Appl Microbiol Biotechnol (1998) 49: 682±690 Ó Springer-Verlag 1998

ORIGINAL PAPER

C. AÊkerberg á K. Hofvendahl á G. Zacchi

B. Hahn-HaÈgerdal

Modelling the in¯uence 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 ¯our

Received: 28 October 1997 / Received revision: 3 February 1998 / Accepted: 6 February 1998

Abstract A kinetic model of the fermentative produc- medical sutures and clips for wound closure and self-
tion of lactic acid from glucose by Lactococcus lactis ssp. degradable prosthetic devices (Kharas et al. 1994).
lactis ATCC 19435 in whole-wheat ¯our has been de- Lactic acid occurs naturally in two optical isomers and,
veloped. The model consists of terms for substrate and since elevated levels of the D-()) form are harmful to
product inhibition as well as for the in¯uence of pH and humans (Expert Committee on Food Additives 1967),
temperature. Experimental data from fermentation ex- L-(+)-lactic acid is the preferred isomer. Polylactic acid
periments under di€erent physical conditions were used with di€erent properties can be produced, depending on
to ®t and verify the model. Temperatures above 30 °C the optical composition of the lactic acid used for po-
and pH levels below 6 enhanced the formation of by- lymerisation (Lipinsky and Sinclair 1986; Kharas et al.
products and D-lactic acid. By-products were formed in 1994). The synthetic production of lactic acid results in a
the presence of maltose only, whereas D-lactic acid was racemic mixture of the two isomers, while fermentative
formed independently of the presence of maltose al- production can yield either form alone or a racemate,
though the amount formed was greater when maltose depending on the organism, substrate and growth con-
was present. The lactic acid productivity was highest ditions used. The control of the optical composition and
between 33 °C and 35 °C and at pH 6. In the concen- the reduction of by-product formation is essential for
tration interval studied (up to 180 g l)1 glucose and 89 g cost-e€ective fermentative production of L-lactic acid,
l)1 lactic acid) simulations showed that both substances since puri®cation constitutes a considerable fraction of
were inhibiting. Glucose inhibition was small compared the total production cost (Evangelista et al. 1994).
with the inhibition due to lactic acid. Cheap raw materials, such as whey, molasses, starch
waste and beet- and cane-sugar have been used for the
fermentative production of lactic acid (Vickroy 1985;
Atkinson and Mavituna 1991). Enzymatically hydro-
Introduction lysed whole-wheat ¯our, containing both gluten and
bran, has been found to contain all the necessary nu-
Lactic acid is produced by fermentation of whey, for trients for L. lactis 19435 (Hofvendahl and Hahn-
example, or synthetically from substrates such as lacto- HaÈgerdal 1997a). Hydrolysis is performed in two enzy-
nitrile (Vickroy 1985; Atkinson and Mavituna 1991). Its matic steps, liquefaction and sacchari®cation, where
technical applications include use as a preservative in fermentation could be integrated with sacchari®cation
food, pharmaceuticals and cosmetics, and the produc- (simultaneous sacchari®cation and fermentation) and a
tion of polylactic acid, a biodegradable polyester used in kinetic model of the entire process is required to opti-
mise the production of L-lactic acid. The model param-
eters for sacchari®cation and fermentation must be
C. AÊkerberg á G. Zacchi (&)
Department of Chemical Engineering 1, determined individually to evaluate whether the two
Lund Institute of Technology/Lund University, steps should be performed simultaneously or separately.
P.O. Box 124, SE-221 00 Lund, Sweden A number of kinetic models for the fermentation of
Tel.: +46 46 2228297 glucose to lactic acid have been proposed (Luedeking
Fax: +46 46 2224526
e-mail: Guido.Zacchi@kat.lth.se and Piret 1959; Mercier et al. 1992; Parente et al. 1994).
Models including terms for both substrate and product
K. Hofvendahl á B. Hahn-HaÈgerdal
Department of Applied Microbiology,
inhibition have been suggested (GoncËalves et al. 1991;
Lund Institute of Technology/Lund University, Venkatesh et al. 1993), as well as a model considering
P.O. Box 124, SE-221 00 Lund, Sweden only product inhibition (Yeh et al. 1991; Cachon and
 683

DivieÁs 1994; Gadgil and Venkatesh 1997). The hydro- volume was 2 l. The pH was measured with a pH electrode (Schott
lysis and fermentation steps have di€erent pH and GeraÈte H63, Germany) and was kept constant at 4.0, 5.0 or 6.0 by a
pH meter and titrator (pHM61 and TTT80, Radiometer, Copen-
temperature optima, and models considering the pH and hagen, Denmark) with the addition of 200 g l)1 NaOH (Akzo No-
temperature dependence of the lactic acid production bel, Eka Nobel, Bohus, Sweden). The stirring rate was 350 rpm, and
rate have been developed for Lactobacillus casei fer- the temperature 30 °C, 33.5 °C, 37 °C, or 40 °C. To ensure anaer-
menting whey (Rincon et al. 1993). A number of models obic conditions, N2 was bubbled through the fermentor at a rate of
150 ml min)1 or 500 ml min)1, controlled by a rotameter (Sho-Rate,
considering only the e€ect of pH on the production rate Brooks, Instrument N. V. Veenendaal, The Netherlands). A sterile
have been proposed for Lactococcus lactis ssp. lactis anti-foaming agent (Silicone antifoam, Kebo Lab, SpaÊnga, Sweden)
biovar diacetylactis (Cachon and DivieÁs 1994), Lacto- was added with a sterile syringe, via a septum, as required. In two
bacillus bulgaricus (Venkatesh et al. 1993; Gadgil and fermentation experiments lactic acid (sodium salt solution, ICN
Venkatesh 1997) and Lactobacillus delbrueckii (Yeh Biomedicals Inc., Aurora, Ohio, USA) was added, either 80 g l)1 at
the time of inoculation or 35 g l)1 12 h after inoculation.
et al. 1991) using lactose or glucose. In the present
study, the kinetics were determined for fermentations at
various pH and temperature values and initial glucose Analysis
and lactic acid concentrations. Based on these data, an
Double measurements of the cell mass of the inoculum as dry
empirical model consisting of rate expressions for cell weight were performed. The samples were ®ltered through 0.2-lm
growth, product formation, starch degradation and membrane ®lters (Supor-200, Gelman Science, Ann Arbor, Mich.,
substrate consumption was developed. For optimisation USA) and dried in a microwave oven at low power (420 W) for
purposes, terms for pH and temperature in¯uence were 15 min. Owing to the high particle content of the ¯our, dry-weight
analyses of the cell mass during fermentation could not be carried
also included in the model. out. The amount of carbon dioxide in the outgoing gas was mea-
sured by photoacoustic spectroscopy (BruÈel & Kjñr type 1308;
BruÈel & Kjñr, Nñrum, Denmark).
Materials and methods Sterile samples were collected regularly, and analysed with
HPLC as described previously (Hofvendahl and Hahn-HaÈgerdal
1997b). D- and L-lactic acid concentrations were measured enzy-
Inoculum preparation and microorganism used matically as described (Hofvendahl and Hahn-HaÈgerdal 1997a).
For the propagation of the inoculum, a ¯our-free, reference me-
dium (Hofvendahl and Hahn-HaÈgerdal 1997b) was used. For Modelling
plates, 20 g l)1 agar (Merck) was added. Lactococcus lactis ssp.
lactis ATCC 19435 (L. lactis 19435) (American Type Culture The rates of cell growth, product formation and substrate con-
Collection, Rockville, Md., USA) stored at )80 °C was plated on sumption were expressed by an unstructured model for batch fer-
an agar plate and incubated at 30 °C for 48±72 h. A single colony mentation. The parameters in the model were ®tted to experimental
was transferred to another plate and incubated for another 24 h. A data, obtained from the fermentation experiments, by non-linear
single colony was then transferred to 5 ml liquid medium and in- least-squares ®tting.
cubated in a Gallenkamp INR-200 orbital incubator (Leicester,
UK) for 12±24 h at 140 rpm. The cells were harvested by centrif-
ugation (8000 g, 2 min, 4 °C), resuspended in 100 ml fresh medium Cell growth
and incubated in the same way for another 6±12 h. The bacteria
were harvested by centrifugation (9000 g, 10 min, 4 °C; Beckman For fermentation with L. delbrueckii, it has been observed that
J2-21, Beckman Instruments Inc., Fullerton, Calif., USA) and re- both substrate and product inhibition is of importance (GoncËalves
suspended in 100 ml NaCl 9 g l)1 (Across Organics, N.J., USA), et al. 1991). In the present study, the cell growth rate, rx (g l)1 h)1),
corresponding to 5% of the total working volume of the fermentor. was described by Monod kinetics, including terms for both types of
This suspension was used to inoculate the fermentor. inhibition.
lmax
Sg
X
rx ˆ …1 ÿ Kp
P  †n …1†
Media composition …Sg †2
Sg ‡ Ks ‡ Ki
Two di€erent fermentation media were used, both containing where X, Sg and P represent the cell, substrate and product con-
whole-wheat ¯our suspended in water. The hydrolysed ¯our medi- centrations (g l)1) respectively, lmax is the maximum speci®c
um contained ¯our at a concentration of 120 g l)1 or 240 g l)1, growth rate (h)1), Ks the saturation parameter (g l)1), Ki the sub-
which was completely hydrolysed to glucose by the sequential ac- strate inhibition parameter (g l)1), Kp the parameter representing
tion of the enzymes Termamyl 120 L (Novo Nordisk, Bagsvaerd, the pH dependence of product inhibition (l g)1) and n the para-
Denmark) at 95 °C for 30 min and SAN Super 240 L (Novo meter used to describe product inhibition.
Nordisk) at 55 °C for 16±20 h (Hofvendahl and Hahn-HaÈgerdal The cells are inhibited by the glucose and lactic acid concen-
1997a). This resulted in glucose concentrations of 89 g l)1 and trations prevailing in the fermentor, and thus concentrations not
about 170 g l)1 respectively. The supplemented ¯our medium con- adjusted for dilution due to NaOH addition (marked*) were used
tained glucose added to a 240 g l)1 ¯our suspension to act as a when calculating the glucose and lactic acid inhibition.
substrate at concentrations around 45 g l)1 (Hofvendahl and
Hahn-HaÈgerdal 1997b). In some fermentation experiments the
¯our-free medium used for inoculum preparation was used. Product formation

The lactic acid formation rate, rp, (g l)1 h)1) was modelled using the
Fermentation Luedeking-Piret expression (Luedeking and Piret 1959):
rp ˆ a
rx ‡ b
X …2†
Cultivation was carried out in a 3-l Chemoferm FLC-B-3 fermentor
(HaÈgersten, Sweden), coupled to a Chemoferm LMS 500 control where a is a growth-associated constant for product production
unit, controlling the temperature and rate of stirring. The working and b is the non-growth-associated constant for product produc-
684

tion (h )1). The lactic acid concentration in the model represents the temperature dependence of the parameters can be described with
total lactic acid concentration, and thus no distinction was made the following expression:
between the D and L isomers. At temperatures higher than 30 °C ÿE1 ÿE2
the formation of the by-products acetic acid, formic acid and eth- parameter ˆ A1
e RT ÿ A2
e RT …8†
anol increased. Acetic acid was the major by-product, and its for- )1
where E1 and E2 have the dimensions J mol and R is the universal
mation rate, rpa (g l)1 h)1), was modelled as a measure of the total
gas constant (J mole)1 K)1). The model was evaluated in the
by-product formation.
temperature interval 30±37 °C. The temperature was raised above
rpa ˆ aa
rx ‡ ba X …3† the standard conditions of 30 °C since the optimal temperature for
the enzymatic hydrolysis of wheat starch is 55 °C. The upper limit
where aa is the growth-associated constant for by-product pro- was chosen to be 37 °C because, at 40 °C, the unstructured model
duction and ba the non-growth associated constant for by-product could not account for all by-products formed. The high by-product
production (h)1) formation also renders this temperature cost-ine€ective. Since the
In the hydrolysed ¯our medium, no by-products were found, temperature interval is relatively small, the temperature parameter
and rpa was set to zero. T, in Eq. 8, was expressed as:
T ˆ T1 ÿ Tref …K†
Starch consumption
where T1 is the temperature (K) under investigation. The reference
A limited amount of glucose was produced by unintentional hy- temperature, Tref, was chosen as 300 K to increase the e€ect of
drolysis of the supplemented ¯our medium. This degradation, rst temperature in Eq. 8.
(g l)1 h)1) was expressed as:
rst ˆ ÿkst
Sst …4†
)1
Results
where Sst is the concentration of starch (g l ), de®ned as the sum of
non-soluble polysaccharides and soluble oligosaccharides larger
than glucose, and kst is an empirical parameter for starch degra- The kinetic parameters in Eqs. 1, 2, 4 and 5 were de-
dation (h)1). The hydrolysed ¯our medium contained no residual termined by minimising the objective function
starch. X X
Q1 ˆ …Xcalc ÿ Xexp †2 ‡ …Pcalc ÿ Pexp †2
X …9†
Substrate consumption ‡ …Sg;calc ÿ Sg;exp †2
The substrate consumption rate, rsg (g l)1 h)1), was described using using experimental data from fermentation experiments
the expression in the ¯our-free medium with initial glucose concentra-
1 1 1 tions between 40 g l)1 and 130 g l)1 (Table 1) exp and
rsg ˆ ÿ
rp ÿ
rpa ÿ
rx ÿ rst ÿ m
X …5† calc represent experimental and simulated data, respec-
Yp Ypa Yx
tively. The ¯our-free medium was used to determine the
describing the glucose consumption during fermentation, resulting
in cell mass, lactic acid and by-product formation, and the glucose cell mass, which could not be determined in the whole-
formation from the unintentional starch degradation. A certain wheat ¯our medium because of the particle content.
amount of glucose is also used for maintenance energy, described Experimental and simulated data from one fermentation
with the m áX term. Yp and Yx are the stoichiometric parameters are presented in Fig. 1. A relationship between lactic
(g/g substrate) describing the theoretical yield of lactic acid and cell
mass respectively, i.e. Yp ˆ 1 g lactic acid/g glucose and acid concentration and dry weight of cells has previously
Yx ˆ 0.79 g cell mass/g glucose, using an experimentally deter- been observed in a medium free from particles (Ho-
mined cell mass composition of CH1.52O0.45N0.22; Ypa is the by- fvendahl and Hahn-HaÈgerdal 1997b). Assuming that the
product yield coecient. Ypa is the by-product yield coecient bacteria produced the same amount of lactic acid per cell
(g by-product/g substrate) and m is the maintenance energy para-
meter (g substrate/g cell mass á h).
mass in whole-wheat ¯our medium as in ¯our-free me-
dium, the parameters a and b in Eq. 2, describing the

pH dependence Table 1 Parameter values in the fermentation model under stan-

dard conditions (pH 6.0 and 30 °C) for fermentation in a particle-
The parameters for maximum speci®c growth rate, lmax, and free medium. lmax maximum speci®c growth rate, Ks saturation
product inhibition, Kp in Eq. 1 are dependent on pH. The pH parameter in Monod expression, Ki substrate inhibition parameter,
dependence of lmax and Kp was expressed according to Sinclair Kp parameter representing pH dependence of the product inhibi-
(1989): tion, n parameter describing product inhibition, a growth-asso-
lm ciated constant for product production, b non-growth-associated
lmax ˆ …6† constant for product production, m maintenance energy parameter
1 ‡ …kl1 =‰H ‡ Š† ‡ kl2 ‰H ‡ Š
(g substrate/(g cell mass á h)).
Kpm
Kp ˆ …7† Parameter Value
1 ‡ …kp1 =‰H ‡ Š† ‡ kp2 ‰H ‡ Š
)1
where [H+] ˆ 10)pH, and lm, kl, Kpm and kp are kinetic param- lmax (h ) 0.403
eters describing the e€ect of pH on lmax and Kp. Ks (g l)1) 0.790
Ki (g l)1) 164
Kp (l g)1) 1.60 ´ 10)2
n 2.06
Temperature dependence
a 13.2
b (h)1) 6.45 ´ 10)2
There are a number of temperature-dependent parameters in the
m (g g)1 h)1) 3.70 ´ 10)3
kinetic model: a, b, aa, ba and kst, and it was assumed that the
 685

conditions and the corresponding results are summa-

rised in Table 2. In all fermentation experiments the
molar amount of carbon dioxide produced was two to
three orders of magnitude lower than the amounts of the
other by-products measured (data not shown). The data
used for modelling were adjusted for dilution caused by
NaOH addition required to maintain a stable pH.
However, all concentrations given in text and tables are
the concentrations prevailing in the fermentor.

In¯uence of glucose inhibition, lactic acid inhibition and

pH on the kinetics
Fig. 1 Experimental and simulated data for a fermentation in ¯our-
free medium at pH 6, 30 °C. Experimental data for lactic acid (m), The e€ect of glucose and lactic acid inhibition was in-
glucose (j) and cell mass (d), and simulated results for lactic acid,
glucose and cell mass (б). The experimental data have been adjusted
vestigated by varying the initial glucose concentration
for dilution due to NaOH addition between 44 g l)1 and 182 g l)1 resulting in lactic acid
concentrations up to 86 g l)1 (Fig. 2). In one fermenta-
tion in supplemented ¯our medium, 35 g l)1 lactic acid
relationship between the growth rate and the product was added 12 h after inoculation (Fig. 2A). The in¯u-
formation rate, were set to the same values for fermen- ence of pH was investigated in fermentation experiments
tation experiments in a whole-wheat ¯our medium as at pH 6.0, 5.0 and 4.0 in hydrolysed ¯our medium
those in the ¯our-free medium. In all subsequent fer- (Fig. 2B±D) and in supplemented ¯our medium (data
mentation experiments one of the whole-wheat ¯our not shown).
media, hydrolysed ¯our or supplemented ¯our, was used. The maximal volumetric lactic acid productivity in-
Fermentation experiments were carried out under creased with decreasing initial glucose concentration and
di€erent physical conditions in supplemented ¯our me- was considerably inhibited at lower pH values (Table 2).
dium. To control the sugar concentration, di€erent However, the productivities of lactic acid and by-prod-
amounts of glucose were added, and the whole-wheat ucts decreased throughout the whole fermentation pro-
¯our contributed nutrients. However, a sudden change cess. When 35 g l)1 lactic acid was added 12 h after
in glucose consumption and lactic acid production rates inoculation (Fig. 2A), the lactic acid productivity de-
was observed in all fermentation experiments. Therefore, creased from 2.1 g l)1 h)1 to 0.40 g l)1 h)1. No glucose
some of the fermentation experiments were repeated in consumption or lactic acid production was observed
hydrolysed ¯our, since previous observations have in- when 80 g l)1 lactic acid was present from the start of the
dicated that nutrients were released during the enzy- fermentation (data not shown). The lactic acid concen-
matic hydrolysis of whole-wheat ¯our (Hofvendahl and tration at the end of the exponential phase decreased with
Hahn-HaÈgerdal 1997a, b). The experimental data were decreasing pH, but increasing the initial glucose con-
used to determine the parameters in the kinetic model. centration from about 80 g l)1 to about 170 g l)1 did not
Unless otherwise stated, the pH was 6.0 and the tem- result in signi®cantly higher lactic acid concentrations at
perature 30 °C (standard conditions). The physical the end of the exponential phase (Table 2).

Table 2 Results of fermentation experiments in whole-wheat ¯our ¯our medium, Suppl supplemented ¯our medium, LA prod LA
media. Std standard conditions, Glc glucose, Temp temperature, produced, end expon at the end of the exponential phase, Q max-
LA lactic acid, LA add 35 g/l LA added after 12 h, Hydr hydrolysed imal volumetric LA productivity

Variable Medium pH Temp Glucose Cell mass LA prod L-LA Q

(°C) initial initial (end expon) (end expon) (g l)1 h)1)
(g l)1) (mg l)1) (g l)1) (% of total LA)

Std Hydr 6.0 30 182 11 86 99 2.9

Glc Hydr 6.0 30 79 6 70 99 4.0
Glc, pH Hydr 5.0 30 174 12 20 99 0.42
Hydr 5.0 30 85 3 21 96 0.83
pH Hydr 4.0 30 170 3 7 97 0.23
Std Suppl 6.0 30 47 20 39 99 2.2
Glc Suppl 6.0 30 72 10 49 98 2.1
Temp Suppl 6.0 33.5 44 15 50 90 2.8
Suppl 6.0 37 40 17 31 96 2.3
Suppl 6.0 40 43 20 32 82 1.5
LA add Suppl 6.0 30 44 13 42 99a 2.1
a
Before lactic acid addition
686

Table 3 Parameter values in the fermentation model under stan-

dard conditions (pH 6.0 and 30 °C) for fermentation experiments
in hydrolysed ¯our and supplemented ¯our media parameters as in
Table 1

Parameter Value
)1
Ks (g l ) 0.790
Ki (g l)1) 164
n 2.36
a 13.2
b (h)1) 6.45 ´ 10)2
m (g g)1 h)1) 6.78 ´ 10)3

Experimental data from Fig. 2B±D were used to de-

termine the parameters for growth, glucose and lactic
acid inhibition, and to determine how these parameters
were in¯uenced by pH. Thus, Ks, Ki, n, m, lm, kl1, kl2,
Kpm, kp1 and kp2 were determined by minimising the
objective function
X X
Q2 ˆ …Pcalc ÿ Pexp †2 ‡ …Sg;calc ÿ Sg;exp †2 …10†
using the values of a and b in Table 1 exp and calc
represent experimental and simulated data, respectively.
The parameter values are summarised in Tables 3 and 4.

In¯uence of temperature on the kinetics

The in¯uence of temperature on product formation was

investigated between 30 °C and 40 °C in fermentation
experiments in supplemented ¯our medium (Fig. 3, 4).
The maximal volumetric lactic acid productivity was
highest at 33.5 °C: 2.8 g l)1 h)1. At 33.5 °C and above,
Fig. 2A±D Experimental and simulated data for fermentation exper- lactic acid formation decreased and the formation of
iments at 30 °C, performed in supplemented ¯our medium (A) and
hydrolysed ¯our medium (B±D); pH 6 with and without 40 g l)1 lactic
equimolar amounts of acetic acid and ethanol increased
acid added after 12 h (A), pH 6 in high and low concentrations of ¯our (Fig. 4). In addition, a number of unidenti®ed com-
(B), pH 5 in high and low concentrations of ¯our (C) and pH 4 (D). pounds were detected in the chromatograms. Formic
The experimental data are for lactic acid (m, n) and glucose (j, h). acid production, on the other hand, decreased with in-
Simulated results are represented by lines. j±j, m±m No lactic acid creasing temperature, and diminished as the fermenta-
added (A) and high concentrations of ¯our (B±D); h- - -h, n- - -n
lactic acid added (A) and low concentrations of ¯our (B±C). The tion continued. At higher temperatures, more of the
experimental data are adjusted for dilution due to NaOH addition lactic acid produced was in the D form (Table 2).
When 20 g l)1 glucose was consumed, the rates of
glucose consumption and lactic acid production de-
The initial glucose concentration did not in¯uence creased suddenly, e.g. Fig. 3A at 10 h. At this point, the
which isomer of lactic acid that was produced at pH 6 maltose liberated by starch degradation diminished, and
(Table 2). However, at pH 5 a lower initial glucose the maltose concentration remained constant at around
concentration resulted in increased D-lactic acid forma- 10 g l)1. Maltose was consumed when the glucose con-
tion. At initial glucose concentrations of around 40, 80 centration became suciently low: 5 g l)1; see, for ex-
or 180 g l)1, a lower pH increased D-lactic acid forma- ample, Fig. 3A at 27 h. Maltose was not modelled
tion. In supplemented ¯our medium, where maltose was
present, the D-lactic acid content, 42% and 10% of total
amount of lactic acid, at pH 4 and 5, was higher than in
hydrolysed ¯our medium (Table 2). Table 4 Parameter values in the expression for the pH dependence
By-product formation was not observed in the hy- of lmax and Kp in hydrolysed ¯our medium (Eqs. 6, 7)
drolysed ¯our medium (Fig. 2B±D), whereas in the lmax Kp
supplemented ¯our medium (Fig. 2A) a slight increase
in acetic acid and ethanol, and in some cases formic acid, lm (h)1) 1.10 Kpm (l g)1) 0.154
was observed. The by-product formation increased with kl1 (M) 9.42 ´ 10)7 kp1 (M) 1.02 ´ 10)5
kl2 (M)1) 1.27 ´ 105 kp2 (M)1) 3.82 ´ 104
decreasing pH (data not shown).
 687

Fig. 3A±D Experimental data and simulations of fermentation Fig. 4A±D Production of acetic acid, ethanol and formic acid. A±D
experiments at pH 6, performed at 30 °C (A), 33.5 °C (B), 37 °C as in Fig. 3. Experimental data for acetic acid (´), ethanol (s) formic
(C), and 40 °C (D). Symbols as in Fig. 1; experimental data for acid (+) and simulated results for acetic acid (Ð). The experimental
maltose (r) and simulated results using two (±±) and one (


) value data have been adjusted for dilution due to NaOH addition
of lmax. The experimental data have been adjusted for dilution due to
NaOH addition
Table 5 Parameter values in the expression for the temperature
dependence of the parameters a, b, aa, ba and kst and values of the
because, in an industrial process, the whole-wheat ¯our parameters Ypa, lmax, before and lmax, after in supplemented ¯our
medium. aa growth-associated constant for by-product formation,
is either totally hydrolysed to glucose prior to fermen- ba non-growth-associated constant for by-product formation, kst
tation (hydrolysed ¯our medium), or sacchari®cation empirical starch degradation parameter, Ypa by-product yield
and fermentation are performed simultaneously, result- coecient (g by-product/g substrate), before/after before or after
ing in the continuous production of glucose by starch drop in product formation rate. Parameters as in Table 1
hydrolysis. Thus, only experimental data obtained at Parameter A1 A2 E1 E2
glucose concentrations higher than 5 g l)1 were used to
®t the model for fermentaiton in supplemented ¯our a 287 8.88 ´ 103 76.9 376
medium. b (h)1) 1.77 82.4
The objective function Q2 was minimised using the aa 2.88 53.9
ba (h)1) 2.97 ´ 10±2 543
experimental data shown in Fig. 3A±C and 4A±C, and kst (h)1) 7.52 ´ 10±2 0.590 151 343
assuming that Ks, Ki, Kpm, kp1, kp2, n and m had the Ypa (g g)1) 0.169
same values as when modelled for fermentation under lmax, before (h)1) 0.525
standard conditions (Tables 3, 4). The parameter Ypa, as lmax, after (h)1) 0.216
well as the temperature dependence of a, b, aa, ba and
kst, was determined and A1, A2, E1 and E2 are presented Comparison of simulation and experimental data
in Table 5. No stoichiometric value was used for Ypa
since it accounts for all by-products formed, and not Parameter values for Ks, Ki, n and m (Table 3) and all
only the acetic acid. To account for the sudden change in parameter values summarised in Tables 4 and 5 were
productivity after 20 g l)1 glucose had been consumed, used for simulations. The resulting curves are shown in
di€erent values of lmax were determined before and after Fig. 2±4, together with the experimental data. To illus-
the drop (Table 5). trate the di€erence between using one or two values of
688

lmax in supplemented ¯our medium, simulations were substrate concentration, and thus the cell growth rate, is
also performed using only lmax, before (Table 5), shown zero.
in Fig. 3A±C. The model was veri®ed by simulating data The sudden change in product formation rate ob-
from fermentation experiments not used for model pa- served in supplemented ¯our medium did not occur in
rameter determination, e.g. those in Fig. 2A. The model the fermentation of hydrolysed ¯our medium. The drop
showed that both substrate and product inhibition are in product formation rate in the supplemented ¯our
present, the latter being dominant. The simulated max- medium may be due to nutrient limitation, and indicates
imum speci®c growth rate, lmax, was highest at pH 6.0: that nutrients are released from the ¯our during enzy-
0.53 h)1, in agreement with the observed lactic acid matic hydrolysis, in accordance with previous studies
productivity in the hydrolysed ¯our medium (Table 2). (Hofvendahl and Hahn-HaÈgerdal 1997a, b), as well as
Simulating the temperature dependence of a, b, aa and with work in progress. The value of lmax for fermenta-
ba con®rmed that maximum lactic acid productivity tion in hydrolysed ¯our medium was the same (0.53 h)1)
occurred 33.5 °C (Table 2) and the increase in acetic as for the supplemented ¯our medium before the drop in
acid formation at higher temperatures (Fig. 4). In ad- product formation rate. It could be observed that sim-
dition, the simulations showed that glucose formation ulations in supplemented ¯our medium using only this
due to starch degradation was highest at 35 °C. value of lmax did not ®t experimental data well, and
therefore two values of lmax were used (Fig. 3A±C).
For L. delbrueckii, susbtrate inhibition is of major
Discussion importance (GoncËalves et al. 1991), and both experi-
mental data and simulations of L. lactis 19435 showed
The model developed in the present study describes that substrate inhibition is of importance in the glucose
statisfactorily the kinetics of lactic acid fermentation in concentration interval investigated: 40±82 g l)1. For
whole-wheat ¯our medium with L. lactis 19435. The example, at an initial glucose concentration of 100 g l)1
model was veri®ed by simulating data from fermentation the cell growth rate decreased by 37% compared with
experiments not used for model parameter determination fermentation without substrate inhibition. Lactic acid
(e.g. Fig. 2A). It could be observed that simulated data ®t concentrations above 70 g l)1 and 110 g l)1 inhibited the
experimental data satisfactorily except after lactic acid growth of Streptococcus cremoris (Bibal et al. 1988) and
addition. This could be due to the fact that lactic acid Lactobacillus plantarum (Giraud et al. 1991). Simula-
added to the fermentation and lactic acid produced from tions showed that the growth of L. lactis 19435 was
the bacteria do not inhibit the cell growth to the same totally inhibited by 74, 16 and 32 g l)1 lactic acid in
extent. The values of the model parameters are speci®c hydrolysed ¯our medium at pH 6, 5 and 4 respectively.
for this bacterial strain. This model considers the e€ect of When modelling, it was observed that this inhibition
both product and substrate inhibition on the kinetics, could not be related to the concentration of the undis-
while models presented previously often consider the sociated acid alone. Neither could the inhibition be ac-
e€ect of product inhibition only (Yeh et al. 1991; Cachon counted for by the total concentration of lactic acid.
and DivieÁs 1994; Gadgil and Venkatesh 1997). A number Thus, it was concluded that both the undissociated and
of models have been presented that include the in¯uence the dissociated forms of lactic acid contributed to the
of pH (Yeh et al. 1991; Venkatesh et al. 1993; Cachon inhibition. This has also been reported for L. bulgaricus
and DivieÁs 1994; Gadgil and Venkatesh 1997). Few and L. rhamnosus growing on lactose and glucose, res-
models consider the e€ect of temperature on the kinetics pectively (Venkatesh et al. 1993; GoncËalves et al. 1997).
(Rincon et al. 1993), whereas the model developed in this Growth of lactic acid bacteria is notably slower below a
study includes both pH and temperature. In addition, the certain pH: pH 5 for Lactococci (Nannen and Hutkins
acetate formation rate was included as a measure of the 1991; Cachon and DivieÁs 1994) and pH 3.4 for
total formation of by-products. Similar expressions for L. plantarum (Giraud et al. 1991). The present study
by-product formation have not been presented before. showed that the maximum speci®c growth rate, lmax, of
The value of the parameter a in Eq. 2 was found to be L. lactis 19435 in hydrolysed ¯our medium was only
much higher than b (13.2 compared with 0.064 h)1) in- slightly higher at pH 6 than at pH 5: 0.53 and 0.47 h)1
dicating that the production of lactic acid was growth- respectively. However, since the e€ect of product inhi-
associated. The parameter m, in Eq. 5, representing the bition was much greater at pH 5 than at pH 6: Kp ˆ
maintenance energy, was higher for the ¯our media than 0.064 l g)1 and 0.014 l g)1 respectively, the product
for the ¯our-free medium containing, among other formation rate was substantially lower at pH 5. At pH 4,
things, yeast extract and tryptone. Thus, in the ¯our lmax was much lower, 0.080 h)1, than at the higher pH
media more glucose was used for products besides cell values, causing the lowest product formation rate.
mass, lactic and acetic acid, than was used in the ¯our- Simulations showed that 32 g l)1 lactic acid inhibited cell
free medium. The model is not valid when all substrate growth completely at pH 4. However, since this con-
has been consumed. Because of the non-growth-associ- centration was never reached and total inhibition of cell
ated term in Eq. 2 (báX) the simulation results in an growth was not observed in fermentation at pH 4 in this
increase of the product concentration, although the study, this value is dicult to verify.
 689

L. lactis switches from homolactic to mixed-acid between 33 °C and 35 °C. However, D-lactic acid and
fermentation when grown under glucose-limited condi- by-product formation increased at temperatures above
tions (Fordyce et al. 1984; SjoÈberg et al. 1995) or when 30 °C, which would increase the cost of puri®cation.
grown on galactose or lactose (Thomas et al. 1980; To simulate the simultaneous sacchari®cation of
Garrigues et al. 1997), or maltose (Lohmeier-Vogel et al. starch to glucose and further fermentation to lactic acid,
1986; Qian et al. 1994; Hofvendahl and Hahn-HaÈgerdal the model developed in this study will be combined with
1997b). In mixed-acid fermentation, pyruvate is metab- a kinetic model for the enzymatic sacchari®cation of
olised into lactic acid, ethanol, acetic acid and either wheat starch. The model will be used to determine op-
formic acid or carbon dioxide (Fordyce et al. 1984). In timal operating conditions and to determine whether
the present study, maltose was only present in the sup- sacchari®cation and fermentation should be performed
plemented ¯our medium and consumption only started simultaneously or separately.
when the glucose concentration had decreased to 5 g l)1,
in accordance with previous observations (Qian et al. Acknowledgements We wish to thank Prof. A. H. Stouthamer for
1997; Hofvendahl and Hahn-HaÈgerdal 1997b). This re- his advice, and Elahe Behtoye and Daniel Kronhall for technical
assistance. This work was supported by the Swedish National
sulted in mixed-acid fermentation, with lactic acid being Board for Industrial and Technical Development.
the major product, and formic acid, acetic acid and
ethanol being by-products. The present study showed
that temperatures above 30 °C and pH levels below 6
also enhanced by-product formation in supplemented References
¯our medium. When the temperature was increased,
only acetic acid and ethanol formation increased in Atkinson B, Mavituna F (1991) Industrial microbial processes. In:
Atkinson B, Mavituna F (eds) Biochemical engineering and
equimolar amounts, whereas the amount of formic acid biotechnology handbook, 2nd edn. Macmillan, New York, pp
decreased with increasing fermentation time. Carbon 1181±1183
dioxide formation was two to three orders of magnitude Bibal B, Goma G, Vassier Y, Pareilleux A (1988) In¯uence of pH,
lower than acetic acid and ethanol formation, con®rm- lactose and lactic acid on the growth of Streptococcus cremoris:
ing that carbon dioxide is only produced under aerobic a kinetic study. Appl Microbiol Biotechnol 28: 340±344
Bobillo M, Marshall VM (1992) E€ect of acidic pH and salt on acid
conditions (Cocaign-Bousquet et al. 1996). Pyruvate end-products by Lactobacillus plantarum in aerated, glucose-
may still be metabolised to formic acid, acetic acid and limited continous culture. J Appl Bacteriol 73: 67±70
ethanol, but the detection of formic acid may be masked Cachon R, DivieÁs C (1994) Generalized model of the e€ect of pH
by compounds produced in higher concentrations at on lactate fermentation and citrate bioconversion in Lac-
tococcus lactis ssp. lactis biovar. diacetylactis. Appl Microbiol
higher temperatures. Biotechnol 41: 694±699
A change from the formation of almost exclusively L- Cocaign-Bousquet M, Garrigues C, Loubiere P, Lindley ND (1996)
lactic acid to a higher proportion of D-lactic acid was Physiology of pyruvate metabolism in Lactococcus lactis.
seen at pH 4 and 5, as well as at 33.5±40 °C, resulting in Antonie Van Leeuwenhoek 70: 253±267
Cooper RA (1984) Metabolism of methylglyoxal in microorgan-
3%±42% of the lactic acid being in the D-form. Similar isms. Annu Rev Microbiol 38: 49±68
observations were made when L. lactis 19345 was grown Evangelista RL, Mangold AJ, Nikolov ZL (1994) Recovery of
without pH control (Hofvendahl and Hahn-HaÈgerdal lactic acid by sorption: Resin evaluation. Appl Biochem Bio-
1997a) and for L. casei (HjoÈrleifsdottir et al. 1990) and L. technol 45/46: 131±144
plantatrum (Bobillo and Marshall 1992), when grown Expert Committee on Food Additives (1967) Lactic acid. WHO
Food Addit Ser 29: 144-148
under starvation conditions or at acidic pH. It has been Fordyce AM, Crow VL, Thomas TD (1984) Regulation of product
suggested that this is because the enzymes D- and L-lac- formation during glucose or lactose limitation in nongrowing
tate dehydrogenase have di€erent pH optima (Mou et al. cells of Streptococcus lactis. Appl Environ Microbiol 48: 332±337
1972; Gordon and Doelle 1975; Bobillo and Marshall Gadgil CJ, Venkatesh KV (1997) Structured model for batch cul-
ture growth of Lactobacillus bulgaricus. J Chem Technol Bio-
1992), or that L-lactate dehydrogenase exclusively is ac- technol 68: 89±93
tivated by fructose 1,6-bisphosphate (Mou et al. 1972). Garrigues C, Loubiere P, Lindley ND, Cocaign-Bousquet M (1997)
However, the latter would imply that the mixed-acid Control of shift from homolactic acid to mixed-acid fermenta-
formation observed under carbon limitation should also tion in Lactococcus lactis: predominant role of the NADH/
NAD+ ratio. J Bacteriol 179: 5282±5287
result in enhanced D-lactic acid formation, and this has Giori GS de, Valdez GF de, Ruiz Holgado AP de, Oliver G (1985)
so far not been observed. In Escherichia coli, for example, E€ect of pH and temperature on the proteolytic activity of
D-lactic acid can be formed from dihydroxyacetone lactic acid bacteria. J Dairy Sci 68: 2160±2164
phosphate via methylglyoxal (Cooper 1984), and this Giori GS de, Valdes GF de, Ruiz Holgado AP de, Oliver G (1986)
pathway might be present in L. lactis as well. E€ect of pH and temperature on diacetyl production by lactic
acid bacteria. Milchwissenschaft 41: 80±81
The in¯uence of temperature on lactic acid produc- Giraud E, Lelong B, Raimbault M (1991) In¯uence of pH and
tion has not been widely investigated. Earlier studies on initial lactate concentration on the growth of Lactobacillus
L. lactis have mainly been concerned with the in¯uence plantarum. Appl Microbiol Biotechnol 36: 96±99
of temperature on proteolytic activity and diacetyl for- GoncËalves LMD, Xavier AMRB, Almieda JS, Carrondo MJT
(1991) Concomitant substrate and product inhibition kinetics in
mation (de Giori et al. 1985, 1986). The experimental lactic acid production. Enzyme Microb Technol 13: 314±319
data obtained and the model developed in the present GoncËalves LMD, Ramos A, Almeida JS, Xavier AMRB, Carrondo
study showed that lactic acid productivity was optimal MJT (1997) Elucidation of the mechanism of lactic acid growth
690

inhibition and production in batch cultures of Lactobacillus Parente E, Ricciardi A, Addario G (1994) In¯uence of pH on
rhamnosus. Appl Microbiol Biotechnol 48: 346±350 growth and bacteriocin production by Lactococcus lactis subsp.
Gordon GL, Doelle HW (1975) Production of racemic lactic acid in lactis 140NWC during batch fermentation. Appl Microbiol
Pediococcus cerevisae cultures by two lactate dehydrogenases. Biotechnol 41: 388±394
J Bacteriol 121: 600±607 Qian N, Stanley GA, Hahn-HaÈgerdal B, RaÊdstroÈm P (1994) Puri-
HjoÈrleifsdottir S, Seevaratnam S, Holst O, Mattiasson B (1990) ®cation and characterisation of two phosphoglucomutases from
E€ects of complete cell recycling on product formation by Lactococcus lactis subsp. lactis and their regulation in maltose-
Lactobacillus casei ssp. rhamnosus in continuous cultures. Curr and glucose-utilizing cells. J Bacteriol 176: 5304±5311
Microbiol 20: 287±292 Qian N, Stanley GA, Bunte A, RaÊdstroÈm P (1997) Product for-
Hofvendahl K, Hahn-HaÈgerdal B (1997a) L-lactic acid production mation and phosphoglucomutase activities in Lactococcus lac-
from whole wheat ¯our hydrolysate using strains of Lactobacilli tis: cloning and characterization of a novel phospoglucomutase
and Lactococci. Enzyme Microb Technol 20: 301±307 gene. Microbiology 143: 855±865
Hofvendahl K, Hahn-HaÈgerdal B (1997b) The e€ect of nutrients Rincon J, Fuertes J, Moya A, Monteagudo JM, Rodriguez L
and a-amylase inactivation on the fermentative lactic acid (1993) Optimization of the fermentation of whey by Lactoba-
production in whole wheat ¯our hydrolysate by Lactococcus cillus casei. Acta Biotechnol 13: 323±331
lactis ssp. lactis ATCC 19435. In: Campbell GM, Webb C, Sinclair CG (1989) Microbial process kinetics. In: Bu'Lock J,
McKee SL (eds) Cereals: novel uses and processes. Plenum, Kristiansen B (eds) Basic biotechnology. Academic Press,
New York, NY, pp 219±227 London, pp 102±103
Kharas GB, Sanchez-Riera F, Severson DK (1994) Polymers of SjoÈberg A, Persson I, Quednau M, Hahn-HaÈgerdal B (1995) The
lactic acid. In: Mobley DP (ed) Plastics from microbes: micro- in¯uence of limiting and non-limiting growth conditions on
bial synthesis of polymers and polymer precursors. Hanser, glucose and maltose metabolism in Lactococcus lactis ssp. lactis
Munich, pp 93±137 strains. Appl Microbiol Biotechnol 42: 931±938
Lipinsky ES, Sinclair RG (1986) Is lactic acid a commodity Thomas TD, Turner KW, Crow VL (1980) Galactose fermentation
chemical? Chem Eng Progr 82: 26±32 by Streptococcus cremoris: pathways, products, and regulation.
Lohmeier-Vogel EM, Hahn-HaÈgerdal B, Vogel HJ (1986) Phos- J Bacteriol 144: 672±682
phorus-31 NMR studies of maltose and glucose metabolism in Venkatesh KV, Okos MR, Wankat PC (1993) Kinetic model of
Streptococcus lactis. Appl Microbiol Biotechnol 25: 43±51 growth and lactic acid production from lactose by Lactobacillus
Luedeking R, Piret EL (1959) Transient and steady state in con- bulgaricus. Process Biochem 28: 231±241
tinuous fermentation. Theory and experiment. J Biochem Mi- Vickroy TB (1985) Lactic acid. In: Blanch HW, Drew S, Wang DIC
crobiol Technol Eng 1: 431±450 (eds) The practice of biotechnology: commodity products, vol 3,
Mercier P, Yerushalmi L, Rouleau D, Dochain D (1992) Kinetics Pergamon Press, Elmsford, NY, pp 761±776
of lactic acid fermentation on glucose and corn by Lactobacillus Yeh PL-H, Bajpai RK, Iannotti EL (1991) An improved
amylophilus. J Chem Technol Biotechnol 55: 111±121 kinetic model for lactic acid fermentation. J Ferment Bioeng 71:
Mou L, Mulvena DP, Jonas HA, Jago GR (1972) Puri®cation and 75±77
properties of nicotinamide adenine dinucleotide-dependent D-
and L-lactate dehydrogenases in a group N Streptococcus.
J Bacteriol 111: 392±396
Nannen NN, Hutkins RW (1991) Intracellular pH e€ects in lactic Note added in proof: Further investigations showed that the D-
acid bacteria. J Dairy Sci 74: 741±746 lactic acid formation observed was not produced by L. Lactis 19435