A chloride-inducible acid resistance mechanism in Lactococcus lactis and its regulation


  • Jan Willem Sanders,

    1. Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands.
    Search for more papers by this author
  • Kees Leenhouts,

    1. Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands.
    Search for more papers by this author
  • Jan Burghoorn,

    1. Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands.
    Search for more papers by this author
  • Jan Roel Brands,

    1. Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands.
    Search for more papers by this author
  • Gerard Venema,

    1. Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands.
    Search for more papers by this author
  • Jan Kok

    1. Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Kerklaan 30, 9751 NN Haren, The Netherlands.
    Search for more papers by this author

Gerard Venema, E-mail g.venema@biol.rug.nl; Tel. (50) 363 2093; Fax (50) 363 2348.


Previously, a promoter was identified in Lactococcus lactis that is specifically induced by chloride. Here, we describe the nucleotide sequence and functional analysis of two genes transcribed from this promoter, gadC and gadB. GadC is homologous to putative glutamate-γ-aminobutyrate antiporters of Escherichia coli and Shigella flexneri and contains 12 putative membrane-spanning domains. GadB shows similarity to glutamate decarboxylases. A L. lactis gadB mutant and a strain that is unable to express both gadB and gadC was more sensitive to low pH than the wild type when NaCl and glutamate were present. Expression of gadCB in L. lactis in the presence of chloride was increased when the culture pH was allowed to decrease to low levels by omitting buffer from the medium, while glutamate also stimulated gadCB expression. Apparently, these genes encode a glutamate-dependent acid resistance mechanism of L. lactis that is optimally active under conditions in which it is needed to maintain viability. Immediately upstream of the chloride-dependent gadCB promoter Pgad, a third gene encodes a protein (GadR) that is homologous to the activator Rgg from Streptococcus gordonii. gadR expression is chloride and glutamate independent. A gadR mutant did not produce the 3 kb gadCB mRNA that is found in wild-type cells in the presence of NaCl, indicating that GadR is an activator of the gadCB operon.


Lactococcus lactis is a Gram-positive mesophilic non-differentiating microorganism. It gains energy from the fermentation of sugars to (mainly) lactic acid, resulting in the acidification of its environment. This is one of the ways in which the organism can prevent food spoilage and one of the reasons why L. lactis is widely used in the dairy industry. Depending on sugar availability and the buffering capacity of the environment, lactococcal cells will spend a large part of their growth cycle at low pH. The question as to how L. lactis can survive such harsh conditions has excited increased interest in recent years. The limiting pH for growth in milk is 4.3 (Piard and Desmazeaud, 1991). Resting cells are much better adapted to low pH than actively dividing cells (Hartke et al., 1994). Acid resistance can be improved significantly by pre-exposure to mild (pH 5–6) acid conditions (Hartke et al., 1996). Similar results were obtained when HCl was used instead of lactic acid (Rallu et al., 1996). Similarly, pre-exposure to UV radiation can improve cell survival at low pH (Hartke et al., 1995). Enhanced survival at low pH induced by starvation or mild acid conditions points to the existence and the activation of acid adaptation mechanisms. This notion is supported by the recent observation of Rallu et al. (1996) that L. lactis MG1363 is very sensitive to low pH when protein synthesis is blocked during an adaptation step at mild acid conditions. Thirty-three acid-induced polypeptides were detected in a two-dimensional analysis of lactococcal cells exposed to pH 5.5 (Hartke et al., 1996). A major role in the maintenance of a constant cellular pH (pHi) is fulfilled by the F0F1-(H+)-ATPase (Kobayashi et al., 1986). L. lactis mutants with a reduced F0F1-ATPase activity were shown to be more acid sensitive (Tomita et al., 1996). The arginine deiminase pathway (ADI) was found to be active at very low pH. As one of the end products of this pathway is ammonia, ADI activity results in an increase in the extracellular pH (pHo) and, therefore, in enhanced survival of low-pH conditions (Marquis et al., 1987; Casiano-Colón and Marquis, 1988). Amino acid antiport in combination with amino acid decarboxylation can be seen as a more general system for pH homeostasis, as an acid is exchanged for a more basic amine and because, in addition, a proton is consumed intracellularly by the decarboxylation (Molenaar et al., 1993). In Escherichia coli and Shigella flexneri, a putative glutamate–γ-aminobutyrate (GABA) antiporter has been shown to confer acid resistance in media supplemented with glutamate (Hersh et al., 1996; Waterman and Small, 1996). The genes for these antiporters were linked to genes encoding glutamate decarboxylases. The E. coli Cad system consists of a lysine–cadaverine antiporter (CadB) and a lysine decarboxylase (CadA). Expression of CadAB is induced at low pH in the presence of lysine (Meng and Bennett, 1992). In a study of regulatory mechanisms involved in acid resistance in L. lactis, a number of acid-resistant mutants was found (Rallu et al., 1996). The identity of the affected genes suggests that a relationship exists between the stringent response control circuit and acid resistance and confirms the involvement of the ADI pathway in acid resistance.

In a previous study, a chloride-dependent chromosomal promoter of L. lactis was identified (Sanders et al., 1997) using a random lacZ integration strategy. One integrant expressed increased levels of lacZ with increasing Cl concentrations in the culture medium. Expression from this promoter was cation independent, and no expression was detectable in the presence of other anions or at a high osmotic pressure. Here, we report on the genetic analysis of the regions surrounding this promoter and show that a regulator is involved in Cl-dependent gene expression. The promoter drives the Cl-dependent expression of a putative amino acid antiporter and glutamate decarboxylase that together operate in glutamate-dependent acid stress resistance in L. lactis.


Nucleotide sequence of the chloride-inducible NS3 locus

In a previous study (Sanders et al., 1997), a chloride-dependent chromosomal promoter (Pgad) was identified in L. lactis using a lacZ integration strategy. The nucleotide sequence of the chromosomal region surrounding the lacZ integration site in the strain in question, L. lactis NS3, was determined, and the data are presented in Figs 1 and 2. Upstream of Pgad, an open reading frame (ORF) could encode a protein of 276 residues with a calculated molecular weight of 32 990. Its deduced amino acid sequence is homologous (25% identity and 18% similarity) to Rgg, a positive regulator involved in the expression of glucosyltransferase in Streptococcus gordonii (Sulavik et al., 1992; Sulavik and Clewell, 1996). This ORF was named gadR because of its regulatory function (see below). GadR also shows 19% identity with PlcR, a positive regulator of the Bacillus thuringiensis phosphatidylinositol-specific phospholipase C gene (Lereclus et al., 1996). Interestingly, gadR is highly similar, but not identical, to a previously identified ORF with unknown function associated with pip in L. lactis C2 (Geller et al., 1993). gadR is preceded by a possible ribosome-binding site and is immediately followed by a 21 bp inverted repeat (IR; ΔG[25°C] =−35.0 kcal mol−1; Tinoco et al., 1973; see Fig. 2). The putative product of a truncated ORF that stops 594 bp upstream of gadR showed 43% identity to the C-terminal end of RNaseHII of E. coli, an enzyme involved in the specific degradation of the ribonucleotide moiety of DNA/RNA hybrids (Itaya, 1990). L. lactis rnhB is followed by a 9 bp IR (ΔG[25°C] = −7.6 kcal mol−1) and a 8 bp IR (ΔG[25°C] =−10.6 kcal mol−1) that may function as rho-independent transcriptional terminators (Fig. 1). Two other weak inverted repeats are present in the long non-coding region between rnhB and gadR (not shown).

Figure 1.

. Gene organization in the NS3 locus. Lollipop, inverted repeat; P, promoter. Relevant restriction enzyme sites are indicated. The black arrows indicate the positions of lacZ fusions in the chromosome of the indicated strains (lacZ not drawn to scale).

Figure 2.

. Nucleotide sequence and expression signals of the intergenic regions in the NS3 locus. Basepair numbering is as in Fig. 1. Parts of the deduced amino acid sequences of the genes are given below the sequence. Facing arrows, inverted repeats; −10 and −35 promoter sequences are underlined and in boldface; vertical arrows, transcription start points; rbs, ribosome-binding site (nucleotides indicated in lower case). Stop codons are indicated with asterisks; start codons are in bold face. Relevant restriction enzyme sites are indicated above the sequence. The position to which primer NS3-12 anneals is double underlined.

The original Sau3A fusion site in the lacZ integrant L. lactis NS3 is located in an ORF of 503 codons immediately downstream of Pgad. This ORF was named gadC, as its deduced amino acid sequence (Fig. 3) is homologous to GadC of S. flexneri (51% identity and 17% similarity; Waterman and Small, 1996) and E. coli GadC (XasA; Hersh et al., 1996). The GadC proteins are homologous to a number of amino acid antiporters, including the lysine–cadaverine antiporter CadB of E. coli. Lactococcal GadC has a deduced molecular weight of 55 369, a pI of 9.73 and is highly hydrophobic. The hydrophobic residues are clustered in 12 domains (Fig. 3), whose locations coincide with those of the hydrophobic domains in S. flexneri GadC (as predicted by a number of topology-predicting computer programs). This suggests that GadC is an integral membrane protein. A conserved domain found in glutamate-transporting proteins is also present in L. lactis GadC (Fig. 3; Waterman and Small, 1996). Two ATG codons are present at the 5′ end of gadC. On the basis of the average spacing between ribosome-binding sites and start codons in L. lactis (4–12 bp; Van de Guchte et al., 1992), the second was tentatively indicated as the gadC start codon. gadC is separated by 19 bp from another ORF of 466 codons of which the deduced protein (with a molecular weight of 53 849 and a pI of 4.87) is homologous to glutamate decarboxylases. Highest homology is found with glutamate decarboxylase from Synechocystis spp. (48% identity and 15% similarity; Kaneko et al., 1996). It is also highly similar to the glutamate decarboxylases, GadA and GadB, of E. coli (43% and 44% identity respectively; Smith et al., 1992). In E. coli and S. flexneri, gadB is linked to the gadC genes. The lactococcal gene was therefore named gadB. Whereas in E. coli and S. flexneri the putative antiporter genes are preceded by gadB, the gene order in L. lactis is gadCB. No possible transcription signals could be identified in or near the 19 bp intergenic region between gadC and gadB, which suggests that they form an operon. Downstream of gadB, a 16 bp IR (ΔG[25°C] = −23 kcal mol−1) followed by a stretch of Ts may function as a rho-independent transcription terminator (Fig. 2). This IR is followed by an ORF in the opposite orientation, and it may therefore function as a dual transcription terminator of both gadCB and this ORF of unknown function.

Figure 3.

. Alignment of the deduced translation products of gadC from L. lactis (L.l. GadC) and S. flexneri (S.f. GadC). Asterisks, identical amino acids; periods, similar amino acids. Putative membrane-spanning domains are double underlined. Residues of a putative glutamate-binding box (Waterman and Small, 1996) are boxed.

Chromosomal map location

Two rare restriction enzyme sites, NotI and ApaI, are separated by only 2065 bp in the nucleotide sequence of the NS3 locus (Fig. 1). In the genetic map of L. lactis MG1363 (Le Bourgeois et al., 1995), there is only one position at which NotI and ApaI are that closely linked, namely between ldh and leu-ilv. Therefore, the gad operon is located between these two gene clusters.

GadR regulates the expression of gadCB positively

Its homology with known positive regulators and the close proximity of gadR to Pgad suggests that GadR is involved in the Cl-dependent expression of gadCB. To investigate this possibility, gadR was inactivated by single cross-over integration of pNS3i3, resulting in strain MGNS3i3. The effect on the transcription of gadC was analysed by Northern hybridization using the 5′ end of gadC as a probe. A 3 kb mRNA was detected only in RNA isolated from L. lactis MG1363 grown in the presence of NaCl (Fig. 4). No signal was found with RNA from MGNS3i3 grown in either the absence or the presence of NaCl, indicating that GadR is indispensable for transcription from Pgad. The size of the mRNA corresponded to the size of gadC plus gadB. The smaller hybridizing RNAs are probably breakdown products of the 3 kb transcript. The same pattern of hybridization was observed in blots using RNA from independent isolations. The 1.6 kb hybridizing band may be a transcript encoding gadC only. The faint bands seen in MGNS3i3 and in MG1363 without NaCl may be caused by background expression from Pgad or by aspecific hybridization. Alternatively, these bands may be an artifact of the Northern blot procedure related to the position of rRNA (Huang et al., 1992; Sulavik and Clewell, 1996).

Figure 4.

. Autoradiograph of a Northern blot of 30 μg of total RNA from L. lactis MG1363 and L. lactis MGNS3i3 (gadR ) grown in the absence (−) or presence (+) of 0.5 M NaCl. The probe was a [α-32P]-dCTP-labelled 450 bp XbaI–Sau3A fragment encoding the 5′ end of gadC. The sizes of an RNA marker are given in the left margin. The arrow indicates the position of the 3 kb gadCB transcript.

Expression of gadCB is enhanced at low pH and by glutamate

Expression of lactococcal gadC was studied in L. lactis NS3, the strain carrying a single-copy gadC ::lacZ fusion in its chromosome (Fig. 1). The experiment was done several times, and a typical example is given in Fig. 5. A modified M17 medium (mM17) was used, in which β-glycerophosphate was omitted to lower the buffering capacity of the medium. mM17 also lacked soytone. β-Galactosidase activity in NS3 was still induced in this modified medium in the presence of 0.3 M NaCl and 2% β-glycerophosphate buffer, albeit at a level fivefold lower than that in standard G1/2M17 (data not shown). In the experiments described in Fig. 5, NaCl (0.3 M) was present in all cultures. In the absence of buffer and depending on the presence of glutamate, the pH of the cultures dropped to values of 4.0–4.5 in the stationary phase, while cultures growing in buffered mM17 reached a pH of 5.5, irrespective of added glutamate. lacZ expression in buffered mM17 was low in the early stages of exponential growth and increased to an optimum in the late logarithmic phase of growth (Fig. 5A). Expression of lacZ was increased 10-fold in mM17 containing no buffer (Fig. 5B). The addition of 5 mM glutamate to cultures stimulated lacZ expression, but optimal expression was obtained with 50 mM glutamate. Glutamate at 50 mM resulted in a 1.5- to twofold additional increase in the expression of lacZ, both in unbuffered and in buffered mM17 broth (Fig. 5). No β-galactosidase activity was detectable in the absence of NaCl, whether or not glutamate was present and independent of the culture pH (data not shown). Therefore, expression of gadCB in batch culture is highest at the onset of the stationary growth phase in the presence of NaCl and glutamate, and at low pH.

Figure 5.

. pH- and glutamate-dependent β-galactosidase expression in L. lactis NS3 (gadC ::lacZ ). Strain NS3 was grown in mM17 medium with (A) or without (B) 2% β-glycerophosphate. β-Galactosidase activity (U mg−1) was followed during growth (fat solid lines) either in the presence of 0.3 M NaCl (▴) or in the presence of 0.3 M NaCl plus 50 mM glutamic acid (▪). Note that the scales of the left y-axes in A and B are different. The optical densities and pH values of the cultures are indicated with dotted and thin lines respectively.

gadR is constitutively expressed

Expression of gadR was studied in strain MGNS32 carrying a single-copy gadR ::lacZ transcriptional fusion (see Fig. 1) under the same growth conditions as used for NS3. β-Galactosidase activity in exponentially growing MGNS32 in buffered mM17 was 6.0 U mg−1 and decreased to 4.0 U mg−1 in the stationary growth phase. In the absence of buffer, 1.5 U mg−1 was measured during exponential growth, and this dropped to 0.5 U mg−1 in the stationary phase. β-Galactosidase activity was independent of the presence or absence of 0.3 M NaCl and was also the same in the presence and absence of 50 mM glutamate. In other words, NaCl-dependent and glutamate-dependent expression from Pgad is not regulated by variations in the level of transcription of gadR. Another lacZ fusion, located immediately downstream of the 21 bp IR, was used to show that transcription of gadR is effectively terminated by this IR. No β-galactosidase activity could be detected in strain MGNS31 carrying this fusion (see Fig. 1). Transcription of gadR starts from a G-residue 117 bp upstream of the AUG start codon (1, 6Figs 1 and 6). The gadR transcription start point is the same in the absence or presence of NaCl (data not shown). The gadR promoter consists of canonical −35 and −10 hexanucleotides separated by 18 bp (Fig. 2).

Figure 6.

. Primer extension using 5 μg of RNA isolated from L. lactis and primer NS3-12 (see Fig. 2). The sequence ladder was obtained using the same primer and plasmid pNS3d as a template. The −10 hexanucleotide of the promoter and the transcription start site of gadR are indicated in bold in the sequence shown on the left.

gadCB confers acid resistance

The ability of L. lactis MG1363 to survive acid stress was tested in the presence and in the absence of gadCB expression. When MG1363 was grown in 1/2M17 to an OD600 of 0.5 (pH 6.5) and acid challenged with lactic acid at pH 3.5 in MS15 medium for 2 h, the viability of the cells decreased dramatically (106-fold, see Table 1). Viability was almost unchanged upon incubation for 2 h in MS15 at pH 6.5. Lactic acid was much more deleterious than hydrochloric acid in this respect. The addition of 1 mM glutamate during challenge did not affect the viability at pH 3.5. When gadCB expression was induced during growth with 0.3 M NaCl, however, lactic acid challenge in the presence of 0.3 M NaCl and 1 mM glutamate reduced viability only 200-fold. Under the latter conditions, omission of glutamate reduced viability 106-fold to the same level as seen in cells in which gadCB was not expressed. As the level of gadCB expression is only slightly affected by the presence of glutamate (see Fig. 5), these results indicate that glutamate is essential for conferring acid resistance.

Table 1. . Acid resistance of L. lactis MG1363 under different conditions of growth and challenge.a Acid resistance was measured at pH 3.5 and at pH 6.5. To set the challenge medium to pH 3.5 either lactic acid or HCl was used. a. All cultures were pregrown in 1/2M17 broth with or without NaCl to an OD600 of 0.5, during which the pH did not drop below 6.5.Thumbnail image of

To confirm the direct involvement of gadCB in acid resistance, the gadR mutant, L. lactis MGNS3i3 (which does not express gadCB, see Fig. 4), was subjected to acid challenge. Survival of this strain in the presence of NaCl and glutamate was reduced to a level similar to that of the wild type in the absence of NaCl and glutamate (Table 2). An insertion mutant in gadC (strain MGNS3i4) and an insertion mutant in gadB (strain MGNS3i7) were even more acid sensitive in the presence of NaCl and glutamate. The former mutation has, most probably, a polar effect on gadB, which would result in the absence of both GadC and GadB. The acid resistance of MGNS3i4 and MGNS3i7 was also lower than the wild type in the absence of NaCl and glutamate. These data show that gadCB, which is transcribed from the chloride-dependent promoter Pgad, is involved in glutamate-dependent acid resistance in L. lactis.

Table 2. . Acid resistance of L. lactis gadCB mutants. a. At pH 3.5 with lactic acid.b. Cultures were pregrown to an OD600 of 0.5 in 1/2M17.c. Cultures were pregrown to an OD600 of 0.5 in 1/2M17 with 0.3 M NaCl and challenged in the presence of 0.3 M NaCl and 1 mM glu.d. Insertions in gadC will, most likely, have a polar effect on gadB.Thumbnail image of


A new type of acid resistance mechanism, recently described for S. flexneri and E. coli, has been found and characterized in L. lactis. It consists of a putative glutamate–GABA antiporter, GadC, and the glutamate decarboxylase, GadB. In a model by Waterman and Small (1996) for S. flexneri GadBC, the putative membrane protein, GadC, is proposed to be involved in the antiport of glutamate and GABA, while the glutamate decarboxylase GadB converts internalized glutamate to GABA with the simultaneous consumption of a proton and production of one molecule of CO2. According to this model, the net result of the combined action of GadBC is the removal of a proton from the cytosol, which would, thus, lead to an increase in the internal pH. In addition, extracellular glutamate is exchanged for the more alkaline GABA. An alternative function for gadCB may be the generation of metabolic energy. Recently, Higuchi et al. (1997) have shown the generation of ATP by glutamate metabolism in resting Lactobacillus cells. This was assumed to occur through a similar mechanism of coupled electrogenic antiport and decarboxylation reactions that generate a proton motive force. An increase in the internal pH and, thus, the formation of a proton motive force in Lactobacillus buchneri was shown to result from a similar mechanism, involving histidine decarboxylation and histidine/histamine antiport (Molenaar et al., 1993).

Mutants of L. lactis carrying either a disrupted gadC or gadB are acid sensitive, which indicates that GadCB confers acid resistance to this organism. This is in agreement with the acid sensitivity of E. coli and S. flexneri gadC mutants that were identified in random transposon mutagenesis studies. Remarkably, no gadB mutants were reported in these studies, despite the essential role of glutamate decarboxylation in the proposed acid resistance mechanism. In the case of E. coli, this can be explained by the presence of two homologous glutamate decarboxylases (Smith et al., 1992). An insertion in lactococcal gadB resulted in a severe loss of acid resistance, indicating that glutamate decarboxylation is an indispensable step in GadCB-mediated acid resistance in L. lactis and supporting the proposed model. These data also indicate that L. lactis has no second glutamate decarboxylase that is sufficiently active under the tested conditions to take over the role of GadB. We also presume that, in the lactococcal GadCB, the antiporter is essential to provide acid resistance, as interruption of the gene by Campbell integration resulted in acid sensitivity. However, as gadC and gadB are part of one operon, the integration in gadC might have a polar effect on the expression of gadB and thus confer an acid-sensitive phenotype. The GadB- and GadCB-negative mutants were more acid sensitive than the wild type, even in the absence of NaCl. This can be explained by a background level of gadCB expression in the wild type in 1/2M17 caused, most probably, by traces of chloride present in this medium (4 mM in 1/2M17, unpublished information). In addition, 1/2M17 contains some glutamate that could be processed by GadCB in the wild type. The observed difference in sensitivity to hydrochloric acid and lactic acid is a result of the permeability of the membrane for the protonated weak organic acid (Kashket, 1987; Cook and Russel, 1994).

Cells containing NaCl-induced GadC and GadB are still acid sensitive when glutamate is absent, indicating that glutamate is an essential component of the acid resistance mechanism in L. lactis. The same has been observed for GadBC-mediated acid resistance in E. coli and S. flexneri (Hersh et al., 1996; Waterman and Small, 1996).

In the presence of chloride, a 3 kb mRNA was formed that hybridized with the 5′ end of gadC. This observation and the absence of a putative terminator downstream of gadC suggests that gadC and gadB form an operon and that the putative terminator downstream of gadB is functional. In S. flexneri and E. coli, gadB and gadC are also genetically linked, but it is not known whether they are expressed together. It is also not known how the genes are regulated in these organisms. In L. lactis, chloride-dependent expression of gadCB is activated by GadR. gadR expression levels are independent of the NaCl and glutamate concentration of the medium, indicating that Cl-dependent and glutamate-dependent expression of gadCB is not controlled via transcriptional regulation of its activator gene gadR. The observed lower expression of gadR ::lacZ in non-buffered medium compared with buffered medium would contradict the role of gadR as a positive regulator in Cl-dependent gadCB expression. However, this result is probably an artifact caused by the instability of β-galactosidase at low pH (Tenu et al., 1971), as the lactococcal cytosol acidifies in an acid environment (Cook and Russel, 1994). GadR may interact directly with Pgad or indirectly via other (unknown) factors. It is tempting to speculate, owing to its location close to the −35 position of Pgad (see Fig. 2), that the 21 bp IR plays a role in transcription initiation from Pgad (Sanders et al., 1997). Chloride may affect gadCB expression by an allosteric effect on the possible regulator–DNA complex, as was shown in vitro for LacR (Brenowitz and Jamison, 1993). Although other possibilities cannot be dismissed, it is conceivable that gadR acquires affinity for Pgad through a conformational change under conditions of low pH and the presence of chloride ions.

Expression of lactococcal gadCB is higher in non-buffered media than in pH-stabilized media. This is of significance, as gadCB functions in low pH resistance. A similar observation was made for S. flexneri glutamate-dependent acid resistance. S. flexneri required adaptation at mild acid conditions to acquire full resistance to subsequent lethal acid pH (Lin et al., 1995). The expression of one or both E. coli glutamate decarboxylases is induced at acidic pH as well (Yoshida et al., 1993). The best studied acid resistance mechanism is E. coli cadBA. This lysine-dependent system consists of a lysine–cadaverine antiporter (CadB) and a lysine decarboxylase (CadA). cadBA expression is induced by the presence of lysine at low pH (Meng and Bennett, 1992). cadBA is regulated by CadC, a membrane protein with an intracellular DNA-binding region and periplasmic pH and lysine signalling sites (Dell et al., 1994). In addition, the level of CadBA is controlled by LysP and cadaverine (Neely et al., 1994). gadCB expression in L. lactis is influenced by both low pH and medium components. Medium-dependent expression may be caused by varying levels of glutamate, the amino acid conferring acid resistance through GadCB. The addition of glutamate to growing cultures resulted in higher β-galactosidase levels in L. lactis NS3 (gadC ::lacZ ). Therefore, glutamate may be a component of the inducing signal, by analogy with lysine in E. coli cadBA induction. Whether or not glutamate is essential for Cl-dependent gadCB expression is unknown, as in all Cl induction experiments (performed in 1/2M17), a basal amount of glutamate was present.

S. flexneri gadC is expressed in the stationary phase and is under the control of σS (Waterman and Small, 1996). In L. lactis, expression of gadCB is initiated in the late exponential phase, as was the case in S. flexneri, and highest expression was observed at the onset of the stationary phase. This pattern of gene expression was seen in both buffered and non-buffered media. However, growth phase dependency and pH dependency of gadCB expression could not be distinguished in the batch cultures that were studied. The decrease in β-galactosidase activity in the stationary growth phase was also observed for independent lacZ fusions to other genes (not shown) and could be the result of degradation of the enzyme by intracellular peptidases or, as mentioned before, of inactivation by the low intracellular pH. As gadCB expression was monitored with a transcriptional gadC ::lacZ fusion, regulation by chloride, low pH and glutamate is most likely at the transcriptional level. The sensing and regulating mechanism that couples data on pH, glutamate and chloride concentration and, possibly, growth stage to gadCB expression remains to be elucidated. It is possible that glutamate levels and acidity are sensed simultaneously as the level of glutamic acid, as this level is pH dependent because of the low pKa (4.25) of glutamate.

GadCB provides acid resistance under very specific conditions in the three organisms in which it has been studied, namely at neutral pH (E. coli ; Hersh et al., 1996), in stationary phase (S. flexneri ; Waterman and Small, 1996) or in the presence of chloride (L. lactis). Why L. lactis gadCB is expressed in the presence of chloride is completely unclear. High NaCl levels, which present an osmotic stress, could provide cross-protection against severe acid challenges. A system with reverse properties, providing NaCl-dependent acid sensitivity, has been described for E. coli (Rowbury et al., 1994). Remarkably, glutamate-dependent acid resistance in S. flexneri is enhanced in the presence of NaCl. This is not caused by the higher osmotic pressure of the medium (Waterman and Small, 1996). Although the authors conclude that the sodium ion was involved, it may well be that gadC expression in S. flexneri is also stimulated by chloride. An analogous stress protection mechanism induced by similar conditions in such diverse organisms as L. lactis and S. flexneri may point towards a common stressful natural environment. Both organisms can enter the gastrointestinal tract where they are faced with the highly acidic and chloride-rich conditions of the stomach. A high level of hydrochloric acid is produced by gastric cells to hydrolyse food peptides and as a barrier against (pathogenic) bacteria. L. lactis is able to pass the gastrointestinal tract (Klijn et al., 1995), albeit in low numbers. The gadCB acid resistance system may play a significant role under these conditions. gadCB may also be important for the survival of lactococcal cells during cheese production as high levels of both NaCl and glutamate are present in cheese.

Experimental procedures

Bacterial strains, plasmids and growth conditions

The bacterial strains and plasmids used in this study are listed in Table 3. L. lactis was grown at 30°C in twofold diluted M17 broth (Terzaghi and Sandine, 1975), containing 0.5% glucose and 1.9% β-glycerophosphate (1/2M17); 1/2M17 plates contained 1.5% agar. Modified M17 (mM17) contained no β-glycerophosphate and no soytone. β-Glycerophosphate was added to a final concentration of 2%, where indicated. Erythromycin (Em) and chloramphenicol were used at final concentrations of 5 mg ml−1, while spectinomycin was used at a final concentration of 100 mg ml−1. 5-Bromo-4-chloro-3-indolyl-β-D-galactopyranoside (Xgal) was used at a final concentration of 0.008%. E. coli was grown in TY broth at 37°C with vigorous agitation or on TY medium supplemented with 1.5% agar. Em was used at 100 mg ml−1 for E. coli.

Table 3. . Bacterial strains and plasmids.Thumbnail image of

General DNA techniques

DNA work was performed essentially as described by Sambrook et al. (1989). DNA was introduced by electrotransformation in E. coli (Zabarovsky and Winberg, 1990) and L. lactis (Holo and Nes, 1989). The Pharmacia T7 sequencing kit was used for DNA sequencing. Oligonucleotides were synthesized with an Applied Biosystems 392A DNA synthesizer. DNA sequences were analysed using the PC/Gene sequence analysis program (IntelliGenetics). Protein homology searches against the GenBank were carried out using the BLAST program (Altschul et al., 1990). Protein sequence alignments were carried out with the PALIGN program of PC/Gene using the structure genetic matrix. Transmembrane segments were predicted using the method of Klein et al. (1985).

β-Galactosidase assays

Cell extracts were prepared by vigorous shaking of cells in the presence of glass beads (Van de Guchte et al., 1991). β-Galactosidase activity was determined as described by Miller (1972). Protein concentrations were determined by the method of Bradford (1976) with BSA as the standard.

RNA analysis

RNA was isolated from exponentially growing L. lactis cultures (optical density at 600 nm of 0.5) as described previously (Van Asseldonk et al., 1993). The synthetic oligonucleotide NS3-12 (5′-GGTTTATAACAGTTCATCAAGTACACTTTTTAATCAGTC-3′) was used for primer extension as described previously (Sanders et al., 1997). The primer extension product was analysed on a sequencing gel next to the products of a sequence reaction using the same primer. Northern hybridizations were carried out at 40°C in a buffer containing 50% formamide, 7% SDS, 2% blocking reagent (Boehringer Mannheim), 5 × SSC, 50 mM sodium phosphate, pH 7, and 0.1% N-lauryl sarcosine. A 450 bp XbaI–Sau3A fragment encoding the 5′ end of gadC was used as a probe by labelling with [α32P]-dCTP using the Boehringer Mannheim random primed DNA labelling kit.

Cloning of gadCB of L. lactis

The region downstream of the lacZ fusion site in L. lactis NS3 was obtained by inverse polymerase chain reaction (PCR) amplification of a 0.55 kb Sau3A–XbaI fragment from L. lactis MG1363 chromosomal DNA (see Fig. 1). The PCR product was cloned in pORI19 using E. coli EC101 as a host, resulting in pNS3i4. In a subsequent inverse PCR, the adjoining 0.9 kb XbaI–Sau3A chromosomal DNA fragment was amplified and cloned in pORI19 using L. lactis LL108 as the cloning host. This construct was named pNS3i5 and used for single cross-over integration, resulting in strain MGNS3i5. Chromosomal DNA of MGNS3i5 was digested with EcoRI to clone sequences downstream of the insertion site of pNS3i5 by plasmid rescue. The chromosomal EcoRI fragments were circularized by self-ligation, and the ligation mixture was used to transform L. lactis LL108. Transformants contained a 4.8 kb plasmid (pNS3i6), which is pORI19 containing a 2.4 kb XbaI–EcoRI fragment from the chromosome (Fig. 1). The sequence of the 2.4 kb PstI–Sau3A DNA fragment located upstream of the lacZ insertion site in L. lactis NS3 was determined from pNS3d. The nucleotide sequence described in this paper has been assigned GenBank accession number AF005098.

Construction of insertion mutants

An internal XbaI–HindIII fragment of gadR was cloned in the integration vector pORI19S using the RepA+E. coli helper strain EC101. The resulting plasmid, pNS3i3, was used to disrupt gadR in L. lactis MG1363 as described previously (Law et al., 1995). The proper chromosomal location of the integrated plasmid was confirmed by Southern hybridization, and the strain was named MGNS3i3. Plasmid pNS3i7 was constructed by cloning a 500 bp AsnI fragment (internal to gadB ) in pORI19 using E. coli EC101 as a host. Single cross-over integration of pNS3i4 and pNS3i7 in the L. lactis MG1363 chromosome resulted in strains MGNS3i4 (gadC ) and MGNS3i7 (gadB ) respectively.

Acid resistance test

Cells of an exponentially growing culture of L. lactis in 1/2M17 with or without 0.3 M NaCl were harvested, washed with water and resuspended in an equal volume of MS15 (Cocaign-Bousquet et al., 1995) without glucose and glutamate but containing the same amount of NaCl as the culture. Glutamate was added to a final concentration of 1 mM where indicated. The pH of MS15 was adjusted to 3.5 with either lactic acid or hydrochloric acid before cell resuspension. After incubation for 2 h at 30°C, the number of viable cells was determined by plating onto 1/2M17 plates. The percentage of acid-resistant cells was calculated from the number of colony-forming units (cfu) after acid treatment divided by the number of cfu at the moment of harvest.


This work was supported by Unilever Research Laboratorium, Vlaardingen, The Netherlands. J.K. is the recipient of a fellowship from the Royal Netherlands Academy of Arts and Sciences (KNAW). We thank Dr Aat Ledeboer and Dr Wouter Musters for helpful discussions and Henk Mulder for preparation of the figures.