Present address: Janet Blatny, Norwegian Defence Research Establishment, PO Box 25, N-2027 Kjeller, Norway.
Vincent G.H. Eijsink, Department of Chemistry, Biotechnology and Food Science, Agricultural University of Norway, Chr. M. Falsensvei 1, PO Box 5003, N-1432 Ås, Norway (e-mail: firstname.lastname@example.org).
Aims: To use promoters and regulatory genes involved in the production of the bacteriocin sakacin P to obtain high-level regulated gene expression in Lactobacillus plantarum.
Methods and Results: In a plasmid containing all three operons naturally involved in sakacin P production, the genes encoding sakacin P and its immunity protein were replaced by the aminopeptidase N gene from Lactococcus lactis (pepN) or the β-glucuronidase gene from Escherichia coli (gusA). The new genes were precisely fused to the start codon of the sakacin P gene and the stop codon of the immunity gene. This set-up permitted regulated (external pheromone controlled) overexpression of both reporter genes in L. plantarum NC8. For PepN, production levels amounted to as much as 40% of total cellular protein.
Conclusions: Promoters and regulatory genes involved in production of sakacin P are suitable for establishing inducible high-level gene expression in L. plantarum.
Significance and Impact of the Study: This study describes a system for controllable gene expression in lactobacilli, giving some of the highest expression levels reported so far in this genus.
Lactic acid bacteria (LAB) are important micro-organisms, because of their many roles in the fermentation and preservation of food and because of their presence in the human intestine (Axelsson 1998; Ross et al. 2002). As many LAB have the generally regarded as safe (GRAS) status and some LAB show probiotic effects on animal and human health (Mercenier et al. 2003), they are important organisms for biotechnological applications. LAB have potential as food-grade cell factories and as delivery vehicles for antigens, antibodies and growth factors.
Many LAB produce anti-microbial peptides called bacteriocins, whose production is often regulated via a quorum-sensing system (Kleerebezem et al. 1999; Nes and Eijsink 1999; Eijsink et al. 2002; Quadri 2002). In the case of class I bacteriocins (lantibiotics such as nisin), the bacteriocin itself acts as pheromone, which activates a two-component regulatory system consisting of a histidine kinase receptor and a cognate response regulator. For class II bacteriocins, there is one known case of a bacteriocin that induces its own production (Kleerebezem and Quadri 2001). However, in most cases strains producing class II bacteriocins secrete a separate pheromone peptide with no or little bacteriocin activity, whose gene is co-transcribed with the genes encoding the cognate two-component regulatory system (see Fig. 1). In both systems, the phosphorylated response regulator acts as a transcription activator, enhancing transcription of all operons involved in bacteriocin production.
The genes and promoters involved in nisin production have been used to develop a nisin-controlled expression (NICE) system for efficient, regulated overproduction of heterologous proteins in lactococci (de Ruyter et al. 1996), lactobacilli (Pavan et al. 2000), and other Gram-positive bacteria (Eichenbaum et al. 1998). Similar systems may be developed on the basis of genes and promoters involved in the production of class II bacteriocins. For example, Axelsson et al. (1998) have established a two-plasmid system for overproduction of heterologous bacteriocins in Lactobacillus sakei, based on genes and promoters involved in the production of sakacin A. Studies with the aminopeptidase N reporter gene (pepN from Lactococcus lactis) have shown that this system is not suitable for high-level protein production in Lactobacillus plantarum (G. Mathiesen, V.G.H. Eijsink and L. Axelsson, unpublished observations). Recently, regulated bacteriocin promoters have been used to construct one-plasmid inducible expression systems (Axelsson et al. 2003; Sørvig et al. 2003). These new systems showed promising results with the gusA reporter gene, but in all cases, expression levels in L. plantarum were ca 10 times lower than the levels obtained with the NICE system (Pavan et al. 2000; Sørvig et al. 2003).
Our previous studies on overproduction of sakacin P in lactobacilli (Hühne et al. 1996; I.M. Aasen and L. Axelsson, unpublished observations) showed that the highest production levels were obtained in strains harbouring all operons naturally involved in sakacin P production, in their natural organization (plasmid pMLS114; Fig. 1). Therefore, pMLS114 was used as a starting point for establishing regulated expression of aminopeptidase N from Lactococcus lactis (pepN) and β-glucuronidase from Escherichia coli (gusA) in L. plantarum. The resulting gene expression system allowed controlled gene expression in lactobacilli, at levels that are among the highest ever described for lactic acid bacteria.
Materials and methods
Bacterial strains, plasmids and standard genetic techniques
Table 1 gives an overview of the bacterial strains and plasmids used in this study. Escherichia coli cells were grown in shaking flasks at 37°C in brain–heart infusion (BHI) medium (Oxoid Ltd, Basingstoke, UK). Lactobacillus plantarum NC8 was grown in MRS medium (Oxoid Ltd) at 30°C without shaking, unless stated otherwise. For plates, media were solidified by adding 1·5% (w/v) agar. Antibiotic concentrations were: ampicillin – 150 μg ml−1 (E. coli); kanamycin – 50 μg ml−1 (E. coli); erythromycin – 200 μg ml−1 (E. coli) and 5 μg ml−1 (lactobacilli).
Table 1. Strains and plasmids used in this study
Source or reference
E. coli DH5α
Host strain for plasmids
E. coli JM109
Host strain for plasmids
E. coli TOP10
Host strain for ‘TOPO-cloning’ of PCR fragments
L. plantarum NC8
Host strain (silage isolate) for pMLS114 derivatives
pMLS114 derivative; gusA replaces sppA and spiA; deletion in sppIP, Emr
All cloning steps were conducted according to standard procedures as described in Sambrook et al. (1989). PCR reactions were performed in a PTC-100 Programmable Thermal Controller (MJ Research, Inc., Waltham, MA, USA) with Pfu DNA polymerase (Promega Corp., Madison, WI, USA), as recommended by the polymerase supplier. Oligonucleotides were purchased from Medprobe (Oslo, Norway). PCR fragments were subcloned using the TOPO system provided by Invitrogen (Carlsbad, CA, USA). All PCR-derived DNA fragments were sequenced using the BigDye Terminator Cycle Sequencing Kit, according to the manufacturer's recommendations, and the ABI Prism 377 DNA sequencer (Perkin Elmer/Applied Biosystems, Foster City, CA, USA).
Chemically competent E. coli JM109 (Promega Corp.), or DH5α (Invitrogen) were transformed applying the protocol provided by the supplier. Lactobacilli were transformed according to the protocol of Aukrust et al. (1995).
Plasmid DNA from E. coli and lactobacilli was isolated using the QIAprep Miniprep Kit (Qiagen, Venlo, The Netherlands). Lactobacillus cells were incubated for 25 min at 37°C with 5 mg ml−1 lysozyme, 15 U ml−1 mutanolysin and 100 μg ml−1 RNase (all Sigma, St Louis, MO, USA) before adding the lysis buffer from the QIAprep kit.
Extraction and purification of DNA from agarose gels was performed using the QIAquick Gel Extraction Kit (Qiagen).
All derivatives of plasmid pMLS114 (Fig. 1; Hühne et al. 1996), were based on a precise exchange of the two genes in the bacteriocin operon (sppA + spiA; Fig. 1) by either pepN (Strøman 1992) or gusA (Schlaman et al. 1994). Thus, the new gene was inserted by making both a translational fusion with the ATG start codon of sppA and a stop codon fusion with spiA. The constructs were made using recombinant PCR (Higuchi 1990) and standard subcloning steps. The primers used at the fusion points are shown in Table 2. In plasmid pGM1 (Fig. 1), the bacteriocin operon is replaced with pepN; plasmid pMLS114 was used as template for PsppA and the region downstream of the stop codon of spiA, whereas plasmid pSTO10 (Table 1) was used as a template for the pepN gene. To make pepN expression dependent on externally added peptide, a deletion was introduced in the sppIP gene in pGM1, yielding pGM4 (Table 1; Fig. 1). The primers used at the deletion point are shown in Table 2. pGM4-GUS is analogous to pGM4, containing the gusA instead of the pepN gene. The primers used at the fusion points are shown in Table 2. The template for the gusA gene was pKRV3a-gus (Table 1). All constructs were transformed to L. plantarum NC8.
Table 2. Primers used at fusion and deletion points
*F, forward direction; R, reverse direction.
†Start codons and stop codons are underlined; nucleotides adjacent to the deletion in the sppIP gene are printed in bold face.
Fusion of PsppA–pepN
Fusion of PsppA–pepN
Stop codon fusion pepN–spiA
Stop codon fusion pepN–spiA
Fusion of PsppA–gusA
Fusion of PsppA–gusA
Stop codon fusion gusA–spiA
Stop codon fusion gusA–spiA
Deletion of sppIP
Deletion of sppIP
Gene expression studies
Over night cultures of L. plantarum NC8 harbouring pepN-containing plasmids were diluted 100-fold (to O.D.600ca 0·1) in MRS with appropriate antibiotics. If induction was desirable, 25 ng ml−1 pheromone peptide (IP-673) (Molecular Biology Unit, University of Newcastle-upon-Tyne, UK) was added to the medium and the cultures were grown to an O.D.600 of 2·3–2·5. In dose–response experiments the pheromone concentration was varied from 0 to 100 ng ml−1. Cells were harvested by centrifugation and disrupted by glass beads (106 μm and finer, G-4649; Sigma) essentially as described by van de Guchte et al. (1991). The resulting protein extract was used to assay aminopeptidase activity using l-lysine p-nitroanilide (Sigma) as substrate (Exterkate 1984). The protocol for determining PepN activity (Exterkate 1984) was modified in the sense that reactions were conducted at 30°C (instead of 37°C) and in 0·1 m Tris–HCl, pH 8·5 (instead of a 0·1 m sodium phosphate buffer). PepN activities were measured three times, using three independent cultures. Protein concentrations were determined using the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA) with bovine serum albumin as a standard. The crude protein extracts were analysed on 12% SDS polyacrylamide gels and the amounts of PepN were quantified as a percentage of the total intracellular protein content by scanning the gels with a densitometer (Gel Doc 1000; Bio-Rad).
Lactobacillus plantarum NC8 cells harbouring pGM4-gus were induced at an O.D.600 of ca 0·3 by addition of pheromone (IP-673) to a final concentration of 50 ng ml−1. β-Glucuronidase activities were determined using a modified β-galactosidase assay (Miller 1972) as described by Axelsson et al. (2003). GUS activities were calculated as described by Miller (Miller 1972) and expressed as Miller Unit equivalents (MU). The induction protocol used for GUS production differed slightly from the induction protocol used for PepN production; when applied to PepN production, the GUS protocol yielded similar results (Fig. 2, Table 3; G. Mathiesen, unpublished observations).
Table 3. PepN activity in recombinant Lactobacillus plantarum NC8 strains
Increase upon induction
Pheromone (25 ng ml−1)
n.a., not applicable.
*PepN activity is expressed as μmol p-nitroanilide min−1 mg−1 protein ± s.d. (n = 3).
†Lactobacillus plantarum NC8 (pGM1) did not require external pheromone for induction of PepN production and addition of pheromone did not affect the outcome of the experiment.
2·7 ± 0·6
0·16 ± 0·09
4·1 ± 0·4
In pMLS114-derived pGM1 (Fig. 1), the pheromone autoinduction loop is intact, meaning that PepN production does not require externally added pheromone. pGM4 (Fig. 1) is a pGM1 variant with a deletion in the pheromone gene (sppIP), meaning that cells harbouring pGM4 need externally added pheromone for activation of pepN expression. Table 3 shows that the highest PepN levels were reached in induced cultures of L. plantarum NC8 (pGM4). SDS-PAGE analysis of cell-free extracts from L. plantarum NC8 harbouring pGM1 or pGM4 (Fig. 2) shows the expected PepN band at ca 95 kDa. Scanning of SDS-PAGE gels containing cell-free extracts of induced L. plantarum NC8 (pGM4) harvested at an O.D.600 of 2·3–2·5 showed that PepN levels amounted to more than 40% of the total intracellular protein content. Noninduced cells of L. plantarum NC8 (pGM4) produced 25-fold less PepN than induced cells (Fig. 2, Table 3).
Figure 3 shows a clear dose–response relationship between the amount of PepN produced by L. plantarum NC8 (pGM4) and the amount of peptide pheromone added to the culture. PepN activity increased almost linearly with pheromone concentration in the 0·5–15 ng ml−1 range, and maximum PepN production was achieved at 25 ng ml−1. Maximum PepN levels were obtained in cultures grown at 30°C, whereas growing the cultures at 20, 25 or 37°C reduced PepN production by 80, 30 and 55%, respectively (results not shown). Induction reduced the growth rate L. plantarum NC8 (pGM4) but not of L. plantarum NC8 (pMLS114) used as a control. Thus, the growth rate reduction was due to the production of PepN and not to the presence of peptide pheromone (results not shown).
Similar studies conducted with pGM4-gus confirmed the usefulness of pMLS114-derived vectors for high level expression. Maximum GUS activity was in the range of 700 MU equivalents, which is four to five times higher than the levels recently obtained (Sørvig et al. 2003; see below for further discussion). GUS levels in noninduced cells were ca 80-fold lower than GUS levels in induced cells (results not shown).
The development of pMLS114-based constructs for gene expression was initiated because previous studies had shown that Lactobacillus strains harbouring this plasmid yielded the highest levels of sakacin P that we have ever obtained in our laboratories (Aasen et al. 2000; I.M. Aasen and L. Axelsson, unpublished observations). The natural organization of the spp gene cluster and promoters in pMLS114 was expected to be important because previous studies of the NICE system had indicated that promoter activity depends in a subtle manner on the levels of all proteins involved in signal transduction, as well as on the ratios between them (Kleerebezem et al. 1997; Pavan et al. 2000). Thus, we attempted to create expression plasmids in which the natural spp gene cluster was largely intact.
Sørvig et al. (2003) have recently created a series of vectors that are more easy-to use than pMLS114, but which are much more ‘engineered’ in terms of promoters, terminators and gene organization. Using the sakacin P-based expression vector pSIP401 to express GUS in L. plantarum NC8, Sørvig et al. (2003) found GUS levels that were about one-fifth of the levels presented here, despite the fact that GUS was translationally fused to the same promoter as in pGM4-gus (note that pSIP401 and pMLS114 are based on the same L. plantarum replicon). Thus, it seems that the use of a pMLS114-like arrangement of genes and promoters indeed is favourable for expression. GUS levels obtained using the two-plasmid NICE system (Kleerebezem et al. 1997) in L. plantarum NC8 (Sørvig et al. 2003) were ca 10 times higher than the levels obtained with the pSIP401 derivative, that is ca two times higher than the levels reported here for pGM4-GUS. In all three cases considerable levels of background expression were observed. Pavan et al. (2000) have shown that a variant of the NICE system based on a L. plantarum host strain with the nisRK genes integrated into the chromosome yields higher expression levels and lower background activity. Similar improvements are conceivable for the sakacin P-derived systems described here.
The results obtained with PepN show that the pMLS114-derived expression vectors are suitable for heterologous expression at very high levels. To our knowledge, the highest PepN levels in lactobacilli reported so far amount to 28% of total cellular protein. These levels were obtained by expressing the pepN gene under the control of the constitutive L. brevis slpA (surface S-layer) promoter in L. plantarum (Kahala and Palva 1999). The level of ca 40% of total cellular protein described in the present study approaches the level of 47% reported for overexpression of pepN in its natural host, Lactococcus lactis, using the NICE system (de Ruyter et al. 1996).
While the pMLS114-derived expression vectors yield high levels of gene expression, they do have some drawbacks. One of these is the lack of versatility of the vector: every expression construct needs to be made by recombinant PCR, using specially designed primers. Another drawback concerns the rather high background expression which presumably precludes the use of this system for expression of genes that encode toxic proteins.
The data presented here show that regulated promoters and regulatory genes from the sakacin P system permit controlled overproduction of proteins in L. plantarum. As lactobacilli are food grade organisms and colonize the intestine, applications of Lactobacillus-based cell factories in food and for medical purposes are conceivable. We are currently in the process of further optimizing sakacin P-based and related expression systems, with focus on reduction of background expression and the construction of smaller, more versatile vectors that are equally effective as the pMLS114 derivatives described here.
This work was supported by a strategic grant from The Research Council of Norway (no. 134314/140). We thank Per Strøman and Eric Johansen at Chr Hansen A/S (Denmark) for providing us with plasmid pSTO10 and Kathrine Hühne and Lothar Kröckel at the Federal Institute for Meat Research (Germany) for supplying pMLS114.