98

reviews

High cell-density culture of

Escherichia coli
SangYup Lee

Escherichia coli is the most widely used prokaryotic system for the synthesis of
heterologous proteins. Once an optimal expression system has been constructed,
protein production can be enhanced by increasing the production of protein per
cell per unit time (specific productivity), or by increasing the cell concentration per
unit time (cell productivity). Various high cell-density culture (HCDC)techniques have
been developed for growing recombinant and non-recombinant E. coli strains in
fed-batch cultures at concentrations greater than 100 grams (dry cell weight)
per liter. This article reviews the problems encountered in HCDC of E. coil, and
discusses various solutions. Feeding strategies for HCDCof E. coil, and the results
obtained using them, are also described.

Recombinant human proteins have become impor-
tant as biological pharmaceuticals, and the annual
market growth is expected to increase at a rate of
5-15% per annum. The combination of recombinant
DNA technology and large-scale culture processes has
enabled these proteins to be produced in quantities
that might otherwise have been difficult, if not imposs-
ible, to obtain from natural sources. Escherichia coli is
the most commonly used host for protein production
mainly because it is such a well-characterized system.
Although E. coli cannot be used to produce some large,
complex proteins (e.g. containing multiple disulfide
bonds and, in particular, unpaired thiols), or proteins
that require post-translational modification for activity,
other proteins, such as interferons, interleukins, colony-
stimulating factors, growth hormones, insulin-like
growth factors and human serum albumin, are among
those that have been successfully produced using
recombinant E. coli (tkef. 1). Many of these proteins are
produced in the form of insoluble, biologically inactive
inclusion bodies, from which biologically active pro-
teins can only be recovered by complicated and costly
denaturation and refolding processes 2. However, recent
advances in our understanding of protein folding and
translocation, and of the roles of molecular chaperones
and foldases in these processes, have made possible the
design of recombinant E. coli strains that accumulate
proteins in soluble form, and strains that secrete pro-
teins3. These advances will further strengthen the domi-
nant status orE. coli as a host for protein production.
S. 2 Lee is at the Department of Chemical En2ineering, Kowan
Advanced institute of Sciena" and Technology (KAIST), Taejon
305-701, Korea.
A primary goal of fermentation research is the cost-
effective production of desired products using high
productivity (i.e. the amount of product formed per
unit volume per unit time) techniques. To date,
biotechnology companies have been producing pharma-
ceutical proteins on a small scale - typically a few tens
of kilograms per annum - and have still made profits
by selling at a high price 4. However, when there is a
competing drug, or a competing company producing
the same protein, this situation changes, and reduc-
tion of production costs of these proteins becomes
more important. For example, bovine somatotropin is
expected to have an annual market of up to 100 tonnes
in the USA alone, and low production costs are criti-
cal to its success 4. Because most proteins are intra-
cellularly accumulated in recombinant E. coli, pro-
ductivity is proportional to the final cell-density and
the specific productivity (i.e. the amount of product
formed per unit cell mass per unit time). High cell-
density culture (HCDC) techniques for culturing
E. coil have been developed to improve productivity,
and also to provide advantages such as reduced culture
volume, enhanced downstream processing, reduced
wastewater, lower production costs and reduced
investment in equipment.
Fed-batch processes have most often been used to
obtain high cell-density5-7. Fed-batch culture uses
inocula growing at the maxinmm specific growth rate
that can be sustained using the nutrients initially
present in a fermenter; various regimes of nutrient
feed are then imposed until fermentation is complete.
Cell concentrations of greater than 50 grams dry cell
weight per liter [g(DCW) V] can be routinely obtained
by fed-batch culture both of non-recombinant and

-IBTECH MARCH 1996 (VOL 14) Copyright 1996, Elsevier Science Ltd. AH rights reserved. 0167 - 7799/96/$15.00
 99

reviews

Table 1. High cell-density culture of non-recombinant Escherichia coli

Final cell
Temperature Culture method/ Cultivation mass
Host strain Medium a (C} feeding mode time (h) [g(DCW) 1-1] Ref.

B54125 Defined 36.5 Exponential 10 86 24

TG1 Defined 28 Exponential 24 128 71
TG1 Defined 28 Exponential 44 148 71
(glycerol)
W3350 Defined 37 Specific growth-rate 21 53 11
control
TG1 Defined 28 Specific growth-rate 35 110 10
control
B(OSU333) Defined 32 Glucose concentration 9 65 36
control
W3110 Defined 37 Constant feeding with 23 174 16
(glycerol) dialysis
W Defined 37 pH-stat 36 105.4 57
(sucrose)
ATCC10536 Semi-defined 37 DO-stat 12 110.2 13
B Semi-defined 36 DO-stat 11 125 18
ATCC8739 Defined 30 DO-stat 14 111.4 54
ATCC8739 Defined 37 DO-stat 12 105 54

a Exceptwhere indicated, glucosewas usedas the carbon source.

recombinant E. coli (Refs 5,6). The development of
H C D C techniques for E. coli has led to efficient, high-
level production of various proteins and non-protein
products such as amino acids s and poly(3-hydroxy-
butyrate) (Ref. 9). However, this technique has sev-
eral drawbacks, including: substrate inhibition, limited
oxygen transfer capacity, the formation of growth-
inhibitory by-products, and limited heat dissipation.
This article reviews various approaches to solving
these problems, and the development of growth media
and feeding strategies for H C D C of E. coll.
Development of growth media
Some nutrients, including carbon and nitrogen
sources, can inhibit cell growth when they are present
above a certain concentration. This explains why just
increasing the concentrations of nutrients in batch-
culture media does not yield high cell-densitT. In
general, E. coli growth is inhibited when the follow-
ing nutrients are present above certain concentrations
(shown in brackets): glucose (50gl 1), ammonia
(3 gl 1), iron (1.15 gl<), magnesium (8.7 gl-I), phos-
phorous (10gl <) and zinc (0.038gl -~) (Ref. 6).
Therefore, H C D C s are started with concentrations
below the inhibitory thresholds, and nutrients are
added as necessaW to maintain high growth-rates.
There are three types of media: defined, complex
and semi-defined. Defined media are generally used
to obtain high cell-density, as the nutrient concen-
trations are known and can be controlled during
culture. Nutrients in complex media, such as peptone
and yeast extract, can val2v
.- in composition and quality,
which makes fermentation with these complex media
less reproducible. However, semi-defined or complex
media are sometimes necessary to boost product for-
marion. To grow cells to a high density; it is necessary,
to design a balanced nutrient medium that contains
all the necessary components for supporting cell
growth, while avoiding inhibition. It is desirable to
make the feed-solution as simple as possible by includ-
ing sufficient non-carbon and non-nitrogen nutrients
in the starting medium. Riesenberg et al. 1 developed
a defined medium that excluded glucose, nlagnesium
and a nitrogen source from the initial medium, and
used this to obtain 110g(DCW)l t of E. coil TG1
in a fed-batch culture. Ammonium hydroxide is
frequently used as a nitrogen source in H C D C s
because it can also be used as a base to control culture
pH.
A balanced, defined medium has been developed
by considering the mass composition of dry E. coli
cells and the yield coefficients of the medium con>
ponents 1-13. In another approach, a pulse-injection
technique has been used to design an optimized
medium by challenging the culture with various
amino acids, vitamins, salts and trace elements in a
steady-state chemostad 4. The development of semi-
defined and complex media has required empirical
and trial-and-error-based processes. The optimization
of any fermentation medium is a labor-intensive
process, owing to the large number of nutrient com-
binations to be tested. The Plackett-Burman tech-
nique can be used to help design optimized media ts.
Maximum cell concentration
Several approaches have been used to determine the
maxinmm concentration of cells in a tightly packed
fermenter, l<iesenberg f' calculated that under these

TIBTECHMARCH1996 (VOL 14)

 100

reviews

Table 2. High cell-density culture of recombinant Esherichia coli

Temperature Culture method/ Culture

Host strain Mediuma (C) feeding mode time (h)

MC1061 Complex 30 Constant 24

K12 Semi-defined 30-42 b Stepwise increase 15
MC1061 Complex 30 Stepwise increase 26
AM-7 Defined 30-37-42 Stepwise increase 50
X90 Defined 37 Exponential 31.2
B Semi-defined 37 Exponential 12
JE5505 Semi-defined 30 Exponential (Stepwise increase) 46
KA197 Defined 30-40 Exponential-constant 21c
TG1 Defined 28 Specific growth-rate control 23
RV308 Defined 26 Specific growth-rate control 29
BMH-71-18 Defined 37 Specific growth-rate control 15
TG1 Defined 28 Specific growth-rate control _e
(Exponential)d
TG1 Defined 30-38 Specific growth-rate control 7f
(Exponential)
W3] 10 Semi-defined 37 Specific growth-rate control 26
(Exponential)
C600 Defined 36 pH-stat 26
XL1-Blue Complex 37 pH-stat 39
XL1-Blue Semi-defined 37 pH-stat 51
W Defined (sucrose) 37 pH-stat 48
DH1 Semi-defined 30 DO-stat 21
M5248 Semi-defined 30-42 DO-stat 21
JMI03 Semi-defined 37 DO-stat with cross-flow filtration 20
AT2471 Defined 38.5 Balanced DO-stat 56
HW21-2 Defined 30 Glucose concentration control -
MM-294 Defined 30-37 Glucose concentration control -
IF03301 Defined 37 Glucose concentration 9
control (glucose powder)
HBI01 Semi-defined 37 Acetate concentration 20
monitoring
JMI03 Defined 37 Fuzzy neural network 23

HBI01 Semi-defined 28 Membrane cell recycling 29

aExceptwhenindicated,glucosewas usedas the carbon source.

bTemperaturechangedfor the inductionof temperature-sensitiveXPL or kP R promoter.
Timetakenfrom 4g(DCW)I 1.
Specific growth-ratecontrolledby exponentialfeeding.
eNot reported.
fTime after induction.
gPoly(3-hydroxybutyrate).

conditions; the maximum cell concentration ofE. coli
is 400 g(DCW)1-1. Markl et al. 16 calculated that in
cultures containing tightly packed cells 3 b~m long and
1 Ixm in diameter, only 25% of the culture would be
culture medium. Considering that the dry weight of
cells is 20-25% of the wet weight, and that the den-
sity of the ceils is slightly greater than that of water, the
maximum cell concentration is 160-200 g(DCW) 1-1
(Ref. 16). A maximum cell concentration of
~200g(DCW)1-1 appears reasonable, as two of the
highest cell-densities obtained to date [174 g(DCW) 1-1
for E. coli W3110 in a dialysis reactor 16, and 175.4
g(DCW) 1-1 for recombinant E. coli producing poly(3-
hydroxybutyrate) (R.ef. 17)] approach this value.
The viscosity of culture broth increases sharply
when the cell concentration exceeds 200 g(DCW) 1-1,
and the culture broth almost loses its fluidity above
220g(DCW) 1-i (Ref. 18). This finding supports the
theory that the maximum attainable cell density of
E. coil is -200 g(DCW) ~1.
Problems with H C D C
Acetate is produced when E. coli is grown under
anaerobic or oxygen-limiting conditions; however,
E. coli cultures growing in the presence of excess
glucose can also produce acetate even under aerobic
conditions 19-21.Acetate is produced when carbon flux
into the central metabolic pathway exceeds the
biosynthetic demands and the capacity for energy
generation within the ce112,22,23; saturation of the
tricarboxylic acid (TCA) cycle and/or the electron
transport chain may be the main cause2,2s. A high

IBTECHMARCH1996 (VOL 14)

 101

revigWS

Final cell mass Product

[g(DCW) I-z] Product concentration Ref.

0D525= 120 Human growth hormone 1.08 g 1-1 26

20 Human insuling-like growth factor-1 600 mg I-z 69
0D525= 11 Human growth hormone 1.75 g I-t 26
68 Human (z consensus interferon 5.6 g 1-1 70
92 Trypsin 56 mg Iq 53
55 Human interleukin 113 2.15 g I-z 50
26 Human leukocyte interferon 1 x 1091U1-1 34
77 ProteinA-I~-galactosidase fusion 19.2 g l t 47
60 Human interferon e l 5.5 1081Ug(DCW)-z 33
50 Mini-antibody 1.04 g H 67
58 Human interferon cd 1.26 1091U1-1 12
95.5 Aprotinin-13-galactosidase fusion 2.85 x 106UI 1 52

40 Human parathyroid hormone 338 mg 1-1 52

OD6o o = 100 Bovine somatotropin 2.9 g I-] 35

63 13-isopropylmalate dehydrogenase 16.00 U g(protein) -1 59

101.4 PHBg 81.2 g 1-1 58
175.4 PHB 65.5 g 1-1 17
124.6 PHB 34.3 g 1-1 57
0D660 = 134.4 Human Proapo A-I 6.0 g 1-1 55
59.5 Human interleukin 2 1.2 g 1-1 56
125 Bacillus thuringiensis toxin 6.6 g 1-1 28
36 Phenylalanine 46 g 1-1 8
0D68o= 90 Human interleukin 2 - 29
0D68o= 75 Human interleukin 2 3.3 g H 41
102 E. coil tryptophan synthase 1.6 x 105U g(protein) -1 46

21 Human epidermal growth factor 60 mg 1-1 68

84 6-galactosidase 4600 U OD6oo-1 65

145 Penicillin acylase 6.0 U mg< 38

concentration of acetate (i.e. above 5 gl 1 at p H 7)
reduces growth rate, biomass yield and maximum
attainable cell densities in HCDCs (P,.eg 19,20,24--26).
The protonated form of acetate interferes with ATP
production by reducing the ApH contribution to the
proton-motive force 19,26. Furthermore, acetate has
been reported to have a greater detrimental effect on
recombinant cells than on non-recombinant cells2v,
and recombinant-protein production is significantly
reduced by acetate acculnulation 26,28,29. Even though
the exact mechanism o f this harmful effect has not
been elucidated, it is possible that acetate may repress
the synthesis o f D N A , R N A , proteins and lipids 3.
The detrimental effect of acetate is exacerbated by
salts that accumulate as a result of the acid and base
used for pH control 26. In general, acetate formation is
dependent on the medium used and the specific
growth rate. For example, acetate forms in complex
and defined media when the specific growth rate
exceeds 0.2 or 0.35h -1, respectively13,2,31. A study
with a different E. coli strain showed that acetate was
produced in a defined medium when the specific
growth rate was higher than 0.14 h -1 (R_e 32). There-
fore, the critical specific growth rate that leads to
acetate formation varies among strains and depends
on the medium used.
The production o f acetate is greater in fed-batch
cultures than in batch culture owing to the extended
culture period. A number of strategies have been
developed to reduce acetate formation in fed-batch

TIBTECHMARCH1996(VOL14',
 102
reviews
culture by controlling the specific growth rate by
limiting essential nutrients such as sources of carbon
or nitrogen 1,11,-~3-3(' In a different approach, the
acetate concentration has been significantly reduced
by removing the culture broth in situ using dialysis 3v
or by recycling the culture medium 3s. However, these
processes are difficult to scale-up, and considerable
amounts of nutrients can be wasted.
Acetate formation can also be reduced or prevented
by altering the formulation of the medium. Acetate is
not produced when glycerol is used as a carbon
source 22. High cell-densities might be achieved rela-
tively easily using glycerol, but one recent report has
detailed the formation of acetate during H C D C orE.
col; TG1 with glycerol as the carbon source 71. The
lower rate of transport of glycerol into the cell, com-
pared with that of glucose, appears to lead to a reduc-
tion in the flux of carbon through glycolysis, greatly
reducing acetate formation. It should also be noted
that glycerol is more expensive than glucose, and
that cells grow more slowly on glycerol than on
glucose -~2,71. The addition of certain amino acids (e.g.
Gly and Met) can increase the growth rate and the
production of recombinant proteins by alleviating the
harmful effects of acetate > .
Temperature is an important variable that can be
used to control cell metabolism. By lowering the cul-
ture temperature from 37C to 26-30C, nutrient
uptake and growth rate can be reduced, thus reduc-
ing the formation of toxic by-products and the gen-
eration of metabolic heat. Lowering culture tem-
perature also reduces cellular oxygen demand,
which enables higher cell-densities to be obtained
without the need for pure oxygen. Furthermore,
it is possible to reduce the formation of inclusion
bodies for some proteins by growing recombinant
cells at lower temperatures 2. These advantages have
persuaded some researchers to use lower temperatures
for H C D C of E. coli (Refs 26,29,33,34,38,52,55,71).
It is also possible to reduce acetate formation
through metabolic engineering. The enzymes
involved in the formation of acetate from acetyl-CoA
are phosphotransacetylase (Pta) and acetate kinase
(Ack) (Ref. 40). Mutant strains of E. coli that are
defective in Pta and/or Ack have been developed 41,42,
and the Pta negative mutant has been grown to high
cell-density" to produce enhanced levels of human
interleukin 2 (Ref. 41) and lipase 42. In other studies,
acetate formation has been reduced by diverting the
flow of catabolic carbon to the production of less-
harmful metabolites such as ethanol 43 or acetoin 44. An
E. coli strain with a modified glucose uptake rate was
constructed by inactivating enzyme II of the glucose
phosphotransferase system; this mutant produced less
acetate, grew to high cell-densiw, and exhibited
enhanced protein production 45.
Oxygen often becomes limiting in HCI)Cs owing
to its low solubiliw. The saturated dissolved oxygen
(DO) concentration in water at 25C and latin
is ~ 7 m g l 1 but oxygen supply can be increased
by increasing the aeration rate or agitation speed.
IBTECHMARCH1996 (VOL 14)
Oxygen-enriched air or pure oxygen has also been
used to prevent oxygen limitation. However, pure
oxygen is expensive and increases production costs
when used in large quantities. Cells can also be
cultured under pressurized conditions to increase oxy-
gen transfer 4(',47. As oxygen consumption increases
with growth rate, the oxygen demand of cells can be
reduced by lowering the growth rate. It has been
possible to obtain high cell-densities of various E. coli
strains by combining these strategies. The expression
~" q '
of a ~ttreo.alla hemoglobin in a recombinant E. coli
improved microaerobic cell growth and the overall
production of recombinant proteins 4s.
Another physical limitation of H C D C is the reduced
mixing efficiency of the fermenter, a problem that
increases with fermenter size. Cells close to the
injection port are exposed to high concentrations of
nutrients, whereas at other locations, cells are starved
of substrate 49. Therefore, it is important to study the
mixing patterns in a fermenter, and to find ways to
improve mixing.
Carbon dioxide can also affect cell growth in
H C D C . High partial-pressure of CO~ (>0.3ann)
decreases growth rate and stimulates acetate for-
mation 18,24. Therefore, the pressurized culture regime
used to increase oxygen transfer may also enhance the
detrimental effect of C O 2. Heat generation can also
become a problem in H C D C , especially in large-scale
fermentations. The rate of heat dissipation decreases
in larger fermenters, as a result of the reduced ratio of
cooling-jacket surface area to fermenter volume 32,5.
The major sources of heat are the mechanical energy
of agitation and metabolic heat from cells. These
problems can be partially solved by growing cells at a
reduced specific growth rate.
Strategies for HCDC
The method of nutrient feeding (Box 1) is critical
to the success of H C D C , as it not only affects the
maximum attainable cell concentration, but also cell
productivity. Product formation can also be affected
by varying feeding strategy2<51. HCDCs are normally
carried out under nutrient (e.g, carbon) limiting con-
ditions, but simple feeding strategies such as constant-
rate feeding 1(',2<s1, a stepwise increase of the feeding
rate,,l 1,15.24,26,34 and exponential f e e d i n g 24,34,35,47,51 53
have been used to obtain high cell-densities o f E . coli
in fed-batch cultures.
In constant-rate feeding, concentrated nutrients are
fed into the fermenter at a predetermined rate. Owing
to the increase in culture volume and cell concen-
tration in the fermenter, the specific growth rate con-
tinuously decreases, and the increase in cell concen-
tration slows down over time. The stepwise (or
gradual) increase of feeding rate can enhance cell
growth by supplying more nutrients at higher cell
concentration. Cells can grow exponentially during
the entire culture period if the feed rate of the growth-
limiting substrate is increased in proportion to cell
growth. The exponential-feeding method has been
developed to allow cells to grow at constant specific
 103

reviews

growth rates47,52,53,71;it also provides the advantage
that acetate production can be minimized by control-
ling the specific growth rate below the critical value
of acetate formation. The feeding rate that allows
exponential growth with constant specific growth rate
can be calculated using the following equation, which
is derived from a simple mass balance with the assump-
tion o f constant cell yield on substrate:
in which M s is the mass-flow rate of the carbon source
(g h<), F is the feed-flow rate (1h-l), Sv is the substrate
concentration in the feeding solution (gP~), b~ is the
specific growth rate (h 1), Yws is the cell yield on car-
bon substrate [g(DCW) gl], m is the specific mainte-
nance coefficient [gg(DCW)-I h I], X is the cell con-
centration [g(DCW)I 1], I/is the culture volume (1),
and t0 is the time at which feeding is started.
Exponential feeding is a simple but efficient method
that has been successfully used for the HC1)C of
several non-recombinant and recombinant E. coli
strains; the specific growth rate is usually maintained
at between 0.1 and 0.3h < to avoid acetate for-
marion 47,s2,s3,71. Cell concentrations of 128 g(I)CW) 1-1
and 148 g(DCW) 1-1 have been achieved by exponen-
tial-feeding, with glucose and glycerol, respectively, as
carbon sources71.
More-sophisticated feeding methods with feedback
control-schemes have also been developed, Indirect
feedback control-schemes that couple feeding with
measurement of various physical parameters, such as
1)O (Refs 18,54-56), pH (1Zef~ 17,57-59), microbial
heat (' and C O 2 evolution rate (CER; Ref. 11), have
been developed. The DO-stat method is based on
the finding that the D O increases sharply when the
substrate is depleted. Therefore, the substrate con-
centration can be maintained within a desired range
by automatically adding a predetermined amount
of nutrient when the D O rises above the preset
value t3. A balanced-DO-star method, which guaran-
tees sufficient oxygen supply and prevents overfeed-
ing of nutrients, has also been developed ~'1. The
pH-stat method is based on the observation that the
pH changes when the principal carbon substrate
becomes depleted s7,62. When the carbon substrate is
exhausted, the pH begins to rise, mainly as a result
of the increase in the concentration of ammonium
ions excreted by cells('2. In a defined medium, the
DO-stat responds more rapidly to nutrient depletion
than the pH-stat. When complex carbon-nitrogen
substrates, such as yeast extract or peptone, are
used together with carbohydrate substrates, the D O
change is not as large when the carbon source is
depleted, as the cells continue to utilize the complex
substrates ('2. Therefore, the pH-stat method is more
suitable when semi-defined or complex media are
used. Cells produce C O 2 during growth, and the
C E R is roughly proportional to the carbon-source
consmnption rate. Therefore, nutrient feeding can be

B o x 1. F e e d i n g m e t h o d s in f e d - b a t c h culture

Without f e e d b a c k control
Constant f e e d i n g - feeding nutrient at a predetermined (constant) rate. The specific growth rate continuously
decreases.
Increased feeding - feeding nutrient at an increasing (gradual, stepwise, or linear) rate. The decrease in specific
growth rate can be compensated.
Exponential feeding - feeding nutrient at an exponential rate. Constant specific growth rate can be achieved.
[The term 'specific growth rate (1~)control' is used when the specific growth rate is maintained at a constant level. This
can be achieved by exponential feeding in most cases, but may deviate when unexpected conditions arise during
culture. When this occurs, feedback control is required to achieve constant I~.]

With f e e d b a c k control
Indirect f e e d b a c k control
DO-stat - feeding nutrient when there is a rise in the concentration of dissolved oxygen (DO), which results from
depletion of the substrate.
pH-stat - feeding nutrient when the pH rises as a result of depletion of the principal carbon source.
Carbon dioxide evolution rate (CER) - this is estimated on-line using a mass spectrometer, and is used to
control nutrient feeding. The CER is roughly proportional to the rate of consumption of the carbon source. This method
is most frequently used to control the specific growth rate.
Cell concentration - the nutrient feeding rate is determined from the cell concentration, which is measured on-line
using a laser turbidimeter.

Direct f e e d b a c k control
Substrate concentration control - nutrient feeding is directly controlled by the concentration of the principal
carbon source (e.g. an on-line glucose analyzer is used to control glucose concentration in the fermenter).

TIBTECH MARCH1996 (VOL 141,

 104
repiel/ps
controlled by using C E R data <63, which can be calcu-
lated from the concentration of C O 2 in the gas outlet.
The advantages of maintaining the specific growth
rate below the critical value of acetate formation have
been mentioned previously. Non-recombinant and
recombinant E. coli strains have been grown to high
cell-densities at near-constant specific growth
rates 1I>12,33,3s,s2. Exponential feeding usually allows
cells to grow at the constant specific growth rate s3.
However, the specific growth rate control should be
distinguished from the exponential feeding method,
as the latter cannot compensate for the change of the
specific growth rate that may occur during the culture.
Several methods have been developed for maintaining
the specific growth rate during fed-batch culture.
E. coli K12 W3350 has been grown to a concen-
tration of 53 g(DCW) 1-1 using increased glucose feed-
ing under computer controlll; the feeding rate of the
glucose-salts medium at any time could be calculated
from the amount of glucose fed into the fermenter up
to that point. In another stud?; E. coli TG1 was grown
to a concentration o f 1 1 0 g ( D C W ) E 1 at a constant
specific growth rate by varying glucose feeding with
a pO 2 control loop 1.
Recent developments in on-line sensors and con-
trol strategies 64 have made possible the more-accurate
control of specific growth rates. A fuzzy neural net-
work has been developed to compensate for the expo-
nential feeding rate as determined by the feedforward
control element; this has been used to obtain
84g(DCW) 1-1 o f recombinant E. coli producing
[3-galactosidase 6S. An automated control method of
fed-batch culture, in which the nutrient feed rate is
determined from cell concentration and culture
volume, has been developed using a laser turbidimeter,
which can continuously measure cell concentration 66.
Direct feedback control is also possible by measur-
ing on-line the concentration of the growth-limiting
substrate in the culture broth and adjusting the con-
centration to the preset value by automatic feeding.
For example, an on-line glucose analyzer has been
used successfully to control glucose concentration
within a desired range 29,36,'41,65. The results from the
H C D C of non-recombinant and recombinant E. coli
strains using various feeding strategies are summarized
in Tables 1 and 2, respectively. Various products, such
as amino acids, poly(3-hydroxybutyrate) and recombi-
nant proteins, have been produced successfully to a
high concentration at high productivity using H C D C .
Because the main purpose o f the H C D C o f recom-
binant E. coli is to improve the productivity of the
recombinant product, culture strategies should also
include details of how and when to induce product
formation during culture. This aspect has recently
been reviewed elsewhere 5.
Conclusions
Non-recombinant and recombinant E. coli strains
can both be grown to high density by fed-batch
culture using one of the feeding methods described
above. It is now relatively easy to obtain cell concen-
IBTECHMARCH1996 (VOL 14}
trations >100g(DCW) I 1 by controlling the acetate
concentration below the inhibitory level. As has
recently been reported by several groups 47,52,53,rl, the
exponential-feeding method, which allows the con-
trol of the specific growth rate below the critical value
of acetate formation, appears to be particularly useful,
as it is simple and does not require any special appa-
ratus. The development of the on-line sensor system
and the accurate control algorithms will allow further
enhancement of H C D C techniques.
References
1 Hodgson, J. (1993) Bio/Technology /1,887-893
2 Kane, J. F. and Hartley, 11. L. (1988) "l}ends Biotedmol. 6, 95-101
3 Hockney, IC C. (1994) Trends Biotedmol. 12, 45(>463
4 Datar, R. V., Cartwright, T. and Rosen, C-G. (1993) Bio/Tedmology
1I, 349-.357
5 Yee, L. and Blanch, H. W. (1992) Bio/TFchnolo2y 10, 1550-1556
6 Riesenberg, 1). (1991) Curt. Opin. Biotechnol. 2, 380-384
7 Kleman, G. L. and Strohl, W. R. (1994) Cun. Opir~. Biotedmol. 5,
180-186
8 Konstantinov, K. B., Nishio, N., Seki, T. and Yoshida, T. (1991)
J. Ferment. Bioen2. 5, 350-355
9 Lee, S. Y. and Chang, H. N. 11995) Can.[. Microbiol. 41 (Suppl. 1),
2O7-215
10 Riesenberg, D. ctal. (1991)J. Biotechnot. 20, 17 28
11 Paahne, T., Tilsma, K., Kahru, A., Vanat;du, K. and Vilu, R. (1990)
BiotechnoL Bioeng. 35, 312-.319
12 Yang, X-M., Xu, L. and Eppstein, L. (199212". Bioteclmol. 23, 291 301
13 Cutayar, J. M. and Poillon, D. (1989) BiotedmoL Lett. 11, 155-160
14 Yee, L. and Blanch, H. W. (19931 Biotedmot. Bioetg. 41,221 230
15 Srinivas, M. R. S., Chand, N. and Lonsane, B. K. (19941 Bioprocess
En2. 10, 139-144
16 Marld, H., Zenneck, C., 1)ubach, A. and Ogbonna,J. C. (1993) Appl.
Micn~biol. Biotechnol. 39, 48 52
17 Lee, S. Y. and Chang, H. N. (1994) J. Environ. Polymer De2rad. 2,
169-I 76
18 Mori, H., Yano, T., Kobayashi, T. and Shimizu, S. (1979)J. Chem.
Eng.Jpn 12, 313-319
19 Lull, G. W. and Strohl, W. R.. (1990) Appl. Environ. Microbiol. 56,
1~)04-1011
20 E1-Mansi, E. M. T. and Holms, W. H. (1989)./. Gem Microbiol. 135,
2875 2883
21 Kleman, G. L. and Strohl, W. P,-. (1994) ApFL Envitvn. MicrobioL 60,
3952-3958
22 Holms, W. H. 119861 Curr. Top. Cell. Regul. 28, 69-105
23 Varma, A. and Palsson, B. O. (19941 Appl. Era,iron. Microbiol. 60,
3724--373/
24 Pan, I. G., Rhee, J. S. and Lebeault, J. M. 11987) Biotechnol. Letr 9,
89-94
25 Han, K., Lira, H. C. and Hong, J. 119921 Biotechnol. Bioen2. 39,
663-671
26 Bech Jensen, E. and ('.arisen, S. 0990) Bioteehnol. Bioeng. 36, 1-11
27 Koh, B. T., Nakashimada, U., Pfeiffer, M. and Yap, M. G. S. 11992)
Biotechnol. Lett. 14, 1115-1118
28 Shimizu, M., Iijima, S. and Kobayashi, T. (1992)J. Ferment. Bioeny,.
74, 163-168
29 MacDonald, H. L and Neway, J. O. (1990) Appl. Environ. Micwbiol.
56, 640-645
30 Cherrington, C. A., Hinton, M. and Chopra, I. (1990)d. Appl.
Bacteriol. 68, 69-74
31 Meyer, H. P., Leist, C. and Fiechter, A. (1984)J. Biotechnol. 1,
355-358
32 Fieschko, J. C. (1989) in Biotechnology (Vol. 7b) (Rehm, H-J. and
Reed, G., eds), pp. 1/7 140, VCH Verlagsgesellschaff
33 R.iesenberg, D. et al. (1990) Appl. Microbiol. Biotechnol. 34, 77-82
34 Lee, J-H., Choi, Y-H., Kang, S-K., Park, H-H. and Kwon, [-B.
(1989) Biotechnol. la,tt. 11,695-698
 105

reviews
35 Yoon, S. K., Kang, W. K. and Park, T. H. (1994) Biotec/lnol. Bioeng.
43, 995~99
36 Kleman, G. L., Chahners, J. J., Lull, G. W. and Strohl, W. R. (1991)
AppI. Environ. Microbiol. 57, 910-917
37 Landwall, P. and Holme, T. (1977)/. Ge~. Microbiol. 103, 345-352
38 Lee, Y. L. and Chang, H. N. (1990) Biotechnol. Bioeng. 36, 330-337
39 Ham K., Hong,J. and Lira, H. C. (1993) Biotechnol. Bioeng. 41,316~24
40 Brown, T. D. K., Jones-Mortimer, M. M. and Komberg, H. L.
(1977)J. Gen. Microbiol. 102, 327 336
41 Bauer, K. A., Ben-Bassat, A., Dawson, M., de la Puente, V. T. and
Neway, J. O. (1990)Appl. Environ. Microbiol. 56, 1296-1302
42 Hahm, D. H., Pan, J. and P,.hee, J. S. (1994) Appl. MicrobioL
Biotechnol. 42, 100-107
43 Ingrain, L. O., Conway, T., Clark, D. P., Swell, G. W. and Preston,
J. F. (1987) Appl. Environ. Microbiol. 53, 2420-2425
44 Aristidou, A. A., San, K-Y. and Bennett, G. N. (1994) Biotechnol.
Bioeng. 44, 944-951
45 Chou, C-H., Bennett, G. N. and San, K-Y. (1994) Biotedmol.
Bioeng. 44, 952-960
46 Matsui, T., Yokota, H., Sato, S., Mukataka, S. and Takahashi, J.
(1989) Agric. Biol. Chem. 53, 2115 2120
47 Strandberg, L. and Enfors, S-O. (1991) Biotechno[. Lett. 13, 61~9-614
48 Khosla, C., Curtis, J. E., DeModena, J., Rinas, R. and Bailey,J. E.
(1990) Bio/Tedmology 8, 849-853
49 Neubauer, P., Haggstrom, L. and Enfors, S-O. (1995) Biotechnol.
Bioeng. 47, 139-146
50 Gleiser, 1. E. and Bauer, S. (1981) Biorechnol. Bioen2. 23, 1015-1021
51 Jung, G., Denefle, P., Becquart,J. and Mayaux,J-F. (1988) Ann. Inst.
Pasteur MicwbM. 139, 129-146
52 Helmuth, K., Korz, D.J., Sanders, E. A. and Deckwer, W-D. (1994)
J. Biotechnol. 32, 289-298
53 Yee, L. and Blanch, H. W. (1993) Biotedmol. Bioeng. 41,781-790
54 Eppstein, L., ShevJtz, J., Yang, X-M. and Weiss, S. (1989)
Bio/Tedmolog), 7, 1178 1181
55 Ohta, K., Shibui, T., Morimoto, Y., lijima, S. and Kobayashi, T.
(1993) d. Ferment. Bioeng. 75, 155-157
56 Seo, D. J., Chung, B. H., Hwang, Y. B. and Park, Y. H. (1992)
J. Femwm. Bioeng. 74, 196-198
57 Lee, S. Y. and Chang, H. N. (1993) Biotechm,1. la,u. 15, 971-974
58 Lee, S. Y., Yim, K. S., Chang, H. N. and Chang, Y. K. (1994)
d. Biotechnol. 32, 203-211
59 Mizutani, S., Mori, H., Shimizu, S., Sakaguchi, K. and Kobayashi, T.
(1986) Biotechnol. Bioen2. 28, 204-209
60 Randolph, T. W., Marison, I. W., Martens, D. E. and von 8tockar, U.
(1990) Biotedmol. Bioeng. 36, 678 684
61 Konstantinov, K., Kishimoto, M., Seki, T. and Yoshida, T. (1990)
Biotedmol. Bioenf,. 36, 75(>758
62 8uzuki,T.,Yamane, T.andShimizu,S. (1990)J.Femu'nt.Bioog~.69,292-297
63 Suzuki, T., Yamane, T. and Shimizu, S. (1986)J. bl,rment. Tedlnot.
64, 317-326
64 Lubbert, A. and Simutis, R. (1994) Trends BiotechnoI. 12, 304-311
65 Ye, K.,Jin, S. and Shimizu, K. (1994)J. Ferment. Bioen2. 77,663-673
66 Yamane, T., Hibino, W., Ishihara, K., Kadotaui, Y. and
Kominami, M. (1992) BiotechnoL Bioeng. 39, 55(>555
67 Pack, P. et al. (1993) Bio/Technolog1, 11, 1271-1276
68 Shimizu, N., Fukuzono, S., Harada, Y., Fujimori, K., Gotoh, K. and
Yamazaki, Y. (1991) Biotedmo[. Bioen 38, 37-42
69 Tsai, L. B., Mann, M., Morris, F., Rotgers, C. and Fenton, D. (1987)
J. Ind. Microbiol. 2, 181-187
70 Fieschko,J. and Ritch, T. (1986) Chem. E,g. Comnlun. 45,229-240
71 Korz, D. J., Rinas, U., Helhnuth, K., Sanders, E. A. and Deckwer,
W-D. (1995)J. BiotechnoL 39, 59 65

practical approach that illuminates

Essence engineering
Aroma Biotechnology
by R a ! f G. Berger, Springer-Verlag, I 9 9 5 . D M 1 8 6 . 0 0
ISBN 3 540 58606 7
Research into enzymatic and
microbial methods o f producing
molecules that are important for
giving foods and cosmetics their
distinctive and prized tastes and
smells has flourished recently, and a
n u m b e r o f commercially successful
processes and products have
emerged. This is an important
d e v e l o p m e n t that has generally
b e e n o v e r l o o k e d by those w h o
consider that the only possible
applications for b i o t e c h n o l o g y are
in healthcare. In response, Professor
Berger has written this very
interesting and useful b o o k that
contains a wealth o f information
and references, m a n y o f w h i c h
w o u l d be difficult for anyone but
(x + 240pages)
an expert to locate. It has been
carefully and accurately compiled,
and is well laid-out with a g o o d
index; h o w e v e r , it could benefit
from m o r e illustrations and
formulae o f the molecules
described.
This b o o k has a well-constructed
t h e m e that starts by describing the
variety o f aroma molecules, and
goes on to deal w i t h traditional
fermented foods, de novo syntheses,
biotransformations, genetically
modified biocatalysts, the use o f
plant cells, process design (including
product isolation) and process
i m p r o v e m e n t . Chapter 4, about
laboratory techniques, is particularly
w e l c o m e and valuable, as it brings a
the ' r e v i e w ' content o f the rest o f
the book. Finally the importance o f
performance i m p r o v e m e n t and
'profitability aspects' for
commercial success are mentioned;
h o w e v e r , for this section to be
really useful, a second edition
should contain a more-systematic
and comprehensive account,
perhaps through the commissioning
o f sections from expert co-authors,
w h o could write on other
important aspects such as safety" and
regulato W issues.
H o w e v e r , as may be expected
from a b o o k o f this relatively small
size, close scrutiny reveals that
there are a n u m b e r o f serious
shortcomings. It concentrates on
the use o f microorganisms, and has
a rather p o o r e r coverage o f the use
o f isolated enzymes; it barely
mentions savoury, flavours such as
yeast and protein hydrolysates that
are increasingly being manufilctured
on a large scale through the use o f
enzymes and seconda W
fermentation. Similarly, there is

Copyright 1996, Elsevier Science Ltd. All rights reserved. 0167 7799/96/$15.00 TIBTECH MARCH1996 (VOL 141