Continuous production of l (+)-lactic acid by Lactobacillus casei in two ...

Appl Microbiol Biotechnol (1999) 51: 316±324

Ó Springer-Verlag 1999

ORIGINAL PAPER

J. M. Bruno-BaÂrcena á A. L. Ragout P. R. CoÂrdoba á F. SinÄeriz

Continuous production of L(+)-lactic acid by Lactobacillus casei in two-stage systems

Received: 4 December 1997 / Received revision: 23 February 1998 / Accepted: 14 March 1998

Abstract A two-stage two-stream chemostat system and a two-stage two-stream immobilized up¯ow packed-bed reactor system were used for the study of lactic acid production by Lactobacillus casei subsp casei. A mixing ratio of D12/D2 ˆ 0.5 (D ˆ dilution rate) resulted in optimum production, making it possible to generate continuously a broth with high lactic acid concentration (48 g l)1) and with a lowered overall content of initial yeast extract (5 g l)1), half the concentration supplied in the one-step process. In the two-stage chemostat system, with the ®rst stage at pH 5.5 and 37 °C and a second stage at pH 6.0, a temperature change from 40 °C to 45 °C in the second stage resulted in a 100% substrate consumption at an overall dilution rate of 0.05 h)1. To increase the cell mass in the system, an adhesive strain of L. casei was used to inoculate two packed-bed reactors, which operated with two mixed feedstock streams at the optimal conditions found above. Lactic acid fermentation started after a lag period of cell growth over foam glass particles. No signi®cant amount of free cells, compared with those adhering to the glass foam, was observed during continuous lactic acid production. The extreme values, 57.5 g l)1 for lactic acid concentration and 9.72 g l)1 h)1 for the volumetric productivity, in up¯ow packed-bed reactors were higher than those obtained for free cells (48 g l)1 and 2.42 g l)1 h)1) respectively and the highest overall L(+)-lactic acid purity (96.8%) was obtained in the two-chemostat system as compared with the immobilized-cell reactors (93%).

J. M. Bruno-BaÂrcena (&) á A. L. Ragout á P. R. CoÂrdoba F. SinÄeriz1 PROIMI-MIRCEN, Av. Belgrano y Pje, Caseros, 4000 S.M. TucumaÂn, Argentina Tel.: +54 81 344 888 Fax: +54 81 344 887 e-mail: [email protected] Present address: 1 CaÂtedra de MicrobiologõÂ a Superior, Facultad de BioquõÂ mica, QuõÂ mica y Farmacia-UNT-TucumaÂn, Argentina

Introduction Manufacture of lactic acid by fermentation is an old industrial process (Vick Roy 1985) that has wide-ranging applications, principally in the biomedical ®eld, where the pure L(+)-lactate isomer is required (Lipinsky 1981). For this reason, there is a continuing interest in developing more ecient processes for its production. The process is usually limited by the injury to the cells brought about by the undissociated form of lactic acid that accumulates in the fermented broth. This inhibition was reported by Friedman and Gaden (1970) and promoted the development of a series of processes based on the elimination of the product by ®ltration and concentration of the cells using a retention unit (Vick Roy et al. 1982, 1983; Boyaval et al. 1987). Experiments were later conducted to increase the active biomass through the use of microbial cells immobilized in carrageenate, alginate or polyethyleneimine, which made it possible to employ continuous culture systems operated at high dilution rates (Audet et al. 1988; Boyaval and Guolet 1988; Gouqiang et al. 1991, 1992). However, the type of adhesion and the distribution of cells in the matrix have di€erent e€ects on the physiology and cellular metabolism, causing variable results in the production processes (Cassidy et al. 1996; HjoÈrleifsdottir et al. 1990). Therefore, systems using naturally adherent (¯occulent) cells have been developed (Kischke et al. 1991; Demirci and Pometo 1995), allowing a continuous replacement of active biomass and thus increasing the useful operating time. The main object of the industrial processes for the production of lactic acid is to obtain high yields of the pure isomer with feedstocks low in nitrogenous compounds under conditions of high temperature, low cell mass production and negligible amounts of residual substrate or other biological products. These requirements reduce downstream processing to a minimum. Nitrogenous sources used in lactic acid fermentation include corn steep liquor, malt sprout extract, casein

317

hydrolysate and whey permeate (VahvaselkaÈ and Linko 1987), but the commonest and most e€ective supplement has traditionally been yeast extract (Aeschlimann and von Stockar 1990; Hujanen and Linko 1996). However, it is important to determine the optimum concentration to use, because yeast extract is still expensive and, when used at concentrations higher than 1%, can reduce the optical purity in the fermented broth, as yeast extract contains some D())-lactate. For these reasons, the object of this work was to investigate the e€ect of a reduction in yeast extract concentration in feed streams along with increased glucose concentration on the kinetic parameters and yield in the homolactic fermentation by Lactobacillus casei, a major producer of the nutritionally important isomer L(+)lactate (Motarjemi and Nout 1996), in a two-stage chemostat fermentation system using two feed streams. The system was compared with a two-stage up¯ow packedbed continuous reactor process with two mixed-feedstock streams.

pump into the second stage, which in turn was fed with a medium containing 120 g l)1 glucose without yeast extract, by a third peristaltic pump at various ¯ow rates. The over¯ow from the second stage was pumped out to the waste reservoir. Two-stage continuous lactic acid fermentations using immobilized cells

Materials and methods

The experiments were performed in a two-stage packed-bed reactor consisting of two water-jacketed glass columns (3 cm inner diameter ´ 10 cm height for the ®rst reactor and 3.5 cm inner diameter ´ 14 cm height for the second reactor) ®lled with porous foam glass particles having a mean diameter of 2±4 mm, a porosity of 60% and a density of 0.2 g cm)3 (Trade name Poraver; a kind gift from Dennert Poraver GmbH, Postbauer-Heng, Germany). The columns had an external loop with a vessel with a stirring device, pH electrode and pH control unit, through which the medium could be continuously recycled in each reactor at a ratio of 50:1, while the same ascending velocity was maintained in both reactors. The ®nal active volume was 50 ml and 100 ml respectively. The over¯ow from the ®rst reactor was pumped into the second reactor (Fig. 2), as indicated above for the two-stage continuous stirred fermenters, but the reactor was fed with medium containing 150 g l)1 glucose without yeast extract. The temperature was set at 42 °C in the ®rst stage and at 45 °C in the second stage and the pH was set at 6.0 by titration with 1.5 M and 3 M NH4OH respectively in the two reactors.

Apparatus and operation

Organisms and cultivation medium

Single-stage and two-stage continuous fermentation Growth experiments were conducted in 2 l and 5 l model 210 LH fermenters equipped with controllers for pH, temperature and agitation speed (Inceltech, Toulouse, France). Single-stage experiments were carried out in chemostat mode with a working volume of 600 ml at 37 °C. The pH was maintained at 5.5 by automatic addition of 1.5 M NH4OH. Two reactors were used for the two-stage experiments; the ®rst had a working volume of 600 ml and in the second fermenter the volume was varied to obtain the desired retention time. In the ®rst stage the pH was kept at 5.5 and the temperature at 37 °C; in the second stage the pH was 6.0 and the temperature was either 40 °C or 45 °C. The fermenters were stirred at 200 rpm. The ®rst fermenter was fed with medium by a type-501 peristaltic pump (Watson-Marlow, Falmouth, UK) (Fig. 1) at a dilution rate of 0.17 h)1. The over¯ow of the ®rst stage was pumped by a second

L. casei subsp. casei CRL 686 was obtained from the culture collection of CERELA (Centro de Referencia para Lactobacilos, TucumaÂn, Argentina) and was used for the experiments in singleand two-stage chemostat cultures. An adhesive (¯occulent) phenotype of this strain L. casei ADNOX95 (Culture collection of PROIMI, TucumaÂn, Argentina), isolated from chemostat culture according to Ragout et al. (1996), was used in the two-stage up¯ow packed-bed reactors. In each case the reactors were inoculated with 6% v/v of an overnight culture of the corresponding strain. The basal medium for lactic acid fermentation for all experiments contained (g l)1) yeast extract 10, MgSO4 á 7H2O 0.05, (NH4)2HPO4 2.5, MnSO4 0.005, and glucose in di€erent concentrations. Yeast extract was omitted from the medium fed to the second stage of the two-stage fermentations. The glucose concentration in the feed was 22 g l)1 for the single-stage cultures. For the two-stage chemostat cultures, it was 22 g l)1 for the ®rst reactor and 120 g/l for the second reactor. In the two-stage immobilized

Fig. 1 Schematic diagram of the two-stage chemostat system with two feed streams

Fig. 2 Schematic diagram of the two-stage two-stream up¯ow packed-bed reactors. A alkali, 1 feed pump for complete medium; 2 recycling pump, 3 alkali pump, C pH controller, F air ®lter, IC heat exchanger, L external pH controlling device, M sampling port, R reactor, P pH electrode

318 fermenters, the ®rst reactor was fed with medium containing 35 g l)1 glucose and the second reactor with 150 g l)1 glucose. Glucose and Mn2+ solutions were autoclaved separately and added aseptically to the medium. Analyses Glucose and total lactic acid were determined by HPLC in a chromatograph (Gilson, Villiers le Bel, France) equipped with a refractive-index detector and a Rezex RHM monosaccharide column (300 ´ 7.8 mm, Phenomenex, Calif., USA) and with mobile phase of bidistilled water at a ¯ow rate of 0.6 ml/min. The temperature for chromatographic operation was maintained at 55 °C. D()) and L(+)-Lactic acid were determined by the enzymatic methods of Gawehn and Bergmeyer (1974) and Gutmann and Walhlefeld (1974), respectively. The percentage of isomers was reported as 100 ´ (L isomers)/(L+D isomers). Free biomass was measured by dry cell weight determination. Cells were centrifuged at 5000 g for 10 min, washed with distilled water and dried at 105 °C. Growth was followed by absorbance measurements at 600 nm on a digital spectrophotometer (Metrolab, Buenos Aires, Argentina). The immobilized biomass was determined as the di€erence between the dry weight at 105 °C and the ashes at 550 °C of a measured volume (a portion of the matrix) of the column. Reproducibility All results presented in the text are averages of triplicates di€ering less than 5%. Estimation of parameters for the single and two-stage fermenters Kinetic and yield parameters were calculated from the steady-state mass balance for both fermenters using the equations given in Table 1 (Park and Baratti 1992). Kinetic and yield parameters in one stage could be evaluated by the classical chemostat equations shown in the ®rst column of Table 1. For the second stage, equations must be derived. The input concentrations of biomass, lactic acid and sugar could be estimated by Eqs. 1±3, which averaged the e€ects of two liquid streams entering this stage. Xi2 ˆ X1
F1 =…F1 ‡ F2 †

…1†

Pi2 ˆ P1
F1 =…F1 ‡ F2 †

…2†

Si2 ˆ …S1
F1 ‡ S02
F2 †=…F1 ‡ F2 †

…3†

All terms in these equations are de®ned in Table 1. In addition, the overall parameters were calculated in an attempt to compare the performance of the two-stage fermenters with that of the conventional single-stage chemostat. For this purpose, the two-stage systems were assimilated to a single stage with a total reactor volume equivalent to V1 plus V2. The overall input sugar concentration Table 1 Equations for evaluation of kinetic and yield parameters of the two-stage continuous systems. D: dilution rate, h)1; F: feed ¯ow rate, ml h)1, V: Active reactor volume, ml; P: lactic acid concentration, g l)1; S: sugar concentration, g l)1; X: biomass concentration, g l)1; YP/S: lactic acid yield, g lactic acid, g sugar)1;

(SiG) could be de®ned according to Eq. 4 and the overall parameters were estimated by the equations shown in Table 1. SiG ˆ …S01
F1 ‡ S02
F2 †=…F1 ‡ F2 †

…4†

The speci®c growth rate in the second stage of the two-stream system (l2 ) was estimated using Eq. 5 derived from a mass balance for the biomass in this stage. l2 ˆ D2 ÿ …F1
X1 =V2
X2 †

…5†

Results E€ect of temperature on lactate fermentation A single-stage continuous culture was ®rst carried out in an attempt to determine the e€ect of temperature on biomass and L(+)-lactic acid production. Continuousculture experiments at a dilution rate of 0.17 h)1 and at various temperatures were carried out with the growthlimiting substrate (glucose). The best results in terms of total glucose consumption were obtained at temperatures of 37 °C and 40 °C (Table 2). It can be observed that, with increasing temperature, the lactate concentration increased and biomass concentration decreased. Figure 3 shows the percentage of carbon used for biomass and for lactate when there is a shift in temperature, indicating that the yield of lactic acid could be increased at the expense of biomass formation over a limited range. As far as the L(+)-lactic acid production is concerned, the ratio of L(+) and D()) isomers of lactic acid was not a€ected by temperature (Table 2). The mean value of the L(+) fraction was 95.5%. Overall performance of the two-stage two-stream system The continuous production of lactic acid by L. casei CRL 686 was carried out in a two-stage two-stream system. The ®rst chemostat, run at 37 °C and at a dilution rate of 0.17 h)1, was used to provide high biomass concentration. The second chemostat was fed with a similar culture medium with additional substrate (120 g l)1 glucose) and without yeast extract, to analyse the in¯uence of the mixing ratio (Mr ˆ F1/F1+F2, or D12/D2) between streams, on the ®nal lactic acid conYX/S: biomass yield, g cell, g sugar)1; QP: volumetric productivity, g l)1. Subscripts 0: initial variables; 1: ®rst fermentor variables; 2: second fermentor variables; i: input variables; G: overall variables (see text, Eq. 4)

Parameters

First stage

Second stage

Overall

Dilution rate, D (h)1) Biomass yield, Yx/s (g g)1) Lactate yield, Yp/s (g g)1) Sugar conversion, C (%) Volumetric productivity, Qp (g l)1 h)1)

F1/V1 X1/(S01 ) S1) P1/(S01 ) S1) 100 á (S01 ) S1)/S01 P1 á D1

(F1 + F2)/V2 (X2 ) Xi2)/(Si2 ) S2) (P2 ) Pi2)/(Si2 ) S2) 100 á (Si2 ) S2)/Si2 (P2 ) Pi2) á D2

(F1 + F2)/V1 + V2 X2/(SiG ) S2) P2/(SiG ) S2) 100 á (SiG ) S2)/SiG P 2 á DG

319 Table 2 E€ect of temperature on percentage of carbon used for the generation of biomass and lactate by Lactobacillus casei CRL 686 in continuous culture at pH 5.5, D = 0.17 h)1 and 22.5 g l)1 initial glucose. C carbon recovery T (°C)

C (%)

Lactate (g l)1)

27 32 37 40 42 45 47 50

110 102 105 98 108 105 90

9.27 10.53 15.43 3.14 17.46 0 18.54 0 19.64 2.05 19.40 2.02 14.22 4.86 Culture wash out

Glucose residual (g l)1)

centration with the desired purity and the in¯uence of this coecient on the kinetic parameters. A schematic diagram of the system is given in Fig. 1. Di€erent dilution rates were tested at 40 °C in the second stage, ranging from a D2 ˆ 0.04 h)1 to D2 ˆ 0.17 h)1 (corresponding to 24±6 h retention time) and with di€erent mixing ratios between feeds. The mixing ratio, D12/D2, was maintained between 0, which corresponded to a system without the ®rst reactor, and 1, which corresponded to a system without the second stage. In the ®rst case no microorganisms were entering the fermenter, so no biomass or lactate was produced. Four retention times were analyzed for each mixing ratio assayed (0.25, 0.4, 0.5, 0.65 and 1). Each dilution rate was maintained for at least seven residence times to obtain steady-state conditions. The steady state was corroborated by biomass and product determinations. The contour plots in Fig. 4A,B,C show the trends arising from the experimental values for lactate, glucose consumed and biomass obtained experimentally, for each mixing ratio and retention time. Some of the values corresponding to Mr ˆ 0.5 are shown in Table 3. The in¯uence of the mixing ratio on the parameters obtained during continuous culture show that the maximum consumption of glucose (Fig. 4B) and the highest

Fig. 3 E€ect of changes in temperature on the yield of biomass and lactate produced by Lactabacillus casei CRL 686 in a chemostat at pH 5.5 and D ˆ 0.17 h)1

Yield of lactic acid (g g)1)

L(+)-lactate (%)

Biomass (g l)1)

0.71 0.75 0.79 0.84 0.91 0.90 0.76

96 95 96 95 95 96 96

6.03 6.36 5.98 5.32 4.40 3.80 3.03

production of lactic acid (Fig. 4A) were observed for a mixing ratio of 0.5, corresponding to a retention time of 24 h. At this point there is only 50% of the concentration of yeast extract in the feed, indicating that a lower input of yeast extract could be used without a€ecting the catabolic capacity of the cells. It is therefore possible to generate a broth with half the impurities obtained in the one-step process. The conditions for maximum lactate production are di€erent from the conditions for biomass production, as Fig. 4C shows that the maximum concentration of solids was obtained with an optimum mixing ratio of 0.3±0.4, corresponding to a retention time of 8 h. This fact allowed us to control both the lactic acid production and the solids production by varying the value of Mr. After establishing the conditions with an optimum mixing ratio of 0.5 (which means that the overall system operates at a yeast extract concentration of 5 g l)1), the complete kinetic parameters of three steady states at dilution rates of 0.042, 0.082 and 0.11 h)1 at 40 °C and D12/D2 ˆ 0.5 are shown in Table 3. The values of the speci®c growth rate calculated for the second stage (l2 ) were similar to those for the overall dilution rate of the whole system, which indicates that the cells were growing; however, the reduced biomass yield, as compared with the one-stage system, con®rmed the e€ect of the increment in temperature on the biomass yield (see above, Fig. 3). The high lactic acid concentration (46 g l)1) was attained at the lower dilution rate in the second fermenter, which operated with the lowest residual glucose concentration (6 g l)1) of the three conditions assayed. These values are in agreement with the values for sugar conversion (91%), in which the lactic acid yield was 70%. When the dilution rate was further increased (steady state III), a high concentration of solids (8.87 g l)1) was obtained, and the volumetric productivity of lactic acid (Qp) was 3.08 g l)1 h)1, but the sugar conversion and the lactate concentration decreased, giving values for lactate from 46 g l)1 to 38.5 g l)1. The temperature of the second stage was changed to 45 °C and four steady states were tested. Table 4 shows all the data of the operation at 45 °C with a Mr (D12/D2) ˆ 0.5. Results show that the speci®c growth rate did not vary signi®cantly; however, there

320

yield of 0.25 g g)1. The amount of sugar in the medium remained at 33.5 g l)1 even though the yield of lactate and lactate productivity were 0.77 g g)1 and 2.48 g l)1 h)1 respectively. Analysis of the data of Tables 3 and 4 for the same volumetric productivity of lactate shows the advantageous e€ect of operating the fermenters at 45 °C, because of the reduction in retention time (a 100% increase of the overall dilution rate, DG) and the increase in lactate concentrations and sugar consumption. This fact demonstrated that cell growth was optimal in the ®rst stage while a lower concentration of the nitrogen source allowed the lactic acid production to remain uncoupled from growth, in the second stage. In order to determine whether the optical purity of L(+)-lactic acid was a€ected by the overall dilution rate

Fig. 4A±C E€ect of the mixing ratio (D12/D2) and retention time in the two-stage two-stream chemostat fermenter for a temperature of 40 °C in the second stage using L. casei CRL 686. A Contours of lactic acid concentration, B contours of glucose used, C contours of solids concentration. The numbers on the lines indicate the concentrations (g l)1)

was better sugar utilization since steady state I (D2 ˆ 0.06 h)1) gave a high concentration of lactate (48.5 g l)1) with a 100% glucose consumption (CG ˆ 100%) as compared with the results at 40 °C when, in the best case, at a lower D2 (0.042 h)1), only 91% of the glucose was used. When the dilution rate of the second fermenter was changed to 0.11 h)1 the largest biomass concentration (10.97 g l)1) was reached, corresponding to a biomass

Table 3 Kinetic and yield parameters in a single-stage chemostat fermentation and in the second stage at 40 °C of a two-stage chemostat fermenter with two mixed feeds (22 g l)1 and 120 g l)1 initial glucose) and with a mixing ratio (D12/D2) of 0.5 Parameter

First stage

)1

Flow rate, F2 (ml h ) Dilution rate, D2 (h)1) Biomass concentration, X2 (g l)1) Lactate concentration, P2 (g l)1) Residual glucose, S2 (g l)1) Biomass yield, Yx/s (g g)1) Lactate yield, Yp=…SiG ÿS2 † (g g)1) Sugar conversion, CG (%) Volumetric productivity, Qp (g l)1 h)1) a L(+)-lactate , (%) Speci®c growth rate, l2 (h)1) Overall sugar concentration, SiG (g l)1) Overall dilution rate, DG (h)1) a

100 ´ (L isomers)/(L +

D

isomers)

Steady states I

II

III

102 0.17 5.98 18.6 0 0.26 0.85 100 3.2

100 0.042 6.28 46 6 0.09 0.7 91 2.3

100 0.082 7.42 42.2 12 0.11 0.65 85 2.53

100 0.11 8.87 38.5 14 0.14 0.64 81 3.08

95 0.17 22

95.8 0.02 71

96 0.05 76

96.4 0.07 73.6

0.17

0.03

0.06

0.08

321 Table 4 Kinetic and yield parameters in the second stage at 45 °C of a two-stage chemostat fermenter with two feed streams (22 g l)1 and 120 g l)1 of initial glucose) and with a mixing ratio (D12/D2) of 0.5 Parameter

Steady states

Flow rate, F2 (ml h)1) Dilution rate, D2 (h)1) Biomass concentration, X2 (g l)1) Lactate concentration, P2 (g l)1) Residual glucose, S2 (g l)1) Biomass yield, Yx/s (g g)1) Lactate yield, Yp=…SiG ÿS2 † (g g)1) Sugar conversion, CG (%) Volumetric productivity, Qp (g 1)1 h)1) L(+)-lactate, (%) Speci®c growth rate, l2 (h)1) Overall sugar concentration, SiG (g l)1) Overall dilution rate, DG (h)1)

I

II

III

IV

100 0.066 6.2 48.5 0 0.08 0.68 100 2.42 96.8 0.04 71 0.05

100 0.074 7.1 40.4 12 0.12 0.68 83 2.42 96 0.05 71 0.06

100 0.082 8.09 35.5 20 0.14 0.63 73 2.13 96 0.05 76 0.06

100 0.11 10.97 31 33.5 0.25 0.77 54 2.48 96.4 0.08 73.6 0.08

reactors. The ®rst stage of the reactors was maintained at 42 °C and the second stage at 45 °C, and both reactors were operated with a recirculation loop to reduce the formation of pH gradients and to improve mixing. The reactors were inoculated with 6% (v/v) of an overnight culture. The continuous fermentation started after the cell adsorption become visible in both reactors, a starting-up period in which the cells started to divide and adher to the support, while some cells remained in the bulk medium. The amount of free cells was no higher than 10% of the adherent biomass (Table 5) indicating that a strong adhesion of the cells was achieved in the reactors working at high ascending velocities of 1.5 cm min)1, provided by the recirculation loop. This system is described in Fig. 2 and the assays were performed with various retention times and ¯ow rates in both reactors. As the fermenter volumes were ®xed, the mixing ratio (Mr) could be varied either by splitting the ¯ow from the ®rst reactor or by changing the ¯ow rate of the ®rst reactor. The pH was maintained at 5.5 and the initial glucose in the feed was 35 g l)1. For the second stage the conditions of operation were: Mr (D12/ D2) ˆ 0.5, pH 6.0, and the glucose concentration was

(DG) of the two-stage system, the concentrations of both L(+)- and D())-lactate were measured in the di€erent steady states. The results obtained (Tables 3 and 4) showed that the L(+)-lactic acid production by L. casei CRL 686 was not a€ected by the di€erent overall dilution rates and temperatures assayed. An increase in purity from 95% (single-stage) to 96.8% (two-stage), in the free-cell system, was probably due to the fact that there was a net decrease to 5 g/l yeast extract concentration when a two-step two-stream system was used. Overall performance of two up¯ow packed-bed reactors operated in two-stage mode With the objective of investigating the e€ect of increasing the active biomass through the use of immobilized cells, L. casei CRL 686 was grown in a chemostat culture to select an adherent variant, as described by Ragout et al. (1996). An adherent variant named L. casei ADNOX95, which showed spontaneous adsorption of cells to foam glass particles (Poraver), was obtained and was used in the experiments with the two-stage up¯ow packed-bed

Table 5 Kinetic and yield parameters in the second stage at 45 °C of a two-stage up¯ow packed-bed continuous reactor with two feed streams (35 g l)1 and 150 g l)1 initial glucose) and with a mixing ratio (D12/D2) of 0.5 Parameter

Flow rate, F2 (ml h)1) Dilution rate, D2 (h)1) Immobilized biomass, X2 (g l)1) Free biomass, X2 (g l)1) Lactate concentration, P2 (g l)1) Residual glucose, S2 (g l)1) Lactate yield, Yp=…SiG ÿS2 † (g g)1) Sugar conversion, CG (%) L(+)-lactate, (%) Volumetric productivity, Qp (g l)1 h)1) Overall sugar concentration, SiG (g l)1) Overall dilution rate, DG (h)1)

First stage

18 0.36 43 4.2 22 0.8 0.85 96.8 91 7.8 26 0.36

Steady states I

II

III

7.5 0.15 33 2.9 57.65 9.64 0.70 89.5 93 5.76 92 0.10

13 0.26 33 3 46.5 32.6 0.71 66.4 93 7.9 97 0.17

20 0.4 33 3.3 37.4 39.3 0.67 58.7 93.7 9.72 95 0.26

322

150 g l)1. Table 5 shows the results obtained with different feeds at a mixing ratio of 0.5. Two independent assays, including three steady states each, were evaluated for D2 between 0.15 h)1 to 0.4 h)1. The ®rst time the system was operated for 5 months and the second time for 3 months. Each dilution rate was maintained during at least ten residence times to obtain steady-state conditions. The steady-state was con®rmed by substrate and product determinations. Lactate concentrations diminished from 57.6 g l)1 to 37.4 g l)1 with a volumetric productivity increasing from 5.76 g l)1 h)1 to 9.72 g l)1 h)1 respectively. At the low dilution rate, D2 ˆ 0.15 h)1, the sugar conversion was the highest (89.5%) and, in all the steady states analyzed, the yield of lactate was constant at 70%. An overall analysis of the results obtained from the experiments in two stages, with chemostats at 40 °C and 45 °C, and with packed-bed reactors (shown in Tables 3±5), indicates that, for the same overall dilution rate (DG ˆ 0.10 h)1), the lactate concentration was higher (57.5 g l)1) for the system with immobilized cells than that obtained in the stirred tanks (38.5 at 40 °C and 31 g l)1 at 45 °C). The yield of lactate was virtually the same (0.70 g g)1) as the values of 0.68 g g)1 and 0.65 g g)1 for the stirred reactor. As far as the L(+)lactic acid production is concerned, the results obtained (Table 5) show that the speci®city of L(+)-lactic acid production by the adhesive strain L. casei ADNOX95 was not a€ected by the di€erent overall dilution rates assayed and an increment in purity from 91% (singlestage) to 93% (two-stage) could be obtained by the use of a two-step two-stream system. Considering that the overall initial sugar concentration, SiG in both systems assayed was 95±97 g l)1 for the immobilized cell system and 73±76 g l)1 for the stirred tank, the sugar conversion was improved in the up¯ow packed-bed reactors, since the yield of 89.5% indicates that the overall sugar consumed was the same as that in the experiments at 45 °C in stirred tanks (100%), but the system with immobilized cells allowed the operation with D2 ˆ 0.15 h)1 (that is, a faster sugar utilization because of the presence of a more active biomass) and volumetric productivities in the range of 9.72±5.76 g l)1 h)1 for packed-bed reactors, as compared with the values obtained for free cells (2.42 g l)1 h)1).

Discussion The results presented in this study clearly show that a high production of high-purity L(+)-lactic acid can be maintained while the overall concentration of complex factors (yeast extract) is kept low. Maintenance of the glycolytic activity in free and immobilized cells leads to an increase in lactic acid concentration and a reduction in the amount of nitrogen source. These kinetic and yield parameters for individual stages provide evidence of the advantages of a two-stage continuous fermenter over a

single-stage system because of the lower amounts of impurities and the larger amount of cell mass available for the conversion. In a single-stage continuous culture, the e€ect of temperature on lactate and biomass yields showed (Table 2 and Fig. 3) that it is possible to increase the lactate yield at the expense of biomass formation; this behavior was observed in a temperature range between 40 °C and 45 °C. However, the optimal temperature for biomass yield was lower than 37 °C. Previously in batch culture, Hujanen and Linko (1996) found optimal temperatures of 37 °C and 41 °C for production of lactic acid by some strains of L. casei, which would suggest the same e€ect of temperature. This behavior was con®rmed in a twostep culture process, where the biomass was eciently produced in the ®rst fermenter and the lactic acid concentration of the second fermenter was enhanced from 18.6 g l)1 to 48.5 g l)1 in the chemostat system and from 22 g l)1 to 57.6 g l)1 in packed-bed reactors. The initial results from a single-stage continuous fermentation produced a maximum of 18.6 g/l lactate with free cells and 22 g l)1 with immobilized cells, with volumetric productivities of 3.2 g l)1 h)1 and 7.8 g l)1 h)1 in the steady states, showing high glucose utilization while no signi®cant di€erences were observed in yield. Furthermore, in the two-stage reactors, a higher concentration of lactic acid was produced in both the free-cell (Table 4) and in the immobilized-cell (Table 5) reactors. Temperature and dilution rates did not a€ect the overall production of L(+)-lactic acid. However, the results showed that older cells, such as those present in the immobilized-cell reactors, produced L(+)-lactic acid less optically pure (93%) than that produced by younger free cells, such as those present in chemostat systems (96.8%). But one important parameter, the overall optical purity, increased by 2% in both systems assayed when operated in two stages. This would mean that the cells became older but the L(+)-lactate purity did not decrease; on the contrary, it ®ll because of the reduced amount of yeast extract, the apparent source of D())lactate (HjoÈrleifsdottir et al. 1990; Ferain et al. 1996). The high productivity values obtained with both chemostat and immobilized cells in two-stage systems go together with a reduced concentration of lactate, indicating that high concentrations of lactic acid could only be attained at the expense of a reduction of the lactic acid volumetric productivity, an e€ect previously shown by Park and Baratti (1992) for ethanol production. The results presented above allowed the selection of suitable conditions for high production rates of L(+)lactate in continuous fermentations of L. casei. Interestingly enough, when the steady states showing high glucose utilization are compared, it is possible to obtain a high lactic acid concentration by using di€erent operational schemes. If the parameter of interest is the volumetric productivity then this situation will lead to a drop in the resulting ®nal concentration of lactic acid; if the parameter of interest is the lactate concentration, then the

Glucose 40 g l)1

b

a

Glucose 26 g l)1 Glucose 92 g l)1 95 g l)1 Glucose 100 g l)1

Single packed bed Y. extract 10 g l)1 Two packed beds Overall Y. extract 5 g l)1 Packed-bed Y. extract, 15 g l)1

Values obtained from published results Values of productivity corresponding to the highest sugar conversion

96.8 89.5 58.7 75a

90a

Glucose 12 g l)1

Y. extract, 10 g l)1

Packed-bed

99.2a

Glucose 30 g l)1

Immobilized stirred tank

50

Fluidized-bed

59a

Lactosea 45 g l)1

22 57.6 37.4 51.4a

9.5a

28a

13

18.2a

48.5

100

Glucose 71 g l)1

Lactose 40 g l)1

Overall Y. extract 5 g l)1 Y. extract 10 g l)1, whey permeate 60 g l)1

Two chemostats

18.6 20.6a

29

33

25

55

Glucose 22 g l 100 Lactosea 45 g l)1 51a

99.2

Y. extract 5 g l)1; whey retenate 50 g l)1 Y. extract, 10 g l)1

Y. extract, 10 g l Y. extract 10 g l)1, whey permeate 60 g l)1

Single chemostat Single chemostat

Chemostat )1

70

Lactose 40 g l)1

Y. extract 5 g l)1; whey retenate 50 g l)1 Y. extract, 30 g l)1 )1

100

Glucose 30 g l)1

Y. extract, 10 g l)1

Immobilized alginate (batch) Continuous Chemostat

89

Glucose 80 g l)1

Y. extract, 10 g l)1

Sugar Lactate utilization (%) (g l)1)

Batch Repeated batch bio®lm reactors

Sugar (g l)1)

Nitrogen source

System

Table 6 Lactic acid production by di€erents systems and operational conditions. Y. yeast

7.8b (D = 0.36 h)1) 5.76b (D = 0.15 h)1) 9.72 (D = 0.40 h)1) 20.1 (D = 0.39 h)1)

1.17 (D = 0.11 h)1)

5.2 (D = 0.18 h)1)

13.5 (D = 1.0 h)1)

7.64 (D = 0.42 h)1)

2.42b (D = 0.06 h)1)

3.2 (D = 0.17 h ) 8.27 (D = 0.4 h)1)

)1

14.43 (D = 0.467 h)1)

5.5 (D = 0.22 h)1)

1.60

0.75

Lactate productivity (g l)1 h)1)

L. casei L. casei L. delbrueckii

Not reported

L. casei

L. casei

L. casei

L. helveticus

L. casei

L. casei L. helveticus

L. casei

L. casei

Streptomyces viridosporus L. casei L. casei

Organism

91 93

Not reported

Not reported

Not reported

Not reported

96.8

95 55±70

97

Not reported

95±97

Not reported

L(+)-lactic acid (%)

GoncËalves et al. (1992)

Guoqiang et al. (1992) This work This work

Aeschlimann and Von Stockar (1990) Krishke et al. (1991) Guoqiang et al. (1992)

Krishke et al. (1991) Gonzalez Vara (1996) This work Aeschlimann and Von Stockar (1990) This work

Guoqiang et al. (1991)

Demirci and Pometo (1995)

Reference

323

324

highest concentrations of lactic acid correspond to the lowest values for volumetric productivity. If the parameter of interest is the overall amount of the L(+)lactic acid isomer, then one should use the cultivation of free cells because of the high percentage of L(+)-lactate obtained (96.8% purity), indicating the optimal purity of the culture broth. Table 6 allows the data on lactic acid fermentation, under optimal conditions, by the two-stage two-stream systems to be compared with some earlier reports of di€erent processes. As can be seen from the table, it is clear that the immobilized systems produced results comparable with the results of immobilized reactors reported here when these results are analyzed in terms of medium enrichment (amounts of nitrogen source). The amounts of nitrogen source supplied in the present twostage systems, for chemostat and packed-bed reactors, which were accompanied by substantially high substrate consumption and lactate concentration, are comparable with the high values reported earlier; however, most of the continuous fermentations with very high productivities reported in the literature are usually accompanied by high amounts of the nitrogen source, so a reduction of yeast extract concentration in the feed, as is the case for our results, would signi®cantly lower the total fermentation costs. In conclusion, the results obtained in this study suggest that the yeast extract concentration can be maintained at a low overall level, 5 g l)1, during fermentation in order to obtain high productivity with high lactate concentration in two-step cascade reactors, decreasing the production costs through the use of a simpler medium, and with fewer diculties in downstream processing because of the reduced concentration of yeast extract and the subsequent reduction of the amounts of impurities. Acknowledgements This work was supported by grants from the Swedish Agency for Research Cooperation with Developing Countries (SAREC), Consejo Nacional de Investigaciones Cientõ ®cas y TeÂcnicas, Argentina (CONICET) and Agencia EspanÄola de CooperacioÂn Iberoamericana, EspanÄa (AECI). The authors want to thank Dr. Ricardo Fitzsimons for the HPLC analyses and E. M. Marin for skilled technical assistance.

References Aeschlimann A, Stockar V von (1990) The e€ect of yeast extract supplementation on the production of lactic acid from whey permeate by Lactobacillus helveticus. Appl Microbiol Biotechnol 32: 398±402 Audet P, Paquin C, Lacroix (1988) Immobilized growing lactic acid bacteria with j-carrageenan-locust bean gel. Appl Microbiol Biotechnol 29: 11±18 Boyaval P, Goulet J (1988) Optimal conditions for production of lactic acid from cheese whey permeate by Ca-alginate entrapped Lactobacillus helveticus. Enzyme Microb Technol 10: 725±728 Boyaval P, Corre C, Terre S (1987) Continuous lactic acid fermentation with concentrated product recovery by ultra®ltration and electrodialysis. Biotechnol Lett 9: 207±212 Cassidy MB, Lee H, Trevors JT (1996) Environmental applications of immobilized microbial cells: a review. J Ind Microbiol 16: 79±101

Demirci A, Pometto III AL (1995) Repeated-batch fermentation in bio®lm reactors with plastic-composite supports for lactic acid production. Appl Microbiol Biotechnol 45: 585±589 Demirci A, Pometto III AL, Johnson KE (1993) Lactic acid production in a mixed-culture bio®lm reactor. Appl Environ Microbiol 59: 203±207 Ferain T, Hobbs JN, Richardson J, Bernard N, Garmyn D, Hols P, Allen EN, Delcour J (1996). Knockout of the two ldh genes has a major impact on peptidoglycan precursor syntesis in Lactobacillus plantarum. J Bacteriol 178: 5431±5437 Friedman MR, Gaden EL (1970) Growth and acid production by Lactobacillus delbrueckii in a dialysis culture system. Biotechnol Bioeng 12: 961±974 Gawehn K, Bergmeyer HU (1974) D())-Lactate. Determination with lactate dehydrogenase and NAD. In: Bergmeyer HU, Gawehn K (eds) Methods of enzymatic analysis, 2nd edn. Academic Press, New York, pp 1469±1475. GoncËalves LMD, Barreto MTO, Xavier AMBR, Carrondo MJT, Klein J (1992) Inert supports for lactic acid fermentation a technological assessment. Appl Microbiol Biotechnol 38: 305± 311 Gonzalez Vara A, Pinelli D, Rossi M, Fajner D, Magelli F, Matteuzzi D (1996) Production of L(+) and D()) Lactic acid isomers by Lactobacillus casei subsp casei DSM 20011 and Lactobacillus coryniformis subsp lorquens DSM 20004 in continuous fermentation. J Ferm Bioeng 81: 548±552 Guoqiang D, Kaul R, Mattiasson B (1991) Evaluation of alginateimmobilized Lactobacillus casei for lactate production. Appl Microbiol Biotechnol 36: 306±314 Guoqiang D, Kaul R, Mattiasson B (1992) Immobilization of Lactobacillus casei cells to ceramic material pretreated with polyethyleneimine. Appl Microbiol Biotechnol 37: 305±310 Gutmann I, Walhlefeld AW (1974) L(+)-Lactate. Determination with lactate dehydrogenase and NAD. In: Bergmeyer HU, Gawehn K (eds) Methods of enzymatic analysis, 2nd edn. Academic Press, New York, pp 1464±1468 HjoÈrleifsdottir S, Seevaratnam S, Holst O, Mattiasson B (1990) E€ects of complete cell recycling on product formation by Lactobacillus casei ssp. rhamnosus in continuous cultures. Curr Microbiol 20: 287±292 Hujanen M, Linko YY (1996) E€ect of temperature and various nitrogen sources on L(+)-lactic acid production by Lactobacillus casei. Appl Microbiol Biotechnol 45: 307±313 Krischke W, SchroÈder M, TroÈsch W (1991) Continuous production of L-lactic acid from whey permeate by immobilized Lactobacillus casei subsp. casei. Appl Microbiol Biotechnol 34: 573±578 Lipinsky ES (1981) Chemicals from biomass: petrochemical substitution options. Science 212: 1465±1471 Motarjemi Y, Nout MJR (1996) Food fermentation: a safety and nutritional assessment. Bulletin of The World Health Organization 74: 553±559 Park SC, Baratti JC (1992) Continuous ethanol production from sugar beet molasses using an osmotolerant mutant strain of Zymomonas mobilis. J. Ferment Bioeng 73: 12±21 Ragout AL, SinÄeriz F, Kaul R, Guoqiang D, Mattiasson B (1996) Selection of an adhesive phenotype of Streptococcus salivarius subsp. thermophilus for use in ®xed-bed reactors. Appl Microbiol Biotechnol 46: 126±131 VahvaselkaÈ MI, Linko P (1987) Lactic acid fermentation in milk ultralitrate by Lactobacillus helveticus. In: Neyssel OM, van der Meer RP, Huyben KC (eds) Proceedings of the 8th European Congress of Biotechnology. vol 3, Elsevier, Amsterdam pp. 317±320 Vick Roy TB (1985) Lactic acid. In: Moo Young M (ed) Comprehensive biotechnology. Pergamon, Oxford, p 761 Vick Roy TB, Blanch HW, Wilke CR (1982) Lactic acid production by Lactobacillus delbrueckii in a hollow ®ber fermenter. Biotechnol Lett 4: 483±488 Vick Roy TB, Mandel DK, Dea DK, Blanch HW, Wilke CR (1983) The application of cell recycle to continuous fermentative lactic acid production. Biotechnol Lett 5: 665±670