- Email: [email protected]
Three-stage fermentation and kinetic modeling of bioflocculant by Corynebacterium glutamicum
Three-stage Fermentation and Kinetic Modeling of Bioflocculant by Corynebacterium glutamicum Liang Shen, Zhongtao An, Qingbiao Li, Chuanyi Yao, Yajuan Peng, Yuanpeng Wang, Ruihua Lai, Xu Deng, Ning He PII: DOI: Reference:
S1004-9541(14)00239-0 doi: 10.1016/j.cjche.2014.11.012 CJCHE 154
To appear in: Received date: Revised date:
2 December 2013 24 February 2014
Please cite this article as: Liang Shen, Zhongtao An, Qingbiao Li, Chuanyi Yao, Yajuan Peng, Yuanpeng Wang, Ruihua Lai, Xu Deng, Ning He, Three-stage Fermentation and Kinetic Modeling of Bioflocculant by Corynebacterium glutamicum, (2014), doi: 10.1016/j.cjche.2014.11.012
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Biotechnology and Bioengineering
of
T
Three-stage Fermentation and Kinetic Modeling Bioflocculant by Corynebacterium glutamicum*
IP
Liang SHEN1, Zhongtao AN1, Qingbiao LI1, Chuanyi YAO1, Yajuan PENG1, Yuanpeng WANG1,
1
SC R
Ruihua LAI1, Xu DENG2 ， Ning HE1, **
Department of Chemical and Biochemical Engineering, College of Chemistry and Chemical
Engineering, Xiamen University, the Key Lab for Synthetic Biotechnology of Xiamen City, Xiamen 361005, China 2
College of Life Science, Shenzhen Key Laboratory of Microbial Genetic Engineering, Shenzhen
NU
University, Shenzhen 518060, China
Abstract Fermentation of bioflocculant with Corynebacterium glutamicum was studied by way of
MA
kinetic modeling. Lorentzian modified Logistic model, time-corrected Luedeking-Piret and Luedeking-Piret type models were proposed and applied to describe the cell growth, bioflocculant
D
synthesis and consumption of substrates, with the correlation of initial biomass concentration and initial glucose concentration, respectively. The results showed that these models could well
TE
characterize the batch culture process of C. glutamicum at various initial glucose concentrations from 10.0 to 17.5 g·L-1. The initial biomass concentration could shorten the lag time of cell growth,
CE P
while the maximum biomass concentration was achieved only at the optimal initial glucose concentration of 16.22 g·L-1. A novel three-stage fed-batch strategy for bioflocculant production was developed based on the model prediction, in which the lag phase, quick biomass growth and
AC
bioflocculant production stages were sequentially proceeded with the adjustment of glucose concentration and dissolved oxygen. Biomass of 2.23 g·L-1 was obtained and bioflocculant concentration was enhanced to 176.32 mg·L-1, 18.62% and 403.63% higher than those in the batch process, respectively, indicating an efficient fed-batch culture strategy for bioflocculant production. Keywords bioflocculant, fermentation, Corynebacterium glutamicum, modeling, kinetics
Article history: Received 2 December 2013 Received in revised form 24 February 2014 Accepted 14 May 2014 Available online xxxx
1 INTRODUCTION * Supported by the National Natural Science Foundation of China (21206143, 51378444) and the program for New Century Exellent Talents of Education Ministry of China (ncet-13-0501). ** Corresponding author. E-mail address: [email protected] 1
ACCEPTED MANUSCRIPT Bioflocculant (microbial flocculant), secreted by certain algae, bacteria, fungi as well as yeast, is an extracellular biopolymer, which is able to induce solid particles, cells and colloidal particles in a liquid suspension to flocculate. Contrasted with
T
traditional chemosynthetic flocculant, biofloculant is harmless and biodegradable
IP
with less secondary pollution [1]. Bioflocculant can be produced at high rates and the
SC R
extracellular bioflocculant is easily recovered from the fermentation broth [2, 3]. Nevertheless, the present reports about bioflocculant are mainly focused on the isolation of bioflocculant-producing microorganisms, chemical structures and
NU
properties of bioflocculant, and the mechanisms of flocculation [4-8]. Research on the fermentation process of bioflocculant is still quite limited.
MA
For this purpose, some research papers have probed in using the kinetic models to exploit an efficient strategy for bioflocculant production. Kang et al. [9], Liu et al.
D
[10] and Cui [11] proposed the Logistic equation and Luedeking-Piret equation to
TE
describe microbial growth and bioflocculant synthesis by Aureobasidium pullulans or Enterococcus cecorum, but these models are not appropriate to predict the
CE P
fermentation process [12]. The Andrews model, describing substrate inhibition effect, could be simplified and employed to modify the Logistic equation. Cheng et al. [13] used the Andrews model modified Logistic equation to describe the growth of
AC
1,3-propanediol-producing microorganism Klebsiella pneumoniae. The modified Logistic equation could well describe the cell growth at a constant initial substrate concentration of 50 g·L-1. In a range of substrate concentrations from 20 to 87 g·L-1, the model was also adapted well. However, the value of Cx,max, the maximum biomass concentration, was constantly yielded by parameter estimation at the constant initial substrate concentration of 50 g·L-1. In fact, Cx,max values are dependent on the initial substrate concentration, so the relationship between Cx,max values and initial substrate concentrations should be investigated. In our previous study, bioflocculant REA-11, which was proved to be a polymer composed of galacturonic acid, was obtained from the fed-batch fermentation with Corynebacterium glutamicum [14-16]. We found that the Logistic equation, 2
ACCEPTED MANUSCRIPT time-corrected Gaden’s model and two kinetic models in the form of Luedeking-Piret could well describe the cell growth, bioflocculant synthesis, and consumption of glucose and urea, respectively. It seems that these four models could well characterize
T
the batch culture process of C. glutamicum at various initial biomass concentrations.
IP
However, they are not applicable as the initial glucose concentration changes.
SC R
Therefore, in this paper, kinetic models are constructed at different initial glucose concentrations for the batch fermentation process of bioflocculant by C. glutamicum. Based on the dynamic analysis, a novel three-stage fed-batch
NU
fermentation strategy is proposed to improve the bioflocculant production.
MA
2 MATERIALS AND METHODS 2.1 Materials
D
2.1.1 Microorganism
TE
The strain used in this study was Corynebacterium glutamicum, presently
China).
CE P
preserved at the China Center for Type Culture Collection (CCTCC 201005, Wuhan,
2.1.2 Media
AC
The medium for slant consisted of (per liter): 5 g glucose, 1 g yeast extract, 1 g beef extract, 2 g tryptone, trace FeSO 4 and 15 g agar. The initial pH was adjusted to 7.2.
The seed medium consisted of (per liter): 10 g glucose, 0.5 g yeast extract, 0.5 g urea, 0.1 g KH2PO4, 0.1 g K2HPO4, 0.1 g NaCl, and 0.2 g MgSO4. The initial pH was adjusted to 8.0. The fermentation medium consisted of (per liter): 1 g yeast extract, 1 g urea, 0.1 g KH2PO4, 0.1 g K2HPO4, 0.1 g NaCl, and 0.2 g MgSO4. In carbon source single-factor experiments, the glucose concentration was 10, 12.5, 15, 16.25, 17.5 g·L-1. In fed-batch culture process, the initial glucose was 16.22 g·L-1. The initial pH of all media was adjusted to 8.0.
3
ACCEPTED MANUSCRIPT 2.1.3 Cultivation conditions The slants were incubated at 28C for 16 h. For seed preparation, two loops of cells were inoculated into 100 ml of seed medium in a 250-ml flask and incubated at
IP
T
28C, 120 r·min-1 on a reciprocal shaker until the optical density at 600 nm of seed culture reached 0.6-0.8.
SC R
For batch fermentation, the seed culture was inoculated at 5% (by volume) into 100 ml of the fermentation medium in a 250-ml flask. The inoculated flask was kept on a rotary shaker at 28C, 120 r·min-1 for 48 h.
NU
The fed-batch fermentation was performed on a 2-liter fermentor (Applikon
MA
Biotechnology Applikon BioBundle, Netherlands). 5% (by volume) of seed culture was inoculated into 1400 ml fermentation culture. The three-stage fed-batch strategy was applied: (1) feeding with 60 gL-1 glucose solution (28.5 ml) and 32 g·L-1 urea
D
solution (14.25 ml) at the 6th hour; (2) feeding with 60 gL-1 glucose solution (95 ml)
TE
and 32 g·L-1 urea solution (47.5 ml) at the 10th hour; (3) feeding with 60 g·L-1 glucose
CE P
solution (95 ml) and 32 g·L-1 urea solution (47.5 ml) at the 19th hour. In this process, the agitation was stopped after 31 h till the culture finished and the aeration rate was
AC
kept at 2 L·L-1·min-1 throughout the process.
2.2 Analytical methods 2.2.1 Determination of biomass Cell growth was measured by dry cell mass (DCM). A total of 5 ml of fermented medium was centrifuged at 10000g for 15 min, washed twice with distilled water, and dried at 105°C until a constant mass was achieved. 2.2.2 Determination of glucose Glucose was determined according to reference [17]. Standard sucrose solutions were prepared at concentrations of 0, 0.08, 0.16, 0.24, 0.32 and 0.4 mg·ml-1 separately. Each of the standards was mixed with 1.5 ml 3,5-dinitrosalicylic acid reagent. The optical density of the solutions was measured at 520 nm and the standard curve was
4
ACCEPTED MANUSCRIPT obtained. Then 1 ml of cell-free culture broth was treated in the same way as described above. The sucrose concentration of the culture broth was calculated according to the standard curve.
T
2.2.3 Determination of urea
IP
Urea concentration was measured by paradimethylaminobenzaldehyde (PDAB)
SC R
method [18]. A series of standard urea solutions at concentrations of 0, 0.04, 0.1, 0.15, 0.2, 0.3 and 0.4 mg·ml-1 were prepared; 5.0 ml PDAB solution (0.03 g·ml-1) and 3.0 ml HCl (10 mol·L-1) were added into each of the standard solutions, mixed and left stand
NU
for 15 min. The optical densities of the standard urea solutions were measured at 446 nm and the standard curve was obtained. Then 1 ml of cell-free culture broth was
MA
treated in the same way as described above. The urea concentration of the culture broth was calculated according to the standard curve.
D
2.2.4 Determination of bioflocculant
TE
Bioflocculant concentration was measured according to the methods described by Wang et al. [19] and Dische [20] through the detection of galacturonic acid, which was
CE P
determined by carbazole colorimetry after eliminating sugar with ethanol.
AC
3 RESULTS AND DISCUSSION 3.1 Cell growth in the batch culture of C. glutamicum The cell growth curves of Corynebacterium glutamicum at various initial
glucose concentrations from 10 to 17.5 g·L-1 are shown in Fig. 1. The results show that the highest biomass concentration appears at initial glucose concentration of 16.25 g·L-1 in the stationary phase while the lowest biomass is observed at initial glucose concentration of 10.0 g·L-1.
5
SC R
IP
T
ACCEPTED MANUSCRIPT
MA
NU
Fig. 1 The cell growth in the batch culture of Corynebacterium glutamicum at various initial glucose concentrations (■10 g·L-1; ●12.5 g·L-1; ▲15 g·L-1; ▼16.25 g·L-1; ◄17.5 g·L-1)
3.2 Logistic model and description of cell growth of C. glutamicum Logistic equation was created by Verhulst for human population growth
TE
D
modeling and rediscovered by Pearl and Reed for the same purpose [21]. The Logistic equation is a substrate independent model. It can well describe the inhibition
CE P
of biomass on growth, existing in many batch fermentations [22]. The Logistic equation is
AC
Cx,0
Cx,0 exp( m t ) Cx,0 1 [1 exp( m t )] Cx,max
(1)
where Cx,0 is the initial biomass concentration (g·L-1), Cx,max is the maximum biomass concentration (g·L-1), μm is the maximum specific growth rate (h -1), and t is the fermentation time (h). Curve fittings with Logistic model for cell growth were established according to our previous research (figure not shown) [12]. The parameter values of the Logistic equation are given in Table 1. The maximum biomass concentration could be achieved at the initial glucose concentration of 16.25 g·L-1. Since the model parameter in Table 1 and the batch culture result in Fig.1 are in a good agreement, it could be assumed that the growth of C. glutamicum is a substrate independent and biomass self-inhibition process. This assumption is in conflict with most reports for
6
ACCEPTED MANUSCRIPT C. glutamicum to produce organic acids, in which the fermentation performance was largely dependent on the substrate [23, 24]. In this meaning, Logistic model could not describe precisely the C. glutamicum culture for bioflocculant REA-11 production.
Cx,0
Cx,max
-1
concentration
-1
/g·L
/g·L
0.03
12.50
0.03
15.00
0.02
16.25
0.03
17.50
0.03
µm /h-1
1.4651
0.2103
1.4808
0.2055
1.5501
0.2115
1.6512
0.2022
NU
10.00
SC R
/g·L-1
IP
Initial glucose
T
Table 1 Parameter values of the Logistic equation
1.5465
0.1929
MA
3.3 Modified Logistic equation and description of cell growth of C. glutamicum
D
In a microbial growth medium, the maximum biomass concentration is closely
TE
related to the concentration of carbon source. Thus the relationship between Cx,max and initial glucose concentration should be reflected in the Logistic equation.
CE P
The relationship between Cx,max and initial glucose concentration is assumed as a peak-shaped function (data not shown). The peak-shaped curves are in different
AC
forms, each with its own characteristics such as Gaussian, Logistic peak equation and Lorentzian equation [25]. Fitted with the three models, R2 were calculated to be 0.9944, 0.9956 and 0.9993, respectively, indicating that Lorentzian equation achieved higher accuracy. Thus the Lorentzian equation is chosen to describe the peak-shaped curve, described as Cx,max Cx,max 0
2A w 4(Cs,0 Cs,c )2 w2
(2)
where Cx,max0 is the lowest maximum biomass concentration (g·L-1), Cs,0 is the initial glucose concentration (g·L-1), Cs,c is the initial glucose concentration with the highest maximum biomass concentration (g·L-1), μm is the maximum specific growth rate (h-1), A and w are constant. The parameter values of the equation are calculated as Cx,max0 = 1.4612, Cs,c = 16.2246, w = 2.3072, A = 0.6884, according to the least squares 7
ACCEPTED MANUSCRIPT fitting by Matlab software. Thus the Lorentzian model is established as Cx,max 1.4612
2 0.6884 2.3072 4(Cs,0 16.2246)2 2.30722
(3)
T
The optimal initial glucose concentration for cell growth is calculated to be
Cx,0 exp( m t ) Cx,0 [1 exp( m t )] 1 2 0.6884 2.3072 1.4612 4(Cs,0 16.2246) 2 2.30722
SC R
Cx
IP
16.22 g·L-1. The Lorentzian modified Logistic model is also established.
(4)
NU
where µm is the mean of maximum specific growth rate, µm = 0.2045, with SD = 0.0075.
MA
As shown in Fig. 2, the modified Logistic model could well describe the cell
AC
CE P
TE
D
growth at various initial glucose concentrations.
Fig. 2 Experimental data (■) and simulation (―) of biomass concentration with modified Logistic equation in the batch culture of C. glutamicum [initial glucose concentration/initial biomass concentration, g·L-1/g·L-1: (a) 10.00/0.03; (b) 12.50/0.03; (c) 15.00/0.02; (d) 16.25/0.03; (e) 17.50/0.03]
3.4 Modified Luedeking-Piret equation and description of bioflocculant production process A delay of bioflocculant production was found compared with the cell growth in our previous study [12]. Therefore, a parameter of the lag time, td, is introduced to 8
ACCEPTED MANUSCRIPT modify the Luedeking-Piret model, called mixed model or part-growth-associated model. This modified Luedeking-Piret equation can be described as dCx dP Yp / x C x dt d(t td )
T
(5)
IP
where Yp/x is the product yield coefficient on biomass (mg·g-1), β is the
(h). The integrated form of Eq. (5) is given by
SC R
non-growth-associated product formation coefficient (g·g−1·h−1), and td is the lag time
MA
NU
Cx,0 exp(0.2045 (t td )) Cx,0 exp( 0.2045 td ) Cp Yp / x Cx,0 [1 exp(0.2045 (t td ))] Cx,0 [1 exp( 0.2045 td )] 1 1 2 0.6884 2.3072 2 0.6884 2.3072 1.4612 1.4612 2 2 2 2 [4( C 16.2246) 2.3072 ] [4( C 16.2246) 2.3072 ] s,0 s,0
TE
D
Cx,0 [1 exp(0.2045 t )] 2 0.6884 2.3072 1.4612 / 0.2045 ln 1 2 0.6884 2.3072 [4(Cs,0 16.2246) 2 2.30722 ] 1.4612 [4(Cs,0 16.2246) 2 2.30722 ] Cp,0
(6)
CE P
The parameter values of the equation are as follows: td = 9.4056, Yp/x = 9.0898, and β = 0.3556, with R2 = 0.9738.
AC
Eq.(6) is applied to simulate the experimental results at initial glucose concentration of 15.00 g·L-1 and initial biomass concentration of 0.03 g·L-1 [Fig. 3(a)]. The time-corrected Luedeking-Piret model fits the bioflocculant synthesis very well.
Fig. 3 Experimental data (■) and simulation (―) of bioflocculant production with time-corrected Luedeking-Piret model and substrate consumption with 9
ACCEPTED MANUSCRIPT Luedeking-Piret type models. (a) bioflocculant production; (b) glucose consumption; (c) urea consumption
3.5 Substrate consumption models
T
Carbon substrate such as glucose is used to form cell structures and metabolic
IP
products as well as the maintenance of cell metabolism [12, 22]. The glucose
SC R
consumption model is a Luedeking-Piret type equation as follows, in which the amount of carbon substrate used for product formation is assumed to be negligible. dCs 1 dCx mCx dt Yx / s dt
(7)
NU
where Yx/s is the biomass yield coefficient on glucose (g·g-1) and m is the maintenance
MA
coefficient (g·g-1·h-1). Integrating Eq. (7) gives Eq. (8),
TE
D
Cx,0 exp(0.2045 t ) Cs Cs,0 1/ Yx / s Cx,0 Cx,0 [1 exp(0.2045 t )] 1 2 0.6884 2.3072 1.4612 2 2 [4( C 16.2246) 2.3072 ] s,0
CE P
Cx,0 [1 exp(0.2045 t )] 2 0.6884 2.3072 1.4612 / 0.2045 m ln 1 2 0.6884 2.3072 [4(Cs,0 16.2246) 2 2.30722 ] 1.4612 [4(Cs,0 16.2246) 2 2.30722 ]
AC
(8)
The values of parameters are estimated to be Yx/s = 0.5264 and m = 0.0835, with
R2 = 0.9635.
Besides, the urea consumption in the batch fermentation of bioflocculant is proposed as
dCu 1 dCx dt Yx / u dt
(9)
where Yx/u is the biomass yield coefficient on urea (g·g-1). The integrated form of Eq. (9) is
10
ACCEPTED MANUSCRIPT
(10)
T
Cx,0 exp(0.2045 t ) 1 Cu Cu,0 C x ,0 C [1 exp(0.2045 t )] Yx / u x,0 1 2.3072 1.4612 2 0.6884 2 2 4( C 16.2246) 2.3072 s,0
IP
The parameter value is calculated to be Yx/u = 3.1983, with R2 = 0.9745.
SC R
Eqs. (8) and (10) are used to simulate the experimental results at initial glucose concentration of 13 g·L-1 and initial biomass concentration of 0.03 g·L-1 [Fig. 3(b,c)]. The results show that the Luedeking-Piret type equation successfully described glucose
NU
and urea consumption in the batch culture process of C. glutamicum.
MA
3.6 Prediction of batch fermentation process of bioflocculant by C. glutamicum
D
As suggested by the above results, the four models could describe the batch
TE
fermentation process of bioflocculant REA-11 by C. glutamicum quite well. For
CE P
further verification of the application of the models, we investigate their prediction potentials.
Fig. 4 gives the prediction curves for cell growth of C. glutamicum with the
AC
Lorentzian modified Logistic equation. The cell growth is accelerated at high initial biomass concentration from 0.01 to 0.05 g·L-1 [Fig. 4(a)]. Meanwhile, glucose exhibits an inhibition on the cell growth at concentrations above 16.225 g·L-1 [Fig. 4(b)], which is consistent with the optimal batch culture condition of glucose concentration. Regarding the comprehensive effect of these two factors as shown in Fig. 4(a,b), it seems that the high initial biomass concentration could shorten the lag phase, but the maximum biomass concentration is reached only at the optimal initial glucose concentration of 16.225 g·L-1. Similarly, Fig. 5 gives the prediction curves of bioflocculant production and substrate consumption with the time-corrected Luedeking-Piret model and Luedeking-Piret type models at various initial biomass and glucose concentrations. Results also indicate that the yield of bioflocculant is dominant by the optimized initial glucose concentration, and the high initial biomass 11
ACCEPTED MANUSCRIPT
D
MA
NU
SC R
IP
T
concentration could increase the production rate in the first stage of fermentation.
AC
CE P
TE
Fig. 4 Biomass prediction by Lorentzian modified Logistic equation in the batch culture of C. glutamicum at various initial biomass and glucose concentrations (IBC: initial biomass concentration; IGC: initial glucose concentration)
12
CE P
TE
D
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 5 Prediction of bioflocculant production and substrate consumption with modified Luedeking-Piret model and Luedeking-Piret type models (IBC: initial biomass concentration; IGC: initial glucose concentration)
Kinetic modeling is an essential step in developing a fermentation process since
the models can be used to determine the optimal operation condition for the production of a target metabolite [26-28]. Therefore, a new technical process could be designed based on the models proposed here for the bioflocculant fermentation with C. glutamicum.
3.7 Three-stage fed-batch culture of C. glutamicum for bioflocculant production According to the prediction results from the kinetic models, a three-stage fed-batch strategy was presented for bioflocculant production by C. glutamicum, as
13
ACCEPTED MANUSCRIPT described in the Materials and Methods. 16.22 g·L-1 was chosen as the initial glucose concentration based on the modified Logistic model calculation. As shown in Fig. 6, the first stage (0-6th hour) was determined by the lag time from the prediction curve in
T
Fig. 4. For the second stage (6th-31th hour), glucose was fed to maintain the
IP
concentration about 16 g·L-1 to continuously benefit the cell growth, at the 6th, 10th and
SC R
19th hour. The biomass concentration was quickly increased to a high level about 1.6 g·L-1, which was close to the maximum biomass concentration from the model prediction value (Fig. 4) at the end of the second stage. The third stage (after 31th hour)
NU
was thereafter designed to harvest bioflocculant by manipulating dissolved oxygen (DO). During this process, the agitation was stopped after 31 h till the culture finished
MA
and the aeration rate was kept at 2 L·L-1·min-1 throughout the process. This is ascribed to the important role of DO in bioflocculant production. Our previous study showed
D
that bioflocculant REA-11 was a polygalacturonic acid biosynthesized from
80%
with
DO
TE
phosphate-1-glucose [29]. The metabolic flux to Pentose pathway increased nearly increased
from
10%
to
70%.
The
metabolic
flux
in
CE P
Embden-Meyerhof-Parnas (EMP) pathway almost remained the same while those with the acetic acid and lactic acid synthetic pathway increased and that with REA-11 synthesis
decreased.
This
result
is
consistent
with
the
conclusion
that
AC
adenosine-triphosphate (ATP) flux is more favorable for cell growth but is unfavorable for the synthesis of REA-11 [14].
Fig. 6 Time courses for bioflocculant production with the three-stage fed-batch culture strategy by C. glutamicum 14
ACCEPTED MANUSCRIPT
The results of the overall three-stage fermentation showed that a biomass of 2.23 g·L-1 was obtained and the bioflocculant concentration was enhanced to 176.32 mg·L-1,
T
18.62% and 403.63% higher than those in the batch process, respectively, indicating an
IP
effective fed-batch strategy for bioflocculant production by C. glutamicum (Fig. 6).
SC R
The results suggest the rationality and practicability of the kinetic models.
NU
4 CONCLUSIONS
Lorentzian modified Logistic model was constructed to describe the cell growth in
MA
the batch culture of C. glutamicum with correlation of initial glucose concentration. Time-corrected Luedeking-Piret model and Luedeking-Piret type models were employed to characterize bioflocculant production and substrate consumption. All the
D
four models fit the batch process very well. Model prediction results show that initial
TE
biomass concentration could shorten the lag time of cell growth, while the maximum biomass concentration is achieved only at the optimal initial glucose concentration, i.e.
CE P
16.22 g·L-1 in this study.
Based on the dynamic data from the kinetic models, a three-stage fed-batch
AC
strategy was investigated, in which the fermentation process sequentially passed through the lag phase, quick biomass growth and bioflocculant production stages with adjustment of the glucose concentration and DO. Higher yields of biomass (2.23 g·L-1) and bioflocculant (176.32 mg·L-1) were obtained through this new strategy, implying an applicable bioflocculant production.
NOMENCLATURE A
constant
Β
non-growth-associated product formation coefficient, g·g-1·h-1
Cs,0
initial glucose concentration, g·L-1
Cs,c
initial glucose concentration with the highest maximum biomass concentration, g·L-1
Cu,0
initial urea concentration, g·L-1
15
ACCEPTED MANUSCRIPT initial biomass concentration, g·L-1
Cx,max
maximum biomass concentration, g·L-1
Cx,max0
lowest maximum biomass concentration, g·L-1
m
maintenance coefficient, g·g-1·h-1
t
fermentation time, h
td
lag time, h
w
constant
Yp/x
product yield coefficient on biomass, mg·g-1
Yx/s
biomass yield coefficient on glucose, g·g-1
Yx/u
biomass yield coefficient on urea, g·g-1
μm
maximum specific growth rate, h-1
NU
MA
REFERENCES 1
SC R
IP
T
Cx,0
Salehizadeh, H., Shojaosadati, S.A., "Extracellular biopolymeric flocculants: Recent trends and biotechnological importance", Biotechnol. Adv. 19(5), 371-385 (2001). Jang, J.H., Ike, M., Kim, S., Fujita, M., "Production of a novel bioflocculant by fed-batch
D
2
3
TE
culture of Citrobacter sp." Biotechnol. Lett. 23(8), 593-597 (2001). Xing, J., Yang, J., Ma, F., Wei, L., Liu, K., "Study on the optimal fermentation time and kinetics of bioflocculant produced by bacterium F2", Adv. Mater. Res. 113-116(2379-2384
4
CE P
(2010).
Elkady, M.F., Farag, S., Zaki, S., Abu-Elreesh, G., Abd-El-Haleem, D., "Bacillus mojavensis strain 32A, a bioflocculant-producing bacterium isolated from an Egyptian salt production pond", Bioresour. Technol. 102(17), 8143-8151 (2011).
5
Liu, L.F., Cheng, W., "Characteristics and culture conditions of a bioflocculant produced by
6
AC
Penicillium sp." Biomed. Environ. Sci. 23(3), 213-218 (2010). Liu, W., Wang, K., Li, B., Yuan, H., Yang, J., "Production and characterization of an intracellular bioflocculant by Chryseobacterium daeguense W6 cultured in low nutrition medium", Bioresour. Technol. 101(3), 1044-1048 (2010).
7
Shu, C.H., Hsu, H.J., "Production of schizophyllan glucan by Schizophyllum commune ATCC 38548 from detoxificated hydrolysate of rice hull", J. Taiwan Inst. Chem. Eng. 42(3), 387-393 (2011).
8
Zhang, Z., Xia, S., Zhao, J., Zhang, J., "Characterization and flocculation mechanism of high efficiency microbial flocculant TJ-F1 from Proteus mirabilis", Colloids Surf., B 75(1), 247-251 (2010).
9
Kang, J.X., Meng, S.K., Wu, L., "Model of fermentation dynamics based on bio-flocculant pullulan production by aureobasidium pullulans", J. Harbin Inst. Technol. 37(10), 1370-1372+1409 (2005).
10
Liu, Z.Y., Zhang, T., Zhang, D.Y., Kang, R., Tian, C., Zhang, J.B., "Kinetic models of bioflocculant producing strain under batch fermentation", J. Inner Mongolia Univ. Technol. 26(93-100 (2007).
11
Cui, Y.H., "Fermentation conditions of Aureobasidium pullulans G-58 and its kinetic study", 16
ACCEPTED MANUSCRIPT Ph.D. Thesis, Jiangnan Univ., China (2008). 12
An, Z.T., Li, Q.B., He, N., Chen, R.B., Jiang, X.D., Yang, K., "Kinetic models for bioflocculant fermentation from Corynebacterium glutamicum", CIESC J. 60(12), 3071-3076 (2009).
13
Cheng, K.K., Lin, R.H., Liu, H.J., Liu, D.H., "Kinetic analysis of anaerobic fermentation of
Wu, H., Li, Q., Lu, R., Wang, Y., Zhuang, X., He, N., "Fed-batch production of a
IP
14
T
1,3-propanediol by Klebsiella pneumoniae", Chin. J. Process Eng. 5(425-429 (2005).
bioflocculant from Corynebacterium glutamicum", J. Ind. Microbiol. Biotechnol. 37(11),
SC R
1203-1209 (2010). 15
He, N., Li, Y., Chen, J., Lun, S.Y., "Identification of a novel bioflocculant from a newly isolated Corynebacterium glutamicum", Biochem. Eng. J. 11(2-3), 137-148 (2002).
16
Lu, D., Liu, J., Mao, Z., "Engineering of Corynebacterium glutamicum to enhance
Chem. Eng. 20(4), 731-739 (2012). 17
Miller, G.L., "Use of dinitrosalicylic acid reagent for determination of reducing sugar", Anal.
MA
Chem. 31(3), 426-428 (1959). 18
NU
L-ornithine production by gene knockout and comparative proteomic analysis", Chin. J.
Li, M., Zheng, C.L., Zhang, Y., "Determination of carbamide in waste stripping liquid by colorimetry method of paradimethylaminobezaldehyde", Technol. Development Chem. Ind. 34(40-43 (2005).
Wang, J.H., Shen, Q.P., He, J.B., Li, Y., "Determination of galacturonic acid in health
D
19
foods", Chin. J. Health Lab. Technol. 16(5), 526-527 (2006). Dische, Z., "A new specific color reaction of galacturonic acid", Arch. Biochem. 16(3),
TE
20
409-414 (1948).
Lobry, J.R., Flandrois, J.P., Carret, G., Pave, A., "Monod's bacterial growth model revisited",
CE P
21
Bull. Math. Biol. 54(1), 117-122 (1992). 22
Liu, J.Z., Weng, L.P., Zhang, Q.L., Xu, H., Ji, L.N., "A mathematical model for gluconic acid fermentation by Aspergillus niger", Biochem. Eng. J. 14(2), 137-141 (2003).
23
Kiefer, P., Heinzle, E., Wittmann, C., "Influence of glucose, fructose and sucrose as carbon
AC
sources on kinetics and stoichiometry of lysine production by Corynebacterium glutamicum", J Ind. Microbiol. Biotechnol. 28(6), 338-343 (2002).
24
Jeon, J.M., Thangamani, R., Song, E., Lee, H.W., Yang, Y.H., "Media optimization of Corynebacterium glutamicum for succinate production under oxygen-deprived condition", J. Microbiol. Biotechnol. 23(2), 211-217 (2013).
25
Schreier, P.G., "All peaks aren't Gaussian", Pers. Eng. Instrum. News 51-54 (1991).
26
Feng, J., Zhan, X.B., Wang, D., Zhang, L.M., Lin, C.C., "An unstructured kinetic model to study NaCl effect on volatile ester fermentation by Candida etchellsii for soy sauce production", Biotechnol. Bioproc. Eng. 17(2), 242-249 (2012).
27
Li, B., Chen, X., Ren, H., Li, L., Xiong, J., Bai, J., Chen, Y., Wu, J., Ying, H., "Kinetic models of ribonucleic acid fermentation and continuous culture by Candida tropicalis no.121", Bioprocess Biosyst. Eng. 35(3), 415-422 (2012).
28
Guo, H., Li, H., "On-line batch process monitoring with improved multi-way independent component analysis", Chin. J. Chem. Eng. 21(3), 263-270 (2013).
29
Li, Y., He, N., Guan, H., Du, G., Chen, J., "A novel polygalacturonic acid bioflocculant REA-11 produced by Corynebacterium glutamicum: a proposed biosynthetic pathway and experimental confirmation", Appl. Microbiol. Biotechnol. 63(2), 200-206 (2003). 17
ACCEPTED MANUSCRIPT Graphic Abstract stage II
stage III
MA
NU
SC R
IP
T
stage I
A three-stage fed-batch strategy was proposed for bioflocculant production by C.
D
glutamicum. In the three stages, i.e. the lag phase, the quick biomass growth and the
TE
bioflocculant production stages, the glucose concentration and dissolved oxygen were adjusted according to the prediction results from the modified kinetic models presented
CE P
in this study. Results show that the biomass and bioflocculant production are largely improved in this three-stage fermentation over the conventional fed-batch
AC
fermentation.
18
S1004-9541(14)00239-0 doi: 10.1016/j.cjche.2014.11.012 CJCHE 154
To appear in: Received date: Revised date:
2 December 2013 24 February 2014
Please cite this article as: Liang Shen, Zhongtao An, Qingbiao Li, Chuanyi Yao, Yajuan Peng, Yuanpeng Wang, Ruihua Lai, Xu Deng, Ning He, Three-stage Fermentation and Kinetic Modeling of Bioflocculant by Corynebacterium glutamicum, (2014), doi: 10.1016/j.cjche.2014.11.012
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Biotechnology and Bioengineering
of
T
Three-stage Fermentation and Kinetic Modeling Bioflocculant by Corynebacterium glutamicum*
IP
Liang SHEN1, Zhongtao AN1, Qingbiao LI1, Chuanyi YAO1, Yajuan PENG1, Yuanpeng WANG1,
1
SC R
Ruihua LAI1, Xu DENG2 ， Ning HE1, **
Department of Chemical and Biochemical Engineering, College of Chemistry and Chemical
Engineering, Xiamen University, the Key Lab for Synthetic Biotechnology of Xiamen City, Xiamen 361005, China 2
College of Life Science, Shenzhen Key Laboratory of Microbial Genetic Engineering, Shenzhen
NU
University, Shenzhen 518060, China
Abstract Fermentation of bioflocculant with Corynebacterium glutamicum was studied by way of
MA
kinetic modeling. Lorentzian modified Logistic model, time-corrected Luedeking-Piret and Luedeking-Piret type models were proposed and applied to describe the cell growth, bioflocculant
D
synthesis and consumption of substrates, with the correlation of initial biomass concentration and initial glucose concentration, respectively. The results showed that these models could well
TE
characterize the batch culture process of C. glutamicum at various initial glucose concentrations from 10.0 to 17.5 g·L-1. The initial biomass concentration could shorten the lag time of cell growth,
CE P
while the maximum biomass concentration was achieved only at the optimal initial glucose concentration of 16.22 g·L-1. A novel three-stage fed-batch strategy for bioflocculant production was developed based on the model prediction, in which the lag phase, quick biomass growth and
AC
bioflocculant production stages were sequentially proceeded with the adjustment of glucose concentration and dissolved oxygen. Biomass of 2.23 g·L-1 was obtained and bioflocculant concentration was enhanced to 176.32 mg·L-1, 18.62% and 403.63% higher than those in the batch process, respectively, indicating an efficient fed-batch culture strategy for bioflocculant production. Keywords bioflocculant, fermentation, Corynebacterium glutamicum, modeling, kinetics
Article history: Received 2 December 2013 Received in revised form 24 February 2014 Accepted 14 May 2014 Available online xxxx
1 INTRODUCTION * Supported by the National Natural Science Foundation of China (21206143, 51378444) and the program for New Century Exellent Talents of Education Ministry of China (ncet-13-0501). ** Corresponding author. E-mail address: [email protected] 1
ACCEPTED MANUSCRIPT Bioflocculant (microbial flocculant), secreted by certain algae, bacteria, fungi as well as yeast, is an extracellular biopolymer, which is able to induce solid particles, cells and colloidal particles in a liquid suspension to flocculate. Contrasted with
T
traditional chemosynthetic flocculant, biofloculant is harmless and biodegradable
IP
with less secondary pollution [1]. Bioflocculant can be produced at high rates and the
SC R
extracellular bioflocculant is easily recovered from the fermentation broth [2, 3]. Nevertheless, the present reports about bioflocculant are mainly focused on the isolation of bioflocculant-producing microorganisms, chemical structures and
NU
properties of bioflocculant, and the mechanisms of flocculation [4-8]. Research on the fermentation process of bioflocculant is still quite limited.
MA
For this purpose, some research papers have probed in using the kinetic models to exploit an efficient strategy for bioflocculant production. Kang et al. [9], Liu et al.
D
[10] and Cui [11] proposed the Logistic equation and Luedeking-Piret equation to
TE
describe microbial growth and bioflocculant synthesis by Aureobasidium pullulans or Enterococcus cecorum, but these models are not appropriate to predict the
CE P
fermentation process [12]. The Andrews model, describing substrate inhibition effect, could be simplified and employed to modify the Logistic equation. Cheng et al. [13] used the Andrews model modified Logistic equation to describe the growth of
AC
1,3-propanediol-producing microorganism Klebsiella pneumoniae. The modified Logistic equation could well describe the cell growth at a constant initial substrate concentration of 50 g·L-1. In a range of substrate concentrations from 20 to 87 g·L-1, the model was also adapted well. However, the value of Cx,max, the maximum biomass concentration, was constantly yielded by parameter estimation at the constant initial substrate concentration of 50 g·L-1. In fact, Cx,max values are dependent on the initial substrate concentration, so the relationship between Cx,max values and initial substrate concentrations should be investigated. In our previous study, bioflocculant REA-11, which was proved to be a polymer composed of galacturonic acid, was obtained from the fed-batch fermentation with Corynebacterium glutamicum [14-16]. We found that the Logistic equation, 2
ACCEPTED MANUSCRIPT time-corrected Gaden’s model and two kinetic models in the form of Luedeking-Piret could well describe the cell growth, bioflocculant synthesis, and consumption of glucose and urea, respectively. It seems that these four models could well characterize
T
the batch culture process of C. glutamicum at various initial biomass concentrations.
IP
However, they are not applicable as the initial glucose concentration changes.
SC R
Therefore, in this paper, kinetic models are constructed at different initial glucose concentrations for the batch fermentation process of bioflocculant by C. glutamicum. Based on the dynamic analysis, a novel three-stage fed-batch
NU
fermentation strategy is proposed to improve the bioflocculant production.
MA
2 MATERIALS AND METHODS 2.1 Materials
D
2.1.1 Microorganism
TE
The strain used in this study was Corynebacterium glutamicum, presently
China).
CE P
preserved at the China Center for Type Culture Collection (CCTCC 201005, Wuhan,
2.1.2 Media
AC
The medium for slant consisted of (per liter): 5 g glucose, 1 g yeast extract, 1 g beef extract, 2 g tryptone, trace FeSO 4 and 15 g agar. The initial pH was adjusted to 7.2.
The seed medium consisted of (per liter): 10 g glucose, 0.5 g yeast extract, 0.5 g urea, 0.1 g KH2PO4, 0.1 g K2HPO4, 0.1 g NaCl, and 0.2 g MgSO4. The initial pH was adjusted to 8.0. The fermentation medium consisted of (per liter): 1 g yeast extract, 1 g urea, 0.1 g KH2PO4, 0.1 g K2HPO4, 0.1 g NaCl, and 0.2 g MgSO4. In carbon source single-factor experiments, the glucose concentration was 10, 12.5, 15, 16.25, 17.5 g·L-1. In fed-batch culture process, the initial glucose was 16.22 g·L-1. The initial pH of all media was adjusted to 8.0.
3
ACCEPTED MANUSCRIPT 2.1.3 Cultivation conditions The slants were incubated at 28C for 16 h. For seed preparation, two loops of cells were inoculated into 100 ml of seed medium in a 250-ml flask and incubated at
IP
T
28C, 120 r·min-1 on a reciprocal shaker until the optical density at 600 nm of seed culture reached 0.6-0.8.
SC R
For batch fermentation, the seed culture was inoculated at 5% (by volume) into 100 ml of the fermentation medium in a 250-ml flask. The inoculated flask was kept on a rotary shaker at 28C, 120 r·min-1 for 48 h.
NU
The fed-batch fermentation was performed on a 2-liter fermentor (Applikon
MA
Biotechnology Applikon BioBundle, Netherlands). 5% (by volume) of seed culture was inoculated into 1400 ml fermentation culture. The three-stage fed-batch strategy was applied: (1) feeding with 60 gL-1 glucose solution (28.5 ml) and 32 g·L-1 urea
D
solution (14.25 ml) at the 6th hour; (2) feeding with 60 gL-1 glucose solution (95 ml)
TE
and 32 g·L-1 urea solution (47.5 ml) at the 10th hour; (3) feeding with 60 g·L-1 glucose
CE P
solution (95 ml) and 32 g·L-1 urea solution (47.5 ml) at the 19th hour. In this process, the agitation was stopped after 31 h till the culture finished and the aeration rate was
AC
kept at 2 L·L-1·min-1 throughout the process.
2.2 Analytical methods 2.2.1 Determination of biomass Cell growth was measured by dry cell mass (DCM). A total of 5 ml of fermented medium was centrifuged at 10000g for 15 min, washed twice with distilled water, and dried at 105°C until a constant mass was achieved. 2.2.2 Determination of glucose Glucose was determined according to reference [17]. Standard sucrose solutions were prepared at concentrations of 0, 0.08, 0.16, 0.24, 0.32 and 0.4 mg·ml-1 separately. Each of the standards was mixed with 1.5 ml 3,5-dinitrosalicylic acid reagent. The optical density of the solutions was measured at 520 nm and the standard curve was
4
ACCEPTED MANUSCRIPT obtained. Then 1 ml of cell-free culture broth was treated in the same way as described above. The sucrose concentration of the culture broth was calculated according to the standard curve.
T
2.2.3 Determination of urea
IP
Urea concentration was measured by paradimethylaminobenzaldehyde (PDAB)
SC R
method [18]. A series of standard urea solutions at concentrations of 0, 0.04, 0.1, 0.15, 0.2, 0.3 and 0.4 mg·ml-1 were prepared; 5.0 ml PDAB solution (0.03 g·ml-1) and 3.0 ml HCl (10 mol·L-1) were added into each of the standard solutions, mixed and left stand
NU
for 15 min. The optical densities of the standard urea solutions were measured at 446 nm and the standard curve was obtained. Then 1 ml of cell-free culture broth was
MA
treated in the same way as described above. The urea concentration of the culture broth was calculated according to the standard curve.
D
2.2.4 Determination of bioflocculant
TE
Bioflocculant concentration was measured according to the methods described by Wang et al. [19] and Dische [20] through the detection of galacturonic acid, which was
CE P
determined by carbazole colorimetry after eliminating sugar with ethanol.
AC
3 RESULTS AND DISCUSSION 3.1 Cell growth in the batch culture of C. glutamicum The cell growth curves of Corynebacterium glutamicum at various initial
glucose concentrations from 10 to 17.5 g·L-1 are shown in Fig. 1. The results show that the highest biomass concentration appears at initial glucose concentration of 16.25 g·L-1 in the stationary phase while the lowest biomass is observed at initial glucose concentration of 10.0 g·L-1.
5
SC R
IP
T
ACCEPTED MANUSCRIPT
MA
NU
Fig. 1 The cell growth in the batch culture of Corynebacterium glutamicum at various initial glucose concentrations (■10 g·L-1; ●12.5 g·L-1; ▲15 g·L-1; ▼16.25 g·L-1; ◄17.5 g·L-1)
3.2 Logistic model and description of cell growth of C. glutamicum Logistic equation was created by Verhulst for human population growth
TE
D
modeling and rediscovered by Pearl and Reed for the same purpose [21]. The Logistic equation is a substrate independent model. It can well describe the inhibition
CE P
of biomass on growth, existing in many batch fermentations [22]. The Logistic equation is
AC
Cx,0
Cx,0 exp( m t ) Cx,0 1 [1 exp( m t )] Cx,max
(1)
where Cx,0 is the initial biomass concentration (g·L-1), Cx,max is the maximum biomass concentration (g·L-1), μm is the maximum specific growth rate (h -1), and t is the fermentation time (h). Curve fittings with Logistic model for cell growth were established according to our previous research (figure not shown) [12]. The parameter values of the Logistic equation are given in Table 1. The maximum biomass concentration could be achieved at the initial glucose concentration of 16.25 g·L-1. Since the model parameter in Table 1 and the batch culture result in Fig.1 are in a good agreement, it could be assumed that the growth of C. glutamicum is a substrate independent and biomass self-inhibition process. This assumption is in conflict with most reports for
6
ACCEPTED MANUSCRIPT C. glutamicum to produce organic acids, in which the fermentation performance was largely dependent on the substrate [23, 24]. In this meaning, Logistic model could not describe precisely the C. glutamicum culture for bioflocculant REA-11 production.
Cx,0
Cx,max
-1
concentration
-1
/g·L
/g·L
0.03
12.50
0.03
15.00
0.02
16.25
0.03
17.50
0.03
µm /h-1
1.4651
0.2103
1.4808
0.2055
1.5501
0.2115
1.6512
0.2022
NU
10.00
SC R
/g·L-1
IP
Initial glucose
T
Table 1 Parameter values of the Logistic equation
1.5465
0.1929
MA
3.3 Modified Logistic equation and description of cell growth of C. glutamicum
D
In a microbial growth medium, the maximum biomass concentration is closely
TE
related to the concentration of carbon source. Thus the relationship between Cx,max and initial glucose concentration should be reflected in the Logistic equation.
CE P
The relationship between Cx,max and initial glucose concentration is assumed as a peak-shaped function (data not shown). The peak-shaped curves are in different
AC
forms, each with its own characteristics such as Gaussian, Logistic peak equation and Lorentzian equation [25]. Fitted with the three models, R2 were calculated to be 0.9944, 0.9956 and 0.9993, respectively, indicating that Lorentzian equation achieved higher accuracy. Thus the Lorentzian equation is chosen to describe the peak-shaped curve, described as Cx,max Cx,max 0
2A w 4(Cs,0 Cs,c )2 w2
(2)
where Cx,max0 is the lowest maximum biomass concentration (g·L-1), Cs,0 is the initial glucose concentration (g·L-1), Cs,c is the initial glucose concentration with the highest maximum biomass concentration (g·L-1), μm is the maximum specific growth rate (h-1), A and w are constant. The parameter values of the equation are calculated as Cx,max0 = 1.4612, Cs,c = 16.2246, w = 2.3072, A = 0.6884, according to the least squares 7
ACCEPTED MANUSCRIPT fitting by Matlab software. Thus the Lorentzian model is established as Cx,max 1.4612
2 0.6884 2.3072 4(Cs,0 16.2246)2 2.30722
(3)
T
The optimal initial glucose concentration for cell growth is calculated to be
Cx,0 exp( m t ) Cx,0 [1 exp( m t )] 1 2 0.6884 2.3072 1.4612 4(Cs,0 16.2246) 2 2.30722
SC R
Cx
IP
16.22 g·L-1. The Lorentzian modified Logistic model is also established.
(4)
NU
where µm is the mean of maximum specific growth rate, µm = 0.2045, with SD = 0.0075.
MA
As shown in Fig. 2, the modified Logistic model could well describe the cell
AC
CE P
TE
D
growth at various initial glucose concentrations.
Fig. 2 Experimental data (■) and simulation (―) of biomass concentration with modified Logistic equation in the batch culture of C. glutamicum [initial glucose concentration/initial biomass concentration, g·L-1/g·L-1: (a) 10.00/0.03; (b) 12.50/0.03; (c) 15.00/0.02; (d) 16.25/0.03; (e) 17.50/0.03]
3.4 Modified Luedeking-Piret equation and description of bioflocculant production process A delay of bioflocculant production was found compared with the cell growth in our previous study [12]. Therefore, a parameter of the lag time, td, is introduced to 8
ACCEPTED MANUSCRIPT modify the Luedeking-Piret model, called mixed model or part-growth-associated model. This modified Luedeking-Piret equation can be described as dCx dP Yp / x C x dt d(t td )
T
(5)
IP
where Yp/x is the product yield coefficient on biomass (mg·g-1), β is the
(h). The integrated form of Eq. (5) is given by
SC R
non-growth-associated product formation coefficient (g·g−1·h−1), and td is the lag time
MA
NU
Cx,0 exp(0.2045 (t td )) Cx,0 exp( 0.2045 td ) Cp Yp / x Cx,0 [1 exp(0.2045 (t td ))] Cx,0 [1 exp( 0.2045 td )] 1 1 2 0.6884 2.3072 2 0.6884 2.3072 1.4612 1.4612 2 2 2 2 [4( C 16.2246) 2.3072 ] [4( C 16.2246) 2.3072 ] s,0 s,0
TE
D
Cx,0 [1 exp(0.2045 t )] 2 0.6884 2.3072 1.4612 / 0.2045 ln 1 2 0.6884 2.3072 [4(Cs,0 16.2246) 2 2.30722 ] 1.4612 [4(Cs,0 16.2246) 2 2.30722 ] Cp,0
(6)
CE P
The parameter values of the equation are as follows: td = 9.4056, Yp/x = 9.0898, and β = 0.3556, with R2 = 0.9738.
AC
Eq.(6) is applied to simulate the experimental results at initial glucose concentration of 15.00 g·L-1 and initial biomass concentration of 0.03 g·L-1 [Fig. 3(a)]. The time-corrected Luedeking-Piret model fits the bioflocculant synthesis very well.
Fig. 3 Experimental data (■) and simulation (―) of bioflocculant production with time-corrected Luedeking-Piret model and substrate consumption with 9
ACCEPTED MANUSCRIPT Luedeking-Piret type models. (a) bioflocculant production; (b) glucose consumption; (c) urea consumption
3.5 Substrate consumption models
T
Carbon substrate such as glucose is used to form cell structures and metabolic
IP
products as well as the maintenance of cell metabolism [12, 22]. The glucose
SC R
consumption model is a Luedeking-Piret type equation as follows, in which the amount of carbon substrate used for product formation is assumed to be negligible. dCs 1 dCx mCx dt Yx / s dt
(7)
NU
where Yx/s is the biomass yield coefficient on glucose (g·g-1) and m is the maintenance
MA
coefficient (g·g-1·h-1). Integrating Eq. (7) gives Eq. (8),
TE
D
Cx,0 exp(0.2045 t ) Cs Cs,0 1/ Yx / s Cx,0 Cx,0 [1 exp(0.2045 t )] 1 2 0.6884 2.3072 1.4612 2 2 [4( C 16.2246) 2.3072 ] s,0
CE P
Cx,0 [1 exp(0.2045 t )] 2 0.6884 2.3072 1.4612 / 0.2045 m ln 1 2 0.6884 2.3072 [4(Cs,0 16.2246) 2 2.30722 ] 1.4612 [4(Cs,0 16.2246) 2 2.30722 ]
AC
(8)
The values of parameters are estimated to be Yx/s = 0.5264 and m = 0.0835, with
R2 = 0.9635.
Besides, the urea consumption in the batch fermentation of bioflocculant is proposed as
dCu 1 dCx dt Yx / u dt
(9)
where Yx/u is the biomass yield coefficient on urea (g·g-1). The integrated form of Eq. (9) is
10
ACCEPTED MANUSCRIPT
(10)
T
Cx,0 exp(0.2045 t ) 1 Cu Cu,0 C x ,0 C [1 exp(0.2045 t )] Yx / u x,0 1 2.3072 1.4612 2 0.6884 2 2 4( C 16.2246) 2.3072 s,0
IP
The parameter value is calculated to be Yx/u = 3.1983, with R2 = 0.9745.
SC R
Eqs. (8) and (10) are used to simulate the experimental results at initial glucose concentration of 13 g·L-1 and initial biomass concentration of 0.03 g·L-1 [Fig. 3(b,c)]. The results show that the Luedeking-Piret type equation successfully described glucose
NU
and urea consumption in the batch culture process of C. glutamicum.
MA
3.6 Prediction of batch fermentation process of bioflocculant by C. glutamicum
D
As suggested by the above results, the four models could describe the batch
TE
fermentation process of bioflocculant REA-11 by C. glutamicum quite well. For
CE P
further verification of the application of the models, we investigate their prediction potentials.
Fig. 4 gives the prediction curves for cell growth of C. glutamicum with the
AC
Lorentzian modified Logistic equation. The cell growth is accelerated at high initial biomass concentration from 0.01 to 0.05 g·L-1 [Fig. 4(a)]. Meanwhile, glucose exhibits an inhibition on the cell growth at concentrations above 16.225 g·L-1 [Fig. 4(b)], which is consistent with the optimal batch culture condition of glucose concentration. Regarding the comprehensive effect of these two factors as shown in Fig. 4(a,b), it seems that the high initial biomass concentration could shorten the lag phase, but the maximum biomass concentration is reached only at the optimal initial glucose concentration of 16.225 g·L-1. Similarly, Fig. 5 gives the prediction curves of bioflocculant production and substrate consumption with the time-corrected Luedeking-Piret model and Luedeking-Piret type models at various initial biomass and glucose concentrations. Results also indicate that the yield of bioflocculant is dominant by the optimized initial glucose concentration, and the high initial biomass 11
ACCEPTED MANUSCRIPT
D
MA
NU
SC R
IP
T
concentration could increase the production rate in the first stage of fermentation.
AC
CE P
TE
Fig. 4 Biomass prediction by Lorentzian modified Logistic equation in the batch culture of C. glutamicum at various initial biomass and glucose concentrations (IBC: initial biomass concentration; IGC: initial glucose concentration)
12
CE P
TE
D
MA
NU
SC R
IP
T
ACCEPTED MANUSCRIPT
AC
Fig. 5 Prediction of bioflocculant production and substrate consumption with modified Luedeking-Piret model and Luedeking-Piret type models (IBC: initial biomass concentration; IGC: initial glucose concentration)
Kinetic modeling is an essential step in developing a fermentation process since
the models can be used to determine the optimal operation condition for the production of a target metabolite [26-28]. Therefore, a new technical process could be designed based on the models proposed here for the bioflocculant fermentation with C. glutamicum.
3.7 Three-stage fed-batch culture of C. glutamicum for bioflocculant production According to the prediction results from the kinetic models, a three-stage fed-batch strategy was presented for bioflocculant production by C. glutamicum, as
13
ACCEPTED MANUSCRIPT described in the Materials and Methods. 16.22 g·L-1 was chosen as the initial glucose concentration based on the modified Logistic model calculation. As shown in Fig. 6, the first stage (0-6th hour) was determined by the lag time from the prediction curve in
T
Fig. 4. For the second stage (6th-31th hour), glucose was fed to maintain the
IP
concentration about 16 g·L-1 to continuously benefit the cell growth, at the 6th, 10th and
SC R
19th hour. The biomass concentration was quickly increased to a high level about 1.6 g·L-1, which was close to the maximum biomass concentration from the model prediction value (Fig. 4) at the end of the second stage. The third stage (after 31th hour)
NU
was thereafter designed to harvest bioflocculant by manipulating dissolved oxygen (DO). During this process, the agitation was stopped after 31 h till the culture finished
MA
and the aeration rate was kept at 2 L·L-1·min-1 throughout the process. This is ascribed to the important role of DO in bioflocculant production. Our previous study showed
D
that bioflocculant REA-11 was a polygalacturonic acid biosynthesized from
80%
with
DO
TE
phosphate-1-glucose [29]. The metabolic flux to Pentose pathway increased nearly increased
from
10%
to
70%.
The
metabolic
flux
in
CE P
Embden-Meyerhof-Parnas (EMP) pathway almost remained the same while those with the acetic acid and lactic acid synthetic pathway increased and that with REA-11 synthesis
decreased.
This
result
is
consistent
with
the
conclusion
that
AC
adenosine-triphosphate (ATP) flux is more favorable for cell growth but is unfavorable for the synthesis of REA-11 [14].
Fig. 6 Time courses for bioflocculant production with the three-stage fed-batch culture strategy by C. glutamicum 14
ACCEPTED MANUSCRIPT
The results of the overall three-stage fermentation showed that a biomass of 2.23 g·L-1 was obtained and the bioflocculant concentration was enhanced to 176.32 mg·L-1,
T
18.62% and 403.63% higher than those in the batch process, respectively, indicating an
IP
effective fed-batch strategy for bioflocculant production by C. glutamicum (Fig. 6).
SC R
The results suggest the rationality and practicability of the kinetic models.
NU
4 CONCLUSIONS
Lorentzian modified Logistic model was constructed to describe the cell growth in
MA
the batch culture of C. glutamicum with correlation of initial glucose concentration. Time-corrected Luedeking-Piret model and Luedeking-Piret type models were employed to characterize bioflocculant production and substrate consumption. All the
D
four models fit the batch process very well. Model prediction results show that initial
TE
biomass concentration could shorten the lag time of cell growth, while the maximum biomass concentration is achieved only at the optimal initial glucose concentration, i.e.
CE P
16.22 g·L-1 in this study.
Based on the dynamic data from the kinetic models, a three-stage fed-batch
AC
strategy was investigated, in which the fermentation process sequentially passed through the lag phase, quick biomass growth and bioflocculant production stages with adjustment of the glucose concentration and DO. Higher yields of biomass (2.23 g·L-1) and bioflocculant (176.32 mg·L-1) were obtained through this new strategy, implying an applicable bioflocculant production.
NOMENCLATURE A
constant
Β
non-growth-associated product formation coefficient, g·g-1·h-1
Cs,0
initial glucose concentration, g·L-1
Cs,c
initial glucose concentration with the highest maximum biomass concentration, g·L-1
Cu,0
initial urea concentration, g·L-1
15
ACCEPTED MANUSCRIPT initial biomass concentration, g·L-1
Cx,max
maximum biomass concentration, g·L-1
Cx,max0
lowest maximum biomass concentration, g·L-1
m
maintenance coefficient, g·g-1·h-1
t
fermentation time, h
td
lag time, h
w
constant
Yp/x
product yield coefficient on biomass, mg·g-1
Yx/s
biomass yield coefficient on glucose, g·g-1
Yx/u
biomass yield coefficient on urea, g·g-1
μm
maximum specific growth rate, h-1
NU
MA
REFERENCES 1
SC R
IP
T
Cx,0
Salehizadeh, H., Shojaosadati, S.A., "Extracellular biopolymeric flocculants: Recent trends and biotechnological importance", Biotechnol. Adv. 19(5), 371-385 (2001). Jang, J.H., Ike, M., Kim, S., Fujita, M., "Production of a novel bioflocculant by fed-batch
D
2
3
TE
culture of Citrobacter sp." Biotechnol. Lett. 23(8), 593-597 (2001). Xing, J., Yang, J., Ma, F., Wei, L., Liu, K., "Study on the optimal fermentation time and kinetics of bioflocculant produced by bacterium F2", Adv. Mater. Res. 113-116(2379-2384
4
CE P
(2010).
Elkady, M.F., Farag, S., Zaki, S., Abu-Elreesh, G., Abd-El-Haleem, D., "Bacillus mojavensis strain 32A, a bioflocculant-producing bacterium isolated from an Egyptian salt production pond", Bioresour. Technol. 102(17), 8143-8151 (2011).
5
Liu, L.F., Cheng, W., "Characteristics and culture conditions of a bioflocculant produced by
6
AC
Penicillium sp." Biomed. Environ. Sci. 23(3), 213-218 (2010). Liu, W., Wang, K., Li, B., Yuan, H., Yang, J., "Production and characterization of an intracellular bioflocculant by Chryseobacterium daeguense W6 cultured in low nutrition medium", Bioresour. Technol. 101(3), 1044-1048 (2010).
7
Shu, C.H., Hsu, H.J., "Production of schizophyllan glucan by Schizophyllum commune ATCC 38548 from detoxificated hydrolysate of rice hull", J. Taiwan Inst. Chem. Eng. 42(3), 387-393 (2011).
8
Zhang, Z., Xia, S., Zhao, J., Zhang, J., "Characterization and flocculation mechanism of high efficiency microbial flocculant TJ-F1 from Proteus mirabilis", Colloids Surf., B 75(1), 247-251 (2010).
9
Kang, J.X., Meng, S.K., Wu, L., "Model of fermentation dynamics based on bio-flocculant pullulan production by aureobasidium pullulans", J. Harbin Inst. Technol. 37(10), 1370-1372+1409 (2005).
10
Liu, Z.Y., Zhang, T., Zhang, D.Y., Kang, R., Tian, C., Zhang, J.B., "Kinetic models of bioflocculant producing strain under batch fermentation", J. Inner Mongolia Univ. Technol. 26(93-100 (2007).
11
Cui, Y.H., "Fermentation conditions of Aureobasidium pullulans G-58 and its kinetic study", 16
ACCEPTED MANUSCRIPT Ph.D. Thesis, Jiangnan Univ., China (2008). 12
An, Z.T., Li, Q.B., He, N., Chen, R.B., Jiang, X.D., Yang, K., "Kinetic models for bioflocculant fermentation from Corynebacterium glutamicum", CIESC J. 60(12), 3071-3076 (2009).
13
Cheng, K.K., Lin, R.H., Liu, H.J., Liu, D.H., "Kinetic analysis of anaerobic fermentation of
Wu, H., Li, Q., Lu, R., Wang, Y., Zhuang, X., He, N., "Fed-batch production of a
IP
14
T
1,3-propanediol by Klebsiella pneumoniae", Chin. J. Process Eng. 5(425-429 (2005).
bioflocculant from Corynebacterium glutamicum", J. Ind. Microbiol. Biotechnol. 37(11),
SC R
1203-1209 (2010). 15
He, N., Li, Y., Chen, J., Lun, S.Y., "Identification of a novel bioflocculant from a newly isolated Corynebacterium glutamicum", Biochem. Eng. J. 11(2-3), 137-148 (2002).
16
Lu, D., Liu, J., Mao, Z., "Engineering of Corynebacterium glutamicum to enhance
Chem. Eng. 20(4), 731-739 (2012). 17
Miller, G.L., "Use of dinitrosalicylic acid reagent for determination of reducing sugar", Anal.
MA
Chem. 31(3), 426-428 (1959). 18
NU
L-ornithine production by gene knockout and comparative proteomic analysis", Chin. J.
Li, M., Zheng, C.L., Zhang, Y., "Determination of carbamide in waste stripping liquid by colorimetry method of paradimethylaminobezaldehyde", Technol. Development Chem. Ind. 34(40-43 (2005).
Wang, J.H., Shen, Q.P., He, J.B., Li, Y., "Determination of galacturonic acid in health
D
19
foods", Chin. J. Health Lab. Technol. 16(5), 526-527 (2006). Dische, Z., "A new specific color reaction of galacturonic acid", Arch. Biochem. 16(3),
TE
20
409-414 (1948).
Lobry, J.R., Flandrois, J.P., Carret, G., Pave, A., "Monod's bacterial growth model revisited",
CE P
21
Bull. Math. Biol. 54(1), 117-122 (1992). 22
Liu, J.Z., Weng, L.P., Zhang, Q.L., Xu, H., Ji, L.N., "A mathematical model for gluconic acid fermentation by Aspergillus niger", Biochem. Eng. J. 14(2), 137-141 (2003).
23
Kiefer, P., Heinzle, E., Wittmann, C., "Influence of glucose, fructose and sucrose as carbon
AC
sources on kinetics and stoichiometry of lysine production by Corynebacterium glutamicum", J Ind. Microbiol. Biotechnol. 28(6), 338-343 (2002).
24
Jeon, J.M., Thangamani, R., Song, E., Lee, H.W., Yang, Y.H., "Media optimization of Corynebacterium glutamicum for succinate production under oxygen-deprived condition", J. Microbiol. Biotechnol. 23(2), 211-217 (2013).
25
Schreier, P.G., "All peaks aren't Gaussian", Pers. Eng. Instrum. News 51-54 (1991).
26
Feng, J., Zhan, X.B., Wang, D., Zhang, L.M., Lin, C.C., "An unstructured kinetic model to study NaCl effect on volatile ester fermentation by Candida etchellsii for soy sauce production", Biotechnol. Bioproc. Eng. 17(2), 242-249 (2012).
27
Li, B., Chen, X., Ren, H., Li, L., Xiong, J., Bai, J., Chen, Y., Wu, J., Ying, H., "Kinetic models of ribonucleic acid fermentation and continuous culture by Candida tropicalis no.121", Bioprocess Biosyst. Eng. 35(3), 415-422 (2012).
28
Guo, H., Li, H., "On-line batch process monitoring with improved multi-way independent component analysis", Chin. J. Chem. Eng. 21(3), 263-270 (2013).
29
Li, Y., He, N., Guan, H., Du, G., Chen, J., "A novel polygalacturonic acid bioflocculant REA-11 produced by Corynebacterium glutamicum: a proposed biosynthetic pathway and experimental confirmation", Appl. Microbiol. Biotechnol. 63(2), 200-206 (2003). 17
ACCEPTED MANUSCRIPT Graphic Abstract stage II
stage III
MA
NU
SC R
IP
T
stage I
A three-stage fed-batch strategy was proposed for bioflocculant production by C.
D
glutamicum. In the three stages, i.e. the lag phase, the quick biomass growth and the
TE
bioflocculant production stages, the glucose concentration and dissolved oxygen were adjusted according to the prediction results from the modified kinetic models presented
CE P
in this study. Results show that the biomass and bioflocculant production are largely improved in this three-stage fermentation over the conventional fed-batch
AC
fermentation.
18