INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Lactococcus lactis plays an important role in dairy fermentations, mainly in the production of cheeses. During such fermentation processes, lactose is present at very high concentrations (50 g/liter) and is converted through glycolysis to primarily form lactic acid as well as minor amounts of other compounds (homolactic fermentation). The resulting low pH contributes to the texture and flavor of cheeses and inhibits the growth of other bacterial species. During homolactic fermentation more than 90% of the lactose consumed is recovered as by-products (42), which shows that the glycolytic pathway functions almost exclusively as a catabolic pathway to supply ATP to the cells. When sugar becomes limiting for growth, or in the presence of a less readily metabolized carbon source, the pattern of product formation switches from homolactic to mixed-acid fermentation, i.e., to the formation of formate, ethanol, and acetate with smaller amounts of lactate (43).

The mechanisms responsible for regulation of glycolytic flux and the shift between different fermentation modes in L. lactis have been studied intensively. The concentrations of intermediary metabolites and cofactors are affected by the external sugar concentration (8, 10, 37, 46). In the presence of excess sugar, the concentrations of fructose-1,6-bisphosphate, the triose-phosphates, and pyruvate and the NADH/NAD⁺ ratio are high, whereas the concentrations of phosphoenolpyruvate and inorganic phosphate are relatively low. The glycolytic flux was proposed to be regulated through the level of fructose-1,6-bisphosphate (43), which is known to activate both pyruvate kinase and lactate dehydrogenase. In contrast, when sugar is limiting, the level of fructose-1,6-bisphosphate decreases and the level of inorganic phosphate increases (8, 37, 46), which results in decreased pyruvate kinase activity and increased phosphoenolpyruvate concentration (28, 45, 49). Finally, phosphoenolpyruvate inhibits the activity of phosphofructokinase (7), which provides a mechanism for regulating the glycolytic flux.

Work has been performed to determine the factors that control the flux through glycolysis by applying metabolic control analysis (13, 22). Poolman et al. (35) showed that the activity of glyceraldehyde-3-phosphate dehydrogenase and phosphoglycerate mutase decreased 5- to 15-fold during lactose starvation. Based on inhibitor titration results, they suggested that glyceraldehyde-3-phosphate dehydrogenase had a large amount of control over the glycolytic flux during nongrowing conditions. However, it is still unclear whether the enzyme is therefore probably less important when the flux through glycolysis is high.

In L. lactis the genes pfk, pyk, and ldh, encoding the glycolytic enzymes phosphofructokinase, pyruvate kinase, and lactate dehydrogenase, respectively, are clustered in the las operon (25). This unique organization could in principle allow for coordinated regulation of the expression of these genes, and it was suggested that expression of the las operon might be involved in regulating energy production and lactate flux in L. lactis (5). Indeed, a fivefold up-regulation of expression of the las operon was recently demonstrated at the onset of the stationary growth phase (1). In another study it was suggested that the las operon might be activated by the CcpA protein (26). The physiological role of regulating expression of the las operon is not clear, and it is unknown whether a coordinated expression of glycolytic genes might be important.

In this study we constructed mutants of L. lactis in which expression of the first gene of the las operon, pfk, was altered. We show that a 50% reduction in the expression of pfk dramatically affects the physiology of L. lactis in an unexpected manner, an effect which appears to be caused by the accumulation of upstream metabolites.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. The bacterial strains and plasmids used in this study are listed in Table 1.

Growth media and growth conditions. Escherichia coli strains were grown aerobically at 37°C in Luria-Bertani broth (36). L. lactis strains were cultivated as batch cultures (flasks) without aeration in M17 broth (40) or defined SA medium (17). The cultures were supplemented with 0.2 or 1% glucose, 1% fructose, 2% mannose, or 2% maltose and incubated at 30°C. A slow stir with magnets was used in order to keep the culture homogenous. Growth experiments were also performed in the Biostat Q fermentor system (B. Braun Biotech International, Melsungen, Germany) as either batch or glucose-limited chemostat cultures. L. lactis strains were routinely cultivated in batch fermentation in 1-liter vessels with a working volume of 600 ml of SA medium supplemented with 0.25% glucose. The temperature was set to 30°C, and the cultures were stirred at 200 rpm. The pH was maintained at 6.8 by the addition of 2 M NaOH. The medium in the fermentor was inoculated with an exponentially growing preculture to an initial optical density at 600 nm (OD₆₀₀) ranging from 0.02 to 0.04. Sampling for measurements of biomass, fermentation products, and enzyme activities were performed by pipetting. The cell density was correlated to the cell mass of L. lactis to be 0.36 g (dry weight)/liter of SA medium at an OD₆₀₀ of 1 (32).

Chemostat cultures. The medium used for the chemostat cultures was M17 with a glucose concentration of 0.25%.

Nongrowing cultures. Precultures were grown as batch cultures in 100 ml of SA medium supplemented with 0.2% glucose. In the exponential growth phase at an OD₆₀₀ of 0.5, the cultures were cooled on ice and harvested by centrifugation. The cultures were resuspended in buffer (SA medium without amino acids and vitamins) supplemented with 0.2% glucose to a final OD₆₀₀ of 0.4.

Excision of integrated plasmids. Counterselection was used to isolate las mutants in which the erythromycin resistance plasmid (pHWA182 or pHWA183) was crossed out of the chromosome as follows. A culture was inoculated to an OD₆₀₀ of 0.03 in M17 medium supplemented with 1% glucose without antibiotic selection for the plasmid. At an OD₆₀₀ of 0.150, erythromycin (2 µg/ml) was added, and at an OD₆₀₀ of 0.300, ampicillin (100 µg/ml) was added to the culture. The culture was incubated 15 h before the cells were removed, washed, and diluted on M17 plates. The colonies were tested by replica plating with or without erythromycin. Clones which had lost erythromycin resistance were verified by Southern blot analysis (39) and DNA sequencing of the promoter area upstream of the pfk gene.

Antibiotics. Antibiotics were used at the following concentrations: ampicillin, 100 µg/ml (selection of pBluescript II SK derivatives, pUC7erm, and pMW40 in E. coli); erythromycin, 2 µg/ml (selection for maintenance of pMU2916 and for the recombination event of pHWA182 and pHWA183 in L. lactis) and 200 µg/ml (selection of pAK80 derivatives and pMU2916 in E. coli); chloramphenicol, 50 µg/ml (selection of pCI372 derivatives); and tetracycline, 10 µg/ml (selection of E. coli strain ABLE K).

DNA techniques, DNA isolation, and DNA sequence analysis. All manipulations with recombinant DNA were carried out according to standard procedures as described in reference 36 unless otherwise stated. Chromosomal DNA from L. lactis was purified as previously described (21) with the following modifications. A final concentration of 15.5 mg of lysozyme/ml for 30 min was used for lysis, and the sample was incubated with sodium dodecyl sulfate for 10 min at 37°C followed by 10 min at 65°C. Plasmid DNA from E. coli for analytic purposes was isolated by using an alkaline lysis method and for preparative purposes by using a plasmid kit (Qiagen, Hilden, Germany). Restriction enzymes, T4 DNA ligase, and DNA polymerase Klenow fragment were purchased from Pharmacia Biotech and used as specified by the manufacturer. The restriction enzymes BsaAI and NaeI were purchased from New England Biolabs. Linearized cloning vectors were treated with shrimp alkaline phosphatase (Amersham Life Science) to avoid religation of the vector. PCR products for cloning purposes were amplified using Pfu DNA polymerase (Stratagene), and the verification of plasmid or chromosomal constructs by PCR was performed with Taq DNA polymerase (Pharmacia Biotech).

Rapid isolation of chromosomal DNA from L. lactis. A rapid method for isolating high quality chromosomal DNA was necessary in the screening of the las mutants. The procedure is a modification of the method described by Jensen and Michelsen (19). A sample of 0.9 ml of culture was added to a 2-ml Eppendorf tube containing 0.9 ml of phenol (preheated to 80°C) and 0.6 g of fine glass beads (Sigma) and the mixture was vortexed for 10 s. The sample was placed at room temperature for 30 min. Cell debris was removed by centrifugation, and the water phase was extracted with 0.8 ml of chloroform. Subsequently, the DNA was precipitated by isopropanol precipitation (36).

Transformation. L. lactis was transformed by electroporation as previously described (15) and plated on SR plates (31). E. coli was made competent using CaCl₂ (36) or made electrocompetent with 10% (vol/vol) glycerol.

Plasmid constructions. The plasmids constructed in this research are listed in Table 1. The P_las promoter region and the complete coding region of pfk were cloned from pMU2916 into pRC1 by digestion with EcoRI, resulting in pHWA58. This plasmid was digested with SspI and EcoRI (blunted), and the coding region of pfk was transferred to pUC18 digested with SmaI, resulting in pHWA71. Due to a mutation in the 3' end of pfk in pMU2916, this DNA region in pHWA71 was replaced by a nonmutated DNA region from MG1363, resulting in pMW40 (kindly provided by M. Willemoës). The region upstream of the las operon in MG1363 was obtained by inverse PCR (36). The PCR product was digested with BsaAI giving a 1.071-kb fragment, which was cloned in pCI372 digested with SmaI, resulting in pHWA134. Subsequently, the upstream fragment was transferred from pHWA134 to pBluescript II SK by digestion with BamHI and KpnI, resulting in pHWA149. The DNA region upstream of pfk and a truncated pfk gene (870 bp) were combined by ligating pMW40 and pHWA149 digested with SacI, resulting in pHWA151. In this parental plasmid the synthetic promoters were inserted. Plasmids pCP25 and pCP29 were used as templates to amplify promoter-containing fragments with the primers pAK8/erm (5'TTTCAACTGCCTGGCAC-3') and pAK80 (5'-TCCTTTCAAAGTTACCC-3'). The products were digested with BamHI and XbaI, yielding fragments of 185 and 171 bp. These were inserted into pHWA151 digested with the same restriction enzymes, resulting in pHWA157 and pHWA158. Finally, the erythromycin resistance gene located on a PvuII fragment (from pUC7erm) was inserted in the NaeI site of pHWA157 and pHWA158, resulting in pHWA182 and pHWA183, respectively.

Quantification of glucose and fermentation products by HPLC. Samples of 2 ml were withdrawn from the cultures at different time intervals, filtered through a 0.22-µm filter, and stored at 20°C until quantification of glucose, pyruvate, lactate, formate, acetoin, acetate, and ethanol. The high-pressure liquid chromatography (HPLC) equipment used was from Shimadzu Corporation, Kyoto, Japan, and the program Class VP 5.0 controlled the system. Products were measured as reported in reference 16 with the following modifications. Separation was performed with a Supelcogel C-610H column, 30 cm by 7.8 mm (Supelco, Inc., Bellafonte, Pa.). Supelcoguard C-610H, 5 cm by 4.5 mm, was used as a guard column. The columns were thermostated at 30°C. The mobile phase consisted of 5 mM H₂SO₄, and the flow rate was 0.5 ml/min. All products were detected on the refractive index detector, RID-10A. Pyruvate was detected on the diode array detector at 208 nm. SA medium contains a large amount of acetate, which was subtracted in the calculation of the amount of acetate produced. Detection of glucose by the refractive index detector is disturbed by the presence of pyruvate, which has the same retention time as glucose. This was corrected by measuring the amount of pyruvate in the sample at 208 nm. The catabolic carbon balance was calculated as the recovery of substrates converted into products in terms of molar concentrations (C-moles).

Enzymatic determination of sugar-phosphates. Extracts were prepared from batch cultures at an OD₆₀₀ of 0.3 to 0.9, using quenching in hot phenol; see "Rapid isolation of chromosomal DNA from L. lactis," above. After chloroform extraction, the concentration of sugar-phosphates was measured by recording the increase in NADH fluorescense (48) with the following modification. The buffer contained 50 mM triethanolamine, pH 7.5, instead of imidazolehydrochloride. Glucose-6-phosphate dehydrogenase from yeast was obtained from Boehringer Mannheim GmbH, Mannheim, Germany, and phosphoglucose-isomerase was obtained from Roche Diagnostics GmbH, Mannheim, Germany. Intracellular concentrations of sugar-phosphates (44) were calculated by assuming that 1 g (dry weight) corresponds to 1.67 ml of intracellular volume and an OD₆₀₀ of 1 corresponds to 0.36 g (dry weight) per liter (32).

Measurement of phosphofructokinase, pyruvate kinase, and lactate dehydrogenase activities. Enzyme activities were measured in permeable cells using a method analogous to the standard procedure used to determine -galactosidase activity (29) with the following modifications. Cells from a 5-ml culture at an OD₆₀₀ of 0.5 were harvested and washed twice in 0.2% KCl. The cells were resuspended in 0.5 ml of extract buffer (45 mM Tris-15 mM tricarballylate buffer, pH 7.5, containing 20% glycerol, 4.5 mM MgCl₂, and 1 mM dithiothreitol [10]). The cells were made permeable by adding 5 µl of 0.1% sodium dodecyl sulfate and 12.5 µl of chloroform and vortexing the mixture for 10 s. The cell extract was diluted in extract buffer and used for assaying phosphofructokinase, pyruvate kinase, and lactate dehydrogenase. The activity was determined from the rate of NADH oxidation at A₃₄₀ at 28°C using a Specord M500 spectrophotometer (Zeiss, Zena, Germany). Phosphofructokinase was assayed according to reference 7 with the following modifications. The final concentrations of the components of the reaction mixture were 87 mM Tris-HCl (pH 8.0), 1 mM ATP, 1 mM fructose-6-phosphate, 10 mM MgCl₂, 10 mM NH₄Cl, and 0.2 mM NADH; a mixture of 0.3 U of triosephosphase isomerase, 1 U of glycerol-3-phosphate dehydrogenase, 0.3 U of aldolase, and 1 mM ATP was used to initiate the reaction. The pyruvate kinase activity was determined as described in reference 4. The final concentrations of the components of the reaction mixture were 40 mM triethanolamine-64 mM KCl buffer (pH 7.5), 10 mM MgCl₂, 1 mM fructose-1,6-bisphosphate, 1 mM GDP, 0.2 mM NADH, and 6.3 U of lactate dehydrogenase; 1 mM phosphoenolpyruvate was used to initiate the reaction. Lactate dehydrogenase was measured by using a modified procedure described in reference 3. The final concentrations of the components of the reaction mixture were 50 mM triethanolamine (pH 7.5), 1 mM fructose-1,6-bisphosphate, and 0.2 mM NADH; 10 mM sodium pyruvate was used to initiate the reaction.

Nucleotide sequence accession number. The sequence of the region upstream of the las operon and the 3' end of the nagA gene has been deposited in the NCBI data bank with accession no. AY007718.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of strains with altered expression of las genes. To alter the expression of the las operon, the native promoter P_las was replaced by synthetic promoters by double homologous recombination (Fig. 1). The region upstream of the las operon (Fig. 1A) from L. lactis subsp. cremoris MG1363 (11) was obtained by inverse PCR and shown to contain a gene with 34% homology to the nagA gene in E. coli (33) encoding N-acetylglucosamine-6-phosphate deacetylase. The nagA gene is also found upstream of the las operon in the closely related L. lactis strain IL1403 (2). To construct plasmids for gene replacement, a fragment encoding the N-terminal part of phosphofructokinase without the P_las promoter was obtained from pMW40 and cloned together with the upstream region in an E. coli vector, resulting in pHWA151. Subsequently, the synthetic promoters CP25 and CP29 were inserted in addition to an erythromycin resistance gene, erm, which resulted in pHWA182 and pHWA183. A schematic presentation of pHWA182 is shown in Fig. 1B.

View larger version (14K): [in this window] [in a new window]   FIG. 1.   Strategy used to alter expression of las genes. (A) Schematic presentation of the chromosomal las operon in MG1363 and of the DNA fragments which have been cloned in pHWA182. The bent arrow indicates the native promoter Plas of the operon. The heavy black line indicates the DNA fragment upstream of the operon. The truncated pfk region is indicated by the light gray line. (B) The synthetic promoter CP25 was flanked by the upstream region of the las operon and the truncated region of pfk on the E. coli vector pHWA182. (C) After selection for erythromycin resistance and subsequent counterselection with ampicillin, the native promoter Plas was replaced by CP25 by double homologous recombination between pHWA182 and MG1363, resulting in HWA217.

las mutants have uncoordinated expression of pfk, pyk, and ldh relative to the wild type. To determine the expression levels of the las genes in the las mutants, the phosphofructokinase, pyruvate kinase, and lactate dehydrogenase activities were measured (Table 2). HWA217 and HWA232 had 39 and 60% phosphofructokinase activity compared to MG1363, respectively. The activity of pyruvate kinase was slightly increased in HWA217 (122%) and close to the wild-type level in HWA232 (97%). The lactate dehydrogenase activity of HWA217 was 140%, and HWA232 had an activity near the wild-type level (104%). As seen in Table 2, the differences of activities of the las enzymes compared to those of MG1363 were more pronounced in strain HWA217. In strain HWA232, phosphofructokinase was the most strongly affected enzyme activity, whereas the activities of pyruvate kinase and lactate dehydrogenase were closer to their respective wild-type values.

Growth rate was strongly affected by altered expression of las genes. The growth of the two las mutants was observed in batch fermentation with glucose as a carbon source. A typical growth curve for cultures of MG1363, HWA217, and HWA232 in Fig. 2 shows that the las mutants had a lower growth rate than the wild-type strain MG1363. Strain HWA217 had a specific growth rate of 0.44 h¹, compared to 0.77 h¹ for MG1363. HWA232 grew slightly faster, with a specific growth rate of 0.54 h¹. Thus, the growth rates of HWA217 and HWA232 were reduced to 57 and 70% of the wild-type value. The yields of biomass per mole of glucose for all strains were approximately the same (Table 3).

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Growth curves of las mutants HWA217 and HWA232 and wild-type control strain MG1363. The cultures were grown as batch fermentation cultures in SA medium supplemented with 0.25% glucose, and pH was kept constant at 6.8 by the addition of 2 M NaOH.

Fermentation was mainly homolactic. The fact that growth rate of the las mutants was decreased while the biomass yield on glucose remained unchanged indicated that altered expression of the las genes decreased the glycolytic flux. To investigate the pattern of fermentation in these strains, the end product formation of the las mutants grown in batch cultures of SA medium supplemented with 0.25% glucose was determined (Fig. 3). The main product produced by all strains was lactate (85%). HWA232 and HWA217 produced slightly larger amounts of formate than did the wild-type strain and small amounts of acetate. Trace amounts of acetate are normally produced by the wild-type strain, but this was not detected in this experiment due to the high background level of acetate in the SA medium. Other end products such as acetoin and ethanol were not detected. The same pattern of product formation was seen with 1% glucose (data not shown).

View larger version (14K): [in this window] [in a new window]   FIG. 3.   End-product formation of las mutants. The products produced by the las mutants and MG1363 during batch fermentation were analyzed by HPLC. The strains displayed a homolactic fermentation pattern. Samples were withdrawn from cultures in the stationary growth phase with an OD600 of approximately 1.1, immediately after the glucose had been exhausted. The C-mole recovery ranged between 80 and 83%.

Metabolic fluxes in the las mutants. Samples for analysis by HPLC were withdrawn from a batch fermentation at different time points during growth (Fig. 2). The glycolytic flux was measured as the steady-state consumption rate of glucose. The lactate, formate, and acetate fluxes were measured as the steady-state production rate of the respective products (Table 3). The glycolytic flux was reduced to 62 and 76% of the wild-type level, and the lactate flux was reduced to 68 and 81% in HWA217 and HWA232, respectively. The formate flux was increased to 122% in HWA217 and to 111% in HWA232, with respect to MG1363, but still the flux through the pyruvate formate lyase branch accounted for less than 10% of the total flux to by-products.

Growth characteristics on various sugars. To investigate the reason for the altered physiology of the las mutants, the strains were grown as batch cultures (flasks) with different sugars (Table 4). As expected, the growth rate of the las mutants was reduced in SA medium supplemented with glucose. A similar phenomenon was seen with mannose, which utilizes the same uptake system as glucose, namely, PTS^Man (47).

1

1

1

Complementing the reduced activity of phosphofructokinase. Out of the three enzymes encoded by the las operon, the activity of phosphofructokinase differed most in the las mutants and we therefore attempted to restore the phosphofructokinase activity by complementation. Strains MG1363, HWA217, and HWA232 were transformed with pMU2916, which carries the pfk gene transcribed from the native P_las promoter, resulting in strains HWA254, HWA256, and HWA258, respectively. These strains were grown as batch cultures in SA medium supplemented with 1% glucose, and the activities of phosphofructokinase and the growth rates were determined (Table 5). The pfk gene was overexpressed six- to ninefold in HWA254, HWA256, and HWA258, compared to the wild-type MG1363, and the strains had similar growth rates, close to 0.60 h¹, which was somewhat lower than the usual growth rate observed for MG1363 without any plasmid (Table 5). These experiments were performed in the presence of erythromycin in order to select for the plasmid pMU2916. The reduced growth rate of the transformed strains could be due either to high expression of the pfk gene or to the presence of erythromycin. Indeed, when grown without selection for erythromycin resistance, the wild-type strain carrying pMU2916 (HWA254) had a growth rate of 0.71 h¹ and the las mutants HWA256 and HWA258 had growth rates of 0.70 h¹ and 0.71 h¹, respectively. Thus, there is a significant increase in the growth rates of the mutants when the pfk gene is expressed in trans compared to growth rates of 0.46 and 0.56 h¹ without the plasmid. These observations, together with the results concerning growth on various sugars, indicate that the low activity of phosphofructokinase was the primary cause of slower growth and glycolytic flux in HWA217 and HWA232.

Growth was restored by limiting the uptake of glucose. An explanation for the changed phenotype of the las mutants could be that the low activity of phosphofructokinase resulted in an accumulation of sugar-phosphates, i.e., glucose-6-phosphate or fructose-6-phosphate, which are toxic at elevated levels (9). If this was the case it might be possible to increase the growth rate by limiting the external concentration of glucose. This hypothesis was tested in a chemostat experiment with a limited amount of glucose.

1

1

1

1

1

View larger version (9K): [in this window] [in a new window]   FIG. 4.   Growth of las mutant HWA217 and MG1363 in chemostat culture at various dilution rates. Biomass concentration of the cultures is shown during the experiments as a function of the dilution rate. Growth was initiated in a batch fermentation. Before the cultures entered the stationary growth phase, the chemostat was inoculated at a dilution rate of 0.40 h1. The experiments in which the dilution rate was increased rapidly (experiment A) or slowly (experiment B) are shown with closed or open symbols, respectively. The biomass concentrations of MG1363 and HWA217 are shown as squares and circles, respectively. The solid line indicates the biomass concentration of HWA217 when the dilution rate was increased slowly. This line also shows the biomass concentration of MG1363 in both cases. The broken line indicates the biomass concentration of HWA217 when the dilution rate was increased rapidly. The cultures were grown in M17 supplemented with 0.25% glucose, and pH was kept constant at 6.8 by the addition of 2 M NaOH.

Sugar-phosphates accumulate in las mutants with low phosphofructokinase activity. Several experiments here indicated that the inhibition of growth in the las mutants is due to the accumulation of sugar-phosphates upstream of phosphofructokinase. We measured the intracellular concentration of these metabolites in cells grown as batch cultures (flasks) (Table 6). Indeed, the concentrations of glucose-6-phosphate were found to be increased to 37 and 55 mM in HWA217 and HWA232, respectively, compared to 16 mM in the wild-type strain MG1363. Fructose-6-phosphate was also increased to 3.9 and 6.2 mM in strains HWA217 and HWA232, respectively, compared to 1.5 mM in MG1363.

Glycolytic flux is unaffected in nongrowing cells. If the accumulation of sugar-phosphates is responsible for the inhibition of growth rate and glycolytic flux in the mutants with low phosphofructokinase activity, then the glycolytic flux might be less affected in resting cells where glycolysis is uncoupled from growth. We therefore measured the glycolytic flux of the wild-type and mutant cells resuspended in buffer (Table 7). Homolactic fermentation continued in the cells, and the rates of glucose consumption and lactate production were almost identical to the rates observed for the wild-type strain.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The original aim of this study was to construct strains with modulated expression of the las operon and to estimate the control on the glycolytic flux by the corresponding three enzymes, using the approach of metabolic control analysis (13, 22). Unfortunately, the relative changes in gene expression in the obtained las mutants were not the same for the individual las genes pfk, pyk, and ldh, and the mutants are therefore not suited for such an analysis. Instead, the strains allowed us to investigate the importance of changing the expression of the pfk gene. We found that moderately reduced phosphofructokinase activity has a strong impact on L. lactis physiology; these changes appear to be due to changes in the concentration of upstream sugar-phosphates.

In both las mutants, the phosphofructokinase activity was reduced approximately twofold. Sequencing the pfk region in the las mutants excluded the possibility that mutations had been introduced in the pfk gene. Instead it is likely that the relative functional half-life of the mRNA for each of the three genes had been altered. The original mRNA of the las operon was not conserved in the mutants because 24 bp were missing in the leader region upstream of pfk. This may have affected the stability of mRNA in this region and lowered the functional half-life of the pfk message compared to the other two genes. The slightly altered pyruvate kinase and lactate dehydrogenase activities in one of the two mutant strains can also be explained from small variations in the leader mRNA sequences conferred by the two different synthetic promoter fragments employed.

It is unlikely that the slightly elevated expression of ldh could be the cause of the altered physiology since much larger variations in lactate dehydrogenase activity were found to have no influence on L. lactis growth rate or glycolytic flux (1; H. W. Andersen, M. B. Pedersen, K. Hammer, and P. R. Jensen, unpublished data). In both mutant strains the activity of pyruvate kinase was quite close to the wild-type level, and it is therefore unlikely that this enzyme was the cause of the observed growth defects. Instead, three lines of evidence indicated that the altered phosphofructokinase activity was the direct cause of the changed physiology: (i) the mutants grew like the wild-type strain in medium containing fructose, which bypasses the use of phosphofructokinase; and (ii) overexpressing the phosphofructokinase activity in trans restored the growth rate to the wild-type rate in the las mutants.

When the glycolytic flux in Lactococcus is reduced, the pattern of fermentation shifts to mixed-acid production, which in turn results in an increased supply of ATP (6, 8, 34). In our las mutants, phosphofructokinase appears to limit the glycolytic flux to a similar extent and the mutants might therefore be expected to show a shift to mixed-acid fermentation. However, it was found that the las mutants were mainly homolactic and probably not energy limited. What causes the lower growth rate of these mutants? In principle, the decrease in phosphofructokinase activity may be limiting for anabolic reactions. However, in complex medium the growth rates of the las mutants were even further reduced relative to the wild-type strain.

Another consequence of reduced phosphofructokinase activity could be an accumulation of upstream metabolites, i.e., glucose-6-phosphate. Glucose-6-phosphate has been reported to be toxic at enhanced levels (9). The concentration of glucose-6-phosphate was found to be increased from 2.2- to 3.4-fold in the two las mutants, and the fructose-6-phosphate concentration was increased 2.6- to 4.1-fold. Curiously, we find that the concentration of both sugar-phosphates is higher in the strain which has intermediary phosphofructokinase activity (60%) than in the strain with the lowest phosphofructokinase activity (39%). We suspect that this may be a consequence of slightly higher expression of pyruvate kinase and lactate dehydrogenase in the strain with the lowest phosphofructokinase activity. Both enzymes affect the phosphoenolpyruvate pool: the higher pyruvate kinase activity will directly lead to a lower phosphoenolpyruvate pool and the higher lactate dehydrogenase activity will lead to a lower pyruvate pool and therefore higher pyruvatekinase activity and lower phosphoenolpyruvate pool. A lower phosphoenolpyruvate pool will in turn affect the activity of the phosphoenolpyruvate:phosphotransferase system and the accumulation of the sugar-phosphates.

In the present study it was found that a sudden change in the dilution rate of mutant HWA217 in a chemostat culture resulted in culture washout. In contrast, when the dilution rate was increased slowly, washout of the las mutant was avoided. Under these conditions, the growth rate could be almost doubled compared to the growth rate in batch cultures. An explanation for this phenomenon is that an acceptable pool of sugar-phosphates in the las mutants was achieved by starving the cells for glucose and thereby eliminating a toxic effect. In Streptococcus mutans a toxic effect, which probably arose from some unidentified glycolytic intermediates, was also avoided by limiting the supply of sugar (14). The likely reason that the fast shift from low to high dilution rate did not result in fast growth was that the cells became trapped with high sugar-phosphate pools.

In cells resuspended in buffer we observed almost the same glycolytic flux of the two las mutants as of the wild-type strain. Glycolysis is uncoupled from growth under these conditions, and this result supports the hypothesis that the accumulation of sugar-phosphates inhibits the growth rate and thereby the glycolytic flux in the las mutants. This result also shows that under these specific conditions in nongrowing cells, phosphofructokinase does not control glycolysis.

The fact that a 40% reduction in the phosphofructokinase activity results in a three- to fourfold-higher concentration of the upstream metabolites indicates that the control exerted by phosphofructokinase on these metabolite concentrations is high. It also suggests that it is important for these cells to accurately balance the activity of the glycolytic enzymes, and the placement of the three genes together in the las operon is likely to be a reflection of this. By coregulating several glycolytic genes, the cells may gain an advantage with respect to regulating the glycolytic flux, because the concomitant changes in metabolite concentrations will tend to be smaller, in contrast to a situation in which only a single enzyme activity is altered. Indeed, computer modeling of the consequences of altering phosphofructokinase activity indicated a poor homeostasis of the glycolytic intermediates (41), which is consistent with the observations in the present study.

Does phosphofructokinase control growth rate and glycolytic flux in L. lactis? One of the main observations in this study is that the growth rate, the glycolytic flux, and the lactate flux were decreased proportionally by a twofold reduction of the activity of phosphofructokinase, which indicates that phosphofructokinase is not present in large excess in wild-type L. lactis cells. This result is surprising because enzymes are usually present in large excess. In E. coli, it was found that the H⁺-ATPase, which is essential for growth on succinate minimal medium, has no control over the growth rate of E. coli, and reducing this enzyme activity by one-half has almost no effect on the growth rate (20). In diploid organisms, mutations that lower the cellular enzyme activities by one-half are often found to be recessive with virtually no effect on cell physiology.