JOUIWAL OF BIOSC I~NCI- ANI) BIOIM;INF,~KIN(~

Vol. 96, No. 2. 134- 140. 2003

Kinetics Analysis of Growth and Lactic Acid Production in

pH-Controlled Batch Cultures of Lactobacillus casei KH- 1
Using Yeast Extract/Corn Steep Liquor/Glucose Medium
MI-YOUNG HA,’ SI-WOUK KIM,2 YONG-WOON LEE,’
MYONG-JUN KIM,’ AND SEONG-JUN KIM’*
Department of Civil, Earth and Environmental Engineering, Chonnam National University, 300 Yongbong-Dong,
Buk-Gu. Gwangju 500-757, Korea’ and Department of Environmental Engineering,
Chosun Universiq, Dong-Gu, Gwangju 501-759, Korea2

Received 20 December 2002iAccepted 19 April 2003

This study was performed to determine the optimal conditions of yeast extract, corn steep li-
quor and glucose concentration for the growth and lactic acid production of Lactobacillus casei
KH-1 and to assess the effect of these conditions using a response surface methodology. A Box-
Behnken design was used as an experimental design for the allocation of treatment combination
as 17 pH-controlled batch cultures. The growth and product parameters were estimated by
Gombertz, Leudeking and Piret models from experimental data, and analyzed statistically with
response surfaces. The effects of yeast extract, corn steep liquor and glucose were significant for
the maximum specific growth rate, p,,,, and the maximum biomass concentration, X,,,. The in-
teraction of corn steep liquor and glucose indicated that the positive or negative effect of glucose
on i&ax in corn steep liquor below or above 2.1% could be explained by the glucose-dependent
availability of a nutrient on ,umllx.Although the experiment was achieved in pH-controlled batch
culture for L. casei KH-1, the growth- and nongrowth-associated production rate parameters, a
and b, were significant in the response surface model. The growth and lactic acid production of L.
casei KH-1 were strongly affected by glucose and the importance of the media composition was
demonstrated. The estimated optimal conditions of the growth and lactic acid production of L.
casei KH-1 were 1.276% and 0.697% for yeast extract, 3.505% and 1.708% for corn steep liquor,
and 2.390% and 2.215% for glucose, respectively.

[Key words: Lactobacillus casei, lactic acid, responsesurface methodology, optimization]

Lactic acid has been widely used as an acidulant and pre-
servative agent in foods and as a precursor for the produc-
tion of emulsifiers, such as stearoyl-2-lactylates, for the
baking industries (1). Worldwide annual lactic acid produc-
tion is estimated at 4 x 10’ kg, while alternative applications
such as a monomer for production of biodegradable plastics
and also an environmentally benign chemical and solvent
for various applications may increase lactic acid demand in
future (2). The conditions of fermentation, such as tempera-
ture, pH, the type of growth media, oxygen, and some neu-
tralizers, have a large effect on the growth activity of lacto-
bacilli (3). Among these, the types of growth media play an
important role in the growth activity. Various growth media
for lactic acid bacteria, such as MRS broth, M-17, Eliker‘s
broth, skim milk, and whey permeates, have been widely
used (4). Some factors considered in the choice of growth
medium are cost, the ability to produce a large number of
cells, and harvesting method. Simple synthetic media, with
these factors taken into account, have been studied (5).
Therefore, changes in medium composition have been pro-
* Corresponding author. e-mail: seongjun@chonnam.ac.kr
phone: +82-62-530-1864 fax: +82-62-530-0864
posed as one way to improve the lactate production process
(6-10). Yeast extract (YE) satisfies the nutritional require-
ments of lactic acid bacteria that are commonly used in lac-
tic acid fermentations, but the cost of this raw material con-
tributes significantly to lactic acid production costs (1 l-l 3).
Consequently, many studies have attempted to find cheaper
supplements, which could be used as alternatives or in com-
bination with YE. That is to say minimization of YE addi-
tion is an important goal for process optimization. In this
study, the yeast extract/corn steep liquor/glucose medium
for lactic acid fermentation has been used because it is not
only expensive but also easy to harvest. The homofermenta-
tive Lactobacillus cusei is known to be an L(+)-lactic acid
producer. Furthermore, L. casei is acidotolerant with an
optimum pH of 5.5 and is relatively insensitive to product
inhibition by lactic acid (14).
Descriptive modeling for L. casei growth and lactic acid
production has been reported (15). While descriptive models
may be useful for modeling experimental data and studying
the effects of experimental conditions on culture kinetics,
they do not include any underlying biologic or physiologi-
cal mechanism that limits the interpretation of data. The ki-
netic models that describe growth and production rates as a

134
VOL. 96,2003 KINETICS OF GROWTH AND LACTIC ACID PRODUCTION 135

function of culture conditions have been proposed for Lac-
tobacillus strains (16-l 9). Response surface methodology
(RSM) offers a practical way of studying multiple variables,
since it allows the detection of main and quadratic factor
effects and first order interactions with a much smaller num-
ber of experiments than a full factorial design (20). Also,
the RSM is a collection of statistical techniques for design-
ing experiments, building models, evaluating the effects of
factors, and searching for the optimal conditions of factors
for desirable responses (20,21). The experiments were con-
ducted according to a Box-Behnken design with four de-
scriptive culture parameters: maximum specific growth rate,
maximum biomass concentration, and growth- and non-
growth-associated production rate parameters that were sta-
tistically analyzed with response surfaces. This study pre-
sents a multifactor kinetic analysis to identify the optimal
conditions of these factors and demonstrates the importance
of interactions between culture conditions and growth and
lactic acid production of L. casei KH-1 in pH-controlled
batch cultures.
MATERIALS AND METHODS
Microorganism, growth media, and culture conditions
L. casei KH-1 isolated from the Korean traditional food kimchi
(fermented Korean cabbage flavoured with garlic and red peppers)
was used (22). L. cusei KH-1 was stored at -80°C with glycerol
(28%) and thawed just before the experiments. Precultures with
MRS broth (Merck, Darmstadt, Germany) were cultivated at 37°C
and 150 rpm under anaerobic condition for 18 h. The precultured
broth of 1% (v/v) was inoculated into each of the 17 batch cultures
designed using a Box-Behnken design (Tables 1 and 2). The batch
cultures were anaerobically carried out in a 1.5-Z bioreactor of
working volume 800 ml under the conditions of 37”C, 150 rpm and
pH 5.7 for 24 h.
Analysis The total biomass concentration was determined
by optical density measurements at 620nm (OD,,,), which were
converted to cell dry weight (X) using a calibration curve (X=
OD,,,/1.3243). Cell dry weight (CDW, g/l) was determined after
membrane filtration (0.45 pm, Whatman, Kent, UK) of a known
volume of fermentation liquor and followed by washing and dry-
ing at 80°C for 24 h. The residual glucose concentration (RGC, g/l>
of culture broth was analyzed using a glucose reagent kit (Yong-
Dong, Seoul, Korea). Lactic acid concentration (LAC, g/Z) was
measured by HPLC (Young-Lin, Anyang, Korea) using a prevail
organic acid 5u (Alltech, Deertields, MA, USA) column and a UV
detector (M720; Young-Lin, Anyang, Korea). The mobile phase
was 5 mM H,SO, at a flow rate of 1 ml/min and the wavelength of
the detector was 220 nm. Twenty ul of sample by sequential treat-
ments with centrifugation (10,000 rpm, 15 min), filtering (0.20 urn;
Whatman, Kent, UK) and the addition of H,SO, (98%) was in-
jected into the HPLC.
Descriptive modeling Kinetic batch parameters were esti-
mated with descriptive models for growth. Since bacteria grow
exponentially, it is often useful to plot the logarithm of the rela-
tive population size ~=ln(NIN,)] against time. The growth curve
of L. casei KH-1 was accurately described by a reparameterized
Gombertz equation (Eq. I) (23).
(1)
According to Zwietering et al. (23), the three phases of the growth
curve can be described by three parameters: The maximum spe-
cific growth rate, rumaX, is defined as the tangent in the inflection
point; the lag time, 1, is defined as the x-axis intercept of this tan-

TABLE 1. Actual factor levels corresponding to coded factor levels

Actual factor level at coded factor level of

Factor Symbol
-1 0 1
Yeast extract (YE, %) 4 0.6970 1.3940 2.0909
Corn steep liquor (CSL, %) x, 1.7075 3.4150 5.1225
Glucose (G %) 4 2.0000 3.0000 4.0000

TABLE 2. ‘Treatment combinations” and responses

Coded variable levels Responses

Run
4 x, 4 P,,,, (h-‘1 Tlax WI a b
1 -1 -1 0 0.4770 6.4365 0.6783 0.0889
2 0 0 0 0.8505 7.2617 I .0962 0.058
3 0 1 -1 22.3052 6.7143 1.6449 0.0098
4 1 0 -1 11.0246 6.8953 1.7995 0.0008
5 -1 0 1 0.6306 6.8812 0.0591 0.1136
6 0 0 0 0.9127 7.1283 I .3483 0.0435
7 0 1 1 0.8378 8.2767 0.8546 0.0574
8 0 0 0 0.8201 7.1425 0.9611 0.0601
9 I 0 1 0.7718 8.4674 1.3986 0.0449
10 1 -1 0 0.5024 6.4179 1.4309 0.0571
11 0 0 0 1.2119 6.9656 1.0502 0.0529
12 0 0 0 0.7155 6.4141 -0.2513 0.1184
13 1 1 0 1.6923 7.9429 1.1943 0.0404
14 0 -1 -1 0.5743 5.5164 1.8703 0.0212
15 -1 1 0 0.8288 7.2333 1.5897 0.0172
16 -1 0 -1 0.8399 5.2461 1.2187 0.0475
17 0 -1 1 0.7546 8.6384 0.4225 0.0794
’ All combinations were adjusted to pH 6.1, 37°C and 150 rpm.
136 HA E’f AL

gent; and the asymptote [A=In(N,IN,,)] is the maximal value IABLE 3. Analysis of variation ofthe 2H response surface model
reached. Curves may show a decline. This kind of behavior is for maximum specific growth rate. /I,,,,,’
called the death phase and is not considered in this paper. The non- Source of Degree of Sum of Mean of
linear equations were fitted to growth data by nonlinear regression variation freedom squares squares F-value”
using SigmaPlot software. This is a search method to minimize the (VS) (W (SS) (MS)
sum of the squares of the differences between the predicted and Model 6 s.14 0.X6 13.5Sh
measured values. Residual 10 0.63 0.063
The lactic acid production rate was described by the Luedeking Lack of fit 6 0.45 0.074 1.61n.s
and Piret (LP) model (Eq. 2) (24): Pure error 4 0.19 0.046
Total 17 28.22 I .66
(2) ns., Not sign&it at the P=O.O5 level.
a AL,,,,was estimated from batch culture biomass data with Gombertz
where a and b are the growth- and nongrowth-associated produc- model (Eq. I) adjusted R*=0.82.
tion rate parameters, respectively. ’ Significant at the P=O.Ol level.
The LP model cannot accurately describe the end of batch cul-
tures, since it predicts that lactic acid production continues regard-
less of substrate availability and lactic acid accumulation. There- candidatesof regressionmodel for explanatory variablesare linear,
fore, the data points obtained from the growth phase and the period interaction, quadratic, cubic, and quadratic terms of coded levels
of constant production rate after the end of growth were used to of YE, CSL and G. The a-levels at which every term in the se-
estimate the growth- and nongrowth-associated production rate lected model should be significant were set as I % and 5%.
parameters, u and b.
Experimental design and statistical analysis A Box-
Behnken design was used to allocate treatment combinations in the RESULTS AND DISCUSSION
experiments conducted for 24 h (Table 2). In these experiments,
the responses, such as the maximum specific growth rate (h-l), Maximum specific growth rate p,,, maximum biomass
maximum biomass concentration (g/l), and growth- and non- concentration X,,,, and growth-associated and nongrowth-
growth-associated production rate parameters a and b were esti- associated production rate parameters, a and 6, were deter-
mated for modeling the growth and lactic acid production of L. mined for pH-controlled batch cultures that used descriptive
casei KH-I. The code factor levels were determined as follows:
X,,=(YE-1.3940)10.6970, &,=(CSL-3.4150)/1.7075 and X,,= growth and lactic acid production models. If the lactic acid
(G-3)/2. Table I contains actual factor levels corresponding to production rate was closely dependent on the growth of lac-
coded factor levels. For each factor, a conventional level was set to tic acid bacteria, the maximum specific growth rate, maxi-
zero as a coded level. Response surface linear, 2 factor interaction mum biomass concentration and growth-associated produc-
(2FI) or quadratic model were calculatedfor the different response tion rate parameter were positive when the values were
variables, separately. larger.
Maximum specific growth rates The 2FI response
surface (Eq. 4) was calculated for the maximum specific
growth rates obtained from the batch culture experiments.
The model was highly significant (P<O.Ol) and the lack of
fit was not significant (Table 3). Therefore, the 2FI response
surface model accurately represented the observed p,, (Fig.
Statistical analysis The optimal conditions were estimated la-c). The effects of Xz were highly significant (WO.001)
by data analysis, and response surface plots were generated by and the effects of X, and JY,were also significant (PcO.01)
Design-Expert (ver. 6; Stat-Ease, Minneapolis, MN, USA). The (Table 4). The results indicated that three factors were

06970 10455 I.3940 17424 2.09X 0.6970 10455 13940 17424 20909 17075 25613 34150 42687 51225
X, 5 X2

FIG. 1. Response surfaces for maximum specific growth rate as a function of (a) yeast extract and corn steep liquor concentration fixing glu-
cose 3%, (b) yeast extract and glucose concentration fixing corn steep liquor=3.4150%, (c) corn steep liquor and glucose concentration fixing
yeast extract= 1.3940%. Each closed circle means the corresponding design point.
 KINETICS OF GROWTH AND LACTIC ACID PRODUCTION 137

TABLE 4. Tests for significance of linear and interaction effects TABLE 6. Tests for significance of linear effects of factors in
of factors in the response surface model for the response surface model for maximum biomass concentration, X,,
maximum specific growth rate, p,,,,
Source of Degree of Sum of Mean of
Source of Degree of Sum of Mean of variation freedom squares squares F-value
variation freedom squares squares F-value WS) (W cw (MS)
(VS) (DF) cw (MS) I 1.93 1.93 8.57”
4
4 I 0.83 0.83 13.09” x, I 1.25 1.25 5.54”
x, 1 2.62 2.62 41.46b x, I 7.78 7.78 34.62b
4 1 0.68 0.68 10.76” a Significant at the P=O.Ol level.
4 *x2 I 0.24 0.24 3.73n.s. b Significant at the P=O.OOl level.
x,*x, 1 0.16 0.16 2.59n.s.
x,*x, 1 0.61 0.61 9.69”
ns., Not significant at the P=O.OS level. TABLE 7. Analysis of variation of the linear response surface model
a Significant at the P=O.Ol level. for growth- and nongrowth-associated parameters a and b”
b Significant at the P=O.OOl level.
Source of Degree of Sum of Mean of
variation freedom squares squares F-value
md (W w (MS)
TABLE 5. Analysis of variation of the linear response surface model
for maximum biomass concentration, X,,,a Model a 3 2.55 0.85 3.71b
Residual 13 2.98 0.23
Source of Degree of Sum of Mean of Lack of fit 9 1.40 0.16 0.40n.s.
variation freedom squares squares F-value Pure error 4 1.57 0.39
WS) OF) cm (MS) Total 17 25.37 1.49
Model 3 10.96 3.65 16.25b Model b 3 9.608. IO” 3.203. IO” 5.43b
Residual 13 2.92 0.22 Residual 13 7.661. 10m3 5.893.10A
Lack of fit 9 2.47 0.27 2.45n.s. Lack of tit 9 4.141.10” 4.601.10-’ 0.52n.s.
Pure error 4 0.45 0.11 Pure error 4 3.521 ‘10” 8.802.10A
Total 17 855.00 50.29 Total 17 3.888.10”
n.s., Not significant at the P=O.O5 level. n.s., Not significant at the P=O.O5 level.
aX,, was estimated from batch culture biomass data with Gombertz aa and b were estimated from batch culture biomass data with
model (Eq. 1) adjusted with R2=0.74. Leudeking and Piret model (Eq. 1) adjusted with R2=0.34 and 0.45,
b Significant at the P =O.Ol level. respectively.
b Significant at the P=O.Ol level.
affected essentially for p,,,, each. Also, it was seen that the
effects of pairs of factors were additive since there were no Also, Aeschlimann and Von Stockar (25) found that no sub-
interactions other than the X,xX, interaction. The additive strate inhibition occurred when L. helveticus was cultured in
effects of the two factors meant that the effect of one factor a concentrated whey permeate medium up to 240 g/l, results
on the response did not depend on the level of the other fac- which are different from those of the present study. The
tor. The combined effects of X,xX, are shown in Fig. lc. equation for X,, of L. casei KH- 1 showed a positive linear
The X,xX, interaction resulted in a slight increase of prnax effect and a negative interaction effect, that is,
with the increase of X, at X, being below about 2.1 and the
Y= 7.03 +0.49(X,,) + 0.39(X*,) + 0*99(X,,)
rapid decrease of ,uu,, with the increase of 1, at X, above
about 2.1. The positive or negative effect of X, on ,u~, at X, Production rate parameters Lactic acid production
below or above about 2.1 could be explained by the X,-de- was estimated using the LP model (Eq. 2) and the estimated
pendent availability of a nutrient in corn steep liquor. That is a and b values were used as response variables. The linear
to say the optimal mixture concentration of the factors response surface model was significant for the growth- and
should be determined. The values of the regression coeffi- nongrowth-associated production rate parameters, a and b,
cients were calculated and then an equation for the 2FI and did not have a significant lack of fit. However, the
model could be written as follows: adjusted R square value was poor in the model for a and b
(Table 7). Therefore, the experimental data were reexam-
Y= 1.15 - 0.32(X,,) - 0.57(X,,) + 0.29(X,,)
ined by eliminating experimental data 12 and 15 in order to
-0.24(X,,xX2,)+0.20(X,,xX,,)+0.39(X2,xX,,)
tentatively enhance the correlation coefficient of the re-
Maximum biomass concentration The response sur- sponse model since experiments 12 and 15 generated ex-
face linear model for X,, was highly significant (WO.01) perimental errors from reparameterization of the Gombertz
with no significant lack of fit (Table 5). The effects of all equation. Tables 8 and 10 demonstrate that the parameters a
factors were significant (Table 6) and& was particularly re- and b were significant in the linear and quardratic response
markable in the comparison of X, and X,. The large positive model with no significant lack of fit, respectively. X, and X,
effect of X, on X,, indicated that glucose was a more im- were significant (P~0.01) for the growth-associated pro-
portant factor than yeast extract and corn steep liquor for the duction rate parameter a (Table 9), and Xi, X,, X,2, and X32
growth of L. cusei KH-1. These data suggested that the were significant (PcO.01) for the nongrowth-associated
growth was limited by carbon substrate. Schepers et al. (10) production rate parameter b (Table 11). However, lactic
reported that the growth of L. helveticus was limited by acid production by L. casei KH-1 was much more greatly
nitrogen substrate in whey permeate-yeast extract medium. affected by the growth-associated parameter than the non-
13x I IA FT AL. .I. BIOSCI.Blot kc,..

TABLE 8. Analysis of variation ofthe linear response surface model

for growth-associated parameter (1”
Source of Degree of Sum ol Mean ol
variation freedom squares squares F-value
(VS) (DF) (SS) (MS)
Model 3 2.88 0.96 I 8.64h
Residual II 0.57 0.05 I
Lack of lit 8 0.48 0.060 2.19n.s.
Pure error 3 0.083 0.028
Total I5 22.78
ns., Not signiticant at the P=O.O5 level.
‘a was estimated from batch culture biomass data with LP model
(Eq. 2) adjusted with R'=0.74.
h Significant at the P=O.Ol level.

TABLE 9. Tests for significance of linear effects of factors in

the response surface model for growth-associated parameter a
_~______
Source of Degree of Sum of Mean of
variation freedom squares squares F-value
W) VW (SS) (MS)
X I 1.05 I .os 20.40"
x, I 6.54.10' 6.54.10 o 1.27.10 Ins.
x3 I 1.80 I .80 35.30"
ns., Not significant at the P=O.O5 level. 0 4 8 12 16 20 24
a Significant at the P=O.Ol level. Time (h)
FIG. 2. Comparison of growth and lactic acid production by L.
TABLE IO. Analysis of variation of the linear response surface casei KH-1 between MRS and optimum medium in pH controlled
model for nongrowth-associated parameter b” batch cultures. Closed and open symbols represent MRS and optimum
medium, respectively. Circles, squares, and triangles represent cell dry
Source of Degree of Sum of Mean of weight (CDW), lactic acid concentration (LAC) and residual glucose
variation freedom squares squares F-value concentration (RGC), respectively.
WV (W cw (MS)
Model 9 1.14.10-’ 1.26.10~’ 18.88”
Residual 5 3.34. IO-’ 6.68.10 j Y= 1.11 +0.39(X1,)+9.83. lO~(X*,)-0.47(X,,)
Lack of fit 2 l.70.101 8.51 .lO 3 1.55n.s.
Pure error 3 1.64.10 ’ 5.47.10 i Y=5.36. 10-2-22.56. 10~2(X,,)-5.11. 10m3(X,,)
Total 15 5.20. IO ’ 3.45.10~: +2.70. 1O-2(X3,)+1.36. 10-2(X,,)2+3.88. lO-3(X2,)Z
n.s., Not significant at the P=O.O5 level. - 1.56. lo-*(X3,)‘-6.48. 10~3(X,,xX2,)
ah was estimated from batch culture biomass data with LP model -5.50. 10~3(X,,xX3,)-2.65~10~3(X2,xX3,)
(Eq. 2) adjusted with R*=O.Y2.
h Significant at the P=O.Ol level. Lactic acid production was affected by yeast extract among
nitrogen sources and glucose concentration. X, has a posi-
tive effect while X3 has a high negative effect on the growth-
TABLE I I. Tests for significance of linear effects of factors in associated production rate parameter a. In constrast, X3 has
the response surface model for nongrowth-associated parameter h
a positive effect while YE has a negative effect on the non-
Source of Degree of Sum of Mean of growth-associated production rate parameter b. Lactic acid
variation freedom squares squares F-value bacteria obtain a large amount of energy by means of
W) (DF) ~~~ cw (MS) glycolysis (or Embden-Meyerhof pathway) with lactic acid
4 I 3.50.10 3 3.50. IO-’ 52.34" as the main final product (11, 26), and acid production by
x, I 1.39. IO1 1.3Y.lO-’ 2.09n.s.
I 5x3.10- 5.83.10-j 87.25" these bacteria is considered to be particularly coupled with
4
X,l I 5.94.10” 5.94.10" 8.89" growth (23).
X2? I 4.81.10-’ 4.81 lo-; 0.72n.s. Validating the optimum point of the factors We at-
x?: I 7.74. IO” 7.74. lOA 11.58” tempted to determine the optimal values by Design-Expert
X,*X? I 8.39. IO 5 8.39. 10m5 1.25n.s. based numerical optimization on an objective function
4 *x3 I 1.21.10’ 1.21.10J I .Sln.s. called desirability. The overall desirability (D) is the geo-
&*X3 I 2.81 IO-’ 2.81 IO-’ 0.42n.s. metric mean of all individual desirabilites (dJ in the range
ns., Not significant at the P=O.O5 level. from 0 (least) to 1 (most): D=(d,xdZx...xd,)““= [IId,]““,
a Significant at the P=O.Ol level. where n is the number of responses. The input factors were
automatically determined to be “in range” and the response
growth-associated parameter. The functions of the coded values kept in the experimental boundaries. In cases of neg-
values of the critical variables for a and b, respectively are ative effects between the response values, the specific
as follows. growth and lactic acid production rates were disregarded,
VOL. 96,2003 KINETICS OF GROWTH AND LACTIC ACID PRODUCTION 139

TABLE 12. Effects of MRS and optimum medium on the yields of cell and lactic acid by L. casei KH-1
Substrate utilization Y YP/S Productivity Culture time
(!z.g-‘1 (d?‘) (g.g-‘1 (g.t’.h-‘) (h)
MRS medium (pH controlled) 0.729 0.409 0.834 0.526 24
Optimum medium (pH controlled)b 0.799 0.312 0.811 0.697 24
MRS medium (pH non-controlled) 0.989 0.382 0.948 0.413 48
ab Batch cultures were conducted at 37°C and 150 rpm for 24 h, under anaerobic conditions and the pH controlled at 5.7.
c Batch culture was conducted at 37°C and 150 rpm for 48 h, under anaerobic conditions and without pH regulation. Initial pH was 5.7.

and the factor (YE, CSL and G concentration) and response
01nlax~JLX~ a and b) values were assigned to low and high
values for economical growth and lactic acid production by
L. cusei KH-1, respectively. We found a combination of
factor levels that simultaneously satisfied the requirements
placed on each of several responses. The optimal concentra-
tions of yeast extract, corn steep liquor, and glucose were
chosen 0.997%, 3.408%, and 2.692%, respectively. The
changes in CDW and LAC, and the kinetic analysis for L.
casei KH-1 are shown in Fig. 2. Table 12 presents compari-
sons of growth and lactic acid production by L. casei KH-1
in pH-controlled batch cultures using MRS medium or opti-
mum medium. The optimum medium was more desirable
than the industrial medium for effective lactic acid produc-
tion. The pH-controlled culture was positive not only with
respect to the growth rate but also culture time when we
compared pH-controlled cultures with non-controlled batch
cultures using MRS medium (Table 12). However, the pH-
controlled culture was negative for the production of lactic
acid. In other words, the relationship between the specific
growth rate and production rate was related to a negative
effect. That is to say, we need to determine the optimized re-
sponse values induced by according to the purpose of the
lactic acid fermentation like the production of cell mass or
the production of lactic acid; for the purpose of cell mass
production, cell growth associated responses values have to
come to the highest degree while in the case of lactic acid
production the product associated response values have to
maintain highly in the experimental boundaries. The result-
ing optimum conditions for the purpose of the growth of
lactic acid bacteria were determined that the concentrations
of YE, CSL and G were 1.276%, 3.505% and 2.390%, re-
spectively, and pm, X,,, a and b were 0.994 h-l, 6.371 gll,
1.332% and 0.035%, respectively, with the desirability of
0.523. When the purpose of the fermentation was lactic acid
production, the optimum conditions for YE, CSL, and G
were 0.697%, 1.708% and 2.215%, respectively, and ,u~~,
X,,,, a and b were 0.491 h-l, 5.380 gll, 1.084% and 0.059%,
respectively, with the desirability of 0.762. The results from
the desirabilities indicate that the optimum media are desir-
able for product rather than growth of L. casei KH- 1.
NOMENCLATURE
4 : yeast extract concentration (g/l)
4, : coded yeast extract concentration for response sur-
face modeling
x, : corn steep liquor concentration (g/l>
4, : corn steep liquor concentration for response surface
modeling
x, : glucose concentration (g/l)
coded glucose concentration for response surface
modeling
A : asymptote: A = ln(Xm/XO)
a : growth-associated production parameter (g-lactic
acid g-l-biomass)
b : ntmgrowth-associated production parameter (g-lac-
tic acid g-‘-biomass h-‘)
x : biomass concentration (g-cell dry weight I-‘)
x mix’. maximum biomass concentration (g-cell dry weight
Z-1)
P : specific growth rate: ,u = l/X. dX/dt (h-‘)
P : maximum specific growth rate (h-‘)
Amax: lag time (h)
ACKNOWLEDGMENT
This study was supported by Chonnam National University in
the program 2000.
REFERENCES
1. Vick Roy, T. B.: Lactic acid, p. 761-766. In Moon, M. Y.
(ed.), Comprehensive biotechnology, vol. 3. Pergamon Press,
Oxford, UK (1985).
2. Datta, R, Tsai, S. P., Bonsignore, P., Moon, S. H., and
Frank, J. R.: Technological and economic potential of poly
(lactic) acid and lactic acid derivatives. FEMS Microbial.
Rev., 16,221-231 (1995).
3. Gilliland, S. E.: Concentrated starter cultures, p. 145-157. In
Gilliland, S. E. (ed.), Bacterial starter culmre for food. CRC
Press, Boca Raton (1985).
4. Miiyra-Mlkinen, A. and Bigret, N.: Industrial use and pro-
duction of lactic acid bacteria, p. 65-95. In Salminen, S. and
von Wright, A. (ed.), Lactic acid bacteria. Marcel Dekker,
New York (1993).
5. Stanley, G.: The manufacture of starters by batch fermenta-
tion and centrifugation to produce concentrations. J. Sot.
Dairy Technol., 30,3&39 (1977).
6. Bergere, J. L. and Hermier, J.: Mass production of cells of
lactic streptococci, II. Growth of Streptococcus luctis in
medium at constant pH. Lait (revue generale des questions
laitieres),48, 13-3 1 (1968).
7. Peebles, M. M., Gilliland, S. E., and Speck, N. L.: Prepara-
tion of concentrated lactic Streptococcusstarters.Appl. Mi-
crobiol., 17, 805-810 (1969).
8. Stahouder, J., Jansen, J.A., and Hap, G.: Preservation of
starters and mass production of starter bacteria. Neth. Milk
Dairy J., 23, 182-199 (1969).
9. Oh, S., Rheem, S., Sim, J., Kim, S., and Back, Y.: Optimiz-
ing conditions for the growth of Lactobacillus casei YIT9018
in tryptone-yeast extract-glucosemedium by using response
surface methodology. Appl. Environ. Microbial., 61, 3809-
3814 (1995).
10. Schepers, A. W., Thibault, J., and Lacroix, C.: Lacto-
bacillus helveticus growth and lactic acid production during
pH-controlled batch cultures in whey permeate/yeast extract
140 HA ET AL.
medium. Part 1. Multiple factor kinetic analysis. Enzyme
Microb. Technol., 30, 176-I 86 (2002).
Il. Aeschlimann, A. and Von Stockar, U.: The effect of yeast
extract supplementation on the production of lactic acid from
whey permeate by Lactobacillus helveticus. Appl. Microbial.
Biotechnol., 32, 398402 (1990).
12. Olmos Dichara, A., Ampe, F., Uribelarrea, J. L., Pareilleux,
A., and Goma, G.: Growth and lactic acid production by
Lactobacillus casei ssp. rhamnosus in batch and membrane
bioreactor: influence of yeast extract and tryptone enrich-
ment. Biotechnol. Lett., 19, 709-7 14 (1997).
13. Payot, T., Chemaly, Z., and Fick, M.: Lactic acid produc-
tion by Bacillus coagulans ~ kinetic studies and optimiza-
tion of culture medium for batch and continuous fermenta-
tion. Enzyme Microb. Technol., 24, 191-199 (1999).
14. Bruno Barcena, J. M., Ragout, A. L., Cordoba, P.R., and
Sineriz, F.: Continuous production of L(+)-lactic acid by Lac-
tobacillus casei in two-stage systems. Appl. Microbial. Bio-
technol., 51, 3 16-324 (1999).
15. Pinelli, D., Gonzalez Vara, A., Matteuzzi, D., and Magelli,
F.: Assessment of kinetic models for the production of L- and
D-kiCtiC acid isomers by Lactobacillus casei DMS 20011 and
Lactobacillus coryniformis DMS 20004 in continuous fer-
mentation. J. Ferment. Bioeng., 83,209-212 (1997).
16. Passos, F. V., Fleming, H. P., and Ollis, D. F.: Kinetics and
modeling of lactic acid production by Lactobacillus plan-
tarum. Appl. Environ. Microbial., 60,2627-2636 (1994).
17. Vereecken, K. M. and Van Impe, J. F.: Analysis and practi-
cal implementation of a model for combined growth and
J. BlOScl. BKX-.N(I..
metabolite production of lactic acid bacteria. J. Food Micro-
biol., 73,239-250 (2002).
18. Fu, W. and Mathews, A. P.: Lactic acid production from
lactose by Lactobacillus plantarum kinetic model and effects
of pH, substrate, and oxygen. Biochem. Eng. J., 3, 1633170
(1999).
19. Amrane, A.: Batch cultures of supplemented whey permeate
using Lactobacillus helveticus: unstructured model for bio-
mass formation, substrate consumption and lactic acid pro-
duction. Enzyme Microb. Technol., 28, 827-834 (2001).
20. Montgomery, D. C.: 3rd Design and analysis of experiments.
John Wiley and Sons, New York (1991).
21. Myers, H. R.: Response surfaces methodology. Edwards
Brothers, Ann Arbor, Michigan (1976).
22. Ha, M. Y., Chung, S. Y., and Kim, S. J.: Isolation and char-
acteristics of a homofermentative lactic acid bacterium. Ko-
rean J. Biotechnol. Bioeng., 17, 333-338 (2002).
23. Zwietering, M. H., Jongenburger, I., Rombouts, F. M., and
Van’t Riet, K.: Modeling of the bacterial growth curve.
Appl. Environ. Microbial., 56, 187551881 (1990).
24. Leudeking, R. and Piret, E. L.: A kinetic study of the lactic
acid fermentation; batch process at controlled pH. J. Bio-
them. Microbial. Technol., 1, 3933412 (I 959).
25. Aeschlimann, A. and Von Stockar, U.: The production of
lactic acid from whey permeat by Lactobacillus helveticus.
Biotechnol. Lett., 11, 195-200 (1989).
26. Leh, M. B. and Charles, M.: Lactic acid production by batch
fermentation of whey permeate; a mathematical model. J. lnd.
Microbial., 4, 65570 (1989).