INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Lactococcus lactis is a gram-positive bacterium used to manufacture a variety of fermented dairy products. Over the past 15 years, considerable effort has been put into characterizing industrially important traits of lactococci (i.e., lactose utilization, proteolytic activity, citrate metabolism, exopolysaccharide production, bacteriocinogenic activity, phage resistance, etc.). Consequently, a variety of gene cassettes have been isolated and characterized, some of which can be used to improve the lactococcal cultures of traditional fermentations or to develop novel and innovative products.

Strain improvement can be accomplished using plasmid vectors to transfer and maintain specific traits in bacteria. The selection of appropriate transformants or transconjugants relies on one or more genes coding for selection markers on the vector. Traditionally, antibiotic resistance genes have been used as versatile and efficient dominant selection markers in laboratory research vectors. On legal and ethical grounds, however, transferable genes that confer resistance to substances used in human drug therapy are unacceptable in food applications.

Consequently, alternate selection markers must be used and indeed have been proposed to construct food-grade vectors for L. lactis. Depending on the type of selection, they can be classified into two groups as either dominant or complementation markers (7). Dominant markers do not rely on host-expressed genes and can be used in most wild-type strains of L. lactis. For instance, lactococcal genes that confer immunity to nisin have been used for dominant selection (14, 15, 25, 35, 56). Since nisin is considered a food-grade molecule, it can be used throughout the fermentation processes to promote plasmid maintenance, as long as this bacteriocin is compatible with specific applications. Resistance to heavy metals (cadmium, copper, tin [33-35]) represents a second type of dominant marker. Although useful in the selection step, heavy metals are toxic and obviously cannot be used in food fermentation to maintain plasmids. However, maintenance of the selection pressure throughout the fermentation process may not be necessary, as the vector using cadmium resistance replicates through a theta mechanism (34), which is expected to be stable (30). Unfortunately in most of these dominant markers, the phenotypes used for the selection process occur naturally in numerous lactococcal strains, which consequently limit their host spectrum.

Complementation markers are based on mutations in the host chromosome that can be complemented by plasmid-expressed markers. One clear advantage of complementation systems is that the selection pressure, with carefully chosen genes, can be applied during industrial fermentation to ensure plasmid maintenance. For example, food-grade complementation marker systems have been based on the lacF gene coding for the essential enzyme IIA of the lactose phosphotransferase system of L. lactis (8, 37, 44). Lactose-deficient hosts with mutations or deletions of the lacF gene can be complemented by a plasmid that carries the wild-type enzyme IIA gene. The plasmid can be kept stable in medium having lactose as the main source of carbon (7). Furthermore, two auxotrophic complementation markers are based on altered tRNA genes that suppress nonsense mutations in genes involved in the biosynthetic pathway of purines and pyrimidines. In the first system, an ochre suppressor (supB) complements purine auxotrophic mutants of L. lactis grown in purine-free medium (9). However, supB suppresses both amber and ochre codons (82% of Lactococcus genes), which therefore causes major pleiotropic effects that alter bacterial growth. Alternatively, supD was used to complement pyrimidine auxotrophic hosts constructed by introducing an amber codon into the chromosomal pyrF gene (52). Since supD only suppresses amber codons (10% of Lactococcus genes), pleiotropic effects are significantly reduced. An auxotrophic supD-carrying mutant derived from an industrial strain showed an acidification rate in milk that is comparable to that of the parental strain. The major disadvantage of such complementation systems is that specific mutations must first be introduced, in a food-grade manner, into every recipient host before the complementing plasmids can be applied.

In this study, we present a third and novel strategy for a food-grade cloning system that has the advantages of both the dominant and the complementation strategies. The design consists of two plasmids that enable the dissociation of the dominant antibiotic marker from the vector plasmid. The vector pVEC1 consists exclusively of L. lactis DNA and complements the companion plasmid pCOM1, which bears the dominant marker. Both plasmids are required for the gene transfer step to L. lactis, but only the vector pVEC1 remains in the transformed cells afterward, resulting in a food-grade microorganism. The effectiveness of the design is ascertained by the stable transfer of a phage resistance mechanism to laboratory and industrial strains. This versatile and efficient vector system is potentially applicable to a large number of lactococcal hosts.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, bacteriophages, plasmids, and media. The bacterial strains, bacteriophages, and plasmids used in this study are listed in Table 1. Escherichia coli was grown at 37°C in Luria-Bertani medium (46) with shaking. L. lactis was grown at 30°C in M17 (54). For strain MG1363 and its derivatives and strain SMQ-481, M17 was supplemented with 0.5% glucose. All of the other lactococcal strains were grown in M17 supplemented with 0.5% lactose. Calcium chloride at a final concentration of 10 mM was added for the propagation of lactococcal bacteriophage. When appropriate, ampicillin was added at 50 µg/ml for E. coli. Chloramphenicol and erythromycin were used at a concentration of 5 µg/ml for L. lactis.

Bacteriophage propagation and microbial assays. Phage propagation, titration, and the assay for efficiency of plaquing (EOP) have been described elsewhere (26, 41, 47). One-step growth curves and centers of infection (COI) were assayed as described previously by Moineau et al. (40). The phage burst size was estimated as the ratio between the phage titer at two consecutive latent phases on the growth curves. The efficiency at which COI formed (ECOI) was calculated by dividing the number of COI on resistant strains by the number of COI on wild-type strains. The data presented for each assay represent the average of at least three independent experiments.

DNA isolation, manipulation, and sequencing. Large-scale plasmid DNA purification from E. coli was carried out using the Qiagen kit (Chatsworth, Calif.), and the silica method (J. D. Brown, http://research.bmn.com/tto [T01281]) was used for small-scale preparations. To obtain L. lactis plasmid DNA, the silica method was modified as follows: 1.5 ml of an overnight culture of L. lactis was harvested and suspended in 200 µl of 25% sucrose containing 30 mg of lysozyme/ml. After incubation at 37°C for 15 min, 400 µl of lysis solution (3% sodium dodecyl sulfate [SDS], 0.2 N NaOH) was added, mixed, and incubated at room temperature for 5 min. Three hundred microliters of 3 M CH₃COOK (pH 4.8) was added, mixed by inversion, and incubated on ice for 10 min. The samples were centrifuged for 10 min, and the aqueous phase was transferred to a new tube, avoiding white particles. Two hundred microliters of silica suspension was added and incubated for 5 min at room temperature. After a brief centrifugation, the silica was harvested and the aqueous phase was discarded. The pellet was suspended in an ethanol wash solution (50% ethanol, 0.1 M NaCl, 1 mM EDTA, 10 mM Tris-HCl [pH 7.5]) and centrifuged, and the solution was discarded. The washing step was repeated, and the silica pellet was air dried for 5 min. The pellet was suspended in 50 µl of 10 mM Tris-HCl, pH 8.0, incubated at 55°C for 5 min, and briefly centrifuged. The aqueous phase containing DNA was collected and transferred to a new tube.

Southern transfer and hybridization. Purified DNA was digested with restriction enzymes and separated by electrophoresis on 0.8% agarose gels. DNA fragments were stained with ethidium bromide, photographed under UV illumination, and then transferred onto positively charged nylon membranes (Roche Diagnostics) by capillary blotting (53). Probes were prepared by labeling DNA fragments with the DIG High-Prime labeling kit (Roche Diagnostics). Prehybridization, hybridization, and posthybridization washes as well as detection were performed as suggested in the DIG System User's Guide for Filter Hybridization (Roche Diagnostics). The DIG Easy Hyb buffer and CSPD were used for the hybridization steps and chemiluminescent detection, respectively.

Plasmid stability. Plasmid segregational stability was determined as the fraction of a culture that maintained the tested plasmid after exponential growth for 100 generations without selective pressure (19). Cultures were assayed for plasmid maintenance at 0 and 100 generations by testing 100 randomly selected colonies for the plasmid-borne antibiotic resistance phenotype. The strains SMQ-562(pRL7) and SMQ-652(pRL7) were evaluated for plasmid stability by testing for phage resistance with the cross-streaking method described by Moineau et al. (41). The correlation between phenotype and plasmid content was confirmed by analyzing the plasmid profile of resistant and sensitive colonies. The percentage of plasmid loss per generation was calculated using the formula of Roberts et al. (45). The results are an average of three independent determinations.

Plasmid copy number. Plasmid DNA was linearized by digestion with restriction enzymes, separated by electrophoresis on a 0.8% agarose gel, and photographed under UV illumination with the Gel Doc 1000 photodocumentation system (Bio-Rad, Mississauga, Ontario, Canada) under nonsaturating conditions. The intensity of each plasmid DNA band was estimated using Molecular Analyst image analysis software (Bio-Rad). The relative copy number of each plasmid was evaluated by comparing the intensity of each DNA band and correcting for plasmid size. Experiments were done in triplicate.

Nucleotide sequence accession number. The complete sequence of pCD4 has been submitted to GenBank (accession no. AF306799).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Selection and isolation of a plasmid for food-grade vector construction. L. lactis subsp. lactis MJC15 was isolated from raw milk cheese by Cardinal et al. (2). It carries five plasmids which range from 2.8 to 70 kb in size. To identify the plasmids that most likely replicate through the theta mechanism, a probe specific for the conserved repB gene of lactococcal theta plasmids (19, 49) was used. The probe was prepared by labeling an 896-bp PCR product obtained from the lactococcal theta plasmid pSRQ900 (accession no. AF001314) with primers SM3 (5'-CCTTTTTACCGTAGGTAGG) and SM11 (5'-GTCGTTTCAAAGAAGCGGTT). Four of the plasmids of strain MJC15 (pCD1, pCD2, pCD3, pCD4) hybridized with the probe. The fifth plasmid (pCD5; 2.8 kb) hybridized with a 1,251-bp RsaI fragment of pMG36ct corresponding to the leading-strand initiation and control region of pWV01, a lactococcal plasmid from the pMV158/pE194 family (28). This suggests that pCD5 replicates by the rolling circle mechanism. Two plasmids, pCD4 and pCD5, were found to be highly stable, as neither one could be cured from the host. Plasmid pCD4 was selected for food-grade vector construction for the following reasons. Firstly, vectors derived from theta replicons have shown higher intrinsic structural stability. Secondly, host range is limited, and thirdly, they are most often mutually compatible, which means that they can coexist in relatively high numbers in a single host (29, 30, 36).

Sequence analysis of pCD4. The plasmid pCD4 contained 6,094 bp and had a G+C content of 33.5% (Fig. 1). Five open reading frames (ORFs) and two cis-acting regions were identified in the sequence of pCD4 (Table 2). The first ORF (orfA) encodes a putative protein of 196 amino acids that bears strong similarities with hypothetical proteins found on a number of lactococcal plasmids. In a recent study (1a), bioinformatic evidence was presented suggesting that these proteins could be integrases belonging to a family of tyrosine recombinases that rearrange DNA duplexes by site-specific recombination. Two putative rho-independent terminators were identified within the first 80 bp downstream of orfA (Fig. 1).

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Functional analysis of the replication region of the lactococcal plasmid pCD4. A linear restriction map is shown at the bottom of the diagram. Putative ORFs (black arrows), cis-acting regions (light grey boxes), promoters (arrowheads), and terminators (stem-loops) shown on the restriction map were inferred from sequence analysis. Thick lines above the linear map represent the fragments cloned into pNC1. The names of recombinant plasmids containing these inserts are indicated on the left. The single break in pRL1 corresponds to the HindIII site used for the cloning of the complete pCD4, and the double breaks in pRL20 correspond to the deletion of ClaI fragments. Only those restriction sites used to generate pCD4 derivatives are indicated on the map. The replication phenotype (Rep) and the calculation of the stability of pCD4 derivatives in strain MG1363 (average percent plasmid loss per generation ± standard deviation) are indicated on the right. The vertical lines delimit the core replicon. ND, not determined.

View larger version (51K): [in this window] [in a new window]   FIG. 2.   Sequence alignment of the oriT locus of lactococcal plasmids. Inverted and direct repeats are indicated above the sequence. IR1, IR2, and IR3 correspond to the inverted repeats that were identified on the functional transfer origins of pCI528 and pNZ4000 (36, 55). Inverted repeats IR4, IR5, and IR6 are present on the functional transfer origin of oriT1 and oriT2 (IR5 and IR6) defined on pNZ4000. The bold letters represent the core region of the R64 oriT locus. The NikA-binding site is boxed, and the nick site is indicated by the arrowhead.

View larger version (71K): [in this window] [in a new window]   FIG. 3.   Sequence analysis of the minimal replicon of pCD4. (A) The restriction sites used to generate pCD4 derivatives are highlighted on the DNA sequence. The 35 and 10 boxes of repB promoter and RBSs are underlined. The reverse-complement sequence of the 35 and 10 boxes of the counter-transcript RNA promoter is double underlined. The AT-rich region is boxed, and the GC clusters are indicated with asterisks. Direct repeats within the AT-rich stretch are represented by double arrows, and the 22-bp iterated sequences repeated three and one-half times (iteron) are indicated by thick arrows above the sequence. Inverted repeats are represented by dashed arrows over the sequence. Protein translation of repB and of the truncated orfX is given below the DNA sequence. Amino acids in reverse video in the RepB sequence correspond to consensual residues with a plurality of at least 30 from an alignment of the amino acid sequence of 35 replication initiator proteins of lactococcal plasmids. The sequences used in the alignment were taken from GenBank. Amino acid motifs for the leucine zipper (positions 31 to 45), copy number control (129 to 144), and DNA binding (positions 215 to 241) are indicated with dashed underlines. The numbers on the left correspond to the nucleotide numbering of pCD4 as submitted to GenBank. The amino acid numbering of RepB is indicated on the right. (B) Helical wheel representation of the leucine motif of RepB. Hydrophobic residues are boxed, charged residues are indicated by a + sign, and the other residues correspond to uncharged polar residues.

Functional analysis of the replicon of pCD4. DNA fragments containing the functional replicon may be identified by the replication ability that they confer on L. lactis. Various fragments of pCD4 were cloned into the replicon screening vector pNC1 (Table 2) and were transferred to L. lactis MG1363 by electroporation. Only clones carrying the functional replicon of pCD4 produced chloramphenicol-resistant colonies (Fig. 1). The core replication module of pCD4 was localized to the 1,980-bp NspV/AseI fragment carried by the recombinant plasmid pRL32 (Fig. 3). This fragment includes the ori region and a complete copy of repB. Only the first 29 amino acids of OrfX are coded on this fragment, and the gene for HsdS is absent, showing that OrfX and HsdS are not essential for replication. All pCD4 derivatives tested (Fig. 1) proved to be very stable, as the loss rate was less than 0.02% per generation in L. lactis MG1363 over 100 generations without selection. Although these were found to be segregationally stable, it is still possible that partial or complete deletion of auxiliary factors could affect copy number (19). Measurement of the relative copy number of pCD4, pRL2, pRL5, pRL30, and pRL32 revealed no significant differences, further demonstrating that the absence of such auxiliary factors does not affect copy number.

Construction of the food-grade vector system. Our criteria for the construction of the two-component vector system were as follows: (i) the vector plasmid pVEC1 must be relatively small, stable without selective pressure, and compatible with wild-type lactococcal plasmids and must consist entirely of L. lactis DNA. (ii) The companion plasmid pCOM1 must contain a dominant selection marker gene, and its replication must be pVEC1 dependent although segregationally incompatible with the latter. The plasmid pRL33 has all the characteristics required for the companion plasmid pCOM1. For the convenience of cloning, the functions needed for propagation and selection in E. coli (ampicillin resistance gene, ColE1 origin of replication) were also included (Fig. 4). For pCOM1, chloramphenicol resistance was replaced by erythromycin (pRL33e) to avoid the potential repression of chloramphenicol gene expression by the replication initiator protein (13).

View larger version (16K): [in this window] [in a new window]   FIG. 4.   Circular maps of pVEC1 and pCOM1. Relevant features of the plasmids and restriction sites useful for cloning are indicated. Truncation in genes is indicated by apostrophes.

View larger version (48K): [in this window] [in a new window]   FIG. 5.   Detection by hybridization of pVEC1 and pCOM1 in total DNA extracts from variants of L. lactis MG1363 harboring different combinations of plasmids by hybridization. (A) Agarose gel before Southern transfer. (B) Autoradiogram obtained after hybridization with a probe specific for pCOM1 prepared by labeling pNC1. (C) Autoradiogram obtained after hybridization with a probe specific for pVEC1 obtained by labeling a repB internal fragment of pSRQ900. Lanes: M, 1-kb DNA mass ladder (Gibco/BRL Life Technologies, Burlington, Ontario, Canada); 1, MG1363 total DNA/EcoRI; 2, MG1363(pCOM1, pVEC1)/EcoRI; 3, MG1363(pVEC1)/EcoRI; 4, MG1363(pCOM1, pVEC1)/ClaI; 5, MG1363(pVEC1)/NspV; 6, MG1363(pCOM1, pVEC1)/NcoI/XbaI; 7, MG1363(pVEC1)/NcoI. The molecular size of marker bands is indicated on the left in kilobases, and hybridization signals corresponding to pCOM1 and pVEC1 are indicated on the right.

Determination of the copy number of pCD4. An internal fragment of 302 bp of the single-copy lactococcal gene recA was used to estimate the copy number of pCD4 and its derivatives per chromosome equivalent. The recA fragment was obtained by PCR amplification using genomic DNA of L. lactis SMQ-481 with primers Rec1 (5'-GACCCAGAATATGCAAA AGCACTCGGTG) and Rec2 (5'-CCAACTTTTTCACGCAATTGGTTGATG) and cloning into pRL2, giving pRL8 (Table 1). Plasmid pRL8 was introduced into L. lactis MG1363, and total DNA was submitted to Southern analysis. The recA-specific probe hybridized with two DNA bands, one corresponding to the chromosomal copy and the other to the pRL8-expressed copy of recA (data not shown). Comparison of the intensities of these bands established that pCD4 is a low-copy-number plasmid with 2.3 ± 1.1 (standard deviation) copies per genome equivalent in L. lactis MG1363.

Application of the two-component cloning system for the transfer of phage abortive resistance to industrial strains. The L. lactis W-37 plasmid pSRQ900 contains the gene coding for AbiQ, a phage resistance mechanism effective against lactococcal phages belonging to the 936 and c2 species. A 2.2-kb fragment coding for AbiQ (Table 1) was subcloned into pVEC1 as an EcoRI/NcoI fragment and was introduced into L. lactis MG1363 by cotransformation with pCOM1. An erythromycin-resistant isolate carrying the proper construction (pRL7) was selected, cured of pCOM1, and tested for phage resistance. L. lactis MG1363 harboring pRL7 conferred high resistance against two lactococcal phages belonging to the 936 and c2 species (Table 4). The potential of pVEC1 as a food-grade vector was further demonstrated by transferring pRL7 into two industrial strains. Southern analysis was carried out to verify the total loss of pCOM1 from the industrial strains carrying pRL7. Only pVEC1 remained in the cells after the curing step, and none of the plasmids was integrated into the host chromosome or in the wild-type plasmids. In addition, resident wild-type plasmids were unaffected by the introduction of pRL7 as shown by 100 generations of coexistence. The plasmid pRL7 is highly stable in these hosts, as no loss of the phage resistance phenotype was recorded after growth for 100 generations. The effectiveness of pRL7 in controlling phage infection was demonstrated by a reduction of the burst size and of the ECOI (Table 4). The phenotype conferred by pRL7 also resulted in the severe reduction of EOP.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

A two-component food-grade cloning system was developed for L. lactis, where the uncoupling of the selection marker and the vector enabled the transient use of a non-food-grade but versatile dominant erythromycin resistance marker that is efficient in virtually all lactococcal hosts. The designed system included (i) a replication-deficient companion plasmid (pCOM1) coding erythromycin resistance and harboring a replication origin and (ii) an autonomously replicating vector (pVEC1) composed of only L. lactis DNA and the trans-acting initiator needed to support replication of pCOM1. Cloning is accomplished by first recombining pVEC1 DNA with the desired genes, followed by its coelectroporation with pCOM1 in L. lactis. Media containing erythromycin allow only the growth of cells that contain both plasmids. Curing of the marker plasmid after selection of cotransformants is easily accomplished in antibiotic-free medium. Plasmid pVEC1 is highly stable and can be maintained without selective pressure for many generations.

As stability is a major requirement for this cloning system, pVEC1 and pCOM1 were based on a theta-type replicon. Plasmids using this mode of replication are structurally and segregationally more stable, especially when carrying long DNA inserts (30). In addition, lactococcal vectors of the theta family were shown to have a narrow host range (24, 29, 36, 56), thus limiting transfer to other microorganisms. Stable coexistence of pRL7 with wild-type plasmids for at least 100 generations in two lactococcal hosts agrees with previous observations that multiple theta-type plasmids can coexist in a single lactococcal host, even if they carry highly related replicons (13, 20, 21, 49).

The segregational stability of plasmids relies on control elements that adjust the rate of replication to maintain a constant copy number (reviewed by del Solar et al. [5]). Control of the plasmid copy number is achieved by modulating the intracellular concentration of the initiator protein with negative regulatory circuits that may include antisense RNA, both antisense RNA and proteins, and sites for binding initiator protein. The fact that all replication-proficient derivatives of pCD4 were as stable as the parent plasmid indicates that all of the control elements were contained within the defined minimal replicon.

Plasmid copies that initiate a replication cycle are selected at random from a pool of identical plasmids within the cell (5). This has a major consequence for plasmids that share replication control functions, as they will compete for stable inheritance and are unable to coexist in a stable manner within a given cell in the absence of selective pressure. This incompatibility leads to the rapid segregation of plasmids within the host population. Deriving pVEC1 and pCOM1 from the same replicon ensures strong incompatibility between the plasmids and facilitates the loss of pCOM1. Incompatibility was also shown by the interference in the replication control of pVEC1 and the consequent reduction in copy number when accompanied by pCOM1 in L. lactis. Plasmid pCOM1 can be conveniently used to probe lactococcal hosts for incompatibility with pVEC1 because resident plasmids that support pCOM1 replication in trans would produce erythromycin-resistant colonies. In situations of incompatibility, a different set of two-component vector plasmids could be used. In a study of the lactococcal plasmid pJW563, complementation of a rep-deficient derivative with an incompatible plasmid led to rearrangement of the plasmids in about 50% of the isolates (19). Although a recA-negative derivative of MG1363 was not used in cotransformation assays, this phenomenon was not observed in this study.

The core replicon of pCD4 exceeds the boundaries generally defined for the minimal replicon of lactococcal theta-type plasmids (49). It was shown that the boundaries of the pSC101 minimal replicon are conditional to host and plasmid genes and to sites external to the core replicon (39). Further investigations should reveal binding of host factors needed for replication of pCD4 in L. lactis MG1363 in the region between the NspV and Sau3A1 sites of the pCD4 replicon.

Partial orfX and orfA as well as the oriT region were maintained in pVEC1 to provide convenient cloning sites. The presence of oriT in recombinant plasmids may result in increased transfer of the construction to other lactococci, if trans-acting conjugation factors are expressed by the host. Future generations of two-component vector systems will be constructed without the oriT locus, and a rho-independent terminator will be cloned downstream of repB.

Even though bacteriophages threaten the long-term use of lactococcal starter cultures in the food industry, the introduction of antiphage barriers in lactococcal strains constitutes a promising avenue for limiting their impact. We have shown the potential of the two-component cloning vector system for transferring AbiQ into industrial L. lactis strains and the stable maintenance of the phenotype under laboratory conditions. Field assays will have to be conducted to test the stability of the transferred characters under industrial processing conditions.