Journal of Industrial Microbiology, 11 (1992) 1-6

9 1992 Society for Industrial Microbiology 0169-4146/92/$05.00

Published by Elsevier

SIM 00442

High production of alkaline protease by Bacillus licheniformis in a

fed-batch fermentation using a synthetic medium
Weiying M a o a, Renrui Pan b and David Freedman a
aNew Brunswick Scientific Co., Inc., Edison, NJ, USA, and bDepartment of Biology, University of Science and Technology of China, Hefei,
Anhui, China
(Received 10 March 1992; revision received 15 May 1992; accepted 19 May 1992)

Key words: Protease; Bacillus licheniformis; Synthetic medium; Fed-batch fermentation

SUMMARY

High production (9016 U/m1)of alkalineprotease by Bacilluslicheniformishas been achieved.A 49 ~o increase in productionwas achievedby the method
used as compared with a batch process. By using a syntheticmedium and a fed-batch operation controIledby the Advanced Fermentation Software(AFS)
package,it was found that the keys to highproduction of protease are: (i) to maintain a low concentrationof glucose ( < 0.43 g/l) in the medium;(ii) to control
pH at a certain level (pH 6.50) in the culture; and (iii)to use rough type colonies as the starting culture. Our fed-batch fermentation process successfully
simulates and surpasses ordinary batch fermentation processes. By using ammonium sulfate instead of soy bean flour as the only nitrogen source, an ex-
pected benefit was the eliminationof unpleasant odors caused by natural organic nitrogenouscomponents in the media. This would improvethe industrial
production environment.

INTRODUCTION ammonium, additives (yeast extract, sodium glutamate,

etc.) were also used as nitrogen sources [2-6,12,13]. It is
Bacillus licheniformis is capable of producing a variety generally acknowledged that there is little or no produc-
of products, such as bacitracin, penicillinase, alkaline tion of protease during the exponential growth phase in
phosphatase, ribonuclease, e-amylase and protease [ 5,13 ]. batch culture [4].
These products are widely used in such industries as ap- There were two aims in this study, to get a high pro-
plied chemistry, tannery, food and feed, etc. duction of alkaline protease by Bacillus licheniformis, and
Natural media have been used for a long period of time to eliminate unpleasant odors during the fermentation.
in bacterial studies on protease fermentation using bacte- The first has been achieved in fed-batch fermentation using
rial producers. Although high production has been ob- our own synthetic medium. Preliminary studies on the
tained and media are inexpensive, drawbacks of the media relationship between the fermentation conditions and pro-
are obvious. The relationships among medium composi- tease production have been performed. A simple inorganic
tion, cell growth and protease formation are unclear, owing compound, ammonium sulfate, instead of soy bean flour
to the complicated compositions of media and the mixing was used as the only nitrogen source in the medium. A n
of cell and solid ingredients. Therefore, it is difficult to expected benefit was that the unpleasant odors disap-
compare the various studies of these relationships [ 10]. peared. This would contribute to an improvement in the
Moreover, natural organic nitrogenous components in the industrial production environment.
media produce unpleasant odors during the fermentation.
Since the 1960s, many important results have been MATERIAL AND METHODS
obtained from a series of studies concerning the metabolic
regulation and the kinetics pattern of protease produced Seed culture
by species of Bacillus under conditions of chemostat cul- Bacillus licheniformis PW-109, a commercial produc-
tures and synthetic or semi-synthetic media. Glucose was tion strain, was used in this investigation. This strain was
used as the only carbon source. In addition to inorganic grown on agar slants composed of 10 g/1 beef extract,
10 g/1 peptone and 5 g/1 sodium chloride at 36 ~ for 24 h.
Correspondence to: W. Mao, R&D Laboratory, New Brunswick The seed was developed by transferring the cells grown on
Scientific Co., Inc., Edison, NJ, 08818, USA. the slant into 250-ml flasks, each containing 100 ml me-
dium. Seed flasks were cultivated on a New Brunswick of 0.1 N Folin-Ciocalteu's phenol reagent. The sample
Scientific Co. G-25 Gyrotory Incubator Shaker at 36 ~ was incubated at 40 ~C for 20 rain to develop color and the
and 200 rpm for 7 h. optical density was measured at 680 nm (Perkin-Elmer
Lambda 4B). The slope of the standard curve determined
Fermentation with L-tyrosine was used in the calculation of protease
A 5-1 working volume BioFlo III fermentor (New activity. One unit of protease activity is defined as the
Brunswick Scientific Co., Inc.) was used in this investi- amount that liberates 1 #g L-tyrosine per min under the
gation. This fermentor is equipped with an internal multi- testing conditions described above.
loop controller which regulates temperature, agitation, dis-
solved oxygen, pH and nutrient addition. The AFS Cell concentration
(Advanced Fermentation Software) of NBS was used to The sample was diluted 10- or 40-fold with de-ionized
control glucose addition. This software was executed on water. The supernatant of the sample as the blank was
an IBM model PS/2 30 personal computer. Basic fermen- similarly diluted. The optical density values of both at
tation medium was composed of 6g/1 glucose, 10 g/1 600 nm vs. water were measured in the same spectropho-
(NH4)2SO4, 8 g/1 Na2HPO4, 4 g/1 KHzPO4, 0.5 g/1 tometer. Cell concentration was calculated as the differ-
MgSOe.7H20 , 0.02g/1 CaC12 and fortified with trace ence of OD values between the sample and the blank.
amounts of Cu 2 +, Fe 3 +, Mn 2 +, Zn 2 + and growth factors.
The seed medium was identical to this formula. SAG-471 Glucose concentration
silicone antifoam (300 ppm) was added before steriliza- Glucose concentrations were determined with a com-
tion. Ammonium hydroxide (7.5 N) was used both to reg- mercial enzyme assay kit (Sigma).
ulate pH and as a supplementary nitrogen source. An
NBS MX-3 Biosampler was used to take samples auto- RESULTS AND DISCUSSION
matically.
Effects of initial pH, pH set point and initial concentration of
Alkaline protease estimation glucose on protease production
The samples were centrifuged (Damon/IEC HN-S Some operating conditions and the results for six ex-
centrifuge) at 2500 rpm for 20 rain. The supernatant was periments are tabulated in Table 1. Unless otherwise men-
diluted 100- or 200-fold. One ml & t h e diluted sample was tioned, the following operating conditions were held con-
preheated in a 40 ~ water bath. One ml of 2~o casein stant for a series of fermentation experiments. After
solution (prepared with a pH 11 borax buffer solution) was inoculation, the volume of the medium was about 41 in
added into the sample. Reaction temperature was 40 ~ each experiment. The volume of added glucose was about
After precisely l0 rain, 2 ml of 0.4 M trichloroacetic acid 200 ml. Regardless of what the initial concentration of
was added into the sample to terminate the reaction. A glucose was, the total amount of glucose used in each run
blank correction was necessary for each sample. After was 50 g/1. Ammonium hydroxide was used to regulate pH
centrifugation, 1 ml of the reactant supernatant was taken and the total volumes added were 90-135 ml in those
and mixed with 5 ml of 0.4 M sodium carbonate and 1 ml experiments. Experiments were run at 36 ~ 0.5 v/v per

TABLE 1
Some operating conditions and results for six runs

Run Initial Set Initial Glucose Carbon/ Peak Peak Protease

no. pH point glucose addition nitrogen turbidity activity activity
of pH concentration start time ratio of cells time (U/ml)
(g/l) (h) (OD) a (h)

1 6.78 6.50 20 27 4.27 2.25 44 8096

2 6.49 6.50 16 23 3.64 2.40 44 8296
3 6.53 6.50 6 18 3.60 2.39 36 9016
4 6.94 7.20 16 16 4.42 2.20 34 2120
5 7.00 7.00 16 13 3.90 2.38 32 2920
6 6.84 7.00 6 17 3.56 2.41 40 4168

a 10-fold dilution.
min air flow rate and 400 rpm initial agitation speed, which material in our studies is also inferred to be a certain kind
was also the minimum speed for dissolved oxygen (DO) of polysaccharide. Its actual properties are as yet unclear.
control. The initial DO level was calibrated to 100~o and It could be seen that when the initial concentration of
the set point of the DO control was 30 To. If the DO level glucose was 2 g/l, the protease production would eventu-
dropped below this point, the agitation speed would in- ally be sufficient, as long as the viscous material which
crease automatically to maintain the DO level until a max- formed in the earlier part of the run could be used after
imum speed of 950 rpm was reached. When initial glucose glucose depletion, such as in run No. 1. However, in most
was depleted, the pH value increased. The pH trend was of such high initial glucose cases, in addition to runs No.
a useful measure of glucose depletion. The glucose addi- i-6, especially in runs for which the pH was maintained
tion was actuated by this trend through the user-defined at 7.00, the viscous material could not be used. Once the
algorithms of AFS. An additional flow rate in the vicin- glucose was added, excess viscous material formed, ren-
ity of 0.21 ml/min corresponding to the flow rate reading dering those fermentations unusable. Bacillus polyrnyxa
of 9 ~o for the nutrient pump control was used to maintain has been reported [7] to form a high-viscosity agar-like
the low concentrations of glucose in the medium. polysaccharide at the optimal pH of 7.00-7.50. In some
The six runs can be divided into two groups according runs, if the initial pH was 7.00 and the pH was maintained
to their peak values ofprotease activity. High values were at 7.00, although the initial concentration of glucose was
observed in runs No. 1, 2 and 3 and lower values in runs only 16 g/l, a large amount of the viscous material formed,
No. 4, 5 and 6. The major variations in operating condi- rendering the fermentation unusable. It can be considered
tions between these two groups were the initial pH values that the pH plays another major role in the formation of
and the set points of the pH controller. The initial pH the viscous material.
values in the high production group were lower than those
in the lower production group. The pH set points were
6.50 in runs No. 1, 2, and 3; 7.20 in run No. 4 and 7.00 Effects of colony type, cell mass and carbon/nitrogen ratio on
in runs No. 5 and 6. It is evident that pH plays a major protease production
role in protease production. The optimal pH set point was In the course of our studies, two morphological types
6.50 in our experimental investigation. of colonies were seen on the slants. The rough type was
Runs No. 1, 2 and 3 belong to the high production the high production strain, while the smooth type was the
group, but different values of protease activity were ob- low production strain. Similar results have been reported
tained because of variations in the initial concentration of previously [ 1,4,8,9]. Unstable results were not expected in
glucose. It was found in our studies that protease produc- our studies, thus more attention was paid to seed slants.
tion was inhibited by glucose. Frankena et al. [4] reported As a short-term measure, the criteria for the production
that in glucose-limited chemostat cultures the specific rate strain used in the factory was implemented. Isolation of
of protease production was maximal at a specific growth high production colonies was performed often and the
rate of 0.22. They asserted that exocellular protease was number of transfers of the strain from slant to slant was
subjected to glucose repression above this specific growth conducted as little as possible.
rate. For the same reasons, among runs No. 1, 2 and 3, Table 1 shows that the variations of the peak turbidity
the highest production was obtained from the batch whose of the cultures were not large from batch to batch. Al-
initial concentration of glucose was the lowest. In the though the OD value in run No. 6 was close to the val-
other two batches, which contained a high initial concen- ues in runs No. 2 and 3, the protease activity in this batch
tration of glucose, most of the initial glucose made little or was only about half of those high activities. However, in
no contribution to protease production. order to obtain high protease activity, a certain amount of
The high initial concentration of glucose caused other cell mass is necessary. Nevertheless, just as the experi-
disadvantages. In runs No. 1, 2 and 3, the quality of seeds ment revealed, if fermentation conditions were not appro-
was nearly identical, but higher concentrations of glucose priate, protease production was still poor, even though the
resulted in a longer lag phase, and inevitably a longer amount of cell mass was large.
fermentation time. In runs No. 4 and 5, although initial In this work, the total amount of glucose was fixed, the
concentrations of glucose were also high, the shorter lag amount of ammonium sulfate was also fixed and ammo-
phase was attained by using a larger amount of inoculum. nium hydroxide as a supplemental nitrogen source was
It was observed that under high initial concentrations automatically added for pH control. The average carbon/
of glucose, a large amount of viscous material formed in nitrogen ratios (g/g) of both high and lower production
cultures. Bergey's Manual of Systematic Bacteriology groups, of 3.83 and 3.96, respectively, were very close.
states that levan is produced extracellularly from sucrose However, the carbon/nitrogen ratio in run No. 3 could be
and raffinose by Bacillus licheniformis [ 11]. The viscous adopted for higher production.
 PW109 ( 3 )
1000 50.0 8.00 100.0"

Agil
{; r'~
'iI , /
/ "~, I ,"'-,
.
870 40.0 7.40 80.0 iI ,' "/, .,...,
t
II /'
I i ,.."
I l l

/
e~
/
740 30.0 6.80 60.0 ! /
i pH /.'
.-.., ....... , ......... . .............................. }=,................... ./
...................... .--I',"

610 20.0 6.20 40.0

480 10.0 5.60 20.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . !

350 0.0 5.00 0.0 . . . . . . . . . .

o.o 10. o 2 o,0 30.0 ~ o,0 5 o.o
Agit NutA pH DO2
Time (h)

Fig. 1. Profiles of the DO level, pH value, glucose addition and agitation speed for run No. 3.

Time profile of the highest production run activity for the same run are shown in Fig. 2. As can be
The profiles of the DO level, pH value, glucose addi- seen, the DO level and glucose concentration gradually
tion and agitation speed for run No. 3 are shown in Fig. 1. decreased with the slow growth of cells. A 12-h period of
This is an original drawing plotted by a printer running on lag phase was observed. The cell growth and glucose con-
the AFS software. Profiles of the cell growth (OD values sumption increased substantially during 12 to 16 h. This
by 40-fold dilution), glucose consumption and protease time can be looked upon as the pre-exponential growth
phase. The depleted concentration value of glucose, which
was only 0.38 g/l, was observed at 18 h. Glucose addition
started at this moment due to an increase of the pH caused
10 by the depletion of glucose. The glucose concentrations
were always less than 0.43 g/1 during the glucose addition
period (18 to 37 h), except that the concentrations were a
little bit higher during the first several hours. The expo-
nential growth phase with a doubling time of 2.2 h oc-
curred during 16 to 26 h. The high requirement of oxygen
by cells during this exponential growth phase was indi-
cated by the phenomenon that the DO mostly remained
at low levels, between 25-35 ~o, and the agitation promptly
increased and then remained at high speeds, between 785-
950 rpm. When the reduction of cell growth resulted from
oL ,,J,,,',,
0 5 10 15 20 25 30 35 40 45 50
ioo, the glucose limitation, a decrease of oxygen requirement
occurred during 26 to 30 h in the culture. Thus it was
Time (h)
observed that, although the DO still remained near the set
Fig. 2. Profiles of the cell growth, glucose consumption and pro- point level, the agitation speed dropped markedly. This
tease activity for run No. 3. phenomenon indicates the occurrence of another period
considered as the post-exponential growth phase. The ter- process successfully simulated and remarkably surpassed
mination of glucose addition occurred at 37 h. The sta- the ordinary batch type fermentation processes. Also, the
tionary phase occurred during 30 to 45 h. unpleasant odors were completely eliminated.
The pH was automatically regulated by ammonium
hydroxide, so that the medium pH remained at the set
points, which were 6.48 before and 6.52 after starting to ACKNOWLEDGEMENT S
add glucose, respectively. This tiny change of the set point
was for ease of formulating user-defined equations, when This investigation was a cooperative project between
AFS was used to actuate the nutrient pump. In the later New Brunswick Scientific Co., Inc. (NBS), U S A and the
stage of this run there was evidence of autolysis. The University of Science and Technology of China, People's
termination of glucose addition speeded up this process. Republic of China. The authors gratefully acknowledge
The switch of the energy source, presumably to amino the financial support from NBS in the U S A and New
acids, and cell autolysis caused a marked increase of the Biotech Associates (a joint venture between NBS and
pH. China United Biotechnology Corporation). We thank
From the profile of protease activity in Fig. 2, it can be Mary Nilan for typing the manuscript.
observed that there was little or no formation of protease
during the earlier exponential growth phase. The activity
of protease was only 416 U/ml at 21 h. A rapid increase REFERENCES
of protease activity began at the middle stage of the ex-
ponential growth phase. The protease production lagged 1 Bulthuis, B.A., J. Frankena, G.M. Koningstein, H.W. van
behind the growth of cells by about 8 h. The peak activ- Verseveld and A.H. Stouthamer. 1988. Instability of protease
ity of protease was found to be 9016 U/ml at 36 h. Then production in a rel /rel+-pair of Bacillus licheniformis and
protease activity gradually decreased with cell autolysis. associated morphological and physiological characteristics.
It was felt that the formation of spores might be cor- Antonie van Leeuwenhoek 54:95-111.
related with the production of protease. Through micro- 2 Dawes, I.W. and J.H.M. Thornley. 1970. Sporulation in
scopic observation, a small number of spores was found Bacillus subtilis. Theoretical and experimental studies in con-
during the later part of the exponential growth phase, and tinuous culture systems. J. Gen. Microbiol. 62: 49-66.
3 Dawes, I.W. and J. Mandelstam. 1970. Sporulation of Bacillus
a large number of spores was found during the stationary
subtilis in continuous culture. J. Bacteriol. 103: 529-535.
phase in run No. 3. The same phenomenon was also ob-
4 Frankena, J., H.W. van Verseveld and A.H. Stouthamer.
served in runs No. 1 and 2. Low protease production 1985. A continuous culture study of the bioenergetic aspects
along with a small number of spores existed during the of growth and production of exocellular protease in Bacillus
stationary phase of runs No. 4, 5 and 6. However, we licheniformis. Appl. Microbiol. Biotechnol. 22: 169-176.
cannot affirm that a large number of spores must corre- 5 Hanlon, G.W. and N.A. Hodges. 1981. Bacitracin and pro-
spond to high protease production. Bulthuis et al. [ 1 ] re- tease production in relation to sporulation during exponential
ported that the smooth type of B. licheniformis (not pro- growth of Bacillus licheniformis on poorly utilized carbon and
tease producing) can sporulate also. nitrogen sources. J. Bacteriol. 147: 427-431.
Protease activity of 6064 U/ml was obtained in the 6 Hanlon, G.W. and N.A. Hodges. 1981. Requirement for glu-
batch culture using the same strain at 37 h under the same cose during production of extracellular serine protease by cul-
tures of Bacillus licheniformis. FEMS Microbiol. Lett. 11: 51-
conditions of temperature, aeration, initial agitation speed
54.
and set point of the DO control. The medium was com- 7 Mitsuda, S., N. Miyata, T. Hirota and T. Kikuchi. 1981.
posed of 40 g/1 starch, 30 g/1 soy bean flour, 4 g/1 Na2HPO 4 Studies of polysaccharides produced by microbes I. High-
and 1 g/1 yeast extract. The initial pH was 8.47 and pH viscosity polysaccharide produced by Bacillus polymyxa.
control was not used in this batch fermentation as is done Hakko Kogaku Kaishi. 59: 303-309.
in the real industrial process. The pH profiles in both 8 Nehete, P.N., V.D. Shah and R.M. Kothari. 1985. Profiles of
batch and fed-batch cases were very different. The reasons alkaline protease production as a function of composition of
are understandable. The media corresponding to both the slant, age, transfer and isolate number and physiological
cases have substantial differences. It has already been state of culture. Biotechnol. Lett. 7: 413-418.
noted that the protease production is repressed by glucose. 9 Nehete, P.N., V.D. Shah and R.M. Kothari. 1986. Isolation
of a high yielding alkaline protease variant of Bacillus
If glucose is used as the only carbon source in the medium,
licheniformis. Enzyme Microbiol. Technol. 8: 370-372.
glucose feeding is necessary. Starch is always used as the 10 Priest, F. 1977. Extracellular enzyme synthesis in the genus
carbon source in batch fermentation processes which are Bacillus. Bacteriol. Rev. 41: 711-753.
non-feeding. These processes are still the established ways 11 Sneath, P.H.A. 1986. Endospore-formingGram-positiverods
of industrial protease production in China. Our fed-batch and cocci. In: Bergey's Manual of Systematic Bacteriology,
 Vol. 2 (Sneath, P.H.A., N.S. Mair and M.E. Sharpe, eds.), formation in a recycling fermentation. Antonie van Leeuwen-
p. 1133, Williams & Wilkins, Baltimore. hoek 52: 325-342.
12 van Verseveld, H.W., J.A. Dehollander, J. Frankena, M. 13 Wouters, J.T.M. and P.J. Buysman. 1977. Production of some
Braster, F.J. Leeuwerik and A.H. Stouthamer. 1986. Model- exocellular enzymes by Bacillus licheniformis 749/C in chemo-
ing of microbial substrate conversion, growth and product stat culture. FEMS Lett. 1: 109-112.