water research 44 (2010) 2554–2562

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Kinetic model of thermophilic L-lactate fermentation by

Bacillus coagulans combined with real-time PCR quantification

T. Hidaka a,*, T. Horie a, S. Akao b, H. Tsuno a

a
Department of Urban and Environmental Engineering, Kyoto University, Kyoto-Daigaku-Katsura, Nishikyo-ku, Kyoto 615-8540, Japan
b
Department of Social Systems Engineering, Tottori University, 4-101, Koyama-Minami, Tottori 680-8552, Japan

article info abstract

Article history: A simple L-lactate fermentation of organic wastes at pH 5.5 and 55  C under nonsterile
Received 27 August 2009 conditions using Bacillus coagulans can be suitable for L-lactate fermentation of garbage. A
Received in revised form mathematical model that simulated the lactate fermentation characteristics of B. coagulans
16 November 2009 was developed by focusing on the inhibitory effects of substrate, lactate (product) and
Accepted 11 January 2010 NaCl, and bacterial growth. Basic fermentation experiments were performed using simple
Available online 18 January 2010 substrates to derive fundamental parameters of growth rate and inhibition effects. The
model was then applied to fermentations using simple substrates and artificial kitchen
Keywords: garbage in order to verify its applicability. Microbial concentration, a key state variable of
Lactate fermentation the model was measured using both real-time polymerase chain reaction (PCR) and
Mathematical model traditional methods. The results of these methods were compared for experimental cases
Inhibition in which only soluble substrates were used. B. coagulans concentrations were suitably
Real-time PCR measured using real-time PCR, even when traditional measurement methods for microbial
Bacillus coagulans concentrations cannot be used. The results indicate that the developed model and biomass
Kitchen garbage measurement can be used to evaluate lactate fermentations using both simple and
complex substrates. These proposed methods would be useful for developing a new
bacterial function-based mathematical model for more complex acid fermentations.
ª 2010 Elsevier Ltd. All rights reserved.

1. Introduction production in order to maintain high optical purity. Conse-

quently, sterilization is necessary for these processes, which
Lactate is currently used in food and pharmaceutical indus- increases the cost. Corn, potato, and other starchy plants have
tries. Furthermore, it has gained attention as a raw material been used as substrates for lactate fermentation (Hofvendahl
for polylactide (PLA), a well-known biodegradable polymer. and Hahn-Hagerdal, 2000), although farming of these plants is
The physical properties and biodegradability of PLA are inconsistent because of the potential food crisis occurring
influenced by its composition ratio of D- and L-lactate isomers. worldwide. Consequently, utilization of organic wastes as an
Therefore, optically pure lactate is required in order to alternative to agricultural products is essential, and they have
produce PLA, which is primarily achieved by fermentation been reported as substrates for lactate fermentation (Hof-
rather than chemical synthesis (Hofvendahl and Hahn- vendahl and Hahn-Hagerdal, 2000). Garbage, a type of organic
Hagerdal, 2000). Various bacteria are known to produce either waste has been proposed as a substrate for lactate fermenta-
of the lactate isomers in pure culture; therefore, contamina- tion because of its high carbohydrate and nutrient content
tion by other bacteria must be avoided during D- or L-lactate (Sakai et al., 2000).

* Corresponding author. Tel.: þ81 75 383 3350; fax: þ81 75 383 3351
E-mail address: [email protected] (T. Hidaka).
0043-1354/$ – see front matter ª 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2010.01.007
 water research 44 (2010) 2554–2562 2555

A simple L-lactate fermentation of organic wastes under concentrations were measured using real-time PCR as well as
nonsterile conditions using Bacillus coagulans has been traditional methods for the experimental cases where only
developed (Rosenberg et al., 2005; Michelson et al., 2006; soluble substrates were used. These results were compared to
Sakai and Ezaki, 2006; Akao et al., 2007a,b). Akao et al. (2007b) check the applicability of real-time PCR method. The devel-
reported that operational conditions of pH 5.5 and 55  C were oped model was applied to fermentation experiments using
suitable for L-lactate fermentation of garbage by indigenous glucose and artificial kitchen garbage, and its applicability was
B. coagulans, while other bacteria were eliminated under verified by comparing lactate production and bacterial
these culture conditions. Mathematical models are useful growth. Furthermore, the applicability of real-time PCR to
means to improve the understanding of this particular quantify B. coagulans in the fermentation of kitchen garbage,
process, predict process performance over a range of design where traditional methods cannot be used, was also
and operating conditions, and optimize performance. The confirmed in this model.
International Water Association Anaerobic Digestion Model
No. 1 (ADM 1, Batstone et al., 2002) is widely used to evaluate
anaerobic digestion; however, it does not include lactate 2. Model development
fermentation.
Lactate fermentation kinetics including the inhibitory A simple biochemical transformation model for lactate
effects of substrates and products, maximum product fermentation was developed. State variables and trans-
concentration, and growth of associated bacteria have been formation paths are shown in Fig. 1. The state variables in this
separately evaluated (Ishizaki et al., 1989; Venkatesh et al., model are particulate carbohydrate, soluble carbohydrate
1993; Boonmee et al., 2003; Lin et al., 2004; Nandasana and other than glucose, glucose, L-lactate and B. coagulans (gCOD/L).
Kumar, 2008). Furthermore, NaCl inhibition is also possibly Rate equations for each path are given in Table 1. Trans-
observed in lactate fermentation (Kargi and Dincer, 1999) since formation paths considered in this model are as follows:
kitchen garbage probably contains high concentrations of degradation (solubilization) of particulate carbohydrate (R1),
NaCl. However, mathematical expressions for these effects are degradation (hydrolysis) of soluble carbohydrate (R2), lactate
different depending upon the research area, and there is no fermentation accompanied by growth of B. coagulans (R3), and
model that comprehensively incorporates all of these effects, self-degradation of B. coagulans (R4). Each reaction rate is
which are essential in order to evaluate lactate fermentation expressed as the product of rate constants, which affects the
characteristics and its limits. In addition, material balance of functions of state variables and bacterial concentrations
total Chemical Oxidation Demand (COD) was not considered associated with each path.
like ADM 1 (Batstone et al., 2002). Furthermore, isolated species Degradation of particulate carbohydrate (R1) is usually
were inoculated in these experiments; however, the detailed expressed by first order kinetics or Contois kinetics (Contois,
growth characteristics of B. coagulans under nonsterile 1959). In this study, the microbial concentration at the
conditions have not been designed as yet. beginning of batch experiments was very low, and after the
In evaluations using mathematical models, data for start-up period degradation of carbohydrate was accelerated
microbial concentrations are also important. They have been due to growth of microorganisms. Therefore, degradation of
estimated by measurements of Suspended Solids (SS) particulate carbohydrate was expressed by Contois kinetics.
(Michelson et al., 2006), COD, optical density (Payot et al., 1999; Degradation of soluble carbohydrate (R2) is expressed by
Ishizaki et al., 1989), and protein and total DNA mass (Nishi- a Michaelis–Menten type equation associated with soluble
mura et al., 1996). However, these indices are not appropriate carbohydrate concentration.
for understanding microbial concentrations for those Lactate fermentation from glucose accompanied by the
substrates containing organic particulate materials such as growth of B. coagulans (R3) is known to be inhibited by
garbage and sewage sludge. Recently, quantification of substrate (glucose), product (lactate), and NaCl. Lactate inhi-
microbes was done using quantitative real-time polymerase bition has been reported as either uncompetitive (Ishizaki and
chain reaction (PCR) with specific primer sets for quantifying Ohta, 1989; Ishizaki et al., 1989), noncompetitive (Ohara et al.,
specific microbes. The applicability of real-time PCR quanti-
fication to evaluate methanogenic behavior has been reported
(Yu et al., 2006; Kobayashi et al., 2009), but its application to Particulate carbohydrate
acid fermenting bacteria, including B. coagulans, in lactate (CC, gCOD/L)
fermentation has not been reported. Developing a method to R1
measure bacterial concentrations will be useful in evaluating
acid fermentation mechanisms. Soluble carbohydrate (S C, gCOD/L)
In this study, a mathematical model was developed by R2
NaCl (S N, gNaCl/L)
focusing on the inhibitory effects of substrate, lactate
Glucose (S G, gCOD/L)
(product) and, NaCl and bacterial growth in order to evaluate Inhibition
the lactate fermentation characteristics of B. coagulans under R3
nonsterile conditions. Basic fermentation experiments were (1– Y ) (Y )
performed using glucose, the simplest substrate, in order to L-Lactate (SL, gCOD/L) B. coagulans (X, gCOD/L)
derive fundamental parameters, including growth rate and R4
inhibition effects. B. coagulans was the predominant species in
cultures at pH 5.5 and 55  C (Akao et al., 2007a,b). Microbial Fig. 1 – State variables and transformation paths.
2556 water research 44 (2010) 2554–2562

NaCl inhibition was also designed as noncompetitive (Kargi

Table 1 – Reaction rate and mass balance for model
development. and Dincer, 1999).
The self-degradation expression was given by a first order
Reaction rate
reaction similar to ADM1 (Batstone et al., 2002). Changing
Degradation of particulate carbohydrate (gCOD/(L h)): rates of state variables are expressed as the sum of the rates of
CC =X paths into the state variable (þ) and paths out of the state
R1 ¼ k1 $ $X
KCC þ CC =X
SC variable (), as shown in Table 1.
Degradation of soluble carbohydrate (gCOD/(L h)): R2 ¼ k2 $ $X
KSC þSC
Lactate fermentation (gCOD/(L h)):
!
SG KEI KEIN maxð0; SL SLct Þ
R3 ¼ k3 $ , $
KSG þSG $ð1þ KSG Þ KEI þSL KEIN þSN
$ 1
SLmax SLct
$X 3. Materials and methods
ES

Self-degradation (gCOD/(L h)): R4 ¼ k4 $X

3.1. Reactor and operation
Mass balance

dCC Fermentation experiments were performed in a reactor made

¼ R1
dt of a 1 L working volume glass flask in which anaerobic
dSC conditions and complete mixing were maintained. Opera-
¼ R1  R2
dt tional conditions are summarized in Table 2. Seven batch
dSG experiments (Run 1–Run 7) and one semicontinuous experi-
¼ R2  R3
dt
ment (Run 8) were performed. Culture temperature was set at
dSL
¼ ð1  YÞ$R3 55  C using a hot water bath and pH was automatically
dt
controlled at 5.5 by the addition of 10 N NH4OH in Run 1–Run 7
dX
¼ Y$R3  R4 or 10 N NaOH in Run 8 (Nisshinrika, NPH-660). Headspace of
dt
the reactor was replaced with nitrogen gas before each
k1, Rate constant for R1(gCOD/(gCOD h)); k2, rate constant
for R2(gCOD/(gCOD h)); k3, rate constant for R3 (gCOD/(gCOD h));
fermentation experiment. Sterilization (121  C, 20 min) of the
k4, rate constant for R4 (1/h); KCC , half saturation constant for R1 reactor was carried out before the experiments in order to
(gCOD/L); KSC , half saturation constant for R2 (gCOD/L); KSG , half avoid contamination, but the feedstock was not sterilized.
saturation constant for R3 (gCOD/L); KES , inhibition parameter of Run 1 was performed twice to check reproducibility (Runs 1a
glucose for R3 (gCOD/L); KEI , inhibition parameter of lactate for R3 and 1b). In Run 1–Run6, glucose was used at different
(gCOD/L); KEIN , inhibition parameter of NaCl for R3 (gNaCl/L); SLct ,
concentrations to determine its effect. Lactate and NaCl were
lactate concentration when inhibition of lactate starts (gCOD/L);
also added to determine their effects in Run 2 and Run 5, and
SLmax , maximum lactate concentration (gCOD/L); Y, yield constant
for R3 (gCOD/gCOD). Run 3, respectively. In Run 7, artificial kitchen garbage was
used which was prepared by mixing 14 types of foods, based
on a survey conducted in Japan (Tanigawa et al., 1997). The
kitchen garbage comprised 10% cabbage, 10% potato, 10%
1992), or exponential (Nandasana and Kumar, 2008). In this carrot, 10% radish, 10% Chinese cabbage, 2.5% apple, 7.5%
study, noncompetitive inhibition was adopted and its appli- orange peel, 10% banana peel, 10% boiled rice, 2.5% bread,
cability was checked using the experimental results. 7.5% noodles, 2.5% boiled eggs, 2.5% meat, and 5% fish.
Furthermore, to reflect the fact that lactate fermentation is Materials were mixed, milled into a slurry and stored in
stopped at a certain lactate concentration (Akao et al., 2007b), a freezer until use. This garbage was diluted twofold with
the maximum limit concentration of produced lactate (SLmax) distilled water and used for L-lactate fermentation experi-
and threshold lactate concentration above which the inhibi- ments. In Run 8, once every 48 h, 400 mL of mixed liquor was
tion effect by accumulation of product started (SLct) were also withdrawn and an equal volume of substrate was added in
included in the model equation for R3. This model was similar order to maintain semicontinuous operation for 384 h. The
to the model by Boonmee et al. (2003). Lactate decreases the substrate was a mixture of raw garbage which was diluted
pH of the medium while it is being produced; however, in this twofold with distilled water, and banana peels which was
study pH was maintained at a constant value of 5.5. Thus, diluted three-fold with distilled water. The characteristics of
dissociated and undissociated lactates were not considered this raw garbage and banana peels are summarized in Table 3.
separately, and the total lactate concentration was used. An inoculum was prepared by preliminary culture at pH 5.5
Inhibition by glucose was expressed as Haldane type inhibi- and 55  C for two days with glucose (20 g/L), yeast extract
tion (Haldane, 1930) since this is a type of substrate inhibition. (2.5 g/L), (NH4)2HPO4 (0.25 g/L), MgSO4$7H2O (0.05 g/L),

Table 2 – Experimental condition.

Run 1a 1b 2 3 4 5 6 7 8

Glucose (g/L) 10 10 10 10 50 50 100 0 0

Lactate (g/L) – – 20 – – 6 – – –
NaCl (g/L) – – – 20 – – – – –
Kitchen garbage (L/L) – – – – – – – 0.5 0.25
Banana peel (L/L) – – – – – – – – 0.17
 water research 44 (2010) 2554–2562 2557

in each real-time PCR reaction for each primer pair. Amplifi-

Table 3 – Characteristics of kitchen garbage.
cation of the target RNA gene used the following conditions:
Kitchen garbage Banana peels initial 10 min incubation at 95  C for Taq polymerase activa-
Total Soluble Total Soluble tion; 30 cycles of denaturation at 95  C for 10 s, annealing at
the optimum temperature of each key microbe for 10 s, and
TS (g/L) 190 – 121 –
extension at 72  C for 30 s. The transition rate was 20  C/s for
SS (g/L) 86 – 65 –
VTS (%) 96 – 88 – all segments during cycling. Cycle threshold (Ct) was deter-
COD (g/L) 201 114 79 42 mined by fit point method using LightCycler software (version
Carbohydrate 115 96 44 33 3.5, Roche). Standard 16S rRNA gene concentrations were
(g/L as glucose) measured using a spectrophotometer (NanoDrop ND-1000,
Protein (g/L as albumin) 20.3 3.7 4.5 1.6 Wilmington, DE, USA). Each 16S rRNA gene concentration was
T–N (mgN/L) 3.4 1.2 1.4 1.0
determined from the Ct obtained from regression equations of
pH (–) 5.1 – 5.5 –
the external standards.

MnSO4$4H2O (0.01 g/L), FeSO4$7H2O (0.01 g/L), L-lactate (10 g/L)

and fermented kitchen garbage (Akao et al., 2007b). This 4. Results and discussion
inoculum (10 mL) was added at the beginning of each
experiment. 4.1. Fermentation performances

3.2. Chemical analysis The inoculum, 10 mL of which was inoculated into 1 L

fermenters, contained L-lactate at 28.1 g/L with an OP of nearly
Aliquots of the fermented broth samples were collected peri- 100%. The microbial community was analyzed using random
odically during each batch experiment and analyzed. Total cloning method (Cheon et al., 2008) with a primer set of
lactate and other organic acids were analyzed using High UNIV519F (Lane, 1991) and UNIV1406R (Lane, 1991) to amplify
Performance Liquid Chromatography (HPLC) (Organic acids approximately 900 and 700 bp fragments for most bacteria
analysis system, column; Shima-pack SCR-102H, Shimadzu, and archaea, respectively. Ninety-two clones were analyzed;
Kyoto, Japan). D- and L-lactate were separately analyzed using 100% were closely related to B. coagulans with an average
HPLC (Shimadzu, column; SUMICHIRAL OA-5000 (40  C), similarity of 99%. This indicated that B. coagulans was
detector; UV detection (254 nm), mobile phase; 5% of 2-prop- predominant in the inoculum. The RNA gene concentration
anol and 1 mM CuSO4 solution, flow rate; 0.8 mL/min). Optical was 8.8  109 copies/L.
purity of lactate (OP) was calculated using the following Fig. 2 shows the time course for the results in Run 1a. At 4 h
equation of fermentation, RNA gene and lactate concentrations began
  to rapidly increase, while glucose concentration began to
½L-Lactate  ½D-Lactate
OPð%Þ ¼   :100 (1) decrease. The increase in RNA gene concentration stopped at
½L-Lactate þ ½D-Lactate
7 h and was maintained at 1  1011 copies/L, whereas lactate
Carbohydrate concentration, expressed using glucose continued to increase until 10 h when glucose was depleted.
concentration as standard, was determined using a phenol– Lactate was produced at 8.6 g/L with an OP of nearly 100%.
sulfate method (DuBois et al., 1956). Lactate production ratio A summary of each batch experiment is shown in Table 4.
from glucose (1  Y ) was expressed as a conversion ratio of In Run 1a and Run 1b, nearly the same results were obtained,
initial carbohydrate (g) into produced lactate (g). Carbon which demonstrated the reproducibility of the experiments.
dioxide and methane were analyzed using CGT-7000 (Shi- When 20 g/L of lactate was added at the beginning, the
madzu). COD was analyzed according to the Standard Method fermentation start time was delayed, as shown by comparing
(APHA, 1995). Dissolved samples were prepared by filtration Run 1 and Run 2. Run 1 and Run 4 had an initial glucose
through a cellulose acetate filter (pore size ¼ 0.45 mm). concentration of 10 g/L and 50 g/L, respectively, and fermen-
tation start times and lactate production ratios from glucose
3.3. Molecular biological analysis were nearly the same in both Runs. In Run 6, although the
initial glucose concentration was 100 g/L, only 52 g/L of lactate
DNA was extracted using a DNeasy Tissue Kit (Qiagen, Hilden, was produced, and some glucose remained. The fermentation
Germany). For real-time PCR reactions, LightCycler 1.2 (Roche, start time was also delayed compared to Run 4. This
Mannheim, Germany) was used to quantify the target maximum lactate concentration was similar to results of
microbes and primer pair of BACO186F (50 -gcatggaggaaaaag- Payot et al. (1999), who reported lactate production by
gaa-30 ) and BACO447R (50 -cccggcaacagagtttta-30 ) was used B. coagulans. In Run 3, NaCl at 20 g/L was added at the begin-
(Asahira, 2006). All reactions used 20 mL reaction capillary ning and fermentation start time was delayed, as compared
tubes with the LightCycler FastStart DNA MasterPlus SYBR with Run 1, although the final lactate production ratio was the
Green I (Roche). Each capillary tube was separately loaded same as Run 1. The OP was always > 98.3%. In Run 7 which
with 1 mL of sample RNA gene, followed by addition of 1 mL used kitchen garbage, particulate carbohydrate began to
(final concentration of 0.5 mM) of forward and reverse primers decrease and simultaneously lactate began to increase at 12 h.
along with 2 mL of SYBR Green I Master mix, and PCR-grade RNA gene concentration stopped increasing at 18 h and was
sterile water to a final volume of 10 mL. A negative control maintained at 1  1011 copies/L. The initial carbohydrate
without the corresponding template RNA gene was included concentration was similar to Run 4, but fermentation started
2558 water research 44 (2010) 2554–2562

100

OP (%)
50

0
0 5 10 15

Concentrations (g/L) 12
10
8
6
Glucose
4 Lactate
2
0
0 5 10 15

1.E+12
B. coagulans (copies/L)

1.E+11

1.E+10

1.E+09

1.E+08

1.E+07
0 5 10 15
Time (hr.)

Fig. 2 – Results of Run 1a.

6 h later and ended 60 h later than in Run 4. Finally, around production ratio from carbohydrate (1  Y) was determined to
32 g/L of lactate was produced having an OP of 98.5%. About be 0.88 (gCOD/gCOD) from the average value of Run 1, Run 4,
30% of the carbohydrate in the substrate remained after and Run 5 in which most of the glucose was converted to
fermentation. This difference was because the substrate lactate without initial inhibition. Some parameters were
contained nonbiodegradable organics such as cellulose. determined using Lineweaver–Burk plots (Fig. 3). Here,
However, the lactate production ratio from carbohydrate was increased bacterial concentration between two samplings (ln)
the same as that reported by Akao et al. (2007b). In Run 8, divided by the time was used as the specific growth rate (m),
semicontinuous operation was successfully maintained. RNA and the decrease in glucose concentration between two
gene concentration was maintained at 1  1011 copies/L, and samplings divided by the time was used as the glucose
around 22 g/L of lactate was produced having an OP of 99.3%. consumption rate (V), when the data measurement was
successful. The values of the rate constant (k3) and half satu-
4.2. Model parameters ration constant (KSG) for lactate production were determined
by assuming that there was no inhibition by lactate when
The parameters used in this model are summarized in Table 5. lactate concentration (SL) was 0–10 gCOD/L (Fig. 3a). The
All were determined from the experimental data. The lactate maximum specific growth rate was determined to be 1.54 (1/h),

Table 4 – Summary of each batch experiment.

Run 1a 1b 2 3 4 5 6 7 8

Time when 6 6 20 11 6 7 9 12 –
fermentation started (h)
Lactate concentration 8.6 9.3 29 9.3 45 48 52 35 21.5  0.4
at the end (g/L)
Lactate production 0.86 0.93 0.74 0.89 0.90 0.84 0.52 0.54 0.59
ratio from carbohydrate (–)
OP (%) 100 100 100 100 100 100 98.3 98.5 99.3
 water research 44 (2010) 2554–2562 2559

inhibitory effect. Two theoretical lines for SL ¼ 0–10 and

Table 5 – Model parameters.
SL ¼ 30 gCOD/L were matched on the X axis, which indicated
Parameter Value noncompetitive inhibition. Then, the inhibition parameter of
k1 (gCOD/(gCOD h)) 0.1 lactate for lactate fermentation (KEI) was determined to be 5.11
k2 (gCOD/(gCOD h)) 80 gCOD/L. This value was similar to that of 5.9–12 gCOD/L
k3 (gCOD/(gCOD h)) 12.8 reported by Ohara et al. (1992). When lactate concentrations
k4 (1/h) 0.01 were >40 gCOD/L, the actual values of V1 were higher than
KCC (gCOD/L) 1.0
the calculated values using the above obtained parameter
KSC (gCOD/L) 1.0
values and the equation for a noncompetitive inhibition
KSG (gCOD/L) 3.1
KES (gCOD/L) 180 model (Fig. 3c). This indicated that when lactate concentra-
KEI (gCOD/L) 5.11 tions were >40 gCOD/L, lactate production rates were further
KEIN (gNaCl/L) 30 inhibited. When they were >56 gCOD/L, lactate production
SLct (gCOD/L) 40 stopped in Run 6, although some glucose remained. Therefore,
SLmax (gCOD/L) 56 in this model, two parameters for lactate concentration were
Y (gCOD/gCOD) 0.12
included: SLct and SLmax (determined as 40 gCOD/L and
56 gCOD/L, respectively). These values were similar to those of
which was divided by Y (¼ 0.12) to calculate the value of k3. The Boonmee et al. (2003) for Lactococcus lactis NZ133 at 30  C.
values of k3 and KSG were determined to be 12.8 (gCOD/gCOD h) Inhibition of glucose was expressed as Haldane type inhi-
and 3.1 (gCOD/L), respectively. Reported values of these bition (Haldane, 1930). The maximum glucose consumption
parameters differ depending on fermentation conditions, but rate when glucose concentrations were in the range of 13–
the values determined here were within the range of previ- 62 gCOD/L in Run 4–Run 6 was obtained when the glucose
ously reported values (Payot et al., 1999; Burgos-Rubio et al., concentration was 23.2 gCOD/L (SGmax). The value of the
2000; Skiadas et al., 2000; Nandasana and Kumar, 2008). inhibition parameter (KES) was determined to be 180 gCOD/L,
Inhibition of lactate fermentation by produced lactate was using the equation SGmax ¼ (KSG$KES)1/2. This value was similar
also evaluated using Lineweaver–Burk plot (Fig. 3b). Here, to the value (127 gCOD/L) reported by Burgos-Rubio et al.
when both glucose and lactate concentrations (SG and SL, (2000). The value of the parameter for inhibition by NaCl (KEIN)
respectively) were less than 10 gCOD/L, it was assumed that was determined to be 30 g NaCl/L from the data in Run 3,
there was no inhibition; however, when lactate concentra- which was similar to the value (70 g) proposed by Kargi and
tions were between 28 and 32 gCOD/L there was a constant Dincer (1999).

a 2

1.5
SL ≥ 50
SL=41~48
1 SL=28~32
SL=14~24
1/μ [h r]

SL=0~10
0.5

0
-0.5 0.0 0.5 1.0
1/SG [L/gCOD]

b 4 c 160
120
3 SL=30
1/V [L·hr/gCOD]

80

2 SL=50
1/V [L·hr/gCOD]

4
SL=30
1
2
SL=0
SL=0
0 0
-0.5 0.0 0.5 1.0 -0.5 0.0 0.5 1.0
1/SG [L/gCOD] 1/SG [L/gCOD]

Fig. 3 – Lineweaver–Burk plot for accumulated lactate inhibition depending on glucose concentrations. m is specific growth
rate (1/h), V is measured glucose consumption rate (gCOD/(L h)), SG and SL are concentrations of glucose and lactate,
respectively. Solid lines are calculated by the noncompetitive inhibition model equation when SG is 0, 30, or 50 (gCOD/L).
2560 water research 44 (2010) 2554–2562

acceptable in consideration of ADM1 (Batstone et al., 2002) or

1.E+11 other reports, such as Biazar et al. (2003), who reported a batch
production of lactate under submerged fermentation of
B. coagulans (copies/L)

cheese whey.
The relationship between increased biomass and real-time
1.E+10
PCR analysis is shown in Fig. 4. Increased biomass was
calculated by multiplying the decreases in glucose concen-
Run 1a trations and yield coefficients of B. coagulans (0.12 gCOD/
1.E+09 Run 2 gCOD-removed). Increases in these two parameters were
Run 3 highly correlated, although their ratios were not constant
since they depended on growth phases. This tendency was
similar to a relationship between axenic Microcystis aeruginosa
1.E+08
NIES 102 cell concentrations measured using microscopic
0.01 0.1 1 10
observations and 16S rRNA gene concentrations measured
Increased biomass (gCOD/L)
using real-time PCR in laboratory cultures (Ha et al., 2009). The
Fig. 4 – Relationship between increased biomass and real- average value of this ratio was 6.47  1010 copies/gCOD-
time PCR analysis. biomass. To simply apply the real-time PCR analysis results,
this ratio was used to calculate biomass concentrations in
COD units from the results of real-time PCR.
Values of rate constants for degradation of particulate
carbohydrate (k1) and soluble carbohydrate (k2) were deter- 4.3. Model verification
mined to be 0.1 gCOD/(gCOD h) and 80 gCOD/(gCOD h) from
the data in Run 7. The rate constant for self-degradation of B. Time courses for the experimental and calculated results in
coagulans (k4) was determined to be 0.01 (1/h) from the data in Run 1–Run 6 are shown in Fig. 5. In each Run, good agreements
Run 6 with more than 90 h operation. These values are for glucose consumption and lactate fermentation were
Conc. (gCOD/L)

15 Run 1a
Conc. (gCOD/L)

1
10 0.1 Glucose Y axis, Left
Lactate (gCOD/L)
5 0.01 Y axis, Right
B. coagulans
(gCOD/L)
0 0.001
0 5 10 15
15 Run 1b
Conc. (gCOD/L)

1 40 Run 2
Conc. (gCOD/L)
1
10 30
0.1
20 0.1
5 0.01
10 0.01
0 0.001
0 0.001
0 5 10 15
0 10 20 30 40

15 Run 3 10 60 10
Conc. (gCOD/L)

Conc. (gCOD/L)

1 1
10 40
0.1 Run 4 0.1
5 20
0.01 0.01
0 0.001 0 0.001
0 5 10 15 20 0 10 20 30 40 50

120 10
Conc. (gCOD/L)

60
Conc. (gCOD/L)

50 1 1
40 80 Run 6
30
Run 5 0.1 0.1
20 40
0.01 0.01
10
0 0.001 0 0.001
0 10 20 30 40 50 0 20 40 60 80 100
Time (hr.) Time (hr.)

Fig. 5 – Time courses of the experimental and calculated results in Run 1–Run 6. Plots are experimental data and lines are
simulated results.
 water research 44 (2010) 2554–2562 2561

2009). Changes in biomass concentrations were also measured

Total carbohydrate Y axis, Left and simulated during the entire growth phase. In Run 8,
Lactate (gCOD/L) decrease in carbohydrate concentrations, increase in lactate
Y axis, Right concentrations, and maintained biomass concentrations were
B. coagulans (gCOD/L)
favorably simulated. This showed that this model can be
applied to not only batch experiments, but also to continuous
80 Run 7 10
experiments.
The lag phase for the inoculated bacteria was not consid-
Conc. (gCOD/L)

Conc. (gCOD/L)
60 1
ered, and the measured RNA gene concentrations of the inoc-
ulum were added as initial bacterial concentrations. In some
40 0.1
cases, calculated bacterial concentrations increased earlier
20 0.01 than the experimental results, but glucose consumption,
lactate production, and bacterial growth were simultaneously
0 0.001 calculated throughout the full range of the Runs. These results
0 50 100 150 200 indicated that this model and biomass measurements using
real-time PCR can be used to simulate lactate fermentation
80 Run 8 10 using both simple and complex substrates.
This thermophilic lactate fermentation under nonsterile
Conc. (gCOD/L)

60 1 Conc. (gCOD/L)
conditions was a good example to verify the applicability of
the proposed methods since only B. coagulans was responsible
40 0.1 for the reactions, and time courses for substrate, product, and
bacterial concentrations were measured. Acid fermentation is
20 0.01 more complex than lactate and methane fermentations since
it includes many kinds of products such as lactate, butyrate,
0 0.001 and acetate, and many types of bacterial communities
0 100 200 300 400 500 depending on the operational and bacterial conditions (Lee
Time (hr.) et al., 2008; Park et al., 2008). ADM1 (Batstone et al., 2002)
includes 7 kinds of bacterial degraders that are classified by
Fig. 6 – Time course of the experimental and calculated their functions, such as sugar degraders and acetate
result in Run 7 and Run 8. Plots are experimental data and degraders, and the production ratio of each organic acid is set
lines are simulated results. at a constant level. It is acceptable for evaluating methane
production, but not organic acid species produced in shorter
retention times. With different specific primers, real-time PCR
can simultaneously measure the concentrations of many
obtained between the experimental and calculated results. types of bacteria. These proposed methods would be useful for
Inhibitory effect functions introduced into this model were developing a new bacterial function-based mathematical
favorably simulated under various conditions of noncompet- model for not only lactate fermentation, but also other organic
itive inhibition by lactate in Run 2, inhibition by NaCl in Run 3, acids or hydrogen fermentation, especially with complex
and both substrate inhibition and maximum limit concen- substrates, where many kinds of bacteria compete with each
tration of lactate in Run 6. Exponential growth and degrada- other in the same reactor.
tion of measured biomass concentrations and calculated
biomass concentrations were also well correlated. In Run 7
and Run 8, kitchen garbage and banana peels were used. From 5. Conclusions
the experimental results, 70% of the total carbohydrate was
considered to be biodegradable and remaining 30% was From the basic fermentation experiments using simple
assumed to be nonbiodegradable. Here, 70% of the measured substrates, the growth kinetics of B. coagulans including the
soluble and particulate carbohydrate concentrations of the inhibitory effects of substrate, product, and NaCl for L-lactate
substrate were used in this model calculation. Time courses fermentation under pH 5.5 and 55  C were quantitatively
for the experimental and calculated results in Run 7 and Run 8 shown. B. coagulans concentrations were suitably measured
are shown in Fig. 6. In Run 7, the decrease in carbohydrate using real-time PCR even when traditional methods such as
concentration and increase in lactate concentration were SS, protein, and turbidity cannot be used to measure microbial
adequately simulated. In this study, the initial microorganism concentrations. The developed model can simulate L-lactate
concentrations were lower than the substrate concentrations. fermentation characteristics, even with the previously
This showed that the substrate-microorganism ratio (Contois mentioned inhibition effects and growth of the associated
kinetic model) was a better marker for demonstrating the bacteria. These results indicate that the developed model and
limiting factor in the hydrolysis of particulate kitchen garbage, biomass measurements using real-time PCR can be used to
rather than modeling the substrate concentration as a first evaluate lactate fermentation using both simple and complex
order reaction. This result is in agreement with other studies substrates. These proposed methods can be useful for devel-
that used a Contois model to describe anaerobic hydrolysis of oping a new bacterial function-based mathematical model for
particulate organic wastes (Yasui et al., 2008; Ramirez et al., fermentations of more complex acids.
2562 water research 44 (2010) 2554–2562

community shifts during anaerobic self-degradation of waste-

Acknowledgements activated sludge. Bioresour. Technol. 100 (21), 4981–4988.
Lane, D.J., 1991. 16S/23S rRNA sequencing. In: Stackebrandt, E.,
This work was partially supported by the research funds of Goodfellow, M. (Eds.), Nucleic Acid Techniques in Bacterial
KAKENHI (18560530), and Ministry of the Environment Systematics. Wiley, Chichester, pp. 115–175.
Lee, M., Hidaka, T., Tsuno, H., 2008. Effect of temperature on
(K2125), Japan.
performance and microbial diversity in hyperthermophilic
digester system fed with kitchen garbage. Bioresour. Technol.
99 (15), 6852–6860.
references Lin, J.Q., Lee, S.M., Koo, Y.M., 2004. Modeling and simulation of
lactic acid fermentation with inhibition effects of lactic acid
and glucose. Biotechnol. Bioprocess. Eng. 9 (1), 52–58.
APHA, AWWA, WEF, 1995. Standard Methods for the Examination Michelson, T., Kask, K., Jõgi, E., Talpsep, E., Suitso, I., Nurk, A.,
of Water and Wastewater, 19th ed. American Public Health 2006. L(þ)-Lactic acid producer Bacillus coagulans SIM-7
Association Publications, Washington, DC. DSM 14043 and its comparison with Lactobacillus delbrueckii
Akao, S., Tsuno, H., Cheon, J., 2007a. Semi-continuous L-lactate ssp. lactis DSM 20073. Enzyme Microb. Technol. 39 (4),
fermentation of garbage without sterile condition and analysis 861–867.
of the microbial structure. Water Res. 41 (8), 1774–1780. Nandasana, A.D., Kumar, S., 2008. Kinetic modeling of lactic acid
Akao, S., Tsuno, H., Horie, T., Mori, S., 2007b. Effects of pH and production from molasses using Enterococcus faecalis RKY1.
temperature on products and bacterial community in L-lactate Biochem. Eng. J. 38 (3), 277–284.
batch fermentation of garbage under unsterile condition. Nishimura, F., Tsuno, H., Somiya, I., 1996. Measurement of
Water Res. 41 (12), 2636–2642. attached biomass. Environ. Sanit. Eng. Res. 10 (3), 245–250 (in
Asahira, T., 2006. Quantification of Identified Microbes in Japanese).
Thermophilic Methane Fermentation Reactors Based on 16S Ohara, H., Hiyama, K., Yoshida, T., 1992. Noncompetitive product
rDNA. Master thesis, Kyoto University. inhibition in lactic-acid fermentation from glucose. Appl.
Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Microbiol. Biotechnol. 36 (6), 773–776.
Pavlostathis, S.G., Rozzi, A., Sanders, W.T.M., Siegrist, H., Park, Y., Hong, F., Cheon, J., Hidaka, T., Tsuno, H., 2008.
Vavilin, V.A., 2002. Anaerobic Digestion Model No.1. Comparison of thermophilic anaerobic digestion
International Water Association (IWA), Publishing, characteristics between single-phase and two-phase systems
London, UK. for kitchen garbage treatment. J. Biosci. Bioeng. 105 (1), 48–54.
Biazar, J., Tango, M., Babolian, E., Islam, R., 2003. Solution of the Payot, T., Chemaly, Z., Fick, M., 1999. Lactic acid production by
kinetic modeling of lactic acid fermentation using Adomian Bacillus coagulans – kinetic studies and optimization of
decomposition method. Appl. Mathem. Comput. 144 (2–3), culture medium for batch and continuous fermentations.
433–439. Enzyme Microb. Technol. 24 (3–4), 191–199.
Boonmee, M., Leksawasdi, N., Bridge, W., Rogers, P.L., 2003. Batch Ramirez, I., Mottet, A., Carrère, H., Déléris, S., Vedrenne, F.,
and continuous culture of Lactococcus lactis NZ133: Steyer, J., 2009. Modified ADM1 disintegration/hydrolysis
experimental data and model development. Biochem. Eng. J. structures for modeling batch thermophilic anaerobic
14 (2), 127–135. digestion of thermally pretreated waste activated sludge.
Burgos-Rubio, C.N., Okos, M.R., Wankat, P.C., 2000. Kinetic study Water Res. 43 (14), 3479–3492.
of the conversion of different substrates to lactic acid using Rosenberg, M., Rebros, M., Kristofikova, L., Malatova, K., 2005.
Lactobacillus bulgaricus. Biotechnol. Prog. 16 (3), 305–314. High temperature lactic acid production by Bacillus coagulans
Cheon, J., Hidaka, T., Mori, S., Koshikawa, H., Tsuno, H., 2008. immobilized in LentiKats. Biotechnol. Lett. 27 (23–24),
Applicability of random cloning method to analyze microbial 1943–1947.
community in full-scale anaerobic digesters. J. Biosci. Bioeng. Sakai, K., Murata, Y., Yamazumi, H., Tau, Y., Mori, M.,
106 (2), 134–140. Moriguchi, M., Shirai, Y., 2000. Selective proliferation of lactic
Contois, D.E., 1959. Kinetics of bacterial growth: relationship acid bacteria and accumulation of lactic acid during open
between population density and specific growth rate of fermentation of kitchen refuse with intermittent pH
continuous cultures. J. Gen. Microbiol. 21 (1), 40–50. adjustment. Food Sci. Technol. Res. 6 (2), 140–145.
DuBois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., Sakai, K., Ezaki, Y., 2006. Open L-lactic acid fermentation of food
1956. Colorimetric method for determination of sugars and refuse using thermophilic Bacillus coagulans and fluorescence
related substances. Anal. Chem. (Wash.) 28 (3), 350–356. in situ hybridization analysis of microflora. J. Biosci. Bioeng.
Ha, J.H., Hidaka, T., Tsuno, H., 2009. Quantification of toxic 101 (6), 457–463.
microcystis and evaluation of its dominance ratio in blooms Skiadas, I.V., Gavala, H.N., Lyberatos, G., 2000. Modelling of the
using real-time PCR. Environ. Sci. Technol. 43 (3), 812–818. periodic anaerobic baffled reactor (PABR) based on the
Haldane, J.B.S., 1930. Enzymes. Longmans, London. retaining factor concept. Water Res. 34 (15), 3725–3736.
Hofvendahl, K., Hahn-Hagerdal, B., 2000. Factors affecting the Tanigawa, N., Takemoto, T., Ohki, H., Kawasaki, T., 1997. Detailed
fermentative lactic acid production from renewable resources component of garbage. J. Jpn. Waste Manage. Assoc. 50, 116–
(1). Enzyme Microb. Technol. 26 (2–4), 87–107. 119 (in Japanese).
Ishizaki, A., Ohta, T., 1989. Batch culture kinetics of L-lactate Venkatesh, K.V., Okos, M.R., Wankat, P.C., 1993. Kinetic model of
fermentation employing Streptococcus sp. IO-1. J. Ferment. growth and lactic acid production from lactose by Lactobacillus
Bioeng. 67 (1), 46–51. bulgaricus. Process Biochem. 28 (4), 231–241.
Ishizaki, A., Ohta, T., Kobayashi, G., 1989. Batch culture growth model Yasui, H., Goel, R., Li, Y.Y., Noike, T., 2008. Modified ADM1
for lactate fermentation. J. Ferment. Bioeng. 68 (2), 123–130. structure for modelling municipal primary sludge hydrolysis.
Kargi, F., Dincer, A.R., 1999. Salt inhibition effects in biological Water Res. 42 (1–2), 249–259.
treatment of saline wastewater in RBC. J. Environ. Eng. 125 Yu, Y., Kim, J., Hwang, S., 2006. Use of real-time PCR for group-
(10), 966–971. specific quantification of aceticlastic methanogens in
Kobayashi, T., Yasuda, D., Li, Y., Kubota, K., Harada, H., Yu, H., anaerobic processes: population dynamics and community
2009. Characterization of start-up performance and archaeal structures. Biotechnol. Bioeng. 93 (3), 424–433.