Fed-Batch Operation of Recombinant
Escherichia coli Containing trp Promoter
with Controlled Specific Growth Rate
Sung Kwan Yoon and Whan Koo Kang
Lucky Biotech Research Institute, P.O.Box 10, Daejeon, Korea
Tai Hyun Park*
Department of Genetic Engineering, Sung Kyun Kwan University,
300 Chunchun-Dong, Suwon, 440-746, Korea
Received June 15, 1993/Accepted December 5, 1993
A feb-batch operation for the production of bovine soma-
totropin (bST) under the control of tryptophan promoter
in Escherichia coli was investigated. The plasmid used
contains a two-cistron system and altered codon usage
for higher expression of bST. Specific growth rate is
an important parameter in the fermentation, because it
affects the production of growth-inhibitory organic acids
and the expression of recombinant protein. The feeding
rate was adjusted to keep the specific growth rate con-
stant in this study. The variable growth yield expressed
as a function of time was used for the calculation of the
feeding rate. The growth yield decreases during the fer-
mentation as product expression is induced. The specific
growth rate was well controlled; however, intracellular
bST concentration decreased at high cell concentrations.
This is considered to be due to degradation by proteases.
The decrease was prevented by an exponential feeding
of the yeast extract as an organic nitrogen source. 0 1994
John Wiley 84 Sons, Inc.
Key words: fed-batch operation recombinant E. coli
trp promoter growth rate, controlled
INTRODUCTION
Fed-batch operation has been widely used for Escherichia
coli fermentation. Various feeding strategies for the fed-
batch operation have been proposed to obtain high cell
concentration and high expression of recombinant pro-
tein. These include the feeding strategies using pH-stat,16
DO-stat," measurement of glucose con~entration~,'~ and
acetic acid concentration18 in the medium, and control of
specific growth rate.'*12*15 The specific growth rate is an
important parameter because the production of growth-
inhibitory organic acids such as acetic acid and the ex-
pression of recombinant protein depend on the specific
growth rate.9,14,20
Control of the specific growth rate at an appropriate
value can give desired cell growth unless the dissolved
oxygen in the medium is limited. However, intracellular
concentration of the recombinant protein decreases at high
cell concentration, even if the product exists as an inclusion
body. This is considered to be due to degradation by
* To whom all correspondence should be addressed.
Biotechnology and Bioengineering, Vol. 43, Pp. 995-999 (1994)
0 1994 John Wiley 81 Sons, Inc.
proteases. Protease is induced under stressful conditions
such as the depletion of carbon or organic nitrogen source,
high temperature, and exposure to e t h a n ~ l . ' ~Protease
,~~
activity can be repressed and, consequently, the productivity
is improved by feeding organic nitrogen."
Fed-batch operation is applied to the production of
bovine somatotropin (bST) in this study. The bST is a
polypeptide consisting of 190 or 191 amino acids6 Tran-
scription of the bST gene is under the control of the
tryptophan promoter in our system. The tryptophan pro-
moter is regulated by tryptophan and tryptophan analog,
such as 3P-indoleacrylic acid (J.AA).l3 The cloned gene
expression is automatically induced in a fed-batch operation
by feeding tryptophan-deficient medium.
In this article, we study a feeding strategy with which
the specific growth rate is controlled at the set point and
the intracellular concentration of recombinant bST does not
decrease even at high cell concentration.
MATERIALS AND METHODS
Bacterial Strain and Growth Condition
Escherichia coli W3110/ptrphsBST was used for this re-
search. The plasmid, ptrphsBST, contained a two-cistron
system for higher expression of bST.17 Transcription was
regulated by the trp promoter, and codon usage was altered
in the first 54 nucleotides to reflect the bias of highly
expressed E. coli genes.2
The M9 medium (100 mL) with 5 g/L yeast extract and
10 g/L glucose was inoculated with 0.2 mL of glycerol
stock, and was incubated overnight. This seed culture was
transferred to 5 L or 6 L of medium in a 14-L fermentor
(Chemap CF 2000) for fed-batch operation. The medium
composition is shown in Table I. Temperature and air flow
rate were maintained at 37°C and 1 vvm, respectively.
NH40H solution was used to keep pH at 7.0. Dissolved
oxygen in the medium was maintained at concentrations
above 20% by regulating the stirrer speed and the pressure
inside the fermentor.
CCC 0006-3592/94/0100995-05
Table I. Medium composition in dual feed fed-batch fermentation.

Composition Starting medium (g/L) Feeding medium 1 (g/L) Feeding medium 2 (g/L)

Glucose 2 400 400

KH2m4 15 - -
MgSO4 . 7H2O 4 - 15.5
Yeast extract 1 - 100
FeS04 . 7 H 2 0 - 0.2 0.2
CaClz . 7H20 - 1.35 1.35
MnS04 * 5 H 2 0 - 0.075 0.075
CoClz . 6 H z O - 0.013 0.013
ZnS04 . 7 H 2 0 - 0.075 0.075
Cuc12 . 5 H 2 0 - 0.013 0.013
NazMoO4 . 2H20 - 0.013 0.013

Fermentation and Control rate described by Eq. (1). The set point of the specific
growth was 0.15 h-' and the growth yield (Y) was assumed
In the early phase of the fermentation, feed was not added
to be 0.714 (ODL/g glucose) which was chosen by the
until the glucose in the medium was depleted and the
total amount of cell produced per glucose consumed in
concentration of dissolved oxygen started to increase. Glu-
previous fed-batch experiments. The specific growth rate
cose concentration was maintained at zero to prevent acetic
is well controlled at 0.15 h-l, as shown in Figure 1.
acid formation. Specific growth rate ( p ) was controlled
However, the growth yield decreases as the fermentation
at a constant value with an exponential feeding of the
is carried out. The difference of the assumed Y from the
glucose solution. The feeding rate was determined by a
real value causes the calculated cell concentration, indicated
mass balance equation of the cell and the substrate, and
by the solid line, to deviate from the experimental values.
is represented by the following equation when S = 0 and But the final cell concentration and the final prediction
ds/dt = 0. match well, because the overall growth yield, which was
F = (pXoVo/Y SF)epLt (1) experimentally obtained, was used for the constant growth
yield. It should be noted that the specific growth rate was
where p is the set point of the specific growth rate and
well controlled even though incorrect growth yield was
SF is the substrate concentration in the feeding medium.
used. Mathematical simulation confirms that the controlled
The growth yield (Y) may be constant or variable. The
specific growth rate approaches the set point in this control
calculated value of the flow rate was converted to an
analog signal which operated a feeding pump (Cole-Parmer,
Ismatec Reglo 100).

Analytical Procedure a
Cell density was measured by a spectrophotometer(Perkin-
Elmer, Coleman 575) at 600 nm. Glucose concentration
was measured using a Sigma Glucose Kit (procedure no.
510) or a strip (Glucopat, Kyoto Daiichi Kagaku Co.).
Tryptophan concentration was determined by the method of
Kupfer and Atkin~on.~ Concentration of bST per unit cell /

was determined by SDS-polyacrylamidegel electrophoresis

(SDS-PAGE) and Coomassie blue staining. A densitometer
(Bio-rad, Model 620) with an interference filter was used
0
to scan the bST band on the gel for quantitative analysis.
The standard curve, which is the correlation between the /A A
0
amount of bST and the peak area read by the densitometer,
was obtained in every SDS-PAGE and showed a good
linear correlation. The contribution of the native E . coli
protein, which may overlap with the bST band, to the signal 0 5 10 15 20 25 30 35
measured by the densitometry, was negligible.
Tlme(hrs)

RESULTS AND DISCUSSION Figure 1. Control of specific growth rate in fed-batch operation using
constant growth yield: (0)cell concentration, (A) growth yield (r), ( 0 )
Fed-batch operation was carried out using the starting specific growth rate ( p ) ,and set point of /I ( . . . .) Solid lines represent
and feeding medium 112 shown in Table I and the flow the calculated values.

996 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 43, NO. 10, APRIL 25, 1994
strategy even with the use of an incorrect growth yield. 120
The specific growth rate and the growth yield decrease
rapidly in the late phase of the fermentation. This is
caused by a depletion of the dissolved oxygen in the
medium because the dissolved oxygen was not controlled
in this experiment. A drastic decrease in the dissolved
oxygen was observed toward the end of the fermentation.
The low values of the growth yield in the late phase
caused an underestimation of the overall growth yield of
0.714 (ODL/g glucose). The glucose concentration was
maintained well at zero throughout the fermentation. The
growth yield is not constant as shown in Figure 1. To
express the growth yield as a function of time, it is assumed D
that the decreasing rate of the growth yield is linearly
proportional to the difference between the growth yield and
the minimum growth yield (Y,). The decreasing rate of the
growth yield is expressed by the following equation
dY/df = (l/ty)(Ym- Y ) Y(to) = Yo (2)
0 5 10 15 20 25 SO

Tlme(hrs)
y = -(y, - Yo)e-('-'O)/'Y + ym (3) Figure 2. Control of specific growth rate in fed-batch operation using
variable growth yield: (0) cell concentration, (A) growth yield (Y),
where t y is a time constant. This is an empirically fitted (0)specific growth rate (p), and set point of p (. . ..) Solid lines
equation. Rearrangement of this equation gives the follow- represent the calculated values.
ing equation, which is linear with respect to time.
ln(Y - Y,) = - ( t - tO)/ty+ ln(Y0 - Ym) (4) volume increases continuously as the cell density increases
According to the experimental data, the parameter values exponentially.
are Y, = 0.95 and YO= 1.65. By plotting ln(Y - Y,) versus The decrease of the specific bST concentration at high
t , other parameter values, fy = 8 and to = 8, are obtained cell density can be due to plasmid instability, product degra-
from the slope and the intercept. Eq. ( 3 ) becomes dation by protease, intervening processes of transcription
and translation, as well as dilution by growth. However,
Y = 0.7e-('-')/' + 0.95 (5)
where t is the time after starting the fermentation. If t is
the time after starting the feeding, Eq. (5) becomes the '20"17 r\
d

following equation E
4
Y = 0.7e-('-W/' + 0.95 (6) h

4 ?t,
0 a
This equation is useful when the dissolved oxygen is not v
v
E E
limited. The solid line for Y in Figure 1 represents a result 0 0
c
4
calculated by Eq. (5). c
4

The same experiment as in Figure 1 was carried out c 2

c
E C
using variable Y expressed by Eq. (6). The set point of the 8 8
U U
specific growth rate was 0.14 h-' and the glucose feeding c E
0 0
was started 4 h after starting the fermentation. Results u V
*
r.
are shown in Figure 2. The solid lines for X and Y are
the calculated results. The specific growth rate was well
V
8
5U
4
controlled at the set point and the calculated cell density u
4

U
agreed well with the experimental data. The production 8
Ia
of bST is shown in Figure 3. bST produced in E . coli rA

exists as an inclusion body. The specific bST concentration

based on cell mass reaches a maximum at OD = 40 then
decreases. Optical density can be used as a valid indicator of
increase in cell mass during induction, because the linear Time(hrs)
relationship between dry cell weight and optical density Figure 3. Profile of cell concentration and specific bST concentration in
is stable following a relatively brief transition phase after specific growth rate-controlled fed-batch operation: (0) cell concentra-
the i n d ~ c t i o nThe
. ~ bST concentration based on the culture tion, and (0)specific bST concentration.

YOON, KANG, AND PARK FED-BATCH OPERATION WITH CONTROLLED GROWTH RATE 997
the plasmid stability test showed that the plasmids were
stable. A protease inhibitor, PMSF (phenyl methane sul-
' 1 4 0
fonyl fluoride) was added to the medium to examine I
whether the decrease of the specific bST concentration
was due to the degradation by protease. The experiment
was carried out in a shaker flask containing M9 medium
.-.
4
3 -
0
with 2 g/L casamino acid. When the OD reached 0.5, v
c
20 ,ug/mL of 3P-indoleacrylic acid (IAA)was added to -4
1

induce the expression. Cell growth curve and specific E

c
c 2-
bST concentration are shown in Figure 4. The specific 0
c
bST concentration decreases rapidly when the cell growth
reaches a stationary phase. The PMSF was added 7 h -
V
0
after starting the experiment. This is a late exponential V
phase. The specific bST concentration profiles are shown
in Figure 4 when the PMSF concentrations in the media
are 0.5 mM and 2 mM. The decreasing rate of the specific
bST concentration was slower with the addition of 0.5 mM
' !#
PMSF than without the addition. With the addition of 2 mM
0
PMSF, the specific bST concentration could be maintained 0 5 10 15 20 25 SO
constantly. These results imply that the decrease of the
Time(hrs)
specific bST concentration was due to product degradation
by the protease. It can be more clearly demonstrated if Figure 4. Effect of PMSF on decrease of specific bST concentration.
we take samples before and after the decrease in the Closed symbol represents the cell concentration and open symbol repre-
bST production and immunoblot with polyclonal anti-bST sents the specific bST concentration.
antiserum in an effort to identify degradation products.
At the stationary phase caused by the depletion of the
was supplied when the cell concentration reached OD = 35.
carbon source in the batch operation, protease is induced The growth yield was constantly maintained at Y = 1.28
and degrades the intracellular bST. The degradation can be (OD L/g glucose) with the feeding of the yeast extract.
prevented by the addition of the protease inhibitor, PMSF. Thus, constant growth yield was used to calculate the
However, this process is not practical because PMSF is
feeding rate in this range and the specific growth rate was
expensive.
well controlled at the set point. Cell growth and specific
The temperature effect on the degradation was observed
next to examine whether the protease activity could be
inhibited by lowering the temperature. The temperature was
shifted from 37°C to 30°C, 2 5 T , and 20"C, respectively,
7 h after starting the experiment. The growth temperature
did not affect the stability of the bST (data not shown).
As described earlier, the degradation of the recombinant
protein by protease is known to be caused by culturing
the cell under stressful conditions, such as the depletion of
carbon or organic nitrogen source, high temperature, and
exposure to ethanol. It is considered that the decrease of
the specific bST concentration in the fed-batch operation
shown in Figure 3 is due to the deprivation of the organic
nitrogen source at high cell density, whereas that in batch
operation is due to the deprivation of the carbon source in
the stationary phase.
The same experiment as shown in Figures 2 and 3 was
carried out with the exponential addition of yeast extract
as an organic nitrogen source to prevent the decrease of
the specific bST concentration at high cell concentration.
Initial culture volume was 6 L and the set point of the 0 5 10 15 20 25 30
specific growth rate was 0.14 h-l. Medium composition is
shown in Table I. Feeding medium 1, containing glucose Time(hrs)
and trace metal, was supplied in the first feeding period; Figure 5. Profile of cell concentration and specific bST concentration
while feeding medium 2, containing all the components in in fed-batch operation with exponential feeding of yeast extract: (0)cell
feeding medium 1 together with yeast extract and MgS04, concentration, and ( 0 )specific bST concentration.

998 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 43, NO. 10, APRIL 25, 1994
bST concentration is shown in Figure 5. The specific bST
concentration does not decrease at high cell concentration,
whereas it decreases at higher cell concentrations than
-
OD = 30 40 in Figure 3. The addition of yeast extract,
an organic nitrogen source, prevents the induction of the
protease and, consequently, prevents the degradation of
the bST. The overall rate of the protein degradation was
observed to increase many-fold when a carbon or organic
nitrogen source such as amino acids was depleted.8 Tryp-
tophan concentration was zero and repression of the trp
promoter was not observed during the exponential addition
of the yeast extract. A similar experiment was carried out
with the addition of ammonium phosphate, an inorganic
nitrogen source; however, the decrease of the specific bST
concentration could not be prevented.
References
1. Allen, B. R., Luli, G. W. 1987. A gradient-feed process for E . coli
fermentations. BioPharm. Nov. 38-41.
2. George, H. J., L’italien, J. J., Pilacinski, W.P., Glassman, D.L.,
Krzyek, R. A. 1985. High-level expression in Escherichia cofi of
biologically active bovine growth hormone. DNA 4: 273-281.
3. Goff, S. A,, Goldberg, A. L. 1985. Production of abnormal proteins
in E. coli stimulates transcription of lon and other heat shock genes.
Cell 41: 587-595.
4. Halst, O., Hakanson, H., Miyabayashi, A,, Mattiasson, B. 1988.
Monitoring of glucose in fermentation processes using a commercial
glucose analyser. Appl. Microbiol. Biotechnol. 28: 32-36.
5. Hwang, S., Feldberg, R. 1990. Effect of inclusion body production
on culture turbidity and cell dry weight in growing bacterial cultures.
Biotechnol. Prog. 6: 48-50.
6. Kane, J. F., Bogosian, G. 1987. Bovine somatotrophin production:
selecting the best host-vector system. BioPharm Nov. 29-51.
7. Kupfer, D., Atkinson, D. 1964. Quantitative method for determination
of indole, tryptophan, and anthranilic acid in the same aliquot. Anal.
Biochem. 8: 82-94.
8. Mandelstam, J. 1960. The intracellular turnover of protein and nucleic
acids and its rate in biochemical differentiation. Bacteriol. Rev. 24:
289-308.
YOON, KANG, AND PARK FED-BATCH OPERATION WITH CONTROLLED GROWTH RATE
9. Meyer, H.-P., Leist, C., Fiechter, A. 1984. Acetate formation in
continuous cultures of Escherichia cofi K12 D1 on defined and
complex media. J. Biotechnol. 1: 355-358.
10. Mitzutani, S., Iijima, J., Morikawa, M., Shimizu, K., Matsubara, M.,
Ogawa, Y., Izumi, R., Matsumoto, K., Kobayashi, T. 1987. On-line
control of glucose concentration using an automatic glucose analyzer.
J. Ferment. Technol. 65: 325-331.
11. Mori, H., Yano, T., Kobayashi, T., Shimizu, S. 1979. High density
cultivation of biomass in fed-batch system with DO-stat. J. Chem.
Eng. Jpn. 12: 313-319.
12. Paalme, T., Tiisma, K., Kahru, A,, Vanataln, K., Vilu, R. 1990.
Glucose-limited fed-batch cultivation of Escherichia coli with
computer-controlled fixed growth rate. Biotechnol. Bioeng. 35:
312-319.
13. Park, T.H., Seo, J.-H., Lim, H.C. 1989. Analysis of kinetic pa-
rameters for the 3P-indoleacrylic acid effect on trp promoter in
Escherichia cofi. Biotechnol. Lett. 11: 87-92.
14. Riesenberg, D., Menzel, K., Schulz, V., Schumann, K., Veith, G.,
Zuber, G., Knorre, W.A. 1990. High cell density fermentation of
recombinant Escherichia coli expressing human interferon alpha 1.
Appl. Microbiol. Biotechnol. 34: 77-82.
15. Riesenberg, D., Schulz, V., Knorre, W.A., Pohl H.-D., Korz, D.,
Sanders, E.A., Ross, A,, Deckwer, W.-D. 1991. High-cell density
cultivation of Escherichia coli at controlled specific growth rate. J.
Biotechnol. 20: 17-28.
16. Robbins, J. W. Jr., Taylor, K. B. 1989. Optimization of Escherichia
cofi growth by controlled addition of glucose, Biotechnol. Bioeng.
3 4 1289- 1294.
17. Schoner, B. E., Hsiung, H. M., Belagaje, R. M., Mayne, N. G., Schoner,
R. G. 1984. Roles of mRNA translational efficiency in bovine growth
hormone expression in Escherichia coli. Proc. Natl. Acad. Sci. USA
81: 5403-5407.
18. Shimizu, N., Fukuzono, S., Fujimori, K., Nishimura, N., Odawara,
Y. 1988. Fed-batch cultures of recombinant Escherichia cofi with
inhibitory substance concentration monitoring. J. Ferment. Technol.
66: 187-191.
19. Tsai, L.B., Mann, M., Morris, F., Rotgers, C., Fenton, D. 1987.
The effect of organic nitrogen and glucose on the production of
recombinant human insulin-like growth factor in high cell density
Escherichia coli fermentation. J. Ind. Microbiol. 2: 181 -187.
20. Zabriskie, D. W., Wareheim, D.A., Polansky, M. J. 1987. Effects
of fermentation feeding strategies prior to induction of expression
of a recombinant malaria antigen in E. cofi J. Ind. Microbiol.
2: 87-95.
999