental conditions. The latter was
seen in D-stats of Lactococcus lactis where pH was first
decreased from 6.3 to 5.4 and then increased back to 6.3
with a very slow rate of change (0.02 U h21) resulting in hysteresis of metabolism: activation of arginine degradation with
a decrease of pH to 5.4 was not deactivated at the end of the
experiment after pH had been increased back to the initial
pH of 6.3 (Lahtvee et al., 2009). Additionally, a very long changestat experiment due to slow acceleration risks the of emergence of unwanted mutations in the cell culture, similar to
1710
what is seen in chemostats (Ferenci, 2007; Gresham &
Hong, 2015; Harder et al., 1977; Helling et al., 1987; Novick
& Szilard, 1950). Importantly, it has been shown using
whole-genome DNA sequencing that no mutations arose
during a regular-length changestat study of cell physiology
in glucose-limited E. coli A-stat cultures if appropriate acceleration is used (Nahku et al., 2011).
Generally, the rate of change of an environmental parameter
supporting the characterization of physiological states in a
changestat experiment comparable with a chemostat depends
on mmax of the organism under study: the higher it is, the faster
the conditions can be changed. For an A-stat experiment,
appropriate acceleration can be estimated based on mmax of
the organism to be in the range of 0.01–0.04|mmax (Kasemets
et al., 2003). This is supported by several studies where mmax
was reached for different micro-organisms in A-stats (Adamberg et al., 2009; Barbosa et al., 2003; Kask et al., 1999; Paalme
et al., 1995, 1997a, b; Valgepea et al., 2010). A thorough study
of the effect of acceleration on the growth of L. lactis showed
that mmax of 0.59 h21 was reached in an A-stat whilst
maintaining the quasi steady state using accelerations
v0.005 h22 (Adamberg et al., 2009). Another study concluded that an acceleration of 0.001 h22 is the fastest that
can be used for approaching steady-state culture characteristics during A-stats with yeasts (van der Sluis et al., 2001).
Generally, A-stat data collected from experiments using accelerations in the range of 0.007–0.050|mmax are comparable
with chemostat data at various levels for many different organisms (Albergaria et al., 2000, 2003; Girbal et al., 2000; Lahtvee
et al., 2011; Nahku, 2012; Valgepea et al., 2010, 2011; van der
Sluis et al., 2001; van Dijken et al., 2000). That being said, one
is still advised to determine the exact rate of change for each
case where detailed characterization of a novel organism is
required (e.g. the last step in optimization of growth
conditions).
Next, we will highlight the literature applying changestats
for fundamental research of metabolism, bioprocess optimization and other purposes (summarized in Table 1).
Accelerostat (A-stat)
The first changestat – the A-stat – was developed in the
early 1990s at the National Institute of Chemical Physics
and Biophysics in Tallinn, Estonia by the research group
of Professor Raivo Vilu whilst studying the growth of
E. coli (Paalme et al., 1995). The A-stat has been the
most used method among changestats (Table 1) mainly
because it enables the study of the effects and dynamics
of m, one of the most important physiological parameters,
on cell metabolism. The A-stat is based on the stepwise
chemostat technique, but instead of increasing D stepwise
after achieving the initial steady state in the chemostat, a
continuous increase of D at constant acceleration is started
in the A-stat according to the equation: D5D0+aD|t,
where D0 is the initial D, aD is acceleration of D and t is
the time after start of acceleration of D. Identical to the
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Microbiology 161
Advanced continuous cultivation methods
Table 1. Literature of changestats of special interest with main results
Main results
Reference
Accelerostat (A-stat)
Escherichia coli
† Saturation of respiration capacity proposed to trigger acetate overflow at m ¼ 0.38 h21
† Disruption of acetate recycling in the phosphotransacetylase–acetyl-CoA synthetase
(PTA–ACS) node proposed to trigger acetate overflow at m ¼ 0.27 h21; A-stat and
chemostat comparable at transcriptome level; m-dependent metabolome, transcriptome
and proteome
† Acetate overflow postponed and reduced fourfold in a mutant strain with coordinated
activation of PTA– ACS and tricarboxylic acid cycles
† Cells achieve faster growth by increasing catalytic and translation rates of proteins
Saccharomyces cerevisiae
† Inter-laboratory study determined mcrit values for various different laboratory yeasts to
select a reference S. cerevisiae strain
† Mutant strain with derepressed glucose control displays 5% higher mcrit
† Supplementing mineral medium with oleic acid increases mcrit by 8%
Hanseniaspora guilliermondii
† Two-phase switch from fully respiratory to respiro-fermentative growth with acetate
excretion preceding ethanol; optimal range of m for high biomass yield
Zygosaccharomyces rouxii
† Crabtree effect detected under aerobic conditions; effect of acceleration on growth
characteristics also a function of the rate of change in environmental substrate
concentrations
Lactococcus lactis
† Continuous shift from mixed acid fermentation to homolactic fermentation with
faster growth; m-dependent metabolome, transcriptome and proteome
† Cells achieve faster growth by increasing catalytic and translation rates of proteins
Corynebacterium glutamicum
† Direct coupling between m and demeton-S-methyl biodegradation in co-metabolism
with fructose; A-stat comparable with De-stat
Dunaliella tertiolecta
† Optimized kinetic parameters for photobioreactor design and bioprocess conditions
for the production of vitamins and carotenoids
Deceleration-stat (De-stat)
Escherichia coli
† Determination of maintenance energy requirements near zero-growth conditions
† Stress response less apparent during a smooth change in nutrient availability compared
with nutrient shifts under fed-batch conditions
Rhodobacter capsulatus
† Optimization of photosynthetic efficiency for the production of hydrogen from acetate and
light energy by the determination of important components of the light energy balance
Saccharomyces cerevisiae
† Stronger stress response toward rapid changes in substrate concentration and D in the
chemostat compared with gradual changes in the De-stat
Thalassiosira pseudonana and Phaeodactylum tricornutum
† Up to 94% of time can be saved compared with chemostats using 10 times faster
decelerations than required for maintaining the quasi steady state whilst losing only
5 % in accuracy to estimate maximal biomass productivity rate
Dilution rate stat (D-stat)
Zygosaccharomyces rouxii
† Decreasing the threonine/methionine ratio in the ingoing medium proved that
methionol synthesis occurs only in the Ehrlich pathway
† Saccharomyces uvarum and Saccharomyces cerevisiae
† Dynamic effects of temperature on m, biomass yield and byproduct profiles; metabolic
events accompanying the transition of metabolism from carbon to nitrogen limitation
http://mic.sgmjournals.org
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Paalme et al. (1995)
Valgepea et al. (2010)
Peebo et al. (2014)
Valgepea et al. (2013)
van Dijken et al. (2000)
Klein et al. (1999)
Marc et al. (2013)
Albergaria et al. (2003)
van der Sluis et al. (2001)
Lahtvee et al. (2011)
Adamberg et al. (2012)
Girbal et al. (2000)
Barbosa et al. (2003), (2005)
Paalme et al. (1997b)
Teich et al. (1999)
Hoekema et al. (2006)
Nisamedtinov et al. (2008)
Hoekema et al. (2014)
van der Sluis et al. (2002)
Kasemets et al. (2003)
1711
K. Adamberg, K. Valgepea and R. Vilu
Table 1. cont.
Main results
Reference
Lactic acid bacteria
† Byproduct patterns of fermentative metabolism for various lactic acid bacteria depend
on the galactose/arginine ratio in the cultivation medium
Escherichia coli
† Maximal acetate co-utilization with glucose maintained up to D ¼ 0.2 h21, but totally
lost between D ¼ 0.45 and 0.5 h21
Yarrowia lipolytica
† Best nitrogen/carbon ratio in the cultivation medium for lipid production without
carbon loss into citric acid between 0.021 and 0.085 N-mol/C-mol
Auxo-accelerostats
Lactic acid bacteria
† Water activity is the most important environmental parameter affecting growth in
cheese-like conditions
† Determination of pH and temperature optima for achieving fastest growth of different
lactic acid bacteria
Saccharomyces cerevisiae
† Dose–effect curves (IC50) five times higher when the inhibitory aliphatic monocarboxylic
acid is added smoothly to the growth environment compared with when added rapidly
† Similar threshold levels for stress responses toward increased temperature, ethanol, salt
or organic acid concentration causing the decrease of mmax for a recombinant and
laboratory strain
Adaptastat
Escherichia coli
† Enables the study of cell metabolism near mmax under substrate limitation and the first
to produce nutrient-limited growth on two substrates
chemostat, the experimenter can control m of the cells in
the A-stat through D under steady-state representative conditions. The A-stat produces higher resolution data, and is
much more time- and resource-efficient compared with
chemostat approaches (Fig. 2).
Acetate overflow metabolism in E. coli
As the A-stat enables the precise determination of m-dependent metabolic switch-points and follows the metabolic
events around the switch-point with high resolution, there
has been a lot of interest in applying the A-stat for studying
acetate overflow metabolism in E. coli (i.e. excretion of acetate
above a certain m when growing on glucose).
Initially, saturation of respiratory chain capacity was proposed to cause acetate accumulation in cultivation broth
at high m (Paalme et al., 1995, 1997b). Later, the first
transcriptome study in an A-stat suggested that gene
expression patterns of the main acetate-producing (pta,
ackA, poxB) and -consuming enzymes (acs) might play a
role in the regulation of acetate overflow (Nahku et al.,
2010). Indeed, a systems biology study proposed that acetate overflow is triggered by downregulation of the enzyme
acetyl-CoA synthetase (ACS). This phenomenon can be
explained by disruption of acetate recycling in the phosphotransacetylase–acetyl-CoA synthetase (PTA–ACS)
node above m50.27 h21 [corresponding to a specific glucose
1712
Adamberg et al. (2006)
Valgepea et al. (2010)
Ochoa-Estopier & Guillouet (2014)
Laht et al. (2002)
Adamberg et al. (2003)
Kasemets et al. (2006)
Kasemets et al. (2007);
Nisamedtinov et al. (2008)
Tomson et al. (2006)
uptake rate qS54.2 mmol (g dry cell weight)21 h21], leading
to a decreasing capability of acetate reassimilation (Valgepea
et al., 2010). This hypothesis was recently proven as coordinated activation of the PTA–ACS and tricarboxylic acid
(TCA) cycles postponed the start of acetate overflow up to
qS56.0 mmol (g dry cell weight)21 h21 and reduced
carbon flow to acetate fourfold at mmax (Peebo et al.,
2014). Importantly, this overflow-reduced E. coli strain
does not accumulate any other detrimental byproduct
besides acetate and showed identical mmax compared with
the wild-type. Application of an A-stat approach was instrumental in these studies for proposing a novel hypothesis and
engineering more efficient strains as it enabled the precise
detection of the switch-point of acetate overflow and monitoring of metabolic events at high resolution.
Crabtree effect in yeast
The A-stat approach has been widely used due to its ability
to study m-dependent metabolic switch-points with high
resolution for detailed characterization of the Crabtree effect
in yeast (De Deken, 1966) where cells switch from fully
respiratory to respiro-fermentative growth above a certain m
(in Crabtree-positive yeasts), termed the critical m (mcrit).
One significant inter-laboratory study determined mcrit values
using A-stats for various different laboratory yeasts to
select a reference Saccharomyces cerevisiae strain amenable to
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Microbiology 161
Advanced continuous cultivation methods
experimental techniques used in genetic, physiological and
biochemical engineering research (van Dijken et al., 2000).
The A-stat approach has also been used for mcrit screening in
a study of an enological yeast strain, Saccharomyces uvarum,
used in wine production, which led to a purely oxidative
fed-batch process (Albergaria et al., 2000) and in a study of
S. cerevisiae for the development of a mathematical model
enabling the quantitative description of transient changes in
metabolism (Herwig et al., 2001). Two studies have investigated the effects of cultivation medium composition on mcrit
of S. cerevisiae: whilst one study concluded that supplementing glucose-based mineral medium with ethanol had no
effect on mcrit (Paalme et al., 1997a), the other study showed
an 8 % increase of mcrit on medium supplemented with oleic
acid (Marc et al., 2013). In addition, a S. cerevisiae doubleknockout strain engineered for derepression of glucose control displayed a 5 % higher mcrit compared with the wildtype, possibly explained by elevated respiratory capacity
(Klein et al., 1999). Interestingly, detailed m-dependent
characterization of the non-Saccharomyces yeast Hanseniaspora guilliermondii (Albergaria et al., 2003) and S. cerevisiae
(Kasemets et al., 2003) revealed a two-phase switch from
fully respiratory to respiro-fermentative growth with the
excretion of acetic acid preceding that of ethanol. Notably,
the same two-phase switch was seen in a series of 20 chemostats of S. cerevisiae (Postma et al., 1989).
Other studies of cell metabolism
The A-stat approach has also been used for m-dependent
characterization of several other metabolic processes in various organisms, e.g. energy and amino acid metabolism,
and regulation of gene expression. A-stat studies of these
and other metabolic processes are discussed next.
One of the first studies using A-stats determined several key
m-dependent physiological characteristics of E. coli growing
on several carbon sources (e.g. glucose, acetate, succinate
and Casamino acids) (Paalme et al., 1997b). Interestingly,
although the maximum specific respiration rates (qO2)
were very similar for all substrates, mmax values varied from
*0.2 to *0.8 h21. The authors concluded that mmax is
dependent on growth yield, and respiratory and glycolytic
capacity of the strain. In addition, analysis of A-stat data
with a stoichiometric flux model indicated that ATP production in E. coli exceeds the energy (theoretically) needed
for biomass synthesis twofold, consistent with earlier calculations (Stouthamer, 1973). Notably, the initial calculations
by Paalme et al. (1997b) were advanced by including the
futile PTA–ACS cycle detected in Valgepea et al. (2010) to
the model network, which revealed that energy spilling in
E. coli is dependent on m and strongly declines after the
start of acetate overflow (Valgepea et al., 2011). The apparent
discrepancy between reduced ATP spilling with rising m and a
constant biomass yield could be explained by the concomitant increase of carbon flux (wasting) to byproducts,
mainly acetate and pyrimidine synthesis pathway intermediates (Valgepea et al., 2011). The high-resolution data of
http://mic.sgmjournals.org
the A-stat was instrumental for detecting the dynamics of
metabolism in these studies.
Two studies of lactic acid bacteria show that the A-stat
approach is suitable for describing m-dependent dynamics of
amino acid metabolism. Firstly, metabolism of L. lactis shifts
from mixed-acid fermentation, with extensive consumption
of arginine, serine, asparagine and alanine, to homolactic fermentation, with balanced amino acid consumption, gradually
with the increase of m from 0.1 to 0.6 h–1 (Lahtvee et al., 2011).
Similarly, a continuous decrease of amino acid consumption
and increase of lactic acid production with faster growth is
seen in A-stat cultures of Lactobacillus plantarum (Kask
et al., 1999). Both of these studies support fine-tuning of
growth media for lactic acid bacteria, thus lowering costs of
expensive defined rich media used in bioprocesses and
enabling more accurate studies of amino acid metabolism
through their lower residual concentrations.
Integration of high-resolution steady-state cultures with
-omics measurements and metabolic modelling has strong
potential for leading to a more accurate and quantitative
understanding of metabolism at the whole-cell level. Such
a systems biology approach was recently applied for E. coli
(Valgepea et al., 2010) and L. lactis (Lahtvee et al., 2011)
by coupling A-stat cultivation with transcriptome, proteome and metabolome analyses, and metabolic flux analysis. Both studies represent a unique dataset of global
regulation of metabolism with rising m. For instance, shifts
in substrate utilization patterns (e.g. respiratory and fermentative) could be linked with gene expression dynamics
(see above). Also, the high mRNA/protein correlations
(R up to 0.8) observed in A-stats imply that the state of the culture for analysis (steady state versus non-steady state) could be
an important factor for mRNA/protein correlation determination. Further computational analysis of these data combined with absolute quantitative proteome analysis (Arike
et al., 2012) reveals that both micro-organisms with very
different metabolism, i.e. E. coli and L. lactis, use the same
principles in the regulation of gene expression levels
(mRNA, protein and metabolic fluxes). Furthermore, combining high-resolution A-stats with high-throughput -omics
analyses led to novel biological discoveries. In more detail, it
was shown using absolute quantitative transcriptomics and
proteomics that cells achieve faster growth through post-transcriptional control of protein abundances (changes in protein
levels are not strictly determined by changes in mRNA levels).
Secondly, and even more importantly, higher flux throughput
supporting faster growth is achieved through the increase of
apparent in vivo catalytic rates of enzymes (Adamberg et al.,
2012; Valgepea et al., 2013).
High-resolution m-dependent characterization of Zygosaccharomyces rouxii (a yeast used in soy sauce production)
metabolism shows that this yeast species also displays the
Crabtree effect under aerobic conditions and the effects of
acceleration on culture characteristics at specific D values
are not only a function of the metabolic adaptation rate of
the yeast, but also of the rate of change in environmental substrate concentrations during the A-stat (van der Sluis et al.,
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
1713
K. Adamberg, K. Valgepea and R. Vilu
2001). Another study investigating the evolution of physiological growth parameters of the yeast H. guilliermondii in an
A-stat detected an optimal range of m for achieving the highest biomass yield and further revealed that this non-Saccharomyces yeast continues to increase its oxygen uptake even
after switching from fully respiratory to respiro-fermentative metabolism (Albergaria et al., 2003).
Lastly, Lindley and colleagues used A-stats for investigating
the high efficiency of demeton-S-methyl (a pesticide and a
chemical warfare agent analogue) biodegradation by Corynebacterium glutamicum in co-metabolism with fructose
and its dependence on m (Girbal et al., 2000). They discovered a direct coupling between the specific demeton-Smethyl consumption rate and m in the range of 0.1–
0.4 h21. Following high-cell-density cultures demonstrated
the potential of using the latter bacterium for degradation
of demeton-S-methyl in industrial processes.
All the latter studies demonstrate the benefits of applying
high-resolution A-stats for determining m-dependent
switches and dynamics in metabolism. These examples are
also useful for bioprocess development due to the industrial
relevance of these micro-organisms. Next, direct application
of A-stat for bioprocess optimization is reviewed.
Bioprocess development
The Dutch group led by Professor R. H. Wijffels has used the
A-stat approach for the optimization of kinetic parameters of
photobioreactors for microalgae cultivation (Barbosa et al.,
2003) and bioprocess conditions for vitamin and carotenoid
synthesis by Dunaliella tertiolecta (Barbosa et al., 2005).
Notably, the A-stat yields comparable data with chemostats
even in such systems where the limiting nutrient is not
evenly distributed within the bioreactor, and is thus suitable
as a fast and accurate tool for the determination of kinetic
parameters and optimization of cultivation conditions (e.g.
for maximal biomass volumetric productivity) in specific
photobioreactors (Barbosa et al., 2003, 2005). Notably, the
benefits of the A-stat approach were evident in optimization
of production yields of vitamins C and E, and carotenoids
lutein and b-carotene in microalgae D. tertiolecta, as optimal
yields were achieved at different light intensities for each
product (Barbosa et al., 2005).
Deceleration-stat (De-stat)
The changestat method deceleration-stat (termed De-stat
in this review) is in principle an A-stat with the only difference that D is decreased at a constant rate (opposite to the
A-stat). Although the first study to implement De-stat was
Paalme et al. (1997b) for the determination of maintenance
energy requirements in E. coli, the De-stat approach has
been most extensively used by Professor R. H. Wijffels’
group. An important advantage of the De-stat compared
with the A-stat is the time saved because of the shorter
time period needed to attain the initial steady state at
higher m and the use of a faster rate of change of D
1714
(Hoekema et al., 2014). Notably, a De-stat has been
shown to yield comparable results with A-stats in the
study of demeton-S-methyl biodegradation capability by
C. glutamicum (Girbal et al., 2000).
Wijffels and colleagues first used the De-stat approach for
optimizing the photosynthetic efficiency of the purple nonsulphur bacterium Rhodobacter capsulatus for the production
of hydrogen from acetate and light energy by determining
important components of the light energy balance: biomass
growth and maintenance, generation of hydrogen, and photosynthetic heat dissipation (Hoekema et al., 2006). They saw
that culture characteristics were strongly dependent on m.
In another De-stat study, the maximum photosynthetic
yield of green microalgae was investigated, revealing that
photobioreactor designs should provide relatively high
specific light supply rates as biomass yield at high biomass
densities decreases at low light supply rates due to high
demands for biomass maintenance (Zijffers et al., 2010).
Additionally, Professor R. H. Wijffels’ group has developed
a simulation for estimating maximal productivity of algal biomass based on the De-stat data, including cases where deceleration exceeded the rate at which the quasi steady state
could be maintained (Hoekema et al., 2014). They showed
that up to 94 % of the time can be saved compared with chemostats using 10 times faster decelerations than required for
maintaining the quasi steady state whilst losing only 5 % in
accuracy to estimate the maximal biomass productivity rate.
Two studies have utilized the De-stat approach to specifically
study stress responses. Firstly, Neubauer and colleagues studied the stringent and general stress response during entry of
E. coli into glucose starvation (Teich et al., 1999). Whilst they
observed clear responses of corresponding response regulators
(ppGpp and s S components) to nutrient shifts under fedbatch conditions, the concerted reaction of ppGpp and s S
was less apparent during a smooth change in nutrient availability in De-stats. Secondly, the response of S. cerevisiae to a
sudden or gradual decrease in glucose concentration and D
has been followed based on the expression of the general
stress response protein Hsp12p (Nisamedtinov et al., 2008).
Similar to Teich et al. (1999), Nisamedtinov et al. (2008)
detected a stronger stress response, through protein Hsp12p
expression, toward rapid changes in substrate concentration
and D in a chemostat compared with gradual changes in Destat. Application of the De-stat approach in both studies led
to the suggestion that stress response mechanisms in cells
have adapted to rapid changes in environmental conditions.
Dilution rate stat (D-stat)
The D-stat approach is the most powerful if it is planned to
study the impact of an environmental parameter (including
cultivation medium composition) on cell physiology at constant m. A range of values at high resolution can be screened
within one experiment, thus more potentially detecting optimal growth conditions (Fig. 1). Although the D-stat was first
used to study the Ehrlich pathway in the yeast Z. rouxii (van
der Sluis et al., 2002), D-stat characteristics were first
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Microbiology 161
Advanced continuous cultivation methods
thoroughly formulated (Kasemets et al., 2003) and later
most extensively practiced by the groups of Professor
R. Vilu and Professor T. Paalme in Tallinn, Estonia. The concept of the D-stat is simple: one environmental parameter
(e.g. temperature, pH or cultivation medium composition)
is smoothly changed whilst D and other parameters are
kept constant according to the equation: N5N0+a|t,
where N is the parameter being changed, N0 is the initial
value of the parameter being changed, a is the rate of
change of parameter N and t is time (Kasemets et al., 2003).
The D-stat approach is suitable for determining critical values
and limits of both environmental parameters (e.g. temperature, pH) and composition of cultivation medium (e.g. relative nitrogen or vitamin concentration) that affect m. For
instance, D-stats with a smooth increase in temperature
detected the dynamic effects of temperature on m, biomass
yield and byproduct profiles of S. uvarum (Kasemets et al.,
2003). The same study also utilized a D-stat to determine
the metabolic events accompanying the transition of metabolism from carbon to nitrogen limitation in S. cerevisiae by
smoothly decreasing the nitrogen/carbon ratio in the ingoing
medium. Similarly, analysis of the effect(s) of pH and/or temperature on L. lactis growth physiology showed dependence of
both biomass yield and lactate production on pH and temperature (Lahtvee et al., 2009). Additionally, substrate consumption and product formation patterns, and gene
expression showed no difference between increasing or
decreasing pH within the range of 5.4–6.3. In addition to
using a De-stat approach (see above), Paalme and colleagues
also applied a D-stat to study the gradual stress response of
S. cerevisiae to salt, ethanol or high temperature through protein Hsp12p expression (Nisamedtinov et al., 2008). High-resolution D-stats at D50.09 h21 exactly detected the threshold
values for the stress response for each treatment: expression of
Hsp12p increased above temperature 34 uC, salt concentration 185 mM or ethanol concentration 0.54 mM.
The D-stat approach is possibly the most useful, especially
regarding bioprocess development, for studying the consumption patterns of different substrates. In such cases, after
steady state in a chemostat has been achieved on one
medium, feeding of another medium with a different composition [added, removed or modified concentration of component(s)] is started and increased in time whilst
simultaneously decreasing the feeding rate of the initial
medium to keep D constant. These D-stats have yielded the
following main results: (i) decreasing the threonine/methionine ratio in the ingoing medium of Z. rouxii cultivations
revealed that methionol synthesis occurs in the Ehrlich pathway (van der Sluis et al., 2002); (ii) byproduct patterns of fermentative metabolism for various lactic acid bacteria depend
on the galactose/arginine ratio in the cultivation medium
(Adamberg et al., 2006); (iii) maximal acetate consumption
of *35 mmol (g dry cell weight)21 by E. coli under glucose
limitation is maintained up to D50.2 h21,whilst co-utilization is totally lost between D50.45 and 0.5 h–1 (Valgepea
et al., 2010); and (iv) the best ratio nitrogen/carbon ratio in
the cultivation medium for producing lipids without carbon
http://mic.sgmjournals.org
loss into citric acid by Yarrowia lipolytica is between 0.021
and 0.085 N-mol/C-mol (Ochoa-Estopier & Guillouet, 2014).
Auxo-accelerostats
The A-stat and D-stat approaches are suitable for high-resolution study of steady-state cell physiology, but only under
nutrient limitation. To study the effects of environmental parameters and medium components on the metabolic switchpoints and optimal growth conditions of cells at maximal
specific growth rates, the auxo-accelerostat approach is the
most suitable (Fig. 1). The classical method for studying
cells in the steady state under substrate excess conditions is
the turbidostat (Bryson & Szybalski, 1952) where cells are
forced to grow with mmax by adjusting D based on biomass
concentration (with optical density as the indicator) using
the following simple logic: if the optical density is below or
above the set-point value, D is decreased or increased, respectively. In addition to culture turbidity, cells can be forced to
grow with mmax through feedback from pH (pH-auxostat;
Martin & Hempfling, 1976), CO2 (CO2-auxostat; Watson,
1969), electrical capacitance of the culture (permittistat;
Markx et al., 1991) or other parameters. However, turbidostat
and auxostats are not suitable for high-resolution studies of
steady-state growth space under nutrient excess as, similar
to the chemostat, the culture has to be stabilized after every
shift in environmental conditions. Therefore, auxo-accelerostats were developed (Kasemets et al., 2003) for achieving
higher resolution through smoothly changing one environmental parameter under nutrient excess conditions.
An auxo-accelerostat culture is controlled online by the experimenter through a parameter that is either directly (e.g. optical
density) or indirectly [e.g. pH, percentage of dissolved oxygen
concentration in bioreactor (pO2%), percentage of CO2 in the
off-gas] related to cell growth. The equation for operating an
auxo-accelerostat is the same as for the D-stat (see above).
Several studies have demonstrated the effectiveness of auxoaccelerostats for screening out optimal or inhibitory conditions
for the growth of yeasts and lactic acid bacteria. Applying various algorithms of the auxo-accelerostat for studying the effects
of pH, temperature, salt concentration and water activity on the
growth of the cheese bacterium Lactobacillus paracasei revealed
that water activity is the most important environmental parameter affecting the growth of cells under cheese-like conditions (Laht et al., 2002). Interestingly, the response curve of
slowly growing L. paracasei (mmax50.3 h21) to pH and temperature was rather flat, whilst fast-growing Streptococcus thermophilus (mmax52.2 h21) showed a narrow range of
temperature (44 uC) and pH (6.6) for growth at mmax (Adamberg et al., 2003).
Auxo-accelerostats have also been used to study stress
responses of yeast in detail. Characterizing the toxic effects
of aliphatic monocarboxylic acids (e.g. acetic, formic and
propionic acids) on S. cerevisiae growth using CO2-auxoaccelerostats revealed an immediate decline of m and growth
yield with slowly increasing acid concentrations (Kasemets
et al., 2006). Furthermore, they showed that the dose–effect
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
1715
K. Adamberg, K. Valgepea and R. Vilu
curves (IC50; acid concentration causing a 50 % decrease in
either m or growth yield) are five times higher when the inhibitory acid is added smoothly to the growth environment (as in
the auxo-accelerostat) compared with when added rapidly (by
pulse). Two studies using auxo-accelerostats and metabolic
modelling for studying the effect of smoothly changing
stress conditions (increasing temperature, ethanol, salt or
organic acid concentration) on a laboratory and a recombinant S. cerevisiae strain determined similar threshold levels
of stress responses causing the decrease of mmax (Kasemets
et al., 2007; Nisamedtinov et al., 2008). Interestingly, these
threshold values determined under nutrient excess and mmax
were very similar to those of glucose-limited D-stats at
D50.09 h–1 (Nisamedtinov et al., 2008). In conclusion, the
latter studies highlight that high-resolution auxo-accelerostats
can successfully be used for determination of the quantitative
effects of environmental conditions on growth characteristics.
Adaptastat
To further complement the array of changestat methods, the
adaptastat was developed for studying cells near mmax under
substrate limitation in aerobic cultures a decade ago in the
National Institute of Chemical Physics and Biophysics in
Tallinn, Estonia by Dr Kalju Vanatalu. In an adaptastat, m
of cells is raised stepwise until near mmax through increasing
D according to activation of oxygen consumption by the
micro-organism (Tomson et al., 2006). More specifically,
D is controlled through a feedback loop as follows: after
attaining steady state in a chemostat, feeding (D) is abruptly
increased followed by stop (or reduction) of nutrient inflow.
Next, the time needed to exhaust the residual substrate is calculated by measuring the rise of the dissolved oxygen concentration. The shorter the time, the faster the growth.
Subsequently, D is increased or decreased depending on
the ratio of the feeding pulse and substrate exhaustion durations. The adaptastat is an attractive method as it enables
the conversion of a chemostat culture to near mmax conditions whilst maintaining substrate limitation. The adaptastat is also the first method to produce nutrient-limited
growth on two substrates. The method should be efficient
for isotopic labelling of bacterial cultures and their components produced abundantly at fast growth (e.g. ribosomes
and RNA).
Conclusion
Currently, the number of recombinant cells designed in
silico exceeds the throughput of comprehensive experimental characterization of their metabolism, which is vital for
validating the in silico predictions. More complete description and in silico reconstruction of their phenotypes has
great potential for accelerating systems microbiologybased bioprocess optimization. With this review, we hope
to have presented that a more accurate and quantitative
understanding of metabolism could be achieved through
steady-state GSA utilizing high-resolution changestat cultivation methods.
1716
The changestat methods have several advantages over the
widely used batch and chemostat methods, and provide
unique knowledge about shifts in metabolism and optimal
growth conditions. Thus, we propose that systematic highresolution characterization of the steady-state growth space
of cells using changestats should not only be used for fundamental studies of metabolism, but also incorporated into
the systems microbiology-based metabolic engineering
pipeline (Fig. 3) (Van Dien, 2013). Comprehensive systems
microbiology through coupling advanced continuous cultivation methods, -omics technologies and metabolic modelling could lead to more efficient cell designs. Notably,
remarkable interest in continuous cultivations has recently
emerged in the pharmaceutical industry due to sustained
operation with consistent product quality, reduced equipment size, high volumetric productivity, streamlined process flow, low process cycle times, and reduced capital
and operating costs (Konstantinov & Cooney, 2015).
In the future, more automated control and higher throughput
of changestats is required to achieve their wider acceptance
and use. One approach to accomplish this is the use of iterative
control algorithms during a changestat experiment through
continuous modification of the rate of change of the environmental parameter according to adaptation of cells. This
approach would enable the change of the direction of
growth space scanning and change of several environmental
parameters simultaneously. There already exists an approach
to increase the throughput of changestats: a method enabling
the multiplication of the steady state by a sequential parallel
cultivation system, termed the MD (‘mother–daughter’)
system (Erm et al., 2014). Higher throughput in the MD
-omics
analyses
Changestat
experiment
GSA
Wild-type
cell
In silico
modeling
Superior cell
factory
Genetic
engineering
Fig. 3. Systems microbiology-based pipeline of metabolic engineering of superior cell factories.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Microbiology 161
Advanced continuous cultivation methods
system is achieved by distributing the steady-state culture
from the ‘mother’ vessel to several ‘daughter’ vessels without
disturbing the physiological state of cells, thus allowing changestats to be carried out in each ‘daughter’ vessel after the
transfer. Hopefully, other successful developments will follow.
array enables the spatio-temporal analysis of the yeast proteome. Proc
Natl Acad Sci U S A 110, 15842–15847.
Acknowledgements
transcript stability in the regulation of gene expression in
Escherichia coli cells cultured on glucose at different growth rates.
Nucleic Acids Res 42, 2460–2472.
The authors thank Andrus Seiman for help with preparation of
figures. The financial support for this work was provided by the European Regional Development Fund project EU29994, and Estonian
Ministry of Education and Research institutional research (IUT
1927) and personal (G9192) funding.
Erm, S., Adamberg, K. & Vilu, R. (2014). Multiplying steady-state
culture in multi-reactor system. Bioprocess Biosyst Eng 37, 2361–2370.
Esquerré, T., Laguerre, S., Turlan, C., Carpousis, A. J., Girbal, L. &
Cocaign-Bousquet, M. (2014). Dual role of transcription and
Ferenci, T. (2007). Bacterial physiology, regulation and mutational
adaptation in a chemostat environment. Adv Microb Physiol 53,
169–229.
Girbal, L., Rols, J.-L. & Lindley, N. D. (2000). Growth rate influences
References
Adamberg, K., Kask, S., Laht, T. M. & Paalme, T. (2003). The effect of
reductive biodegradation of the organophosphorus pesticide
demeton by Corynebacterium glutamicum. Biodegradation 11,
371–376.
temperature and pH on the growth of lactic acid bacteria: a pHauxostat study. Int J Food Microbiol 85, 171–183.
Gresham, D. & Hong, J. (2015). The functional basis of adaptive
Adamberg, K., Adamberg, S., Laht, T.-M., Ardö, Y. & Paalme, T.
(2006). Study of cheese associated lactic acid bacteria under
Harder, W., Kuenen, J. G. & Matin, A. (1977). A review. Microbial
carbohydrate-limited conditions using D-Stat cultivation. Food
Biotechnol 20, 143–160.
Adamberg, K., Lahtvee, P.-J., Valgepea, K., Abner, K. & Vilu, R.
(2009). Quasi steady state growth of Lactococcus lactis in glucose-
limited acceleration stat (A-stat) cultures. Antonie van Leeuwenhoek
95, 219–226.
Adamberg, K., Seiman, A. & Vilu, R. (2012). Increased biomass yield
of Lactococcus lactis by reduced overconsumption of amino acids and
increased catalytic activities of enzymes. PLoS One 7, e48223.
Albergaria, H., Duarte, L. C., Amaral-Collaço, M. T. & Gı́rio, F. M.
(2000). Study of Saccharomyces uvarum CCMI 885 physiology
under fed-batch, chemostat and accelerostat cultivation techniques.
Food Technol Biotechnol 38, 33–38.
Albergaria, H., Torrão, A. R., Hogg, T. & Gı́rio, F. M. (2003). Physiological
evolution in chemostats. FEMS Microbiol Rev 39, 2–16.
selection in continuous culture. J Appl Bacteriol 43, 1–24.
Helling, R. B., Vargas, C. N. & Adams, J. (1987). Evolution of
Escherichia coli during growth in a constant environment. Genetics
116, 349–358.
Herwig, C., Marison, I. & von Stockar, U. (2001). On-line
stoichiometry and identification of metabolic state under dynamic
process conditions. Biotechnol Bioeng 75, 345–354.
Hoekema, S., Douma, R. D., Janssen, M., Tramper, J. & Wijffels, R. H.
(2006). Controlling light-use by Rhodobacter capsulatus continuous
cultures in a flat-panel photobioreactor. Biotechnol Bioeng 95,
613–626.
Hoekema, S., Rinzema, A., Tramper, J., Wijffels, R. H. & Janssen, M.
(2014). Deceleration-stats save much time during phototrophic
culture optimization. Biotechnol Bioeng 111, 792–802.
behaviour of Hanseniaspora guilliermondii in aerobic glucose-limited
continuous cultures. FEMS Yeast Res 3, 211–216.
Hoskisson, P. A. & Hobbs, G. (2005). Continuous culture - making a
Arike, L., Valgepea, K., Peil, L., Nahku, R., Adamberg, K. & Vilu, R.
(2012). Comparison and applications of label-free absolute proteome
Huang, C.-J., Lin, H. & Yang, X. (2012). Industrial production of
quantification methods on Escherichia coli. J Proteomics 75, 5437–5448.
Barbosa, M. J., Hoogakker, J. & Wijffels, R. H. (2003). Optimisation of
cultivation parameters in photobioreactors for microalgae cultivation
using the A-stat technique. Biomol Eng 20, 115–123.
Barbosa, M. J., Zijffers, J.-W. F., Nisworo, A., Vaes, W., van
Schoonhoven, J. & Wijffels, R. H. (2005). Optimization of biomass,
vitamins, and carotenoid yield on light energy in a flat-panel
reactor using the A-stat technique. Biotechnol Bioeng 89, 233–242.
Becker, J., Zelder, O., Häfner, S., Schröder, H. & Wittmann, C. (2011).
From zero to hero - design-based systems metabolic engineering of
Corynebacterium glutamicum for l-lysine production. Metab Eng 13,
159–168.
Bordbar, A., Monk, J. M., King, Z. A. & Palsson, B. Ø. (2014).
Constraint-based models predict metabolic and associated cellular
functions. Nat Rev Genet 15, 107–120.
Bryson, V. & Szybalski, W. (1952). Microbial selection. Science 116,
45–51.
Bull, A. T. (2010). The renaissance of continuous culture in the post-
genomics age. J Ind Microbiol Biotechnol 37, 993–1021.
comeback? Microbiology 151, 3153–3159.
recombinant therapeutics in Escherichia coli and its recent
advancements. J Ind Microbiol Biotechnol 39, 383–399.
Ishii, N., Nakahigashi, K., Baba, T., Robert, M., Soga, T., Kanai, A.,
Hirasawa, T., Naba, M., Hirai, K. & other authors (2007). Multiple
high-throughput analyses monitor the response of E. coli to
perturbations. Science 316, 593–597.
Jantama, K., Zhang, X., Moore, J. C., Shanmugam, K. T., Svoronos,
S. A. & Ingram, L. O. (2008). Eliminating side products and
increasing succinate yields in engineered strains of Escherichia coli
C. Biotechnol Bioeng 101, 881–893.
Kasemets, K., Drews, M., Nisamedtinov, I., Adamberg, K. & Paalme,
T. (2003). Modification of A-stat for the characterization of
microorganisms. J Microbiol Methods 55, 187–200.
Kasemets, K., Kahru, A., Laht, T.-M. & Paalme, T. (2006). Study of the
toxic effect of short- and medium-chain monocarboxylic acids on the
growth of Saccharomyces cerevisiae using the CO2-auxo-accelerostat
fermentation system. Int J Food Microbiol 111, 206–215.
Kasemets, K., Nisamedtinov, I., Laht, T.-M., Abner, K. & Paalme, T.
(2007). Growth characteristics of Saccharomyces cerevisiae S288C in
De Deken, R. H. (1966). The Crabtree effect: a regulatory system in
changing environmental conditions:
Antonie van Leeuwenhoek 92, 109–128.
yeast. J Gen Microbiol 44, 149–156.
Kask, S., Laht, T.-M., Pall, T. & Paalme, T. (1999). A study on growth
Dénervaud, N., Becker, J., Delgado-Gonzalo, R., Damay, P., Rajkumar,
A. S., Unser, M., Shore, D., Naef, F. & Maerkl, S. J. (2013). A chemostat
characteristics and nutrient consumption of Lactobacillus plantarum
in A-stat culture. Antonie van Leeuwenhoek 75, 309–320.
http://mic.sgmjournals.org
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
auxo-accelerostat
study.
1717
K. Adamberg, K. Valgepea and R. Vilu
Klein, C. J. L., Rasmussen, J. J., Rønnow, B., Olsson, L. & Nielsen, J.
(1999). Investigation of the impact of MIG1 and MIG2 on the
Nisamedtinov, I., Lindsey, G. G., Karreman, R., Orumets, K.,
Koplimaa, M., Kevvai, K. & Paalme, T. (2008). The response of the
physiology of Saccharomyces cerevisiae. J Biotechnol 68, 197–212.
yeast Saccharomyces cerevisiae to sudden vs. gradual changes in
environmental stress monitored by expression of the stress response
protein Hsp12p. FEMS Yeast Res 8, 829–838.
Konstantinov, K. B. & Cooney, C. L. (2014). White Paper on
Continuous Bioprocessing. May 20-21,
Symposium. J Pharm Sci 104, 813–820.
2014
Continuous
Novick, A. & Szilard, L. (1950). Description of the chemostat. Science
Koseki, S. (2009). Microbial Responses Viewer (MRV): a new
112, 715–716.
ComBase-derived database of microbial responses
environments. Int J Food Microbiol 134, 75–82.
food
Ochoa-Estopier, A. & Guillouet, S. E. (2014). D-stat culture for
Laht, T.-M., Kask, S., Elias, P., Adamberg, K. & Paalme, T. (2002).
studying the metabolic shifts from oxidative metabolism to lipid
accumulation and citric acid production in Yarrowia lipolytica.
J Biotechnol 170, 35–41.
to
Role of arginine in the development of secondary microflora in
Swiss-type cheese. Int Dairy J 12, 831–840.
Lahtvee, P.-J., Valgepea, K., Nahku, R., Abner, K., Adamberg, K. &
Vilu, R. (2009). Steady state growth space study of Lactococcus lactis
in D-stat cultures. Antonie van Leeuwenhoek 96, 487–496.
Lahtvee, P.-J., Adamberg, K., Arike, L., Nahku, R., Aller, K. & Vilu, R.
(2011). Multi-omics approach to study the growth efficiency and
Paalme, T., Kahru, A., Elken, R., Vanatalu, K., Tiisma, K. & Vilu, R.
(1995). The computer-controlled continuous culture of Escherichia
coli with smooth change of dilution rate (A-stat). J Microbiol
Methods 24, 145–153.
Paalme, T., Elken, R., Vilu, R. & Korhola, M. (1997a). Growth
amino acid metabolism in Lactococcus lactis at various specific
growth rates. Microb Cell Fact 10, 12.
efficiency of Saccharomyces cerevisiae on glucose/ethanol media with
a smooth change in the dilution rate (A-stat). Enzyme Microb
Technol 20, 174–181.
Le Marc, Y., Pin, C. & Baranyi, J. (2005). Methods to determine the
Paalme, T., Elken, R., Kahru, A., Vanatalu, K. & Vilu, R. (1997b). The
growth domain in a multidimensional environmental space. Int J
Food Microbiol 100, 3–12.
growth rate control in Escherichia coli at near to maximum growth
rates: the A-stat approach. Antonie van Leeuwenhoek 71, 217–230.
Lee, J. W., Na, D., Park, J. M., Lee, J., Choi, S. & Lee, S. Y. (2012).
Paddon, C. J., Westfall, P. J., Pitera, D. J., Benjamin, K., Fisher, K.,
McPhee, D., Leavell, M. D., Tai, A., Main, A. & other authors
(2013). High-level semi-synthetic production of the potent
Systems metabolic engineering of microorganisms for natural and
non-natural chemicals. Nat Chem Biol 8, 536–546.
Long, Z., Nugent, E., Javer, A., Cicuta, P., Sclavi, B., Cosentino
Lagomarsino, M. & Dorfman, K. D. (2013). Microfluidic chemostat for
measuring single cell dynamics in bacteria. Lab Chip 13, 947–954.
Málek, I. (1958). The physiological state of microorganisms during
continuous culture. In Continuous Culture of Microorganisms:
A Symposium, pp. 21. Prague: Publishing House ASCR.
Marc, J., Feria-Gervasio, D., Mouret, J.-R. & Guillouet, S. E. (2013).
Impact of oleic acid as co-substrate of glucose on short and longterm Crabtree effect in Saccharomyces cerevisiae. Microb Cell Fact 12, 83.
Markx, G. H., Davey, C. L. & Kell, D. B. (1991). The permittistat:
a novel type of turbidostat. J Gen Microbiol 137, 735–743.
Martin, G. A. & Hempfling, W. P. (1976). A method for the regulation
of microbial population density during continuous culture at high
growth rates. Arch Microbiol 107, 41–47.
McCloskey, D., Palsson, B. Ø. & Feist, A. M. (2013). Basic and applied
uses of genome-scale metabolic network reconstructions of
Escherichia coli. Mol Syst Biol 9, 661.
Moffitt, J. R., Lee, J. B. & Cluzel, P. (2012). The single-cell chemostat:
an agarose-based, microfluidic device for high-throughput, single-cell
studies of bacteria and bacterial communities. Lab Chip 12, 1487–1494.
Monod, J. (1950). [The technique of continuous culture, theory and
applications]. Ann Inst Pasteur (Paris) 79, 390–410, (in French).
Nahku, R. (2012). Validation of Critical Factors for the Quantitative
Characterization of Bacterial Physiology in Accelerostat Cultures.
Tallinn: TUT Press.
Nahku, R., Valgepea, K., Lahtvee, P.-J., Erm, S., Abner, K., Adamberg,
K. & Vilu, R. (2010). Specific growth rate dependent transcriptome
profiling of Escherichia coli K12 MG1655 in accelerostat cultures.
J Biotechnol 145, 60–65.
antimalarial artemisinin. Nature 496, 528–532.
Peebo, K., Valgepea, K., Nahku, R., Riis, G., Õun, M., Adamberg, K. &
Vilu, R. (2014). Coordinated activation of PTA-ACS and TCA cycles
strongly reduces overflow metabolism of acetate in Escherichia coli.
Appl Microbiol Biotechnol 98, 5131–5143.
Postma, E., Verduyn, C., Scheffers, W. A. & Van Dijken, J. P. (1989).
Enzymic analysis of the crabtree effect in glucose-limited chemostat
cultures of Saccharomyces cerevisiae. Appl Environ Microbiol 55,
468–477.
Stephanopoulos, G., Misra, J., Hwang, D., Schmitt, W., Alevizos, I.,
Silva, S. & Gill, R. (2002)., Defining biological states and related
genes, proteins and patterns US Patent US 2002/0169562 A1.
Stouthamer, A. H. (1973). A theoretical study on the amount of ATP
required for synthesis of microbial cell material. Antonie van
Leeuwenhoek 39, 545–565.
Sun, J. & Alper, H. S. (2015). Metabolic engineering of strains: from
industrial-scale to lab-scale chemical production. J Ind Microbiol
Biotechnol 42, 423–436.
Teich, A., Meyer, S., Lin, H. Y., Andersson, L., Enfors, S. & Neubauer, P.
(1999). Growth rate related concentration changes of the starvation
response regulators sigmaS and ppGpp in glucose-limited fed-batch
and continuous cultures of Escherichia coli. Biotechnol Prog 15, 123–129.
Tomson, K., Barber, J. & Vanatalu, K. (2006). Adaptastat – a new
method for optimising of bacterial growth conditions in continuous
culture: Interactive substrate limitation based on dissolved oxygen
measurement. J Microbiol Methods 64, 380–390.
Valgepea, K., Adamberg, K., Nahku, R., Lahtvee, P.-J., Arike, L. & Vilu, R.
(2010). Systems biology approach reveals that overflow metabolism of
acetate in Escherichia coli is triggered by carbon catabolite repression of
acetyl-CoA synthetase. BMC Syst Biol 4, 166.
Nahku, R., Peebo, K., Valgepea, K., Barrick, J. E., Adamberg, K. & Vilu,
R. (2011). Stock culture heterogeneity rather than new mutational
Valgepea, K., Adamberg, K. & Vilu, R. (2011). Decrease of energy
variation complicates short-term cell physiology studies of Escherichia
coli K-12 MG1655 in continuous culture. Microbiology 157, 2604–2610.
spilling in Escherichia coli continuous cultures with rising specific
growth rate and carbon wasting. BMC Syst Biol 5, 106.
Nakamura, C. E. & Whited, G. M. (2003). Metabolic engineering for
Valgepea, K., Adamberg, K., Seiman, A. & Vilu, R. (2013). Escherichia
the microbial production of 1,3-propanediol. Curr Opin Biotechnol
14, 454–459.
coli achieves faster growth by increasing catalytic and translation rates
of proteins. Mol Biosyst 9, 2344–2358.
1718
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
Microbiology 161
Advanced continuous cultivation methods
van der Sluis, C., Westerink, B. H., Dijkstal, M. M., Castelein, S. J.,
van Boxtel, A. J., Giuseppin, M. L., Tramper, J. & Wijffels, R. H.
(2001). Estimation of steady-state culture characteristics during
genetic properties of four Saccharomyces cerevisiae strains. Enzyme
Microb Technol 26, 706–714.
acceleration-stats with yeasts. Biotechnol Bioeng 75, 267–275.
at maximum growth rate by control of carbon dioxide production.
J Gen Microbiol 59, 83–89.
van der Sluis, C., Rahardjo, Y. S. P., Smit, B. A., Kroon, P. J., Hartmans, S.,
Ter Schure, E. G., Tramper, J. & Wijffels, R. (2002). Concomitant
Watson, T. G. (1969). Steady state operation of a continuous culture
extracellular accumulation of alpha-keto acids and higher alcohols by
Zygosaccharomyces rouxii. J Biosci Bioeng 93, 117–124.
Yim, H., Haselbeck, R., Niu, W., Pujol-Baxley, C., Burgard, A., Boldt,
J., Khandurina, J., Trawick, J. D., Osterhout, R. E. & other authors
(2011). Metabolic engineering of Escherichia coli for direct
Van Dien, S. (2013). From the first drop to the first truckload:
commercialization of microbial processes for renewable chemicals.
Curr Opin Biotechnol 24, 1061–1068.
Zijffers, J.-W. F., Schippers, K. J., Zheng, K., Janssen, M., Tramper, J.
& Wijffels, R. H. (2010). Maximum photosynthetic yield of green
van Dijken, J. P., Bauer, J., Brambilla, L., Duboc, P., Francois, J. M.,
Gancedo, C., Giuseppin, M. L. F., Heijnen, J. J., Hoare, M. & other
authors (2000). An interlaboratory comparison of physiological and
http://mic.sgmjournals.org
production of 1,4-butanediol. Nat Chem Biol 7, 445–452.
microalgae in photobioreactors. Mar Biotechnol (NY) 12, 708–718.
Edited by: G. Thomas
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Sat, 06 May 2017 20:39:50
1719