Author's personal copy

Biochemical Engineering Journal 40 (2008) 107–115

Modeling of batch fermentation kinetics for succinic

acid production by Mannheimia succiniciproducens
Hyohak Song a,b , Seh Hee Jang a,c , Jong Myoung Park a,b , Sang Yup Lee a,b,c,d,e,∗
a Metabolic and Biomolecular Engineering National Research Laboratory, Korea Advanced Institute of Science and Technology (KAIST),
335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of Korea
b Department of Chemical and Biomolecular Engineering (BK21 Program), Korea Advanced Institute of Science and Technology (KAIST),

335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of Korea

c BioProcess Engineering Research Center, Korea Advanced Institute of Science and Technology (KAIST), 335 Gwahangno,

Yuseong-gu, Daejeon 305-701, Republic of Korea

d Center for Systems and Synthetic Biotechnology, Institute for the BioCentry, Korea Advanced Institute of Science and Technology (KAIST),

335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of Korea

e Department of Bio and Brain Engineering, Bioinformatics Research Center, Korea Advanced Institute of Science and Technology (KAIST),

335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of Korea

Received 17 August 2007; received in revised form 15 November 2007; accepted 24 November 2007

Abstract
Kinetic models are proposed for the batch production of succinic acid from glucose by Mannheimia succiniciproducens MBEL55E. The
models include terms accounting for both substrate and product inhibitions. Experimental data collected from a series of batch fermentations
with different initial glucose concentrations were used to estimate parameters and also to validate the models proposed. The optimal values of the
parameters were approximated by minimizing the discrepancy between the model predictions and corresponding experimental data. The growth of
M. succiniciproducens could be expressed by a modified Monod model incorporating inhibitions of glucose and organic acids accumulated in the
culture broth. The Luedeking–Piret model was able to describe the formation of organic acids as the fermentation proceeded, in which succinic,
acetic, and formic acids followed a mixed-growth-associated pattern. However, unexpectedly, lactic acid fermentation by M. succiniciproducens
was nearly nongrowth-associated. In all cases, the model simulation matched well with the experimental observations, which made it possible to
elucidate the fermentation characteristics of M. succiniciproducens during efficient succinic acid production from glucose. These models thus can
be employed for the development and optimization of biobased succinic acid production processes.
© 2007 Elsevier B.V. All rights reserved.

Keywords: Mannheimia succiniciproducens; Succinic acid; Fermentation; Kinetic models; Inhibition

1. Introduction and growing environmental concerns [1]. Indeed, researchers

Succinic acid, an important four-carbon platform chemical,
is mostly being produced by chemical processes using liquefied
petroleum gas or petroleum oil as a starting material. However,
it has been widely accepted that the current petroleum-based
processes will be soon replaced by fermentation of renew-
able resources due to the limited nature of petroleum reserves
∗ Corresponding author at: Department of Bio and Brain Engineering, and
Bioinformatics Research Center, Korea Advanced Institute of Science and Tech-
nology (KAIST), 335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of
Korea. Tel.: +82 42 869 3930; fax: +82 42 869 3910.
E-mail address: leesy@kaist.ac.kr (S.Y. Lee).
have screened several succinic acid-producing microorgan-
isms, including Anaerobiospirillum succiniciproducens [2,3],
Actinobacillus succinogenes [4,5], and Mannheimia suc-
ciniciproducens [6,7]. Among these, M. succiniciproducens
has been studied more extensively based on its complete
genome sequence, and shown to be an efficient succinic acid
producer [8]. In M. succiniciproducens, phosphoenolpyruvate
carboxykinase carboxylates phosphoenolpyruvate to oxaloac-
etate, which is further converted to succinate by the reductive
tricarboxylic acid cycle and menaquinone system. On the
other hand, pyruvate formed from phosphoenolpyruvate is
converted to acetic, formic, and lactic acids via pyruvate
dissimilation pathways. The central carbon metabolism of

1369-703X/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.bej.2007.11.021
 Author's personal copy

108 H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115

M. succiniciproducens is described in detail by Kim et al.

Nomenclature [8].
In order to develop a cost-competitive fermentative suc-
i degree of product inhibition (dimensionless)
cinic acid production process, strain improvement maximizing
j number of experimental data points (dimension-
the production of succinic acid but minimizing the formation
less)
of by-products through rational metabolic engineering is of
Kd specific death rate (h−1 )
vital importance. We recently developed an improved succinic
KI substrate inhibition constant (g L−1 )
acid producer, M. succiniciproducens LPK7, by genome-based
KS substrate saturation constant (g L−1 )
metabolic engineering, resulting in much increased succinic
ms specific maintenance coefficient (h−1 )
acid production while reduced formation of by-products, such
P product concentration (g L−1 )
as acetic, formic, and lactic acids [9]. In addition, the pos-
PCRIT critical product concentration (g L−1 )
sibilities of cost-effective succinic acid production by M.
rP product formation rate (g L−1 h−1 )
succiniciproducens from inexpensive and abundant feedstocks,
−rs substrate consumption rate (g L−1 h−1 )
including wood hydrolysates and whey were also investigated
rX biomass formation rate (g L−1 h−1 )
[10,11].
rAA acetic acid formation rate (g L−1 h−1 )
Kinetic modeling is regarded as an indispensable step in
rFA formic acid formation rate (g L−1 h−1 )
developing a fermentation process since the models can be used
rLA lactic acid formation rate (g L−1 h−1 )
to determine an optimal operation condition for the production
rSA succinic acid formation rate (g L−1 h−1 )
of a target metabolite. Both structured and unstructured models
S substrate concentration (g L−1 )
have been used for kinetic modeling. Although the former can
X biomass concentration (g L−1 )
explain complex microbial system at the molecular levels, rela-
YX stoichiometric yield coefficient for biomass on
tively simpler unstructured kinetic models have frequently been
glucose (g g−1 )
used for practical applications [12,13]. Several unstructured
YAA stoichiometric yield coefficient for acetic acid on
kinetic models have been proposed to describe the fermenta-
glucose (g g−1 )
tion of glucose to organic and amino acids such as citric, lactic,
YFA stoichiometric yield coefficient for formic acid on
and glutamic acids [14–16]. However, little kinetic study on the
glucose (g g−1 )
fermentative production of succinic acid has been performed to
YLA stoichiometric yield coefficient for lactic acid on
date.
glucose (g g−1 )
In this paper, we present empirical kinetic models to describe
YSA stoichiometric yield coefficient for succinic acid
the batch production of succinic acid by M. succiniciproducens
on glucose (g g−1 )
MBEL55E. A series of batch fermentations with different ini-
Z minimizing objective function of the discrepancy
tial glucose concentrations were conducted in a bioreactor, and
between experimental data and simulated results
the experimental data were used to estimate parameters and also
(dimensionless)
to validate the models. The behaviors of M. succiniciproducens
Greek letters during fermentation were predicted using the models, and the
αAA growth-associated acetic acid formation parame- predictions were compared with the experimental observations.
ter (g g−1 DCW) The models developed were able to successfully explain cell
αFA growth-associated formic acid formation param- growth, succinic acid production, by-products (acetic, formic,
eter (g g−1 DCW) and lactic acids) formation, and glucose utilization. The fermen-
αLA growth-associated lactic acid formation parame- tative characteristics of succinic acid production from glucose
ter (g g−1 DCW) by M. succiniciproducens were discussed in detail.
αSA growth-associated succinic acid formation
parameter (g g−1 DCW) 2. Materials and methods
βAA nongrowth-associated acetic acid formation
parameter (g g−1 DCW h−1 ) 2.1. Organism and growth conditions
βFA nongrowth-associated formic acid formation
parameter (g g−1 DCW h−1 ) Preculture of M. succiniciproducens MBEL55E (KCTC
βLA nongrowth-associated lactic acid formation 0769BP; Korean Collection for Type Cultures, Daejeon, Korea)
parameter (g g−1 DCW h−1 ) [6] was performed in a sealed anaerobic flask containing 250 mL
βSA nongrowth-associated succinic acid formation medium composed of yeast extract 5 g L−1 , NaCl 1 g L−1 ,
parameter (g g−1 DCW h−1 ) CaCl2 ·2H2 O 0.2 g L−1 , MgCl2 ·6H2 O 0.2 g L−1 , and K2 HPO4
μ specific growth rate (h−1 ) 8.709 g L−1 . The medium with CO2 as the gas phase was heat
μm maximum specific growth rate (h−1 ) sterilized at 121 ◦ C for 15 min. The pH of the medium was
adjusted to 7.0 by adding 5N NaOH. Separately sterilized glu-
cose was added into the medium to a final concentration of
50 mM as a carbon source just before inoculation. Inoculation
was done by seeding 2.5 mL glycerol stock culture (15%, w/v,
 Author's personal copy

H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115 109

−70 ◦ C) into the medium. Anaerobic flask cultures were per- addition, the presence of the high concentrations of acids in cul-
formed at 39 ◦ C for 9 h under a CO2 atmosphere in an anaerobic ture broth often causes to cell death [22–24]. The dependence
chamber (ThermoForma, Marietta, OH). of cell growth on products could be described by a generalized
Batch fermentations were carried out at 39 ◦ C in a 5-L LiFlus nonlinear equation suggested by Levenspiel [25]. To incorpo-
GX bioreactor (BioTron Inc., Puchon, Korea) at a working vol- rate both substrate and product inhibitions on the growth of M.
ume of 2.5 L. The initial glucose concentrations in the medium succiniciproducens, additional terms were introduced into Eq.
was varied from 1.0 to 66 g L−1 . The pH was maintained at (1) as follows:
6.5 ± 0.1 by the addition of a 28% (v/v) ammonia solution.  i
The agitation speed and temperature were set to 200 rpm and μm S P
μ= × 1 − ,
39 ◦ C, respectively. The bioreactor was continuously sparged (S + KS + (S 2 /KI )) PCRIT
with industrial grade CO2 (Special gas, Daejeon, Korea) at a when P < PCRIT (2)
flow rate of 0.25 vvm during the whole period of fermentation.
An oxygen trap (Agilent, Waldbronn, Germany) was fitted in the and
outlet line of the gas cylinder to completely remove residual oxy- μ = −Kd , when P ≥ PCRIT (3)
gen, if any, from CO2 gas. All chemicals used in this study were
of reagent grade and were obtained from either Sigma–Aldrich where P is the product concentration (g L−1 ), P CRIT is the critical
(St. Louis, MO) or Junsei Chemical (Tokyo, Japan), except for product concentration at which cell growth fully stops (g L−1 ),
glucose and yeast extract which were purchased from Samyang and Kd is the specific death rate (h−1 ). Exponent i denotes the
Genex (Inchon, Korea) and Becton Dickinson Company (Le degree of product inhibition (dimensionless). M. succinicipro-
Pont de Claix, France), respectively. ducens produces succinic acid as a major fermentation product
and simultaneously forms acetic, formic, and lactic acids as by-
2.2. Analytical procedures products during anaerobic glucose fermentation. Accordingly,
P and PCRIT in Eq. (2) were given as the sum of the amounts of
Quantitative analyses of glucose and organic acids, includ- succinic and the other acids, rather than considering individual
ing acetic, formic, lactic, and succinic acids, were performed acids separately. Subsequently, the final kinetic equations for the
by a high-performance liquid chromatography (Varian ProStar batch growth of M. succiniciproducens could be represented by
210, Palo Alto, CA) equipped with UV/VIS (Varian Prostar the following equations:
  i 
320) and refractive index (Shodex RI-71, Tokyo, Japan) detec- μm S P
tors. A MetaCarb 87H column (300 mm × 7.8 mm, Varian) rX = × 1− X,
(S + KS + (S 2 /KI )) PCRIT
was used to separate the components at 60 ◦ C using 0.01N
H2 SO4 as a mobile phase at a flow rate of 0.6 mL min−1 . when P < PCRIT (4)
Cell growth was monitored by using a spectrophotometer
(Ultrospec3000, Amersham Biosciences, Uppsala, Sweden) at and
the optical density of 600 nm (OD600 ), and dry cell weight rX = −Kd X, when P ≥ PCRIT (5)
(DCW) was determined by the method described previously
[9]. where rX is the biomass formation rate (g L−1 h−1 ) and X is the
biomass concentration (g L−1 ).
2.3. Development of kinetic models The formation of organic acids, including succinic, acetic,
formic, and lactic acids, by M. succiniciproducens could be
The kinetics of fermentation can be normally described by a described by the Luedeking–Piret model [26], in which the
cell growth model (rX ), a product formation model (rP ), and a product formation rate depends on both rX and X:
substrate utilization model (rS ). In general, the Monod model rSA = αSA rX + βSA X (6)
is used to represent bacterial growth under substrate-limited
conditions. Taking into a consideration of the excess substrate rAA = αAA rX + βAA X (7)
inhibition on the growth of M. succiniciproducens, this well- rFA = αFA rX + βFA X (8)
known model can be modified as follows [17]:
rLA = αLA rX + βLA X (9)
μm S
μ= (1) where rSA , rAA , rFA , and rLA represent the formation rates of
(S + KS + (S 2 /KI ))
succinic, acetic, formic, and lactic acids (g L−1 h−1 ), respec-
where μ is the specific growth rate (h−1 ), μm is the maxi- tively. αSA , αAA , αFA , and αLA denote the growth-associated
mum specific growth rate (h−1 ), S is the substrate concentration parameters for the formation of succinic, acetic, formic, and lac-
(g L−1 ), KS is the substrate saturation constant (g L−1 ), and tic acids (dimensionless), respectively. βSA , βAA , βFA , and βLA
KI is the substrate inhibition constant (g L−1 ). However, Eq. denote the non-growth-associated parameters for the formation
(1) cannot be directly applicable to many fermentation systems of succinic, acetic, formic, and lactic acids (h−1 ), respectively.
since cell growth has a tendency to be inhibited gradually by A carbon substrate, glucose in this study, used by M. suc-
the fermentation products as cultivation proceeds [18–21]. In ciniciproducens is converted to form biomass, succinic acid,
 Author's personal copy

110 H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115

and by-products, and also used for the maintenance. Based on Table 1
the carbon mass balance, the consumption rate of glucose can Values of kinetic parameters
be described as follows: ␮m (h−1 ) 1.324
ms (h−1 ) 0.061
1 1 1 1 1
−rS = rX + rSA + rAA + rFA + rLA i (dimensionless) 1.301
YX YSA YAA YFA YLA YAA (g g−1 ) 0.999
+ mS X (10) YFA (g g−1 ) 1.532
YLA (g g−1 ) 0.999
where −rS is the consumption rate of glucose (g L−1 h−1 ).
YX , YSA (g g−1 ) 1.310
YX (g g−1 ) 0.765
YSA , YAA , YFA , and YLA are the stoichiometric yield coefficients
Kd (h−1 ) 0.010
of biomass, and succinic, acetic, formic, and lactic acids on KS (g L−1 ) 1.123
glucose (g g−1 ). mS is the specific maintenance coefficient (h−1 ). KI (g L−1 ) 88.35
PCRIT (g L−1 ) 17.23
2.4. Model parameters estimation αAA (g g−1 ) 0.626
αFA (g g−1 ) 0.665
αLA (g g−1 ) 0
With the experimental data obtained from a series of batch fer- αSA (g g−1 ) 1.619
mentations of M. succiniciproducens, the kinetic parameters of βAA (h−1 ) 0.124
the models were determined by minimizing an objective function βFA (h−1 ) 0.105
of Z: βLA (h−1 ) 0.210
βSA (h−1 ) 0.355
n
 n

ZX = (Xj,exp − Xj,sim )2 , ZP = (Pj,exp − Pj,sim )2 ,
j=1 j=1
was negatively influenced by the high concentration of glucose
n
 in the culture broth. It was found that the maximum specific
or ZS = (Sj,exp − Sj,sim )2 s (11)
growth rate (μm ) of M. succiniciproducens can be achieved is
j=1
1.324√h−1 at a glucose concentration of 9.964 g L−1 (given by
where ZX , ZP , and ZS represent the least-square discrepancies S = KS KI ) in a bioreactor. A higher specific growth rate is
between experimental data and simulated results of biomass, often desirable to grow cells faster to a desirable cell density
product, and substrate, respectively. j denotes the number of as it can reduce the fermentation time and thus improves over-
experimental data points. Subscripts ‘exp’ and ‘sim’ denote all productivity. In this sense, M. succiniciproducens showing a
experimental and simulated, respectively. high specific growth rate can be considered as a good succinic
acid producer.
3. Results and discussion The inhibitory effects of organic acids on microbial growth
have been known to depend on the pH of culture broth, the
3.1. Microbial growth dissociation constants of acids (pKa ), and their molar concen-
trations [27,28]. As mentioned earlier, M. succiniciproducens
In order to investigate the behaviors of M. succiniciproducens produces succinic acid as a major product and acetic, formic,
on the anaerobic fermentation of glucose, batch fermentations and lactic acids as minor ones from various carbon sources under
were carried out using the medium supplemented with varying anaerobic conditions. In this study, the pH was maintained at
concentrations (1, 3, 6, 10, 18, 28, 41, 49, and 66 g L−1 ) of glu- 6.5 during the whole period of fermentation. The dissociation
cose. First, μm and KS in Eq. (2) were estimated from the data
collected from the fermentations with low initial glucose con-
centrations (1, 3, and 6 g L−1 ). During these fermentations, no
glucose and product inhibitions were observed, implying that
S2 /KI and P/PCRIT are close to zero. Consequently, Eq. (2) was
reduced to the classical Monod equation μ = μm S/(S + KS ).
The values of KS and μm were obtained from reciprocal plot-
ting of the Monod equation; 1/μ versus 1/S (r2 = 0.991). Next,
KI was estimated from the fermentations with high initial glu-
cose concentrations (18, 28, 41, 49, and 66 g L−1 ) using the
data collected during the exponential growth phase (until 2.5 h
of inoculation) to avoid the inhibitory effect of products on the
cell growth. In this case, Eq. (2) was reduced to Eq. (1) because
P/PCRIT is close to zero. Reciprocal plotting of Eq. (1), 1/μ ver-
sus S provided KI (r2 = 0.981). Thus, determined values of KS ,
KI , and μm are listed in Table 1. Using these values, the model
predictions were compared with experimental data as shown Fig. 1. Effects of initial glucose concentrations on the specific growth rate of
in Fig. 1. As expected, the growth of M. succiniciproducens M. succiniciproducens MBEL55E. Experimental (symbol); simulated (line).
 Author's personal copy

H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115 111

Fig. 2. Time profiles of biomass (closed circle, solid line) and glucose (open circle, dashed line) in batch cultures of M. succiniciproducens MBEL55E with different
initial glucose concentrations: (a) 10 g L−1 , (b) 18 g L−1 , (c) 28 g L−1 , (d) 41 g L−1 , (e) 49 g L−1 , and (f) 66 g L−1 . Experimental (symbols); simulated (lines).

constants of organic acids produced by M. succiniciproducens
can be considered not much different for modeling purposes:
acetic acid (pKa of 4.75), formic acid (pKa of 3.84), lactic acid
(pKa of 3.86), and succinic acid (pKa1 of 4.21). Thus, the prod-
uct term in Eqs. (2)–(4) could be assumed to be as the sum
of the amounts of individual acids to simplify the models. As
shown in Figs. 2–4, the growth of M. succiniciproducens grad-
ually slowed down with increasing amounts of acids, and the
growth completely ceased when it exceeded 17.23 g L−1 (PCRIT )
in all experiments. Above this critical value, the concentrations
of biomass declined linearly with Kd = 0.001 h−1 as the fermen-
tation proceeded. The parameter i was estimated to be 1.301.
This is not surprising because the inhibition of microbial growth
by organic acids is common and the acids accumulated to high
concentrations finally cause cell lysis, even though the resistance
levels against acids can be different among microorganisms
[22,27,29,30].
3.2. Succinic acid production and by-products formation
The concentration of succinic acid increased proportionally
to increasing the concentration of biomass in the exponential
growth phase, and its production continued even during the
slow growth and stationary phases (Fig. 3). This means that
the production of succinic acid by M. succiniciproducens fol-
lows the Luedeking–Piret model, indicating that the production
rate of a product relies not only on the rate of cell growth (rX )
but also on the concentration of biomass (X) [26]. The non-
growth-associated coefficient for succinic acid production by
M. succiniciproducens (βSA = 0.355 h−1 ) was first determined
from the data collected where no cell growth was observed
(P ≥ PCRIT ). Then, the growth-associated coefficient for suc-
cinic acid production by M. succiniciproducens (αSA = 1.619)
was estimated when P < PCRIT . During the batch culture with
the initial cell concentration of 0.04 g DCW L−1 , the maxi-
mum volumetric productivity of succinic acid was predicted
to be 1.046 g L−1 h−1 at an initial glucose concentration of
18 g L−1 , which is in excellent agreement with the experimen-
tal result of 1.043 g L−1 h−1 (Fig. 3). The predictions of the
model simulation were compared with the experimental data
and are depicted in Fig. 3, showing a good agreement each
other.
In the same manner, the coefficients for the formation of
acetic, formic, and lactic acids by M. succiniciproducens in
 Author's personal copy

112 H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115

Fig. 3. Time profiles of succinic acid concentration in batch cultures of M. succiniciproducens MBEL55E with different initial glucose concentrations: (a) 10 g L−1 ,
(b) 18 g L−1 , (c) 28 g L−1 , (d) 41 g L−1 , (e) 49 g L−1 , and (f) 66 g L−1 . Experimental (symbol); simulated (line).

Eqs. (7)–(9) were estimated, respectively, and the values are
shown in Table 1. Like succinic acid, the formation of acetic
and formic acids by M. succiniciproducens followed a mixed-
growth-associated product formation pattern (Fig. 4). However,
unexpectedly, lactic acid started to be produced only when
the cell growth rate was nearly zero, which means that the
formation of lactic acid by M. succiniciproducens follows a non-
growth-associated product formation mode under the current
experimental condition. Lactic acid fermentation is presented
as a typical mixed-growth-associated product for most microor-
ganisms including Escherichia coli, Lactobacillus, Lactococcus,
and Saccharomyces [15,21,23,27,31]. Lactic acid formation and
succinic acid formation share one common physiological mean-
ing of NADH regeneration during the anaerobic fermentation;
NADH produced during glycolysis needs to be regenerated to
continue fermentation. However, there is a difference in that two
reducing power equivalents are used for succinic acid produc-
tion from phosphoenolpyruvate, while one equivalent is used
for lactic acid production from pyruvate. Production of suc-
cinic acid rather than lactic acid during the cell growth phase
of M. succiniciproducens implies that the pathway involved
in the production of succinic acid is much more favored in
M. succiniciproducens under the given fermentation condition;
anaerobic fermentation of glucose under CO2 atmosphere. M.
succiniciproducens operates branched TCA cycle; a reductive
branch leading to succinic acid formation from oxaloacetate and
an oxidative branch producing succinyl-CoA from citrate. When
the cell growth is slowed down or stopped, the oxidative branch
operation may be reduced or stopped causing less generation of
reducing powers. This may cause imbalance of reducing power
for the production of succinic acid alone, and cells convert some
pyruvate to lactic acid.
3.3. Substrate utilization
The consumption rate of glucose was calculated based on
the carbon mass balance using the stoichiometric yield coef-
ficients and the estimated specific maintenance coefficient
(ms = 0.061 h−1 ). The yield coefficients are shown in Table 1.
In order to calculate the carbon compound incorporated into
 Author's personal copy

H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115 113

Fig. 4. Time profiles of the concentrations of acetic (triangle, dotted line), formic (square, solid line), and lactic (diamond, dashed line) acids in batch cultures of
M. succiniciproducens MBEL55E with different initial glucose concentrations: (a) 10 g L−1 , (b) 18 g L−1 , (c) 28 g L−1 , (d) 41 g L−1 , (e) 49 g L−1 , and (f) 66 g L−1 .
Experimental (symbols); simulated (lines).

the cells from glucose, the four main elements of C, H, O,
and N of M. succiniciproducens cells at exponential growth
phase were quantified using an Elemental Analyzer (Flash
EA1112Series, CE instruments, Wigan, UK); it was found to be
CH1.736 O0.367 N0.240 . M. succiniciproducens fixes one mole of
CO2 to yield one mole of succinic acid via three different CO2 -
fixing metabolic reactions catalyzed by phosphoenolpyruvate
carboxykinase, phosphoenolpyruvate carboxylase, and malic
enzyme [9]. All CO2 used in the production of succinic acid was
assumed to be provided externally in the calculation. Together
with the experimental data, the predicted time profiles of glu-
cose are depicted in Fig. 2. Of particular interest from Fig. 2
is that the discrepancies between the experimental data and the
model simulation results were observed at the stationary growth
phase, and were slightly greater when high initial concentra-
tions (41, 49, and 66 g L−1 ) of glucose were used (Fig. 2(d)–(f)).
For the simplified model development, we did not incorporate
CO2 production by the cells, which occurs through the decar-
boxylation reactions catalyzed by oxaloacetate decarboxylase,
malic enzyme, pyruvate dehydrogenase, and formate dehydrase
in M. succiniciproducens. The loss of carbon in the form of CO2
produced by cells thus seems to contribute to this discrepancy.
Also, the released carbon in the form of cell debris as a result
of cell lysis was not included in the calculation. Additionally,
changes in cell composition [32–34] during the stationary phase
might have caused this discrepancy as we used one cell com-
position determined at exponential growth phase. Nonetheless,
the simple models developed in this study successfully describe
the cell growth, glucose consumption, and products formation,
which can thus be used for optimizing the process for succinic
acid production. For example, the optimal substrate (glucose)
concentration can be selected for the fed-batch and continuous
fermentation operations.
4. Conclusion
In this study, we developed the simple empirical kinetic mod-
els for the batch production of succinic acid from glucose by M.
 Author's personal copy
114 H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115
succiniciproducens MBEL55E. The results of the model simu-
lations showed very good agreement with the experimental data
obtained at varying initial glucose concentrations. The growth
of M. succiniciproducens could be expressed by an expanded
Monod model, including terms for both substrate and product
inhibitions, which were suggested by Andrews [17] and Leven-
spiel [25], respectively. The models allow identification of the
optimal concentrations of glucose and organic acids in a biore-
actor for the cell growth as well as for the production of succinic
acid. The production of succinic acid by M. succiniciproducens
follows the Luedeking–Piret model with the growth- and non-
growth-associated parameter values of 1.619 and 0.355 h−1 ,
respectively. The models could elucidate the fermentative char-
acteristics of M. succiniciproducens and also could explain its
superior ability to produce succinic acid under CO2 atmosphere.
The models proposed here should be useful for developing and
optimizing the succinic acid production processes from renew-
able resources.
Acknowledgement
This work was supported by the Genome-based Integrated
Bioprocess Project of the Ministry of Science and Technol-
ogy through the Korea Science and Engineering Foundation
(KOSEF). Further supports by the LG Chem Chair Professor-
ship and by the Center for Ultramicrochemical Process Systems
(KOSEF) are appreciated.
References
[1] H. Song, S.Y. Lee, Production of succinic acid by bacterial fermentation,
Enzyme Microbial. Technol. 39 (2006) 352–361.
[2] N.P. Nghiem, B.H. Davison, B.E. Suttle, G.R. Richardson, Production of
succinic acid by Anaerobiospirillum succiniciproducens, Appl. Biochem.
Biotechnol. 63/65 (1997) 565–576.
[3] N.S. Samuelov, R. Lamed, S. Lowe, J.G. Zeikus, Influence of CO2 –HCO3 -
level and pH on growth, succinate production, and enzyme activities
of Anaerobiospirillum succiniciproducens, Appl. Environ. Microbiol. 57
(1991) 3013–3019.
[4] S.E. Urbance, A.L. Pometto, A.A. DiSpirito, Y. Denli, Evaluation of
succinic acid continuous and repeat-batch biofilm fermentation by Acti-
nobacillus succinogenes using plastic composite support bioreactors, Appl.
Microbiol. Biotechnol. 65 (2004) 664–670.
[5] M.V. Guettler, D. Rumler, M.K. Jain, Actinobacillus succinogenes sp. nov.,
a novel succinic-acid-producing strain from the bovine rumen, Int. J. Syst.
Bacteriol. 49 (1999) 207–216.
[6] P.C. Lee, S.Y. Lee, S.H. Hong, H.N. Chang, Isolation and characterization
of new succinic acid producing bacterium, Mannheimia succiniciproducens
MBEL55E, from bovine rumen, Appl. Microbiol. Biotechnol. 58 (2002)
663–668.
[7] S.H. Hong, J.S. Kim, S.Y. Lee, Y.H. In, S.S. Choi, J.K. Rih, C.H. Kim, H.
Jeong, C.G. Hur, J.J. Kim, The genome sequence of the capnophilic rumen
bacterium Mannheimia succiniciproducens, Nature Biotech. 22 (2004)
1275–1281.
[8] T.Y. Kim, H.U. Kim, J.M. Park, H. Song, J.S. Kim, S.Y. Lee, Genome-
scale analysis of Mannheimia succiniciproducens metabolism, Biotechnol.
Bioeng. 97 (2007) 657–671.
[9] S.J. Lee, H. Song, S.Y. Lee, Genome-based metabolic engineering of
Mannheimia succiniciproducens for succinic acid production, Appl. Envi-
ron. Microbiol. 72 (2006) 1939–1948.
[10] D.Y. Kim, S.C. Yim, P.C. Lee, W.G. Lee, S.Y. Lee, H.N. Chang, Batch
and continuous fermentation of succinic acid from wood hydrolysate by
Mannheimia succiniciproducens MBEL55E, Enzyme Microbial Technol.
35 (2004) 648–653.
[11] P.C. Lee, S.Y. Lee, S.H. Hong, H.N. Chang, Batch and continuous cul-
tures of Mannheimia succiniciproducens MBEL55E for the production of
succinic acid from whey and corn steep liquor, BioProc. Biosyst. Eng. 26
(2003) 63–67.
[12] J. Luong, A. Mulchandani, A. LeDuy, Kinetics of biopolymer
synthesis: a revisit, Enzyme Microbial Technol. 10 (1988) 326–
332.
[13] F. Garcı́a-Ochoa, M. Garcı́a-León, A. Romero, Kinetics modeling of xan-
than production from sucrose, Chem. Biochem. Eng. 4 (1990) 15–20.
[14] M. Roehr, O. Zehentgruber, C.P. Kubice, Kinetics of biomass formation and
citric acid production by Aspergillus niger on pilot plant scale, Biotechnol.
Bioeng. 23 (1981) 2433–2445.
[15] C. Akerberg, K. Hofvendahl, G. Zacchi, B. Hahn-Hagerdal, Modeling the
influence of pH, temperature, glucose and lactic acid concentrations on
the kinetics of lactic acid production by Lactococcus lactis ssp. lactis
ATCC 19435 in whole-wheat flour, Appl. Microbiol. Biotechnol. 49 (1998)
682–690.
[16] N.S. Khan, I.M. Mishra, R.P. Singh, B. Prasad, Modeling the growth of
Corynebacterium glutamicum under product inhibition in l-glutamic acid
fermentation, Biochem. Eng. J. 25 (2005) 173–178.
[17] J.F. Andrews, A mathematical model for the continuous culture of microor-
ganisms utilizing inhibitory substrates, Biotechnol. Bioeng. 10 (1968)
707–723.
[18] X.M. Ge, F.W. Bai, Intrinsic kinetics of continuous growth and ethanol
production of a flocculating fusant yeast strain SPSC01, J. Biotechnol. 124
(2006) 363–372.
[19] S. Zafar, M. Owais, H.S. Saleemuddin, Batch kinetics and modeling of
ethanolic fermentation, Int. J. Food. Sci. Technol. 40 (2005) 597–604.
[20] A. Richard, A. Margaritis, Empirical modeling of batch fermentation kinet-
ics for poly(glutamic acid) production and other microbial biopolymers,
Biotechnol. Bioeng. 87 (2004) 502–515.
[21] M.-Y. Ha, S.-W. Kim, Y.-W. Lee, M.-J. Kim, S.-J. Kim, Kinetics analy-
sis of growth and lactic acid production in pH-controlled batch cultures
of Lactobacillus casei KH-1 using yeast extract/corn steep liquor/glucose
medium, J. Biosci. Bioeng. 96 (2003) 134–140.
[22] Q. Kong, G.Q. He, F. Chen, H. Ruan, Studies on a kinetic model for butyric
acid bioproduction by Clostridium butyricum, Lett. Appl. Microbiol. 43
(2006) 71–77.
[23] C.B. Youssef, G. Goma, A. Olmos-Dichara, Kinetic modeling of Lacto-
bacillus casei ssp. rhamnosus growth and lactic acid production in batch
cultures under various medium conditions, Biotechnol. Lett. 27 (2005)
1785–1789.
[24] W.J. van Niel, P.A.M. Claassen, A.J.M. Stams (Eds.), Substrate and
product inhibition of hydrogen production by the extreme thermophile,
Caldicellulosiruptor saccharolyticus, Biotechnol. Bioeng. 81 (2003) 255–
262.
[25] O. Levenspiel, The monod equation: a revisit and a generalization to product
inhibit situations, Biotechnol. Bioeng. 22 (1980) 671–687.
[26] R. Luedeking, E.L. Piret, Kinetic study of the lactic acid fermentation:
batch process at controlled pH, J. Biol. Microbiol. Technol. Eng. 1 (1959)
393–424.
[27] N.V. Narendranath, K.C. Thomas, W.M. Ingledew, Acetic acid and lac-
tic acid inhibition of growth of Saccharomyces cerevisiae by different
mechanisms, J. Am. Soc. Brew. Chem. 59 (2001) 187–194.
[28] C. Vasseur, L. Baverel, M. Hébraud, J. Labadie, Effect of osmotic, alka-
line, acid or thermal stresses on the growth and inhibition of Listeria
monocytogenes, J. Appl. Microbiol. 86 (1999) 469–476.
[29] J. Lin, I.S. Lee, J. Frey, J.L. Slonczewski, J.W. Foster, Comparative analysis
of extreme acid survival in Salmonella typhimurium, Shigella flexneri, and
Escherichia coli, J. Bacteriol. 177 (1995) 4097–4104.
[30] Y.S. Park, K. Toda, M. Fukaya, H. Okumura, Y. Kawamura, Production of
a high concentration acetic acid by Acetobacter aceti using a repeated fed-
batch culture with cell recycling, Appl. Microbiol. Biotechnol. 35 (1991)
149–153.
[31] M.L. Shuler, F. Kargi, Bioprocess Engineering: Basic Concepts, 2nd ed.,
Prentice-Hall Inc., New Jersey, USA, 1992.
 Author's personal copy

H. Song et al. / Biochemical Engineering Journal 40 (2008) 107–115 115

[32] M.P. Nandakumar, A. Cheung, M.R. Marten, Proteomic analysis of extra-
cellular proteins from Escherichia coli W3110, J. Prot. Res. 5 (2006)
1155–1161.
[33] J. Pramanik, J.D. Keasling, Stoichiometric model of Escherichia coli
metabolism: incorporation of growth-rate dependent biomass composi-
tion and mechanistic energy requirements, Biotechnol. Bioeng. 56 (1997)
398–461.
[34] E. Cabacungan, R.A. Pieringer, Excretion of extracellular lipids by
Streptococcus mutants BHT and FA-1, Infect. Immun. 27 (1980) 556–
562.