Page 1
ORIGINAL PAPER
Optimization and scale-up of 2,3-butanediol productionby Bacillus amyloliquefaciens B10-127
Taowei Yang • Xian Zhang • Zhiming Rao •
Shenghui Gu • Haifeng Xia • Zhenghong Xu
Received: 30 August 2011 / Accepted: 17 November 2011 / Published online: 26 November 2011
� Springer Science+Business Media B.V. 2011
Abstract The effects of culture conditions on 2,3-
butanediol (2,3-BD) production and its possible scale-up
have been studied. A newly isolated Bacillus amylolique-
faciens B10-127, belonged to GRAS microorganisms and
showed a remarkable 2,3-BD producing potency, was used
for this experiment. Corn steep liquor, soybean meal and
ammonium citrate were found to be the key factors in the
fermentation according to the results obtained from the
Plackett–Burman experimental design. The optimal con-
centration range of the three factors was examined by the
steepest ascent path, and their optimal concentration were
further optimized via response surface methodological
approach and determined to be 31.9, 22.0 and 5.58 g/l,
respectively. The concentration of the obtained 2,3-BD
increased significantly with optimized medium (62.7 g/l)
when compared with unoptimized medium (45.7 g/l) and
the 2,3-BD productivity was about 2.4-fold (The fermen-
tation time was shorten from 72 to 42 h). To observe scale-
up effects, batch fermentation was carried out at various
working volumes. At a working volume of 20.0 l, the final
2,3-BD concentration and yield were 61.4 and 0.38 g/g at
36 h with a 2,3-BD productivity of 1.71 g/l h. This result
shows similar amount of 2,3-BD obtained in lab-scale
fermentation, and it is possible to scale up to larger fer-
mentors without major problems.
Keywords Optimization � Scale-up � 2,3-butanediol �Bacillus amyloliquefaciens B10-127
Introduction
The bio-based bulk chemicals production from renewable
resources has recently attracted increasing attention as it is
a green technology and environment-friendly compared
with chemical processes (Hermann et al. 2007). Microbial
production of 2,3-butanediol (2,3-BD) is one of the
examples. Interest in this bioprocess has been increasing
recently due to that 2,3-BD has large number of industrial
applications and this course would alleviate the depen-
dence on oil supply for the production of platform chem-
icals. As an important starting material, 2,3-BD can be
used to produce valuable derivatives such as methyl ethyl
ketone and 1,3-butadiene (Haveren et al. 2008; Tran and
Chambers 1987). Besides, 2,3-BD has wide application in
transport fuels production, in the manufacturing of printing
inks, perfumes, and fumigants, moistening and softening
agents, explosives and plasticizes, pharmaceutical carriers
(Celinska and Grajek 2009; Syu 2001).
2,3-BD could be produced from carbohydrates via the
mixed acid fermentation pathway by many bacterial spe-
cies such as Klebsiella pneumoniae (Yu and Saddler 1983),
Enterobacter aerogenes (Zeng et al. 1990), Bacillus poly-
myxa (De Mas et al. 1988) and Serratia marcescens (Neish
et al. 1947; Zhang et al. 2010a, b). So far, the most efficient
2,3-BD producers reported are K. pneumoniae, Klebsiella
oxytoca and E. aerogenes (Celinska and Grajek 2009).
Among all these strains, Bacillus amyloliquefaciens was
T. Yang � X. Zhang � Z. Rao (&) � S. Gu � H. Xia
The Key Laboratory of Industrial Biotechnology, Ministry
of Education, Laboratory of Applied Microorganisms and
Metabolic Engineering, School of Biotechnology, Jiangnan
University, Wuxi 214122, Jiangsu Province, People’s Republic
of China
e-mail: [email protected]
Z. Xu
Laboratory of Pharmaceutical Engineering, School of Medicine
and Pharmaceutics, Jiangnan University, Wuxi 214122, Jiangsu
Province, People’s Republic of China
123
World J Microbiol Biotechnol (2012) 28:1563–1574
DOI 10.1007/s11274-011-0960-7
Page 2
rarely reported as a producing strain for 2,3-BD production
compared with other organisms. While the strain generally
recognized as safe (GRAS) microorganism used for
industrial production was much safer than 2,3-BD-pro-
ducing strain K. pneumoniae that cause disease.
Optimization of culture medium is a very important
aspect in the field of food microbiology and fermentation to
improve product yield and reduce process variability, as
well as reducing development time and overall costs
(Kennedy and Krouse 1999; Mu et al. 2009). Many
researchers are therefore studying processes for the pro-
duction of 2,3-BD. In the previous study, several strategies
have been widely used to enhance 2,3-BD production, such
as optimizing medium component, optimizing fermentation
operating conditions and establishing mathematical models
(Alam et al. 1990; Celinska and Grajek 2009; Ghosh and
Swaminathan 2003; Zeng et al. 1994). Nutritional and
physiological factors such as media composition, aeration,
pH, and temperature of the cell culture are essential for
operating fermentative processes. The optimization of
environmental and nutritional conditions for production of
2,3-BD by S. marcescens, K. pneumoniae and K. oxytoca
have been studied (Zhang et al. 2010a, b), respectively, and
the concentration of production showed a considerable
improvement under the optimal situation.
In our previous work, a GRAS strain B. amylolique-
faciens B10-127, capable of producing 2,3-BD effectively
and tolerating glucose up to 300 g/l was isolated. And we
also optimized fermentation operating conditions (Yang
et al. 2011). Compared with other known 2,3-BD produc-
ing strains, the 2,3-BD productivity of this microorganism
was not satisfactory, but B. amyloliquefaciens B10-127 is
superior for its GRAS status that meets safety regulations
for industrial-scale fermentation. In this report, for optimal
production of 2,3-BD, physiological and nutritional factors
have to be studied; however, scale-up for industrial appli-
cation production by Bacillus genus has rarely been stud-
ied. In the present investigation, the newly developed
GRAS strain B. amyloliquefaciens B10-127 was used.
Medium optimization was performed using Plackett-Bur-
man and central composite design to maximize the pro-
duction of 2,3-BD. In order to observe scale-up effects,
different sizes of fermentation culture have been tested.
Materials and methods
Bacterial identification
16S rRNA gene sequence was amplified according to
standard procedures (Sambrook and Russell 2001) and
compared to the sequences in the GenBank database
through BLAST sequence analysis (Yang et al. 2011).
Media and culture conditions
The strain B. amyloliquefaciens B10-127 was maintained
on agar slants containing the following medium: glucose
60 g/l, peptone 10 g/l, yeast extract 5 g/l, NaCl 5 g/l, and
2% agar at pH 6.5. The slants were incubated at 37�C for
14 h, maintained at 4�C and subcultured at 4-week
intervals.
The seed culture was prepared by inoculating a full loop
of cells from freshly prepared slants into 50 ml of the
following medium: glucose 60 g/l, K2HPO4 4 g/l, yeast
extract 5 g/l, corn steep liquor 10 g/l, pH 6.5. The culti-
vation was conducted in 250-ml shake flasks for 10 h with
agitation (160 rpm, reciprocal shaker) at 37�C.
Shaking flask effects of fermentation were investigated
in 250 ml Erlenmeyer flasks containing 50 ml of fer-
mentation medium. The size of inoculum was 4% (v/v).
The basic fermentation medium before optimization
contained (g/l): glucose 150 g/l, K2HPO4 6 g/l, corn
steep liquor 10 g/l, yeast extract 10 g/l, pH 6.5. The
cultures were incubated at 37�C on a rotary shaker at
160 rpm.
Effect of nitrogen sources on 2,3-BD production
The impact of various nitrogen sources on cell growth
and 2,3-BD production were firstly investigated by a tra-
ditional step-by-step replacing experimental procedure.
Sources of nitrogen include organic nitrogen and inor-
ganic nitrogen. The sources chosen for the study were
beef extract, yeast extract, peptone, soybean meal, corn
steep liquor, ammonium critrate, urea, (NH4)2SO4,
(NH4)2HPO4, NH4Cl.
Selection of significant variables by Plackett–Burman
design
The Plackett–Burman (PB) design, an efficient technique
for medium-component optimization (Plackett and Burman
1946), was used to select significantly variables for 2,3-BD
production, and insignificant ones were eliminated to
obtain a smaller, more manageable set of factors. The fitted
first-order model is:
Y ¼ b0 þX
biXi
Y is the predicted response, b0 and bi are constant coef-
ficients, and Xi is the coded independent factors. Each
variable is represented at two levels, high and low, which
are denoted by (?1) and (-1), respectively. A total of
eight parameters were included for selection, Table 2
illustrates the levels of each variable used in the experi-
mental design.
1564 World J Microbiol Biotechnol (2012) 28:1563–1574
123
Page 3
Path of steepest ascent
The method of steepest ascent is a procedure for moving
sequentially along the direction of the maximum increase
in the response (Box et al. 1978). The direction of steepest
ascent is the direction in which 2,3-BD increased rapidly
by increasing or decreasing the condition of the significant
factors. Based on the results of the PB experimental design,
the optimal level scope of each selected factor was
examined by means of this well-known path of steepest
ascent method. The path of steepest ascent was initiated
from the center of the PB design. Experiments were per-
formed along the steepest ascent path until the response did
not increase any more.
Central composite designs and response surface
analysis
The next step in the formulation of the medium was to
determine the optimum levels of significant variables for
2,3-BD production. For this purpose, response surface
methodology (RSM) based on central composite design
(CCD) with five coded levels was employed to determine
the most significant factors screened by PB design for
enhancing 2,3-BD production (Abdul Rahman et al. 2011;
Kennedy and Krouse 1999; Malinowska et al. 2009). The
three independent factors were investigated at five different
levels (-1.682, -1, 0, ?1, ?1.682) and the experimental
design used for study are all shown in Table 6. The
response values (Y) in each trial were the average of the
duplicates. The data obtained from RSM on 2,3-BD pro-
duction were subjected analysis of variance (ANOVA).
The experimental results of RSM were fitted via the
response surface regression procedure, using the following
second-order polynomial equation:
Y ¼ b0 þX
bixi þX
biix2i þ
Xbijxixj
in which Y is the predicted response, xi and xj are inde-
pendent variables, b0 is the intercept, bi is the linear
coefficient, bii is the ith quadratic coefficient, and bij is the
ijth interaction coefficient.
Design-Expert, Version 7.0 (STAT-EASE Inc., Minne-
apolis, USA) was used for the experimental designs and
statistical analysis of the experimental data. The analysis of
variance (ANOVA) was used to estimate the statistical
parameters.
Scale-up production of 2,3-BD
In order to obtain scale-up factors, batch fermentation
experiments were performed at various working volumes
using the optimal culture conditions obtained from the
labscale tests. The working volumes were 3.0, 7.0 and 20.0
in 5, 10 and 30 l vessels, respectively (manufactured by
Biotron). Optimization of nutritional conditions determined
by flask experiments were used for scale-up tests. Agitation
speed, aeration rate and temperature were maintained at
350 rpm, 0.66 vvm and 37�C which detected in our pre-
liminary tests.
Analytical methods
The cell mass concentration was determined by measuring
the OD at 600 nm in a UV-visible spectroscopy system
(UV-2000, UNICO, China). The cell dry weight (DCW)
was calculated from the optical density using calibration
curve for the strain. The composition of fermentation broth
was analyzed using a Agilent 1200 high performance liquid
chromatograph (HPLC) system (Agilent Corp., USA) with
a RID-10A refractive index detector. The stationary and
mobile phases were a Waters SugarPak1 (6.5 mmid 9
300 mm; Waters, USA) and ddH2O at 0.5 ml/min,
respectively. The column temperature was controlled at
30�C. All experiments were repeated at least three times.
Results
Bacterial identification
The bacterium used in this study was isolated from our
preliminary work. After purified several times, the isolate
B10-127 was identified through BLAST analysis of the
partial sequences of 16S rRNA gene. It was 99% identical
with some sequence of B. amyloliquefaciens according to
its 16S rDNA sequence. The sequence was deposited in the
GenBank database with accession no. HQ005359. Based
on these results, strain B10-127 was identified as a strain of
B. amyloliquefaciens and designated as B. amyloliquefac-
iens B10-127 (Yang et al. 2011).
Effect of nitrogen sources on 2,3-BD production
The influences of organic and inorganic nitrogen sources
on cell growth and 2,3-BD production were tested. The
results (Fig. 1; Table 1) had clearly shown that cell mass
and 2,3-BD production were markedly affected by the
addition of nitrogen sources. Among the organic nitrogen
sources investigated, soybean meal resulted in the maxi-
mum 2,3-BD, followed by corn steep liquor (Fig. 1),
whereas ammonium critrate proved to be the best among
the inorganic compounds tested (Table 1). So soybean
meal, corn steep liquor and ammonium critrate were cho-
sen as nitrogen sources for further optimization.
World J Microbiol Biotechnol (2012) 28:1563–1574 1565
123
Page 4
Selection of significant variables by Plackett–Burman
design
The PB design is a powerful method for screening signif-
icant factors. The statistical technique is widely used as a
tool for checking the efficiency of several processes. To
optimize the culture medium in the fermentation of glucose
by B. amyloliquefaciens B10-127 to yield 2,3-BD, the
constituents of the medium were firstly examined. PB
design for a total of eight variables was used to identify
which variables have significant effects on 2,3-BD pro-
duction (Table 2). The medium includes phosphate, suc-
cinic acid, Fe2?, Mn2?, and Mg2?, which significantly
affect 2,3-BD production (Garg and Jain 1995; Syu 2001).
In addition to these factors, the medium also contained
glucose as carbon source and corn steep liquor, soybean
meal and ammonium citrate as nitrogen source according
to the pre-experiments. The upper and lower limits of each
variable were chosen according to the preliminary inves-
tigation of the limits of the variables. The PB experimental
design for 14 trials with two levels for each variable and
the effect of 8 variables on 2,3-BD production are dem-
onstrated in Table 3.
To approach the neighborhood of the optimum response,
the fitted first-order model equation for 2,3-BD production
was obtained from the PB design experiments:
Y ¼ 53:88þ 3:63 X1 þ 2:74 X2 þ 5:56 X3 � 0:17 X4
þ 2:01 X5 þ 0:46 X6 � 0:56 X7 þ 2:09 X8
Statistical testing was carried out using Fisher’s test for
ANOVA according to the experimental data. The
coefficient R2 of the first-order model was 0.964,
indicating that nearly 97% of the variability in the
response could be explained by the model. The value of
the adjusted determination coefficient (Adj R2 = 89.3%)
was also very high to advocate for a high significance of
the model. R2 value of this model higher than 0.9 was
considered as having a very high correlation. So it was
reasonable to use the regression model to analyze the
Fig. 1 Effect of organic
nitrogen sources on 2,3-BD
production. Beef extract (filledsquare), Corn steep liquor (filledcircle), Soybean meal (filledtriangle), Peptone (filledinverted triangle), Yeast extract
(filled star)
Table 1 Effect of Inorganic
nitrogen sources on 2,3-BD
production
Time (h) DCW (g/l) Acetoin (g/l) 2,3-BD
(g/l)
Yield
(g/g Glucose)
Productivity
(g/l h)
Control 84 16.1 7.8 52.1 0.35 0.62
Urea 78 17.9 6.9 54.3 0.36 0.70
Ammonium citrate 66 19.1 4.6 57.2 0.38 0.87
(NH4)2SO4 78 20.4 6.3 53.9 0.36 0.69
(NH4)2HPO4 72 19.6 5.4 55.3 0.37 0.77
NH4Cl 84 15.7 8.9 50.7 0.34 0.60
Table 2 The Plackett–Burman design for screening variables in 2,3-
BD production
Factors (g/l) Variables Low level (-1) High level (?1)
Corn steep liquor X1 10 20
Soybean meal X2 4 10
Ammonium Citrate X3 1 4
K2HPO4 X4 1 4
FeSO4�7H2O X5 0 0.1
MnSO4�7H2O X6 0 0.1
MgSO4�7H2O X7 0.2 0.6
Succinic acid X8 0.1 0.5
1566 World J Microbiol Biotechnol (2012) 28:1563–1574
123
Page 5
trends in the responses. In this model, the F test and
P values were used to identify the effect of each factor on
2,3-BD production. P value below 0.05 indicates that the
model terms are significant. Table 4 shows the effects of
the variables on the response and the significant levels.
Based on the statistical analysis, the factors (P \ 0.05)
such as X1 (corn steep liquor), X2 (soybean meal), X3
(ammonium citrate) and X8 (succinic acid) had the greatest
positive impacts on the production of 2,3-BD, but
compared with X1, X2 and X3, X8 had less positive
effect. X5 (FeSO4) and X6 (MnSO4) were set at their high
levels according to the positive effects although they were
nonsignificant to 2,3-BD production. Factors such as X4
(K2HPO4) and X6 (MgSO4) had negative effects and were
set at their low levels. And then, corn steep liquor, soybean
meal and ammonium citrate were selected for further
optimization to obtain a maximum response.
Path of steepest ascent
The path of steepest ascent was determined to find the
proper direction of changing variables by increasing or
decreasing the value of the main factors. The above results
indicated that corn steep liquor, soybean meal and
ammonium citrate can significantly influenced the 2,3-BD
production compared with other factors. To search the
proper direction of these three factors with the other factors
fixed at zero level, the path of the steepest ascent was
employed. The design and responses of the steepest ascent
experiment are shown in Table 5. It is shown that the
highest response was 61.43 g/l when the concentration of
corn steep liquor, soybean meal and ammonium citrate was
selected to be 30, 20 and 5 g/l, respectively. It suggested
that this point was near the optimal point and this combi-
nation was used as the middle point for the second-order
experiment, i.e., CCD.
Table 3 Plackett–Burman
design for screening of
significant factors affecting
2,3-BD production
Run Variable levels 2,3-BD (g/l)
X1 X2 X3 X4 X5 X6 X7 X8
1 1 -1 1 1 -1 1 1 1 58.3
2 1 1 1 -1 -1 -1 1 -1 60.2
3 -1 -1 -1 -1 -1 -1 -1 -1 36.3
4 1 1 -1 1 1 1 -1 -1 53.2
5 1 -1 -1 -1 1 -1 1 1 52.7
6 -1 1 1 1 -1 -1 -1 1 58.8
7 -1 1 1 -1 1 1 1 -1 59.2
8 1 1 -1 -1 -1 1 -1 1 58.2
9 0 0 0 0 0 0 0 0 60.4
10 -1 -1 -1 1 -1 1 1 -1 39.4
11 -1 -1 1 -1 1 1 -1 1 57.7
12 0 0 0 0 0 0 0 0 59.6
13 -1 1 -1 1 1 -1 1 1 50.1
14 1 -1 1 1 1 -1 -1 -1 62.4
Table 4 Effects and statistical analysis of variables
Variable Coefficient Standard error F value P value
Intercept 53.88 0.75 13.52 0.0118*
X1 3.63 0.75 23.48 0.0084*
X2 2.74 0.75 13.43 0.0215*
X3 5.56 0.75 55.21 0.0018*
X4 -0.17 0.75 0.055 0.8265
X5 2.01 0.75 7.21 0.0550
X6 0.46 0.75 0.38 0.5732
X7 -0.56 0.75 0.56 0.4969
X8 2.09 0.75 7.82 0.0498
R2 = 0.9643, R2 (Adj) = 0.8930
* Significant at 99% confidence degree (P \ 0.05)
Table 5 Experiment design and results of the steepest ascent path
Run Corn steep
liquor (g/l)
Soybean
meal (g/l)
Ammonium
Citrate (g/l)
2,3-BD
(g/l)
Origin 15 8 2 35.5
1 20 12 3 47.4
2 25 16 4 53.2
3 30 20 5 61.4
4 35 24 6 58.5
5 40 28 7 50.9
World J Microbiol Biotechnol (2012) 28:1563–1574 1567
123
Page 6
Central composite designs and response surface
analysis
Response surface optimization is more advantageous than
the traditional single parameter optimization in that it saves
time, space and raw materials (Ryad et al. 2010). Based on
the PB design and the path of steepest ascent, RSM using
CCD was applied to determine the optimal levels of the
three selected variables (corn steep liquor, soybean meal
and ammonium citrate) which significantly influenced the
2,3-BD production. A total of 20 runs were needed for
optimizing the three individual parameters in the current
CCD (Table 6). By applying multiple regression analysis
on the experimental data, the response variable and the test
variables were related by the following quadratic equation:
Y ¼ 61:93þ 1:62 X1 þ 2:32 X2 þ 2:46 X3 � 0:66 X1X2
þ 0:31 X1X3 þ 0:29 X2X3 � 1:92 X21 � 2:19 X2
2
� 2:35 X23
where Y is the predicted 2,3-BD production (g/l); X1, X2
and X3 are the coded values of corn steep liquor, soybean
meal and ammonium citrate, respectively.
On the basis of the experimental values, statistical
testing was carried out using Fisher’s test for ANOVA
(Table 7). The Student’s F test and P values were used as a
tool to check the significance of each coefficient, which
also indicated the interaction strength between each inde-
pendent variable. For any of the terms in the model, a large
regression coefficient and a small P value would indicate a
more significant effect on the respective response variables
(Elibol 2004). Thus, the smaller was the values of P, the
more significant was the corresponding coefficient. As
Table 6 The results of the central composition experiment
Run Coded variable level Real variable level 2,3-BD
(g/l)X1 X2 X3 Corn steep
liquor (g/l)
Soybean
meal (g/l)
Ammonium
Citrate (g/l)
1 -1 -1 -1 25 16 4.0 49.1
2 1 -1 -1 35 16 4.0 52.3
3 -1 1 -1 25 24 4.0 54.2
4 1 1 -1 35 24 4.0 55.5
5 -1 -1 1 25 16 6.0 52.7
6 1 -1 1 35 16 6.0 57.9
7 -1 1 1 25 24 6.0 59.7
8 1 1 1 35 24 6.0 61.5
9 -1.68 0 0 21.6 20 5.0 53.5
10 1.68 0 0 38.4 20 5.0 59.8
11 0 -1.68 0 30 13.3 5.0 52.1
12 0 1.68 0 30 26.7 5.0 59.7
13 0 0 -1.68 30 20 3.32 51.6
14 0 0 1.68 30 20 6.68 59.3
15 0 0 0 30 20 5.0 61.7
16 0 0 0 30 20 5.0 62.3
17 0 0 0 30 20 5.0 61.9
18 0 0 0 30 20 5.0 62.0
19 0 0 0 30 20 5.0 61.5
20 0 0 0 30 20 5.0 62.1
Table 7 Significance test of regression coefficient
Variable Coefficient Standard error F value P value
Intercept 61.93 0.18
X1 1.62 0.12 188.55 \0.0001
X2 2.32 0.12 387.65 \0.0001
X3 2.46 0.12 437.31 \0.0001
X1X2 -0.66 0.15 18.52 0.0016
X1X3 0.31 0.15 4.12 0.0698
X2X3 0.29 0.15 3.49 0.0914
X12 -1.92 0.11 280.46 \0.0001
X22 -2.19 0.11 363.23 \0.0001
X32 -2.35 0.11 418.03 \0.0001
Model 214.15 \0.0001
Lack of fit 3.64 0.0911
R2 = 0.9948, R2 (Adj) = 0.9902
1568 World J Microbiol Biotechnol (2012) 28:1563–1574
123
Page 7
shown in Table 7, the F and P values were 214.15
and \0.0001, respectively. So the test model was statisti-
cally significant at the 99% level of significance.
The regression equation obtained from the ANOVA
showed that the multiple correlation coefficient R2 was
0.9984 (a value [0.75 indicates fitness of the model). This
is an estimate of the fraction of overall variation in the
data accounted by the model, and thus the model is
capable of explaining 99.84% of the variation in response.
The adjusted R2 is 0.9902 and the predicted R2 is 0.9665,
which indicates that the model is good (for a good sta-
tistical model, The closer the R2 value is to 1.00, the
stronger the model is and the better it predicts the
response). ‘‘Adeq Precision’’ measures the signal to noise
ratio. The adequate precision value of the present model
was 41.950, and this also suggests that the model can be
used to navigate the design space. The adequate precision
value is an index of the signal-to-noise ratio, and values of
higher than 4 are essential prerequisites for a model to be
a good fit.
The response surface curves are plotted to explain the
interaction of the variables and to determine the optimum
level of each variable for maximum response. The
response surface curves are shown in Figs. 2, 3 and 4.
Each figure demonstrates the effect of two factors while
the other factors were fixed at zero level. The model
predicted the optimal values of the three most significant
variables were X1 = 0.38, X2 = 0.51 and X3 = 0.58.
Correspondingly, the values of corn steep liquor, soybean
meal and ammonium citrate were 31.9, 22.0 and 5.58 g/l,
respectively. The maximum predicted concentration of
2,3-BD was 63.5 g/l. By optimization of culture condi-
tions, 2,3-BD production was enhanced from 45.7 to
63.5 g/l, and the fermentation time was shorten from 72 to
42 h.
Validation of the second-order polynomial equation
Based on the results of medium optimization, the optimum
composition for 2,3-BD production by B. amyloliquefac-
iens B10-127 is as follows (g/l): glucose 150, corn steep
liquor 31.9, soybean meal 22.0, ammonium citrate 5.58,
K2HPO4 2.5, MgSO4�7H2O 0.3, MnSO4�7H2O 0.05, FeS-
O4�7H2O 0.05, Succinic acid 0.3. To validate the adequacy
of the model equation for predicting maximum 2,3-BD
production, three additional experiments in shake flasks
were performed using the predicted culture conditions.
Under the optimized condition, the 2,3-BD average yield of
62.7 g/l was obtained at 42 h, which was obviously in good
agreement with the model predicted maximum value of
63.5 g/l. Therefore, this result indicated that the optimized
medium favored the production of 2,3-BD.
Scale-up production of 2,3-BD
The maximum 2,3-BD concentration of 62.7 g/l at 42 h
with a 2,3-BD productivity of 1.49 g/l h was obtained by
batch culture in shake flasks, although these results were
new records on 2,3-BD fermentation by GRAS microor-
ganism to our knowledge, it was less efficient than that of
reported ‘‘high-producers’’ (such as K. pneumoniae,
K. oxytoca and E. aerogenes). In order to determine the
scale-up capacity, fermentation was carried out at 5-, 10-
and 30-l fermenter with working volumes at 3.0, 7.0 and
20.0 l, respectively. Optimal media and culture conditions
determined by flask experiments were used for scale-up
tests. The fermentation results are shown in Fig. 5, when
the working volume was increased from 3.0 to 20 l, 2,3-BD
production was slightly reduced and the fermentation time
was extended by about 4 h. However, the cellular growth
Fig. 2 Response surface figure (a) and corresponding contour (b) of
the mutual effects of corn steep liquor and soybean meal on 2,3-BD
production
World J Microbiol Biotechnol (2012) 28:1563–1574 1569
123
Page 8
and 2,3-BD secretion were similar at the different working
volumes. It was also found that 2,3-BD production rapidly
reduced after reaching the maximum value and a con-
comitant increase of acetoin production appeared after the
glucose was depleted. The similar phenomenon was also
found by Zhang et al. (2011).
Discussion
In order to make the production of 2,3-BD economical on
industrial scale, high concentration and productivity and
low cost of fermentation are essential (Garg and Jain 1995;
Ma et al. 2009). To achieve this aim several strategies have
been used such as screening a productive strain, optimizing
fermentation operating conditions and establishing mathe-
matical models, beside the mentioned methods, designing
an appropriate fermentation medium is also crucial
important to improve the efficiency and productivity of the
fermentation process because product concentration, yield,
and cell growth conditions are strongly influenced by
medium composition such as the carbon source, nitrogen
source, inorganic salts, and so on (Garg and Jain 1995; Ma
et al. 2009). However, there is no general defined medium
for 2,3-BD production by different microbial strains
because every microorganism has its own special nutri-
tional requirements depending on its environment. So it is
Fig. 3 Response surface figure (a) and corresponding contour (b) of
the mutual effects of corn steep liquor and ammonium citrate on 2,3-
BD production
Fig. 4 Response surface figure (a) and corresponding contour (b) of
the mutual effects of soybean meal and ammonium citrate on 2,3-BD
production
1570 World J Microbiol Biotechnol (2012) 28:1563–1574
123
Page 9
Fig. 5 Batch fermentation
profiles of Bacillusamyloliquefaciens B10-127 for
the scale-up experiment,
determined using various
working volumes: a 3.0 l,
b 7.0 l, c 20.0 l. Acetoin (filledsquare), 2,3-BD (filled circle),
acetoin ?2,3-BD (filledtriangle), DCW (filled invertedtriangle), Glucose (filled star)
World J Microbiol Biotechnol (2012) 28:1563–1574 1571
123
Page 10
not an easy task to explore a medium that contains all the
main nutritional factors and obtain their optimum levels.
This work primarily aimed at optimizing the process
variables for production of 2,3-BD in using statistical
optimization technique for multivariable effect. Culture
conditions and media composition optimization by a con-
ventional one-at-the-approach led to a substantial increase
in 2,3-butanediol yield. However, this approach not only is
generally time consuming and requires a large number of
experiments to be carried out but also has the limitation of
ignoring the importance of interaction of various parame-
ters (Anvari and Safari Motlagh 2011). More efficient
analytical techniques are based on RSM (Bezerra et al.
2008). This was first proposed by Box and his collaborators
in 1951 as a method to determine the optimal conditions
which maximize or minimize a response (Box and Wilson
1951). It enables a large amount of data to be obtained
from a reduced number of experiments, including the
potential interactions between the studied factors (Bas and
BoyacI 2007). The RSM can be defined as a group of
statistical and technical tools used to study the relationship
between a response of interest and several input variables.
The model has to describe the behavior of a group of data
with a view to making statistical predictions. The aim is the
simultaneous optimization of several factors to lead to the
best performance of a particular system (Bezerra et al.
2008; Ryad et al. 2010).
Production of 2,3-BD has been shown to be sensitive to
repression by different nitrogen sources (Ma et al. 2009;
Zhang et al. 2010a, b). So nitrogen sources were firstly
tested on the growth and 2,3-BD production of B. amylo-
liquefaciens B10-127 by the conventional one-at-the-
approach. In the present investigation, results obtained
showed that soybean meal, corn steep liquor and ammo-
nium critrate resulted in a positive effect on 2,3-BD pro-
duction compared to other nitrogen sources. Then
statistically based experimental designs proved to be a
valuable tool in optimizing the medium for 2,3-BD pro-
duction by the isolated strain B. amyloliquefaciens B10-
127. Among the eight variables tested by PB experiments,
soybean meal, corn steep liquor and ammonium critrate
were identified as the most important components for 2,3-
BD production. Their optimal concentrations were
obtained by using statistical analysis of RSM.
In microbial fermentations, the production costs are
mainly dependent on the nitrogen source cost (Karin and
Barbel 2000), as well as the carbon source cost. The
maximum 2,3-BD yield was usually obtained when cells
were grown in glucose media containing peptone and beef
extract (Alam et al. 1990; Nilegaonkar et al. 1992) or yeast
extract (Perego et al. 2000; Zhang et al. 2010a, b). Soybean
meal, the by-product after extracting most of the oil from
soybeans, which is rich in protein and energy, is an
inexpensive valuable nutrient source available on a large
scale. So soybean meal is an inexpensive valuable and
stimulating media component for 2,3-BD fermentation. It
has an obvious economic advantage as compared to other
organic nitrogen sources including yeast extract and pep-
tone. So soybean meal was demonstrated as a good nitro-
gen source for large-scale 2,3-BD fermentation. Corn steep
liquor is a major byproduct of the corn wet-milling
industry, contains approximately 47% crude protein and is
a low-cost nutrient source available on a large scale
(Parekh et al. 1999). Corn steep liquor is widely used in the
fermentation industry to produce a variety of substances,
such as lactic acid by Streptomyces sp. (Rivas et al. 2004)
and ethanol by Zymomonas mobilis (Silveira et al. 2001).
The preliminary experiment results have clearly shown that
2,3-BD production was markedly enhanced by corn steep
liquor compared with other nitrogen sources culture
(Tables 1, 4). This result is in agreement with Ma (Ma
et al. 2009). What’s more, soybean meal and corn steep
liquor are alternative, low-cost, and high-yield media
component for 2,3-BD fermentation and has an obvious
economic advantage. Ammonium critrate is also a most
important factor that influenced the 2,3-BD accumulation
in this study (Table 4). It is reported that 2,3-butanediol
production can be increased by addition of different
organic acids, because they are intermediate metabolites
for 2,3-BD production (Anvari and Safari Motlagh 2011;
Yu and Saddler 1982). So ammonium critrate offers the
nitrogen source for cell growth and is also very important
to product formation.
The scale-up experiment was performed to observe scale-
up effects for application in industrial processes. As shown
in Fig. 5 there was a slight reduction on 2,3-BD production
at a larger working volume. This phenomenon was also
finely explained by Oh (Oh et al. 2009). When scale-up
experiments were performed, parameters such as tempera-
ture and pH were difficult to maintain at the same value,
especially at all locations within the fermentor. As the
absolute quantity of carbon and nitrogen sources increased,
the production of organic acids and endotoxins was
increased. The surrounding conditions were thus harmful to
cell growth. Therefore, when working volumes increased,
cell and production concentration were decreased. And the
maximum 2,3-BD concentration was about 67.6 g/l at sta-
tionary phase. This result shows similar amount of 2,3-BD
obtained in lab-scale fermentation, and it is possible to scale
up to larger fermentors without major problems.
In conclusion, we conducted a sequential statistical
experimental design to optimize the medium for 2,3-BD
production by B. amyloliquefaciens B10-127. Corn steep
liquor, soybean meal and ammonium citrate had a signifi-
cantly effect on 2,3-BD production and the optimal values
of the three key factors were 31.9, 22.0 and 5.58 g/l,
1572 World J Microbiol Biotechnol (2012) 28:1563–1574
123
Page 11
respectively. After the optimization, the 2,3-BD concen-
tration increased to 66.5 g/l over a fermentation period of
30 h for the fed-batch culture when using the optimized
culture medium and the 2,3-BD productivity reached
2.22 g/l h. It was shown that statistical experimental design
offered an effect and feasible approach for 2,3-BD fer-
mentation medium optimization. This newly isolated
GRAS strain B. amyloliquefaciens B10-127 showed a
higher 2,3-BD production potency than that of K. pneu-
moniae, which had been demonstrated for its potential on
an industrial scale. Furthermore, the optimum medium for
2,3-BD production by this strain mainly composed of
inexpensive nutrient sources (common and cheap inorganic
salts, a small quantity of corn steep liquor, soybean meal,
and ammonium critrate) that are available for efficient and
economical production of 2,3-BD on a large scale.
Although cell concentration and 2,3-BD concentration
were decreased slightly, the scale-up experiment from flask
to fermentor showed the possibility of extended application
to commercial production processes. So B. amylolique-
faciens B10-127 should be an excellent candidate for the
microbial fermentation of 2,3-BD on an industrial scale.
Acknowledgments This work was supported by the Program for
New Century Excellent Talents in University (NCET-10-0459), the
National Natural Science Foundation of China (30970056), the High-
tech Research and Development Programs of China (2007AA
02Z207), the Fundamental Research Funds for the Central Universi-
ties (JUSRP31001), the Program of Introducing Talents of Discipline
to Universities (111-2-06) and a Project Funded by the Priority
Academic Program Development of Jiangsu Higher Education
Institutions.
References
Abdul Rahman MB, Jarmi NI, Chaibakhsh N, Basri M (2011)
Modeling and optimization of lipase-catalyzed production of
succinic acid ester using central composite design analysis. J Ind
Microbiol Biotechnol 38:229–234. doi:10.1007/s10295-010-
0817-3
Alam S, Capit F, Weigandg WA, Hong J (1990) Kinetics of 2,
3-butanediol fermentation by Bacillus amyloliquefaciens: effect
of initial substrate concentration and aeration. J Chem Technol
Biotechnol 41:71–84. doi:10.1002/jctb.280470109
Anvari M, Safari Motlagh MR (2011) Enhancement of 2, 3-butane-
diol production by Klebsiella oxytoca PTCC 1402. J Biomed
Biotechnol 2011:1–7. doi:10.1155/2011/636170
Bas D, BoyacI IH (2007) Modeling and optimization I: usability of
response surface methodology. J Food Eng 78:836–845. doi:
10.1016/j.jfoodeng.2005.11.024
Bezerra MA, Santelli RE, Oliveira EP, Villar LS, Escaleira LA (2008)
Response surface methodology (RSM) as a tool for optimization
in analytical chemistry. Talanta 76:965–977. doi:10.1016/j.
talanta.2008.05.019
Box GEP, Wilson KB (1951) On the experimental attainment of
optimum conditions. J Roy Stat Soc Ser B Method 13:1–45
Box GEP, Hunter JS, Hunter WG (1978) Statistics for experimenters.
Wiley, New York, pp 291–334
Celinska E, Grajek W (2009) Biotechnological production of 2,
3-butanediol—current state and prospects. Biotechnol Adv
27:715–725. doi:10.1016/j.biotechadv.2009.05.002
De Mas C, Jansen NB, Tsao GT (1988) Production of optically active
2, 3-butanediol by Bacillus polymyxa. Biotechnol Bioeng
31:366–377. doi:10.1002/bit.260310413
Elibol M (2004) Optimization of medium composition for actinorho-
din production by Streptomyces coelicolor A3(2) with response
surface methodology. Process Biochem 39:1057–1062. doi:
10.1016/s0032-9592(03)00232-2
Garg SK, Jain A (1995) Fermentative production of 2, 3-butanediol: a
review. Bioresour Technol 51:103–109. doi:10.1016/0960-
8524(94)00136-O
Ghosh S, Swaminathan T (2003) Optimization of process variables
for the extractive fermentation of 2, 3-butanediol by Klebsiellaoxytoca in aqueous two-phase system using response surface
methodology. Chem Biochem Eng Q 17:319–325
Haveren Jv, Scott EL, Sanders J (2008) Bulk chemicals from biomass.
Biofuels Bioprod Bioref 2:41–57. doi:10.1002/bbb.43
Hermann BG, Blok K, Patel MK (2007) Producing bio-based bulk
chemicals using industrial biotechnology saves energy and
combats climate change. Environ Sci Technol 41:7915–7921.
doi:10.1021/es062559q
Karin H, Barbel HH (2000) Factors affecting the fermentative lactic
acid production from renewable resources. Enzyme Microb
Technol 26:87–107. doi:10.1016/S0141-0229(99)00155-6
Kennedy M, Krouse D (1999) Strategies for improving fermentation
medium performance: a review. J Ind Microbiol Biotechnol
23:456–475. doi:10.1038/sj.jim.2900755
Ma C, Wang A, Qin J, Li L, Ai X, Jiang T, Tang H, Xu P (2009)
Enhanced 2, 3-butanediol production by Klebsiella pneumoniae
SDM. Appl Microbiol Biotechnol 82:49–57. doi:10.1007/
s00253-008-1732-7
Malinowska E, Krzyczkowski W, Lapienis G, Herold F (2009)
Improved simultaneous production of mycelial biomass and
polysaccharides by submerged culture of Hericium erinaceum:
optimization using a central composite rotatable design (CCRD).
J Ind Microbiol Biotechnol 36:1513–1527. doi:10.1007/s10295-
009-0640-x
Mu W, Chen C, Li X, Zhang T, Jiang B (2009) Optimization of
culture medium for the production of phenyllactic acid by
Lactobacillus sp. SK007. Bioresour Technol 100:1366–1370.
doi:10.1016/j.biortech.2008.08.010
Neish AC, Blackwood AC et al (1947) Production and properties of 2,
3-butanediol; dissimilation of glucose by Serratia marcescens.
Can J Res 25:65–69
Nilegaonkar S, Bhosale SB, Kshirsagar DC, Kapadi AH (1992)
Production of 2, 3-butanediol from glucose by Bacillus lichen-iformis. World J Microbiol Biotechnol 8:378–381. doi:
10.1007/BF01198748
Oh IJ, Kim DH, Oh EK, Lee SY, Lee J (2009) Optimization and
scale-up of succinic acid production by Mannheimia succinic-iproducens LPK7. J Microbiol Biotechnol 19:167–171. doi:10.4014/jmb.0807.447
Parekh M, Formanek J, Blaschek HP (1999) Pilot-scale production of
butanol by Clostridium beijerinckii BA101 using a low-cost
fermentation medium based on corn steep water. Appl Microbiol
Biotechnol 51:152–157. doi:10.1007/s002530051375
Perego P, Converti A, Borghi AD, Canepa P (2000) 2, 3-Butanediol
production by Enterobacter aerogenes: selection of the optimal
conditions and application to food industry residues. Bioprocess
Eng 23:613–620. doi:10.1007/s004490000210
Plackett RL, Burman JP (1946) The design of optimum multifactorial
experiments. Biometrika 33:305–325. doi:10.1093/biomet/33.
4.305
World J Microbiol Biotechnol (2012) 28:1563–1574 1573
123
Page 12
Rivas B, Moldes AB, Domı́nguez JM JCP (2004) Development of
culture media containing spent yeast cells of Debaryomyceshansenii and corn steep liquor for lactic acid production with
Lactobacillus rhamnosus. Int J Food Microbiol 97:93–98. doi:
10.1016/j.ijfoodmicro.2004.05.006
Ryad A, Lakhdar K, Majda KS, Samia A, Mark A, Corinne AD, Eric
G (2010) Optimization of the culture medium composition to
improve the production of hyoscyamine in elicited Daturastramonium L. Hairy roots using the response surface method-
ology (RSM). Int J Mol Sci 11:4726–4740. doi:10.3390/ijms11
114726
Sambrook J, Russell DW (2001) Molecular cloning: a laboratory
manual, 3rd edn. Cold Spring Harbor Laboratory, New York
Silveira MM, Wisbeck E, Hoch I, Jonas R (2001) Production of
glucose-fructose oxidoreductase and ethanol by Zymomonasmobilis ATCC 29191 in medium containing corn steep liquor as
a source of vitamins. Appl Microbiol Biotechnol 55:442–445.
doi:10.1007/s002530000569
Syu MJ (2001) Biological production of 2, 3-butanediol. Appl
Microbiol Biotechnol 55:10–18
Tran AV, Chambers RP (1987) The dehydration of fermentative 2,
3-butanediol into methyl ethyl ketone. Biotechnol Bioeng
29:343–351. doi:10.1002/bit.260290308
Yang TW, Rao ZM, Zhang X, Lin Q, Xia HF, Xu ZH, Yang ST
(2011) Production of 2,3-butanediol from glucose by GRAS
microorganism Bacillus amyloliquefaciens. J Basic Microbiol.
doi:10.1002/jobm.201100033
Yu EK, Saddler JN (1982) Enhanced production of 2, 3-butanediol by
Klebsiella pneumoniae grown on high sugar concentrations in
the presence of acetic acid. Appl Environ Microbiol 44:777–784
Yu EK, Saddler JN (1983) Fed-batch approach to production of 2,
3-butanediol by Klebsiella pneumoniae grown on high substrate
concentrations. Appl Environ Microbiol 46:630–635
Zeng A, Biebl H, Deckwer W (1990) Effect of pH and acetic acid on
growth and 2, 3-butanediol production of Enterobacter aerog-enes in continuous culture. Appl Microbiol Biotechnol
33:485–489. doi:10.1007/BF00172538
Zeng AP, Byun TG, Posten C, Deckwer WD (1994) Use of
respiratory quotient as a control parameter for optimum oxygen
supply and scale-up of 2, 3-butanediol production under
microaerobic conditions. Biotechnol Bioeng 44:1107–1114. doi:
10.1002/bit.260440912
Zhang L, Sun J, Hao Y, Zhu J, Chu J, Wei D, Shen Y (2010a) Microbial
production of 2, 3-butanediol by a surfactant (serrawettin)-
deficient mutant of Serratia marcescens H30. J Ind Microbiol
Biotechnol 37:857–862. doi:10.1007/s10295-010-0733-6
Zhang L, Yang Y, Sun J, Shen Y, Wei D, Zhu J, Chu J (2010b)
Microbial production of 2, 3-butanediol by a mutagenized strain
of Serratia marcescens H30. Bioresour Technol 101:1961–1967.
doi:10.1016/j.biortech.2009.10.052
Zhang X, Yang T, Lin Q, Xu M, Xia H, Xu Z, Li H, Rao Z (2011)
Isolation and identification of an acetoin high production bacte-
rium that can reverse transform 2,3-butanediol to acetoin at the
decline phase of fermentation. World J Microbiol Biotechnol. doi:
10.1007/s11274-011-0754-y
1574 World J Microbiol Biotechnol (2012) 28:1563–1574
123