Journal of Biotechnology 104 (2003) 155 /172

www.elsevier.com/locate/jbiotec

Industrial production of amino acids by coryneform bacteria

Thomas Hermann *
Degussa AG, Feed Additives, Research and Development Biotechnology, Kantstr. 2, 33790 Halle/Westf., Germany
Received 15 January 2003; received in revised form 12 March 2003; accepted 3 April 2003

Abstract
In the 1950s Corynebacterium glutamicum was found to be a very efficient producer of L-glutamic acid. Since this
time biotechnological processes with bacteria of the species Corynebacterium developed to be among the most
important in terms of tonnage and economical value. L-Glutamic acid and L-lysine are bulk products nowadays. LValine, L-isoleucine, L-threonine, L-aspartic acid and L-alanine are among other amino acids produced by
Corynebacteria . Applications range from feed to food and pharmaceutical products. The growing market for amino
acids produced with Corynebacteria led to significant improvements in bioprocess and downstream technology as well
as in molecular biology. During the last decade big efforts were made to increase the productivity and to decrease the
production costs. This review gives an overview of the world market for amino acids produced by Corynebacteria .
Significant improvements in bioprocess technology, i.e. repeated fed batch or continuous production are summarised.
Bioprocess technology itself was improved furthermore by application of more sophisticated feeding and automatisation strategies. Even though several amino acids developed towards commodities in the last decade, side aspects of the
production process like sterility or detection of contaminants still have increasing relevance. Finally one focus of this
review is on recent developments in downstream technology.
# 2003 Elsevier B.V. All rights reserved.
Keywords: Coryneform bacteria; Amino acids; Bioprocess; Downstream

1. Introduction
The recently finished deciphering of the Corynebacterium glutamicum genome by several amino acid producers marked a further milestone for
industrial use of this microorganism. The history
of the species Corynebacterium as amino acid
producer started in the 1950s when Dr Kinoshita

* Tel.:
/49-5201-711-3299; fax:
/49-5201-711-3512.

E-mail
address:
thomas.hermann@degussa.com
Hermann).

(T.

was the first to discover that C. glutamicum is a

superior amino acid producer (Kinoshita et al.,
1957; Udaka, 1960; Nakayama et al., 1961). Until
this time amino acids were available exclusively by
extraction methods or chemical synthesis. The
increasing demand for L-glutamic acid as a flavour
enhancer combined with the discovery of a microbial L-glutamic acid producer by Kinoshita and his
coworkers started the successful story of C.
glutamicum .
In Japan extracted sea weeds were used as a
source of food seasoning since centuries. In 1908,
Professor Kikunae Ikeda discovered that the

0168-1656/03/$ - see front matter # 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0168-1656(03)00149-4

156

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

major ingredient of this sea weed extract */called

kelp */was mono-sodium glutamate (MSG). In
the first half of the 20th century MSG was
produced by extraction from wheat, soybean and
other plant protein sources after hydrolysis by
concentrated hydrochloric acid. Shortly after Dr
Kinoshitas discovery in 1957, Kyowa-Hakko
started the fermentative production of MSG by
applying Micrococcus glutamicus (later renamed
C. glutamicum ) (Kumagai, 2000).
Nowadays more than 2 million tons of amino
acids are produced per year, most of them by
application of coryneform bacteria in bioprocesses. The annual market growth for most amino
acids is 10% and higher. These numbers are
catalysators for the development of improved
bioprocess and downstream technologies. However, the strains are the core of the production
process and improvements in molecular biology
and functional genomics led to leaps in strain
performance.
The genome projects at the end of the last
century marked a milestone in development. This
was the end of unidimensional developments (i.e.
the exclusive application of screening techniques)
and the start of faster parallel improvements using
DNA chips, proteomics, fluxomics and metabolomics techniques (Fig. 1).

2. Market overview
Current total annual worldwide consumption of
amino acids is estimated to be over 2 million tons.
The annual demand for amino acids like MSG
based flavour enhancers or feed additives comprised mainly of L-lysine, D,L-methionine and Lthreonine is estimated to be significantly higher
than 1 million tons each. The annual demand for
amino acids used in pharmaceutical products
mainly for intravenous and enteral nutrition is
15 000 tons (Kusumoto, 2001).
Around 1.5 million tons L-glutamic acid are
produced per year using coryneform bacteria.
MSG is used in food as a taste enhancer and
because of its own unique flavour called Umami
in Japanese. Prepared food usually contains 0.1 /
0.8% MSG but especially in east Asian dishes a

higher supplementation is common. Despite reports quoting some observers and experts that
MSG could be dangerous for health, a survey by
WHO and FAO showed that most food sweeteners
are safe for the health of humans. At the moment
the glutamic acid market is growing by about 6%
per year. Major producers of MSG are Ajinomoto,
Miwon, Kyowa-Hakko and Cheil-Jedang.
Compared with L-glutamic acid, L-lysine is used
almost completely as a feed additive. Traditional
feedstuffs like corn, wheat or barley are poor in
lysine. In order to increase feed efficiency, either Llysine rich crops like soybean meal or L-lysine is
added to these raw materials. Adding soybean
meal increases the protein level and supplies
additional non-limiting amino acids. The dosed
addition of L-lysine saves raw materials and
reduces nitrogen excretion.
Addition of 0.5% L-lysine increases feed quality
as much as adding approximately 20% soybean
meal. Because most of the other amino components of soybean meal are not used by the animals,
the amount of nitrogen excretion is increased
significantly. Using low protein diets supplemented with crystalline amino acids and applying
feeding strategies, which closely match animal
requirements, can help solve this problem.
In 2001, the world market for L-lysine was
550 000 tons with a growth rate of 7% per year.
Main producers are Ajinomoto, ADM, Kyowa
Hakko, Cheil Jedang, BASF and Degussa through
its Midwest Lysine joint venture with Cargill. LLysine is produced exclusively in bioprocesses
using coryneform bacteria.
Like L-lysine, the amino acid L-threonine is used
almost exclusively as a feed additive. Especially pig
and poultry diets have a high demand of Lthreonine. While i.e. corn germ meal contains
similar amounts of L-threonine and L-lysine,
soybean meal contains almost twice as much Llysine as L-threonine. The increase of L-threonine
concentration from 0.55 to 0.75% in a corn /
sorghum /peanut meal based diet for young broilers increases the breast meat deposition by more
than 15%.
In 2002, the L-threonine world market has a
volume of about 30 000 tons with an approximate
annual growth rate of 15%. Major producers are

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

157

Fig. 1. Market development of biotechnological produced amino acids.

Ajinomoto, ADM and Degussa. L-Threonine is

usually produced by strains of Escherichia coli .
Industrial aromatic amino acid producers are
described either as strains of the species E. coli (i.e.
Gerigk et al., 2002a,b) or as coryneform bacteria
(i.e. Ikeda and Katsumata, 1999). Even though
chemical synthesis was more important in the past,
bioprocesses dominate the market today. Applications for L-tryptophan include (especially) food
and feed purposes. Market size was approximately
1200 tons per year in 2001 with two-digit annual
growth rates. The leading producers in this market
are Ajinomoto, Kyowa Hakko and ADM.
Low-caloric sweetener aspartame is the source
of commercial interest for L-phenylalanine. World
consumption in 2002 was estimated to be 14 000
tons (Budzinski, 2001). In times of increasing

demand for soft drinks and low-caloric food the

market is still growing.
Industrial production of L-glutamine started in
the late 1960s. Currently, it is manufactured for
use as a therapeutic agent against gastroenterologic disorders, improvement of liver and brain
functions, immunoenhancement agent, and
against gastric ulcer and alcoholism. Furthermore
it is applied in cosmetics and as a food additive.
Worldwide annual production using bioprocesses
with coryneform bacteria is approximately 2000
metric tons (Kusumoto, 2001).
Branched chain amino acids are produced in an
amount of a few 100 tons per year each. Even
though several producers of the species E. coli are
mentioned in the literature, the best described
producers belong to the species Corynebacterium .

158

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

While up to now L-cysteine was produced

almost completely by extraction from i.e. hair, an
alternative bioprocess was established using E. coli
recently (Dassler et al., 2000). Several coryneform
strains were described to produce significant
amounts of L-cysteine (Wada et al., 2002).

3. Mechanisms of amino acid production

It is neither the objective of this paper to review
the metabolism of coryneform bacteria nor to
summarise the qualities of described production
strains. Nevertheless knowledge of fundamentals
of their metabolism help to understand the needs
of bioprocess technology. Those who are interested in this field find a number of reviews and
original articles focusing on amino acid biosynthesis in coryneform bacteria in general (Leuchtenberger, 1996; Eggeling and Sahm, 1999; Kumagai,
2000) and on L-glutamic acid (Kimura, 2003), Llysine (Pfefferle et al., 2003), L-isoleucine (Sahm et
al., 1996), L-valine (Radmacher et al., 2002) or Lthreonine (Debabov, 2003) biosynthesis in special.

4. Bioprocess technology
4.1. Standard bioprocesses for production of amino
acids
Bioprocesses for production of amino acids
were developed since the end of the 1950s. In the
beginning the technology served only a small
market but during the years the demand grew
and new applications were found. With the growing market the size of the plants and the bioreactors increased stepwise. Nowadays bioreactor sizes
from 50 to 500 m3 are standard in amino acid
production (depending on the product).
Before reaching the production reactor a few
other steps have to be carried out. Inocula
preparation is the first crucial step in most
bioprocesses. The purpose of this lab procedure
is to obtain a huge amount of stable inocula, which
could be used to run dozens of production batches
under the same defined conditions finally reaching
the same result or yield. Inocula have to be

validated after preparation in terms of sterility

and productivity before being transferred to production. Inocula often influence productivity and
yield of the bioprocess significantly. Due to the
importance of inocula quality (inoculum size and
stability) and quantity (amount of inocula and cell
titer) it has to be tested regularly during its use in
order to avoid decreasing productivity. On the
background of, i.e. fluctuating raw material qualities or different bioreactor specifications this has
to be performed under defined conditions. The
inoculum is afterwards propagated in a so called
seed train. In the case of bioprocesses with coryneform bacteria this means usually 1:10 steps from
seed step to seed step. Depending on the scale of
the production bioreactor this is performed in
shaking flasks or more preferably in bioreactors of
different scale. Usually at least two seed bioreactors are used to obtain a sufficient biomass for
inoculation of a production batch. Often harvesting conditions are crucial for the further course of
the production process. Bioreactors are usually
connected by sterile pipes, which allow a safe and
fast transfer from one vessel to another. Furthermore they have air spargers which are connected
to sterilisable air filters.
Parameters like pH, aeration, feed rate of
carbon source or process temperature are targets
for optimisation. Preconditions of bioprocesses in
scales up to 500 m3 are the increased hydrostatic
pressure in the vessel and a limited agitation speed.
Furthermore pCO2 is usually increased because of
lower aeration in bigger volume bioreactors. This
point might be crucial for scale-up of bioprocesses
with coryneform bacteria because of the significance of anaplerotic reactions (de Graaf et al.,
2001). This is underlined by the fact that coryneform bacteria possess a pyruvate carboxylase and
a phosphoenolpyruvate carboxylase. Together
with phosphoenolpyruvate carboxykinase, malic
enzyme, malate dehydrogenase and the glyoxylate
cycle they contribute to carbon flux through
oxaloacetate a key precursor for the biosynthesis
of the amino acids of the aspartic acid family.
Phosphoenolpyruvate carboxylase was shown to
be essential for L-glutamate production with a
temperature-sensitive mutant of C. glutamicum
(Delaunay et al., 1999). Addition of CO2 to the

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

process air increased the biomass yield and considerably decreased formation of organic acids in
L-lysine production with leucine and homoserine
auxotrophic C. glutamicum (Hadj-Sassi et al.,
1996). Oxygen availability was evaluated by several authors. Recently Shu and Liao (2002)
revealed the relevance of the oxygen transfer rate
coefficient (KLa) on L-phenylalanine production
with C. glutamicum . They showed that increased
oxygen transfer rates (OTR) increase L-phenylalanine productivity and yield (positively) up to 42%
reaching values of 0.37 g l 1 h1 and 14%,
respectively. Yao et al. (2001) showed for L-lysine
production with Brevibacterium lactofermentum
that dissolved oxygen concentrations influence
the process. Inhibition of high pO2 could occur
during the very early growth phase and depressive
effect of low oxygen availability was confined to
the rest of the process. These results suggest that
different stages of the bioprocess require different
pO2 levels. By adapting a pO2 profile the L-lysine
yield was increased from 0.300 to 0.343 g g1.
Whether these effects were due to changed pO2 or
pCO2 was not elucidated.
Process temperature can be changed depending
on the demands of the organism. Elevated temperatures are used to induce L-glutamic acid
production in certain strains (Gourdon and Lindley, 1999; Delaunay et al., 2002). Most textbooks
mention 30/34 8C as the optimum temperature for
cultivating coryneform bacteria. However, strains
with increased thermotolerance were described
recently (Kimura et al., 2002). Those strains are
sometimes preferred for production processes
because of the limited amount of cooling water
available at some production sites.
Standard bioreactors for production of amino
acids are equipped with simple Rushton turbines
or in combination with axial mixing systems.
Stirrer tip speed seems to be a smaller problem
with coryneform bacteria than with some other
biotechnologically relevant organisms. A possible
reason for this might be the thick cell wall
encircling the cell (Marienfeld et al., 1997). Gas
blending can be used to separate effects of
dissolved oxygen from negative influences of the
stirrer tip speed (Pollard et al., 2002). However,
like in all bioreactors, mixing is a problem

159

especially under consideration of bioreactor volumes up to 500 m3. Recently an European

community funded study about scale-up of an E.
coli process with several groups was performed
(Enfors et al., 2001). In a 22 m3 scale bioreactor
equipped with Rushton turbines organic acids like
formate accumulated, indicating oxygen limited
zones, though the dissolved oxygen signal did not
show any oxygen limitation. A reduced biomass
yield in this scale was suggested to be due to
repeated formation and re-consumption of organic
acids originating from overflow metabolism or
mixed acid fermentation (Xu et al., 1999a,b). Later
on a dynamic model of glucose overflow metabolism in batch and fed-batch cultivations of E. coli
under fully aerobic conditions was developed
allowing prediction of physiological reactions
under certain process conditions (Xu et al.,
1999a,b). Overflow metabolism or mixed acid
fermentation induced stress responses of the E.
coli cells, which were determined by analysing
mRNA levels. Stress responses were relaxed when
the cells returned to the substrate limited and
oxygen sufficient compartment of the reactor.
Corresponding analysis in the large reactor
showed that the concentration of mRNA of four
stress induced genes was lowest at the sampling
port most distant from the feed zone. It was
assumed that repeated induction/relaxation of
stress responses in a large bioreactor may contribute to altered physiological properties of the
cells grown in large-scale bioreactors (Schweder et
al., 1999). Production of organic acids like acetate
or lactate could lower the product yield in
bioprocesses additionally by wasting energy
through a transport futile cycle if they are exported
under consumption of energy. Once exported they
could be able to re-enter the cell by passive
diffusion when they are protonated. This process
was examined by Axe and Bailey (1995) for
production of lactate and acetate by E. coli . Byproduct formation of organic acids like acetate,
lactate or pyruvate is also described for bioprocesses with coryneform bacteria (Hadj-Sassi et al.,
1998; Hua et al., 1998; Gourdon and Lindley,
1999).
Improvements were made in the field of understanding the formation of gradients in bioreactors.

160

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

By using computational fluid dynamics a prediction of conditions in large scale bioreactors is

possible (Schmalzriedt and Reuss, 1999; Reuss et
al., 2000; Campolo et al., 2002).

1.

4.2. Processes with increased productivity

Nowadays amino acids are produced in bioprocesses applying either batch, fed batch, repeated
fed batch or continuous production technique. In
batch technology a medium with all nutrients
necessary for growth and production is filled into
the bioreactor. Then the batch is started by
inoculation with biomass from another bioreactor.
During the process the only streams entering the
bioreactor are liquid or gas for pH adjustment, i.e.
ammonia and air. The batch is finished after total
depletion of nutrients. Even if batch technology
seems to be rather old-fashioned it is still used in
many plants all over the world. Batch technology
is applied as a very simple process in terms of
process control (no nutrient supply necessary),
technology (no vessel for feed solution necessary)
or sterility (less feed lines). Major boundary for a
possible intensification of a batch process is the
osmolarity of the initial medium. Especially high
initial concentrations of the carbon source interfere with growth and productivity of coryneform
bacteria (for overview see Morbach and Kramer,
2002; Ruebenhagen et al., 2001). A disadvantage
of most batch processes is the relatively low
productivity due to increased lag phases.
The fed batch process is still the standard
process for manufacturing products from bioprocesses. This is also valid for amino acids. In a fed
batch process only a particular amount of the
vessel is filled with medium before inoculation of
the batch, for example 50%. After consumption of
the initial carbon source more carbon source is fed
into the bioreactor through a separate feed line
until the reactor is completely filled. This feeding
might follow a certain pre-defined feed profile or
run with a feeding strategy allowing to produce
under a defined physiological condition. Basic
characteristic of a fed batch process is, therefore,
the control of nutrient concentrations leading to
higher yield or productivity. Advantages of the fed
batch process could be summarised as follows:

2.

3.

4.

The control of the nutrient concentration

reduces influences of the nutrient on productivity or yield of the process. In amino acid
production high amounts of carbon source are
consumed by the cells in order to transform
them to the desired product. If the nutrient
concentration exceeds a certain level unwanted effects like substrate inhibition or
overflow metabolism might occur. Typical
overflow metabolites are lactate or acetate
(Gourdon and Lindley, 1999; Enfors et al.,
2001).
Due to the feed of one or more nutrient flows
into the bioreactor osmolaric stress for the cell
is reduced. In some processes 20% of the final
mass or more of a batch has its origin in sugar
fed into the process (Ikeda, 2003). By reduction of the initial nutrient concentration and
further control of the nutrient concentration
the lag-phase of the bacteria can be reduced.
Ikeda and Katsumata (1999) describe a Ltryptophane process with an L-phenylalanine
and L-tyrosine auxotrophic strain of C. glutamicum . In auxotrophic strains overfeeding
might result in decreased yields due to overgrowth of cells. Running the process with feed
rates far below the maximum sugar consumption rate of the cell reduces this effect in some
cases. Pfefferle et al. (1993) describe these
effects in a 10 m3 scale bioprocess with a Lleucine auxotrophic S -(2-aminoethyl)-cysteine
resistant L-lysine producer of the species C.
glutamicum . During one production batch
with this strain the sucrose feed rate was
higher than the maximum sucrose consumption rate of the cells leading to a final process
yield of 30.9%. In a second batch the sucrose
feeding was lower than the maximum sucrose
consumption rate leading to a final process
yield of 32.3%. Additionally fed batch processes allow the feeding of further nutrients
during the process. However, in times of high
yields in commercial amino acid production
and high developed molecular biology tools
auxotrophic strains should have a limited
future.
The control of the nutrient supply allows to
run the process under conditions of an opti-

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

mum oxygen transfer into the bioreactor.

Especially during the growth phase of the
bacteria oxygen demand might exceed the
capacities of industrial bioreactors. Under
these conditions oxygen supply might become
insufficient, oxygen limitation itself produces
unwanted partly oxidised by-products. Therefore, one important scale up criterium for a
bioprocess is the ratio of oxygen uptake to
sugar uptake (Cooney et al., 1968).
A further increase in productivity is obtained by
reducing preparation times between batches to a
minimum. Preparation time between fed batch
bioprocesses include several or all of the following
actions.
Inactivation of the process liquid after the
bioprocess, transfer of the process liquid to downstream, cleaning of the empty vessel and all
connections and pipes, calibration of electrodes,
re-filling of the vessel, sterilisation of air filters,
sterilisation of the vessel and pipes and inoculation. This preparation time can take up to 10 h,
thus reducing the productivity of the plant significantly. Depending on the degree of automatisation of the plant, the skills of the co-workers and
quality and quantity of preventive maintenance
this time could be reduced to a minimum (Junker,
2001). Nevertheless in order to increase productivity a repeated fed batch process is often considered. In this kind of process a first batch is
performed under classical fed batch conditions.
At the end of the batch a part of the process liquid
is transferred to downstream while another part
stays in the bioreactor. Latter is used as an
inoculum for the next batch. The amount of
inoculum varies depending on the demands of
the process. In general 60/95% of the process
liquid are withdrawn from the vessel. Afterwards
the vessel is refilled with fresh nutrients to its
starting volume before the next batch starts
regularly. Repeated fed batch bioprocesses are
common, i.e. for fermentative acetic acid production (Arnold et al., 2002). For L-lysine production
with C. glutamicum this kind of process was
described for an S -(2-aminoethyl)-L-cysteine resistant model strain with a feedback resistant aspartokinase. In two successive batches the volumetric

161

productivity was 1.71 and 2.01 g l1

h . Considering a process time of 30 h and a 10 h
preparation time the productivity was increased up
to 15% with this process modification (Research
disclosure, 2000). Recently a repeated fed process
with several L-threonine producers of the species
E. coli was described (Hermann and Rieping,
2002). By applying the model strain E. coli B3996 a volumetric L-threonine productivity of 1.77
g l 1 h1 was achieved in 36 h cultivation time.
Considering (again) a preparation time of 10 h the
process productivity would be decreased to 1.39 g
l1 h1. In six repeats the productivity was
increased by more than 20% reaching 1.69 g l1
h1. Thus, advantages of the repeated fed batch
process are often higher process productivities due
to decreased lag times of the bacteria and higher
plant productivities due to reduced preparation
times between the batches. Compared with the
classical fed batch process other advantages include a decreased number of seed bioprocesses
reducing variable costs up to 10% and a reduction
of investment costs in case of newly erected plants.
Disadvantages are a higher risk of contaminations,
which on the other hand might be covered by more
sophisticated contamination detection methods.
Furthermore sometimes construction of new
strains is necessary for repeated fed batch processes because of increased generation numbers of
the bacteria and instability of the producers as
described for the production of L-phenylalanine
(Kim, 1995). Up to 2000 only a repeated fed batch
processes with C. glutamicum for the production
of L-lysine was described (Pham, 1994; Pham et
al., 1995). In some cases it is even better if only 5 /
20% of the process liquid are taken from the vessel
giving room for further feeding. However, this so
called semi-continuous process is already very
close to the next level of process intensification */the continuous process.
Continuous bioprocesses do have a long tradition in both, industrial scale process intensification
and in lab scale as a tool to examine the
characteristics of a bioprocess. In this type of
process the cells are cultivated until a particular
biomass density, product concentration or quality
is achieved before the bioprocess is switched to the
continuous mode. Hence fresh nutrients are fed to
L-lysine /HCl
1

162

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

the process while the same amount of process

liquid is withdrawn from the vessel keeping the
process volume constant. Two methods are generally applied in continuous processes:
1.

2.

A turbidostat levels the flows into and from

the bioreactor to keep the biomass concentration constant.
In the so called chemostat the concentration of
nutrients, i.e. carbon source or phosphate
source, is constant.

Carbon or phosphate limited continuous cultures are favourable because they allow a tight
control of the process and reduce the risk of
contaminations. Koyoma et al. (1998) describe a
continuous process for production of L-glutamic
acid with Brevibacterium lactofermentum reaching
volumetric productivities of up to 8 g l 1 h 1 and
thus doubling productivities achieved with batch
processes. In 1989, Hirao et al. described a
continuous process for the production of L-lysine
(Hirao et al., 1989). They used a strain based on C.
glutamicum ATCC13032 resistant to S -(2-aminoethyl)-L-cysteine, rifampicin, streptomycin and
6-azauracil reaching productivities of 5.6 g l 1
h1 and a L-lysine /HCl yield on glucose of 0.40 g
g1. The process itself showed a stable L-lysine
production for more than 300 h. Considering a
preparation time of 10 h the fed batch process with
this strain reached a productivity of 1.72 g l 1 h1
(100 g l1 L-lysine /HCl in 48 h process time).
Under these constraints the continuous process
showed a more than 3-fold increase in productivity
compared with the standard fed batch process. In
parts this was achieved by optimisation of the
culture redox potential kinetics. A similar approach was used for L-lysine producing C. glutamicum by Kwong and Rao (1991, 1992) which
added dithiothreithole to optimise the redox
potential of an aerobic growing batch culture. By
this the redox potential was lowered and in parallel
amino acid production rate was increased significantly. de Hollander et al. (1997) described a Llysine bioprocess with coryneform bacteria under
carbon and phosphate double limitation. Compared with a carbon limited process the productivity was increased from 3.18 to 3.75 g l 1 h1 L-

lysine /HCl. Additionally the biomass specific Llysine formation was increased 2.8-fold and the
typical instability of the strains under continuous
conditions was reduced under double limitation.
Major drawbacks for continuous bioprocesses are
mainly the risk of the contaminations in the
culture and possible strain instabilities. Contamination risk is increased in continuous culture
because of additional flows into and from the
vessel. The medium is often sterilised through
continuous sterilisers which have a particular risk
of blockage (Gondorf, 2001) when used with some
salts and nutrient combinations. However, if a
severe contamination of the culture occurs it is
usually detected early enough to allow a save finish
of the bioprocess. The potential loss in these cases
is usually much smaller than the gain in productivity of the plant working in continuous mode.
Strain instabilities could be reduced by process
modifications as mentioned above. On the other
hand continuous lab cultures are used to select
mutants with changed properties, i.e. faster
growth. Azuma et al. (1988) and Azuma and
Nakanishi (1988) describe these effects in continuous bioprocesses for producing L-arginine with
Corynebacterium acetoacidophilum .
Several groups tried to improve continuous
bioprocesses by separating the process into two
or more bioreactors. Usually the growth phase
takes place in a first bioreactor, while production
phases take place in another one. This so called
cascade bioprocess allows to run each phase of
growth under optimised conditions. In processes
with separated growth and production phases as
described for many amino acid producers this
procedure seems to be beneficial. In production
plants such a process modification allows to
include seed fermenters in the production process.
High growth rates lead to low residence times
which are often suitable for these usually smaller
bioreactors. Production phases with their small
growth rates could then be transferred into the
much bigger main reactors increasing the plant
productivity by the volume of the seed fermenters
(Emde, 1997).
Recent developments of repeated fed batch and
continuous processes indicate the tremendous

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

potential in process intensification of fermentative

amino acid production.
A different kind of process intensification for C.
glutamicum was described by Kim et al. (2000). By
addition of a certain amount of bioprocess liquid
obtained, i.e. from an earlier seed culture, to the
running process culturing time in the seed reactor
was decreased by 20% and in the production
reactor by 5%.

4.3. Raw materials for the production of amino

acids
Selection of raw materials is essential for
economic amino acid production. Especially the
carbon source represents a major part of variable
production costs. This explains the sometimes
tight connection between amino acid and sugar
producers which developed over time. Some amino
acid producers are located very close to sugar
plants in order to decrease transport costs or joint
ventures are formed. Depending on the geographical location of the plant carbon sources like
cane molasses, beet molasses, or starch hydrolysates from corn, potato or cassava are used. While
molasses are common in Europe, South America
and China, starch hydrolysate is the most important carbon source in North America. Tapioca
hydrolysate, the starch hydrolysate from cassava is
widespread in South-East Asia. Pure sugars are
usually favourable compared with molasses because of unwanted side reactions and changing
qualities of the complex media components. Some
types of molasses for example contain higher
concentrations of biotin which inhibits corynebacterial L-glutamic acid biosynthesis. Even though
some authors describe processes with ethanol,
acetic acid or n-paraffins as carbon source, these
substrates have significant economical disadvantages compared with sugar carbon sources. Methylotrophic bacteria are able to convert methanol
into L-glutamic acid, L-lysine (Bacillus sp.; Methylobacillus glycogenes ) or L-threonine (M. glycogenes ) (Motoyama et al., 2001, 1994; Schendel et
al., 1990). Similar processes with methanol are
described for production of L-phenylalanine or Lmethionine.

163

Nitrogen sources vary from inorganic pure

ammonia and ammonium salts like ammonium
sulphate to complex organic components like
peptone hydrolysates or corn steep liquor. These
components might be either introduced into the
initial media of the process or dosed into the
bioreactor during the process like ammonia water
or gas which is commonly used for pH regulation.
Application of a positive CO2 pressure is described
for bioprocesses with C. glutamicum . Carbonate
or hydrogen carbonate ions interact as counterions
for produced basic amino acids like L-lysine or Lhistidine (Itoyama et al., 2001). This procedure
reduces the consumption of, i.e. ammonia sulphate
in the bioprocess. The advantage of inorganic
nitrogen sources is the stable quality obtainable
on the market. Peptone hydrolysates from animal
origin are considered problematic in bioprocess
media recipes since the BSE incidents. More
expensive peptone sources like yeast extract have
more defined ingredients. Corn steep liquor is a
cheap and often standardised nitrogen source rich
in amino acids, oligopeptides, vitamins and nucleotides. However, as a by-product of a lactic acid
bioprocess with corn as substrate it has changing
qualities. These fluctuations originate in the source
and age of the corn and the course of the steeping
process. Fluctuation in raw material quality is
well-known as source of bad bioprocess results.
Cheap corn steep liquor qualities often have
increased L-lysine contents leading to reduced
yields in e.g. penicillin production (Banuelos et
al., 2000). Changing concentrations of iron in corn
steep liquor batches were shown to influence Lthreonine production with E. coli significantly
(Okamoto and Ikeda, 2000). It was shown that
by addition of inorganic iron the process got
independent from fluctuating iron concentrations
in the CSL batches. Corn steep liquor contains
significant amounts of inorganic and organic
phosphate, another important ingredient of bioprocess medium recipes. Nevertheless main phosphate sources are inorganic salts like potassium or
sodium phosphate. Same is valid for magnesium,
sulphur and other ions necessary for the growth of
coryneform bacteria. Vitamins like biotin are
usually added into the process in their pure form.
Some organic components have a limited stability

164

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

at higher temperatures which occur i.e. during

sterilisation or under these conditions react with
others (i.e. Maillard reaction). Composition of the
media must, therefore, be adapted to the conditions of sterilisation. Furthermore salts might form
insoluble complexes which limit their availability
for the organism. As mentioned for repeated fed
batch and continuous bioprocesses continuous
sterilisation is usually more favourable than batchwise sterilisation (Jain and Buckland, 1988).
However, quality check of new raw material lots
by either instrumental analytical methods (i.e.
mass spectrometry) or biotests is essential.
4.4. Automatisation and process control
A lot of efforts were made to find ways for
precise control or efficient automatisation of
bioprocesses. There is great demand within the
bioindustry for systems that allow optimisation
and automatic control of bioprocesses. Nowadays
state of the art bioprocesses for production of
amino acids run more or less fully automised with
very little human control. Control of the nutrient
supply is crucial for most industrial scale bioprocesses due to the necessity to run the process under
conditions of an optimum oxygen transfer in the
bioreactor as discussed above. Furthermore prevention of overflow metabolism is important
(Enfors et al., 2001). Excretion of overflow metabolites like acetate, lactate, pyruvate or even
unwanted amino acids like L-alanine are described
for coryneform bacteria by several authors (HadjSassi et al., 1998; Hua et al., 1998; Gourdon and
Lindley, 1999). Typical cause of overflow metabolism is an excess of the carbon source combined
with a growth limitation of the organisms. These
effects are also described for other organisms like
E. coli (Konstantinov, 1996; Xu et al., 1999a). It
was shown that L-phenylalanine producing E. coli
excretes acetate when substrate consumption exceeds a critical glucose uptake rate (Konstantinov
et al., 1991). Xu et al. (1999a,b) describe overflow
metabolism in E. coli growing with growth rates
higher than 0.3 h1. Cocaign-Bousquet et al.
(1996) used chemostat cultivations of C. glutamicum to determine critical growth rates for the
onset of overflow metabolites. By efficient regula-

tion of the carbon concentration in the bioreactor

overflow could be avoided. Miwa et al. (1992)
described an efficient control of by-product formation in industrial fed batch processes with C.
glutamicum by an adapted feed strategy. Therefore, the feeding of the carbon source followed a
certain calculated flow rate. By regularly intermitting this feeding carbon limitation in the bioreactor was detected due to control of pH or pO2
values. Depending on the time until such a leap in
pH or pO2 occurred and the intensity of this
starvation signal feed rate was adapted for the
following time period. A similar feedback control
of glucose feeding in E. coli bioprocesses was
described recently (Akesson et al., 2001). Pfefferle
et al. (1993) applied a similar principle to ensure a
sugar limited L-lysine production process with C.
glutamicum . By this higher yields were reached
and by-product formation was reduced.
Takiguchi et al. (1997) described an online
metabolic pathway analysis which was used to
optimise the control of L-lysine production with C.
glutamicum . Online metabolic flux analysis for
control and monitoring of bioprocesses was reviewed by Shioya et al. (1999). Various optimisation and control methods for bioprocesses based
on deterministic mathematical models have been
developed. This approach has rarely succeeded in
actual production plant operations due to the
difficulty in developing a mathematical model
capable of describing the complex intracellular
reactions of the used microorganisms. Empirical
knowledge of the bioprocess was, therefore,
adapted to fuzzy control systems. Fuzzy control
can utilise empirical knowledge gained from
skilled operators by employing IF THEN rules.
Recently fuzzy control systems and its industrial
application was described for riboflavin production with Bacillus subtilis (Horiuchi and Hiraga,
1999; Honda and Kobayashi, 2000) or acetic acid
production (Arnold et al., 2002). A fuzzy expert
system was used to optimise L-glutamic acid
production by estimating the physiological state
of the culture (Kishimoto et al., 1991a,b).
Multivariate statistical process control of bioprocesses is a helpful tool for detection of process
interferences. Multivariate projection methods are
used to observe interactions among individual

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

variables and to integrate various process stages

into a monitoring system (Albert and Kinley,
2001; Lennox et al., 2001). Compared with other
approaches (i.e. fuzzy technology) multivariate
statistical process control has the advantage that
it is based on process data. Disadvantage of this
technique is that frequent minor changes necessitate the development of adaptive models while
major changes result in invalid models. The
development of a bioprocess knowledge base by
extraction from the operators (implemented in the
form of production rules) is integrated with multivariate data-based methods for fault detection by
Glassey et al. (2000). The industrial benefits
arising from this integrated system include: (1)
reduced variability, (2) increased mean performance levels, (3) reduced operator-training time
and (4) knowledge management in the broader
organisation.
For L-leucine auxotrophic L-lysine producers
the optimisation of the initial L-leucine concentration is crucial. The subsequent optimisation of a Lleucine feed regime resulted in an optimal growth
rate and increased process performance (Shimizu
et al., 1999). Optimisation of the media composition by application of a genetic algorithm and
control of the glucose concentration was used by
Weuster-Botz et al. (1997) to improve the L-lysine
production with C. glutamicum . An evolutionary
strategy for optimisation of fed batch bioprocesses
with real-code genetic algorithms was discussed by
Roubos et al. (1999).
4.5. Process scale up and scale down
Transfer of a lab process into a production
process running in bioreactors with volumes up to
500 m3 or more needs efficient know how of
process scale up (for overview see: Thiry and
Cingolani, 2002). In the past, scale up was often
performed by stepwise increase of bioreactor
volumes, i.e. because of a plant or production
extension. However, some groups worked on the
development of efficient tools for simulating
process scale up. Problems in scale up involve
the increased (hydrostatic) pressure, higher pCO2,
different sterilisation conditions, inhomogeneous
substrate or ammonia ion distribution, shear

165

forces at the stirrer tip, power input and gradient

formation in bioreactors of industrial scale. Latter
includes gradients of pH, dissolved oxygen, ammonia or carbon sources caused by mixing times
up to 100 s and more (Einsele, 1978). For studying
gradient formation in lab scale at least two
principal types of bioreactors were developed: (1)
stirred tank reactors with a second external
volume (i.e. vessel) used to decrease mixing
efficiency and (2) reactors with internal installations inhibiting a sufficient mixing. Several examples were published for the first type of reactor. Xu
et al. (1999a,b) describe a system in which a certain
part of the process liquid from a stirred tank
reactor is bypassed through a plug flow reactor
before reentering the main reactor again. Due to
the lack of aeration and mixing in the plug flow
reactor the part of the process liquid which is in
this reactor is usually oxygen limited. By adapting
the residence time in this tube large scale reactors
with badly mixed zones of different volumes could
be simulated. This effect was intensified by adding
the carbon source to the process liquid just before
entering the plug flow reactor. This system was
used to study physiological responses of the cell to
partial oxygen limitation (Schweder et al., 1999).
By taking samples from different levels of the plug
flow reactor responses of the microorganisms
could be quantified. The same system was evaluated by Amanullah et al. (2001) for simulating
the influence of pH gradients on acetone and 2,3butanediol production with Bacillus subtilis. The
production of these metabolites is sensitive to pH
values between 6.5 and 7.2. Gradient formation
was simulated by adding alkali for pH control
either into the loop or directly into the stirred tank
reactor. It was shown that although biomass
concentration remained unaffected by pH variations, product formation was influenced by residence times in the plug flow reactor of 60 s or
more. A similar simulation approach is to separate
the process into two bioreactors which are connected to a loop (Amanullah et al., 1993). This
technology allows to run the process under two (or
more) different but defined conditions. Schilling et
al. (1999) describe a bioreactor system with internal installations. In this type of reactor mixing is
inhibited by installation of five horizontal disks

166

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

separating the vessel into chambers. Combined

with Rushton turbines axial mixing was reduced
and the apparent mixing time u90 was prolonged
from 10 to 130 s. With a leucine-auxotrophic Llysine producing C. glutamicum the process performance was significantly decreased in this type
of reactor compared with control experiments.
Reduced sugar consumption (/14%), and biomass formation (/7%) resulted in a decrease in Llysine formation (/12%) in the model reactor.
These findings were reflected in less specific
enzyme activity for important key enzymes like
for citrate synthase, phosphoenolpyruvate carboxylase, and aspartate kinase. The reduced specific
activity of citrate synthase correlated with lower
CO2 evolution (/36%) during cultivation. Cellular sensing of gradients and the physiological
response of the organism to it is of increasing
interest. Furthermore knowledge of genotype,
phenotype and physiology of the used microorganism is very valuable for process scale up.
Besides this the development and use of mixing
systems with thorough axial mixing brought improvements in process performance. For scale-up
of rushton turbine-agitated tanks and details on
stirrer design see i.e. Stahl-Wernersson and Tragardh (1999) and Campolo et al. (2002).
4.6. Sterility
A monoseptic bioprocess is often essential to
fulfil specifications of the product and to reach
desired yields or volumetric productivities. Specifications for maximum bacterial contamination
titers depend on the kind of downstream procedure and the application of the product. Common
contaminants differ from plant to plant, but
Bacillus sp. is a common contaminant in amino
acid production processes. Sometimes bacterial
contaminants grow on by-products and increase,
therefore, the purity or even the yield of the
bioprocess. Even if phage contaminations in
bioprocesses with C. glutamicum occur more
seldom compared with E. coli or other bacterial
bioprocesses they have usually higher impact on
plant productivity (Sanders, 1993). Phage contaminations could be reduced either by technical
changes or changes in the production process.

Technical changes include effective sterilisation of

the process air, careful sanitation of the plant and
precise culture-handling practices (Junker, 2001).
Cross contamination between the bioprocess
equipment and upstream processes should be
avoided. Several independent strain lines should
be stored to reduce the risk of repeated phage
contaminations. In case of the first phage contamination the production could be switched
immediately to another strain. Finally phage
resistant strains are an option to reduce contamination risks. Rapid phage detection could be
crucial in solving a phage contamination problem.
Classical methods like plaque assay, spot assay,
and broth lysis have the disadvantage to be time
consuming and labour intensive. Usually they are
only helpful for distinguishing between a phage
contamination and a technical problem like wrong
media composition. Powerful online tools like
multivariate data analysis (Albert and Kinley,
2001) or fuzzy technology could help to detect a
phage contamination early enough to stop the
running process or even more important to avoid
cross contaminations.
4.7. Downstream technology
A cost-efficient downstream process is crucial to
reduce investment and production costs in amino
acid production. Stable product quality parameters like dustiness or bulk density are desired
by the customers. The separation of biomass is
usually the first step of amino acid purification
independent of the later way of downstream.
Removal of the cell mass is accomplished either
by gravitation-based techniques (i.e. centrifugation
or decantation) or by filtration. A significant
amount of product might be lost in this biomass
removal step which has furthermore the disadvantage of (A) high investment costs and (B) biomass
waste streams.
Once the biomass is removed from the product
stream the purification of the product begins.
Typical purification steps are chromatographic
technologies, combined concentration/crystallisation steps or combinations of both. What method
or what sequence of methods is used depends on a
couple of effectors: the physico-chemical proper-

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

ties of the amino acid (i.e. solubility, isoelectric

point), composition of the process liquid (i.e.
quality and quantity of by-products), environmental regulations (i.e. waste liquor treatment) and on
the application of the product (i.e. feed or
pharmaceutical use). Furthermore raw materials
might have an influence as in the case of molasses.
Unused components from molasses lead to a
disadvantageous increased waste flow when using
this substrate.
Ion exchange resins are the base for chromatographic methods. Disadvantages of this common
technique are lower concentrations and increased
waste streams leading to higher costs in wasteliquor treatment. On the other hand chromatographic methods provide usually higher product
qualities. Typically, process liquid separated from
the biomass is acidified, preferably by the addition
of hydrochloric acid or sulphuric acid, to ease
adsorption of the amino acid to the ion exchange
resins. In addition to the main product, various
other cations could be present in the process liquid
which are also bound to the resin. The adsorbed
product is then preferably eluted by an ammonia
solution and the ion-exchange column is regenerated. The product solution obtained in this way is
then concentrated. In the case of L-lysine, L-lysine /
HCl is obtained in crystalline form after neutralisation with hydrochloric acid. In general, a variety
of ion-exchange columns connected in sequence
are necessary for obtaining a pure product. However, the amino acid solution coming from an ion
exchanger has usually a sufficient purity to be
further purified with just one crystallisation step.
If only crystallisation is applied in product purification, often two or more subsequent crystallisation steps are necessary. Lee et al. (2002) used
ion exclusion chromatography to recover L-lysine
from the bioprocess liquid. After microfiltration
permeate was adjusted to the isoelectric point of Llysine and fed to a cation exchange resin, L-lysine
was eluted by feeding with distilled water. This
technology has significant advantages compared
with waste generating ion exchange processes.
Crystallisation techniques like cooling or vacuum crystallisation are economically favourable
but could not be applied for all amino acids. The
solubility of L-glutamine, i.e. is barely affected by

167

temperature as shown by a flat solubility curve

making cooling crystallisation unsuitable for this
product (Kusumoto, 2001). Due to the amphoteric
character of amino acids sufficient crystallisation
depends also on the pH during crystallisation as
shown in the case of L-histidine (Osterberg and
Wadsten, 1999). Advantages of these concentration/crystallisation or ion exchange technologies
are that high product purities are obtained. Disadvantages are high investment costs and considerable high waste streams, i.e. for biomass or salt
solutions.
Especially for the production of L-lysine several
simplified downstream processes were developed.
Some of them are described without biomass
separation. Hofler et al. (1997) produce a granulated lysine sulphate containing biomass just by
evaporating the process liquid with a succeeding
spray granulation step. As mentioned above
advantages are dramatically reduced waste
streams, an increased downstream yield and reduced investment costs. A similar approach led to
a liquid lysine sulphate with just one evaporation
step (Uffmann and Binder, 2002; Binder and
Uffmann, 2002). Both products are positioned in
the L-lysine (feed) market providing the customers
with additional valuable co-products from its
bioprocess. These co-products deliver additional
amino acids, energy and phosphorous as valued in
the nutritional matrix.
These downstream technologies need efficient
process control and low variations in bioprocess
performance (i.e. biomass or by-product formation) to reach specifications which allow successful
product formulation. Biomass containing products
obtained from bioprocesses with coryneform bacteria have the advantage that this species of
organisms is generally regarded as safe.
Fig. 2 compares downstream processes for
production of a biomass containing L-lysine, Llysine /HCl using ion exchange (Oka, 1999) and Lthreonine using crystallisation steps (Ohtani et al.,
1987).
As mentioned above the amount of waste
streams depends on the kind of purification and
on the number of recycling steps. After ion
exchanger steps equal masses of salt (i.e. ammonium sulphate) and amino acid are obtained. These

168

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

Fig. 2. Comparison of downstream processes for production of L-lysine/HCl using ion exchange (Oka, 1999), L-threonine using
crystallisation steps (Ohtani et al., 1987) and a biomass containing L-lysine by spray granulation (Hofler et al., 1997).

salts could be at least in parts recycled but this is

energy consuming and cost intensive. Discharged
mother liquors may be used as fertilisers (Ikeda,
2003). The sometimes significant amount of product contained in mother liquids could be recycled
by electrodialysis as described by Fischer et al.
(2001). This technique has the advantage that
amino acids could be isolated as pure acid or base.
Reactive extraction during the bioprocess might
be helpful to obtain higher product titers especially
in cases of product inhibition. This was shown by
Gerigk et al. (2002a,b) for production of Lphenylalanine with an E. coli strain.

5. Outlook
Applications of amino acids either in food and
pharmaceuticals or in animal feed nutrition is

expected to grow further on during the next years.

While the bigger products like L-glutamic acid will
at least keep their growth rate, smaller products
will develop with increasing speed and will thereby
be supported through the development of the
bigger products and vice versa. New technologies
will be developed in all parts of the L-amino acid
manufacturing process. Process intensifications for
bioprocesses like the mentioned repeated fed batch
or continuous bioprocess technology will be implemented into the production sites. Vice versa
improvements in one part of the production plant
often lead to bottlenecks in other parts. Therefore,
efforts in upstream or downstream technology like
faster or more efficient sterilisation technologies or
biomass separation units will accompany these
developments.
One major catalysator for these improvements is
the recent deciphering of the corynebacterial

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

genome. The availability of the genome data is the

essential step for faster and more efficient improvement of amino acid production strains.
Furthermore, it gives access to a couple of post
genome technologies like expression profiling or
proteome technology. Metabolic flux analysis and
metabolome analysis will lead to a deeper understanding of the metabolism and a faster development of new strains. Some of these techniques will
be furthermore used to monitor batches or will
even allow to control bioprocesses. Data acquired
from these technologies could be used together
with online or offline bioprocess data to develop
more efficient neuronal networks or multivariate
statistical process control.
The efficient combination of strain development
and process development will lead to further leaps
in process performances.

Acknowledgements
I would like to thank K. Huthmacher and W.
Pfefferle for continuous support. Furthermore I
am grateful to M. Rieping, B. Bathe, R. Kelle and
G. Thierbach for helpful discussions and a lot of
suggestions. Furthermore I would like to thank J.
Rolando and G. Himmel for many helpful comments.

References
Akesson, M., Hagander, P., Axelsson, J.P., 2001. Avoiding
acetate accumulation in Escherichia coli cultures using
feedback control of glucose feeding. Biotechnol. Bioeng.
73, 223 /230.
Albert, S., Kinley, R.D., 2001. Multivariate statistical monitoring of batch processes: an industrial case study of fermentation supervision. Trends Biotechnol. 19, 53 /62.
Amanullah, A., Baba, A., McFarlane, C.M., Emery, A.N.,
Nienow, A.W., 1993. In: Nienow, A.W. (Ed.), Bioreactor
and Bioprocess Fluid Dynamics. Mechanical Engineering
Publications, London, p. 381.
Amanullah, A., McFarlane, C.M., Emery, C.M., Nienow,
A.W., 2001. Scale-down model to simulate spatial pH
variations in large-scale bioreactors. Biotechnol. Bioeng.
73, 390 /399.
Arnold, S., Becker, T., Delgado, A., Emde, F., Enenkel, A.,
2002. Optimizing high strength acetic acid bioprocess by

169

cognitive methods in an unsteady state cultivation. J.

Biotechnol. 97, 133 /145.
Axe, D.D., Bailey, J.E., 1995. Transport of lactate and acetate
through the energized cytoplasmic membrane of Escherichia
coli . Biotechnol. Bioeng. 47, 8 /19.
Azuma, T., Nakanishi, T., 1988. Factors affecting L-arginine
production in the continuous culture of an L-arginine
producer of Cornynebacterium acetoacidophilum . J. Ferment. Technol. 66, 285 /290.
Azuma, T., Nakanishi, T., Sugimoto, M., 1988. Isolation and
characterization of stable L-arginine producers from continuous culture broth of Corynebacterium acetoacidophilum .
J. Ferment. Technol. 66, 279 /284.
Banuelos, O., Casqueiro, J., Gutierrez, S., Martin, J.F., 2000.
Overexpression of the lys 1 gene in Penicillium chrysogenum :
homocitrate synthase levels, alpha-aminoadipic acid pool
and penicillin production. Appl. Microbiol. Biotechnol. 54,
69 /77.
Binder, M., Uffmann, K.E., 2002. Aqueous lysine-containing
animal feed supplements and process for the production
thereof. US patent 6,465,025.
Budzinski, A., 2001. Aminosauren, Peptide und die Chemie
dazu. Chemische Rundschau 6, 10.
Campolo, M., Paglianti, A., Soldati, A., 2002. Fluid dynamic
efficiency and scale-up of a retreated blade impeller CSTR.
Ind. Eng. Chem. Res. 41, 164 /172.
Cocaign-Bousquet, M., Guyonvarch, A., Lindley, N.D., 1996.
Growth rate dependent modulation of carbon flux through
central metabolism and the kinetic consequences for glucose-limited chemostat cultures of Corynebacterium glutamicum . Appl. Environ. Microbiol. 62, 429 /436.
Cooney, C.L., Wang, D.I.C., Mateles, R.I., 1968. Measurement
of heat evolution and correlation with oxygen consumption
during microbial growth. Biotechnol. Bioeng. 11, 269 /281.
Dassler, T., Maier, T., Winterhalter, C., Bock, A., 2000.
Identification of a major facilitator protein from E. coli
involved in efflux of metabolites of the cysteine pathway.
Mol. Microbiol. 36, 1101 /1112.
Debabov, V.G., 2003. The threonine story. In: Scheper, T.
(Ed.), Advances in Biochemical Engineering, vol. 79.
Springer, Berlin, pp. 113 /136.
de Graaf, A.A., Eggeling, L., Sahm, H., 2001. Metabolic
engineering for L-lysine production by Corynebacterium
glutamicum . Adv. Biochem. Eng. Biotechnol. 73, 9 /29.
de Hollander, J.A., Eswilder, F.R., Noordover, J.A.C., 1997.
Improvement of amino acid fermentation processses. EU
Patent application EP 0796916.
Delaunay, S., Uy, D., Baucher, M.F., Engasser, J.M., Guyonvarch, A., Goergen, J.L., 1999. Importance of phosphoenolpyruvate carboxylase of Corynebacterium glutamicum
during the temperature triggered glutamic acid fermentation. Metab. Eng. 1, 334 /343.
Delaunay, S., Lapujade, P., Engasser, J.M., Goergen, J.L.,
2002. Flexibility of the metabolism of Corynebacterium
glutamicum 2262, a glutamic acid-producing bacterium, in
response to temperature upshocks. J. Ind. Microbiol.
Biotechnol. 28, 333 /337.

170

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

Eggeling, L., Sahm, H., 1999. L-Glutamate and L-lysine:

traditional products with impetuous developments. Appl.
Microbiol. Biotechnol. 52, 146 /153.
Einsele, A., 1978. Process Biochem. 7, 1.
Emde, F., 1997. Reaktionstechnische Untersuchungen zum
Einsatz kaskadierter Saulenreaktoren im Bereich biochemischer Stoffumwandlungen. Ph.D. thesis, RWTH Aachen,
Germany.
Enfors, S.O., Jahic, M., Rozkov, A., Xu, B., Hecker, M.,
Jurgen, B., Kruger, E., Schweder, T., Hamer, G., OBeirne,
D., Noisommit-Rizzi, N., Reuss, M., Boone, L., Hewitt, C.,
McFarlane, C., Nienow, A., Kovacs, T., Tragardh, C.,
Fuchs, L., Revstedt, J., Friberg, P.C., Hjertager, B.,
Blomsten, G., Skogman, H., Hjort, S., Hoeks, F., Lin,
H.Y., Neubauer, P., van der Lans, R., Luyben, K., Vrabel,
P., Manelius, A., 2001. Physiological responses to mixing in
large scale bioreactors. J. Biotechnol. 85, 175 /185.
Fischer, A., Martin, C., Muller, J., 2001. Verfahren zur
Aufreinigung von aminosaurehaltigen Losungen durch
Elektrodialyse. German patent application DE 19952961
A1.
Gerigk, M.R., Bujnicki, R., Ganpo-Nkwenkwa, E., Bongaerts,
J., Sprenger, G., Takors, R., 2002a. Process control for
enhanced L-phenylalanine production using different recombinant E. coli strains. Biotechnol. Bioeng. 80, 746 /754.
Gerigk, M.R., Maa, D., Kreutzer, A., Sprenger, G., Bongaerts, J., Wubbolts, M., Takors, R., 2002b. Enhanced
pilot-scale fed-batch L-phenylalanine production using recombinant E. coli by fully integrated reactive extraction.
Bioprocess Biosyst. Eng. 25, 43 /52.
Glassey, J., Montague, G., Mohan, P., 2000. Issues in the
development of an industrial bioprocess advisory system.
Trends Biotechnol. 18, 136 /141.
Gondorf, A.H., 2001. Fouling von Fermentationsmedien bei
der kontinuierlichen Kurzzeitsterilisation in kompakten
Warmeubertragern. Fortschr.-Ber. VDI Reihe 3/697. PhD
thesis, Universitat Dortmund, Germany.
Gourdon, P., Lindley, N.D., 1999. Metabolic analysis of
glutamate production by Corynebacterium glutamicum .
Metab. Eng. 1, 224 /231.
Hadj-Sassi, A., Deschamps, A.M., Lebeault, J.M., 1996.
Process analysis of L-lysine fermentation with Corynebacterium glutamicum under different oxygen and carbon
dioxide supplies and redox potentials. Process Biochem.
31, 493 /497.
Hadj-Sassi, A., Fauvart, L., Deschamps, A.M., Lebeault, J.M.,
1998. Fed-batch production of L-lysine by Corynebacterium
glutamicum . Biochem. Eng. J. 1, 85 /90.
Hermann, T., Rieping, M., 2002. Fermentation process for the
preparation of L-threonine. WO Patent application
WO0218543.
Hirao, T., Nakano, T., Azuma, T., Sugimoto, M., Nakanishi,
T., 1989. L-Lysine production in continuous culture of an Llysine hyperproducing mutant of Corynebacterium glutamicum . Appl. Microbiol. Biotechnol. 32, 269.
Hofler, A., Alt, H.C., Klasen, C.J., Friedrich, H., Hertz, U.,
Morl, L., Schutte, R., 1997. Process for the preparation of

an animal feed supplement based on fermentation broth.

European patent application EP 809940.
Honda, H., Kobayashi, T., 2000. Fuzzy control of bioprocess in
Japan. Biotechnol. Annu. Rev. 5, 1 /23.
Horiuchi, J.-I., Hiraga, K., 1999. Industrial application of fuzzy
control to large-scale recombinant vitamin B2 production. J.
Biosci. Bioeng. 87, 365 /371.
Hua, Q., Fu, P.-C., Yang, C., Shimizu, K., 1998. Microaerobic
lysine fermentations and metabolic flux analysis. Biochem.
Eng. J. 2, 89 /100.
Ikeda, M., 2003. Amino acid production processes. Biotechnological manufacture of lysine. In: Scheper, T. (Ed.),
Advances in Biochemical Engineering, vol. 79. Springer,
Berlin, pp. 1 /36.
Ikeda, M., Katsumata, R., 1999. Hyperproduction of tryptophan by Corynebacterium glutamicum with the modified
pentose phosphate pathway. Appl. Environ. Microbiol. 65,
2497 /2502.
Itoyama, T., Kobayashi, M., Mitani, Y., Usui, N., 2001.
Method for producing basic amino acids, e.g. L-lysine, arginine and -histidine, in which CO2 is used to act as a
counter ion to the basic amino acid cations, and so reduce
the cost of raw materials e.g. ammonium sulphate. European patent application 1,182,261
Jain, D., Buckland, B.C., 1988. Bioprocess Eng. 3, 31.
Junker, B., 2001. Technical evaluation of the potential for
streamlining of equipment validation for fermentation
applications. Biotechnol. Bioeng. 74, 49 /61.
Kim, D.I., 1995. Nonmunjip /Sanop Kwahak Kisul Yonguso
(Inha Taehakkyo) 23, 523.
Kim, Y.U., Ko, J.H., Lee, J.H., Lim, S.J., Oh, M.S., Shin,
H.C., Sung, J.S., 2000. Culture method for L-lysine producing strain obtained by artificial mutation and process for
producing L-lysine. Korean patent application KR00 27663.
Kimura, E., 2003. Metabolic engineering of glutamate production. In: Scheper, T. (Ed.), Advances in Biochemical
Engineering, vol. 79. Springer, Berlin, pp. 37 /58.
Kimura, E., Kikuchi, Y., Kawahara, Y., Goto, S., Kurahashi,
O., Nakamatsu, T., 2002. Stress-resistant microorganism
and method of producing fermentative product. US patent
6,338,956.
Kinoshita, S., Udaka, S., Shimono, M., 1957. Studies on the
amino acid fermentation. Part I. Production of L-glutamic
acid by various microorganisms. J. Gen. Appl. Microbiol. 3,
193 /205.
Kishimoto, M., Allsop, P., Moo-Young, M., 1991a. A fuzzy
expert system for the optimization of glutamic acid production. Bioprocess Eng. 6, 163 /172.
Kishimoto, M., Kitta, Y., Takeuchi, S., Nakajima, M.,
Yoshida, T., 1991b. Computer control of glutamic acid
production based on fuzzy clusterization of culture phases.
J. Ferment. Bioeng. 72, 110 /114.
Konstantinov, K.B., 1996. Monitoring and control of the
physiological state of cell cultures. Biotechnol. Bioeng. 52,
271 /289.
Konstantinov, K.B., Nishio, N., Seki, T., Yoshida, T., 1991.
Physiologically motivated strategies for control of the fed

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

batch cultivation of recombinant Escherichia coli for
phenylalanine production. J. Ferment. Bioeng. 71, 350 /355.
Koyoma, Y., Ishii, T., Kawahara, Y., Koyoma, Y., Shimizu,
E., Yoshioka, T., 1998. Method for producing L-glutamic
acid by continuous fermentation. European patent application EP844308.
Kumagai, H., 2000. Microbial production of amino acids in
Japan. In: Scheper, T. (Ed.), Advances in Biochemical
Engineering, vol. 69. Springer, Berlin, Germany, pp. 71 /85.
Kusumoto, I., 2001. Industrial production of L-glutamine. J.
Nutr. 131, 2552S /2555S.
Kwong, S.C.W., Rao, G., 1991. Utility of culture redox
potential for identifying metabolic state changes in amino
acid fermentation. Biotechnol. Bioeng. 38, 1034 /1040.
Kwong, S.C.W., Rao, G., 1992. Effect of reducing agents in an
aerobic amino acid fermentation. Biotechnol. Bioeng. 40,
851 /857.
Lee, I., Lee, K., Namgoong, K., Lee, Y.-S., 2002. The use of ion
exclusion chromatography to achieve a more efficient lysine
recovery from fermentation broth. Enzyme Microb. Technol. 30, 798 /803.
Lennox, B., Montague, G.A., Hiden, H.G., Kornfeld, G.,
Goulding, P.R., 2001. Process monitoring of an industrial
fed-batch fermentation. Biotechnol. Bioeng. 74, 125 /135.
Leuchtenberger, W., 1996. Amino acids */technical production
and use. In: Rehm, H.-J., Reed, G. (Eds.), Biotechnology:
Products of Primary Metabolism, vol. 6. VCH-Verlag,
Weinheim, Germany, pp. 465 /502.
Marienfeld, S., Uhlemann, E.M., Schmid, R., Kramer, R.,
Burkovski, A., 1997. Ultrastructure of the Corynebacterium
glutamicum cell wall. Antonie Van Leeuwenhoek 72, 291 /
297.
Miwa, H., Tamura, K., Koyama, Y., Tsurata, M., Tosaka, O.,
Shimazaki, K., Nakamura, T., Nakayama, T., 1992. Process
and apparatus for regulating the concentration of the
carbon source in the aerobic culture of a microorganism.
French patent 2,669,935.
Morbach, S., Kramer, R., 2002. Body shaping under water
stress: osmosensing and osmoregulation of solute transport
in bacteria. Chembiochem 3, 384 /397.
Motoyama, H, Yano, H, Ishino, S, Anazawa, H, Teshiba, S.,
1994. Effects of the amplification of the genes coding for the
L-threonine biosynthetic enzymes on the L-threonine production from methanol by a gram-negative obligate methylotroph, Methylobacillus glycogenes . Appl. Microbiol.
Biotechnol. 42, 67 /72.
Motoyama, H., Yano, H., Terasaki, Y., Anazawa, H., 2001.
Overproduction of L-Lysine from methanol by Methylobacillus glycogenes derivatives carrying a plasmid with a
mutated dapA gene. Appl. Environ. Microbiol. 67, 3064 /
3070.
Nakayama, K., Kitada, S., Kinoshita, S., 1961. Studies on
lysine fermentation I. The control mechanism on lysine
accumulation by homoserine and threonine. J. Gen. Appl.
Microbiol. 7, 145 /154.
Ohtani, M., Kitahara, T., Akashi, K., 1987. Process for the
purification of L-threonine. French patent 2,588,016.

171

Oka, T., 1999. Amino acids, production processes. In: Flickinger, M.C., Drew, S.W. (Eds.), Encyclopedia of Bioprocess
Technology: Fermentation, Biocatalysis, and Bioseparation.
Wiley, London, p. 89.
Okamoto, K., Ikeda, M., 2000. Development of an industrial
stable process for L-threonine fermentation by an Lmethionine-auxotrophic mutant of Escherichia coli . J.
Biosci. Bioeng. 89, 87 /89.
Osterberg, T., Wadsten, T., 1999. Physical state of L-histidine
after freeze-drying and long-term storage. Eur. J. Pharm.
Sci. 8, 301 /308.
Pfefferle, W., Lotter, H., Friedrich, H., Degener, W., 1993.
Process for the fermentative production of amino acids.
European patent 532,867.
Pfefferle, W., Mockel, B., Bathe, B., Marx, A., 2003. Biotechnological manufacture of lysine. In: Scheper, T. (Ed.),
Advances in Biochemical Engineering, vol. 79. Springer,
Berlin, pp. 59 /112.
Pham, C.B., 1994. Preparation of L-lysine by repeated fed batch
fermentation using mutant strains of auxotrophic homoserine in fermentation medium containing coconut water
enriched by sugar or sugar substrate. Patent Philippines PH
27,995.
Pham, C.B., Matsumura, M., Kataoka, H., 1995. Microb. Util.
Renewable Resour. 9, 539.
Pollard, D.J., Kirschner, T.F., Hernandez, D., Hunt, G.,
Olewinski, R., Salmon, P.M., 2002. Pilot-scale process
sensitivity studies for the scaleup of a fungal fermentation
for the production of pneumocandins. Biotechnol. Bioeng.
78, 270 /279.
Radmacher, E., Vaitsikova, A., Burger, U., Krumbach, K.,
Sahm, H., Eggeling, L., 2002. Linking central metabolism
with increased pathway flux: L-Valine accumulation by
Corynebacterium glutamicum . Appl. Environ. Microbiol.
68, 2246 /2250.
Research disclosure, 2000. Research disclosure 431, 427.
Reuss, M., Schmalzriedt, S., Jenne, M., 2000. Application of
Computational Fluiddynamics (CFD) to Modelling Stirred
Tank Bioreactors. In: Schugerl, K., Bellgardt, K.-H. (Eds.),
Bioreaction Engineering. Springer, Berlin.
Roubos, J.A., van Straten, G., van Boxtel, A.J.B., 1999. An
evolutionary strategy for fed-batch bioreactor optimization;
concepts and performance. J. Biotechnol. 67, 173 /187.
Ruebenhagen, R., Morbach, S., Kraemer, R., 2001. The
osmoreactive betaine carrier BetP from Corynebacterium
glutamicum is a sensor for cytoplasmic K
. EMBO J. 20,
5412 /5420.
Sahm, H., Eggeling, L., Eikmanns, B., Kramer, R., 1996.
Construction of L-lysine-, L-threonine-, and L-isoleucineoverproducing strains of Corynebacterium glutamicum .
Ann. New York Acad. Sci. 782, 25 /39.
Sanders, M.E., 1993. Bacteriophages in industrial fermentations. In: Webster, R.G., Granoff, A. (Eds.), Encyclopedia
of Virology. Academic Press, London, pp. 117 /121.
Schendel, F.J., Bremmon, C.E., Flickinger, M.C., Guettler, M.,
Hanson, R.S., 1990. L-lysine production at 50 8C by

172

T. Hermann / Journal of Biotechnology 104 (2003) 155 /172

mutants of a newly isolated and characterized methylotrophic Bacillus sp. Appl. Environ. Microbiol. 56, 963 /970.
Schilling, B.M., Pfefferle, W., Bachmann, B., Leuchtenberger,
W., Deckwer, W.-D., 1999. A special reactor design for
investigations of mixing time effects in a scaled-down
industrial L-lysine fed-batch fermentation process. Biotechnol. Bioeng. 64, 599 /606.
Schmalzriedt, S., Reuss, M., 1999. Einsatz von CFD zur
Analyse biotechnischer Stoffumsetzungen in ein- und zweiphasig durchstromten Ruhrkesselreaktoren. Chem.-Ing.Tech. 71 (9), 954 /955.
Schweder, T., Kruger, E., Xu, B., Jurgen, B., Blomsten, G.,
Enfors, S.O., Hecker, M., 1999. Monitoring of genes that
respond to process-related stress in large-scale bioprocesses.
Biotechnol. Bioeng. 65, 151 /159.
Shimizu, H., Takiguchi, N., Tanaka, H., Shioya, S., 1999. A
maximum production strategy of lysine based on a simplified model derived from a metabolic reaction network.
Metabol. Eng. 1, 299 /308.
Shioya, S., Shimizu, H., Takiguchi, N., 1999. In: Lee, S.Y.,
Papoutsakis, E.T. (Eds.), Metabolic Engineering. Marcel
Dekker, New York, p. 227.
Shu, C., Liao, C., 2002. Optimization of L-phenylalanine
production of Corynebacterium glutamicum under product
feedback inhibition by elevated oxygen transfer rate.
Biotechnol. Bioeng. 77, 131 /141.
Stahl-Wernersson, E., Tragardh, C., 1999. Scale-up of Rushton
turbine-agitated tanks. Chem. Eng. Sci. 54, 4245 /4256.
Takiguchi, N., Shimizu, H., Shioya, S., 1997. An on-line
physiological state recognition system for the lysine fermen-

tation process based on a metabolic reaction model.

Biotechnol. Bioeng. 55, 170 /181.
Thiry, M., Cingolani, D., 2002. Optimizing scale-up fermentation processes. Trends Biotechnol. 20, 103 /105.
Udaka, S., 1960. Screening method for microorganisms accumulating metabolites and its use in the isolation of Micrococcus glutamicus . J. Bacteriol. 79, 754 /755.
Uffmann, K.E., Binder, M., 2002. Aqueous L-lysine-containing
animal feed supplements and process for the production
thereof. US Patent 6,340,486.
Wada, M., Awano, N., Haisa, K., Takagi, H., Nakamori, S.,
2002. Purification, characterization and identification of
cysteine desulfhydrase of Corynebacterium glutamicum , and
its relationship to cysteine production. FEMS Microbiol.
Lett. 217, 103 /107.
Weuster-Botz, D., Kelle, R., Frantzen, M., Wandrey, C., 1997.
Substrate controlled fed-batch production of L-lysine with
Corynebacterium glutamicum . Biotechnol. Prog. 13, 387 /
393.
Xu, B., Jahic, M., Blomsten, G., Enfors, S.O., 1999a. Glucose
overflow metabolism and mixed-acid fermentation in aerobic large-scale fed-batch processes with Escherichia coli .
Appl. Microbiol. Biotechnol. 51, 564 /571.
Xu, B., Jahic, M., Enfors, S.O., 1999b. Modeling of overflow
metabolism in batch and fed-batch cultures of Escherichia
coli . Biotechnol. Prog. 15, 81 /90.
Yao, H.M., Tian, Y.C., Tade, M.O., Ang, H.M., 2001.
Variations and modelling of oxygen demand in amino
acid production. Chem. Eng. Proc. 40, 401 /409.