Cre-lox-Based System for Multiple Gene Deletions and Selectable-Marker Removal in Lactobacillus plantarum

Abstract

The classic strategy to achieve gene deletion variants is based on double-crossover integration of nonreplicating vectors into the genome. In addition, recombination systems such as Cre-lox have been used extensively, mainly for eukaryotic organisms. This study presents the construction of a Cre-lox-based system for multiple gene deletions in Lactobacillus plantarum that could be adapted for use on gram-positive bacteria. First, an effective mutagenesis vector (pNZ5319) was constructed that allows direct cloning of blunt-end PCR products representing homologous recombination target regions. Using this mutagenesis vector, double-crossover gene replacement mutants could be readily selected based on their antibiotic resistance phenotype. In the resulting mutants, the target gene is replaced by a lox66-P₃₂-cat-lox71 cassette, where lox66 and lox71 are mutant variants of loxP and P₃₂-cat is a chloramphenicol resistance cassette. The lox sites serve as recognition sites for the Cre enzyme, a protein that belongs to the integrase family of site-specific recombinases. Thus, transient Cre recombinase expression in double-crossover mutants leads to recombination of the lox66-P₃₂-cat-lox71 cassette into a double-mutant loxP site, called lox72, which displays strongly reduced recognition by Cre. The effectiveness of the Cre-lox-based strategy for multiple gene deletions was demonstrated by construction of both single and double gene deletions at the melA and bsh1 loci on the chromosome of the gram-positive model organism Lactobacillus plantarum WCFS1. Furthermore, the efficiency of the Cre-lox-based system in multiple gene replacements was determined by successive mutagenesis of the genetically closely linked loci melA and lacS2 in L. plantarum WCFS1. The fact that 99.4% of the clones that were analyzed had undergone correct Cre-lox resolution emphasizes the suitability of the system described here for multiple gene replacement and deletion strategies in a single genetic background.

The development of tools for genetic engineering of gram-positive bacteria is highly valuable for research applications. The classic strategy for obtaining gene deletion variants is based on homologous recombination, using double-crossover integration of heterologous nonreplicating vectors such as pUC (34, 36, 37, 42), pACYC184 (6, 48, 50), or their derivatives in the genome. Several convenient systems that derive from this strategy use conditionally replicating vectors, such as the thermosensitive pG⁺host system (40) and the RepA-dependent lactococcal pORI system (originating from pWV01) (12, 33, 51) and its broad-host-range derivative (42, 51).

In addition to systems that derive from the classic strategy, various site-specific recombination systems such as Flp-FRT (52), Gateway (Invitrogen), ParA-res (31), TnpR-res (13), and Cre-lox are used in mutational strategies. To date, however, these systems have not been available for construction of an unlimited number of mutations in the same genetic background in gram-positive bacteria. For this purpose, the versatile Cre-lox system is a promising candidate. The Cre recombinase is a 38-kDa protein that belongs to the integrase family of site-specific recombinases. It catalyzes cofactor-independent recombination between two of its recognition sites, called loxP. The 34-bp consensus for loxP sites consists of an asymmetrical core spacer of 8 bp, defining the orientation of the loxP site, and two 13-bp palindromic flanking sequences (1, 23). A DNA sequence that is flanked by loxP sites is excised when the loxP sites are convergently oriented, whereas the sequence is inverted when the loxP sites are divergently oriented. Cre recombinase is able to act on both inter- and intramolecular loxP sites, although recombination of intramolecular lox sites is kinetically favorable (32).

The versatile properties of Cre recombinase make it ideal for use in many genetic manipulation strategies. Therefore, the Cre-lox system has been used for a wide variety of eukaryotes such as plants (20), Saccharomyces cerevisiae (54), mice (45, 55), feline cell lines (30), human cell lines (26, 43), and chicken cell lines (5). For example, recombination of intermolecular loxP sites has been used for site-specific integration of transfected DNA into the chromosome (4, 29, 30). Many strategies use recombination of loxP sites to excise the intermediate DNA sequence. This includes work on conditional gene deletions (5) and recombinatorial activation of gene expression (55). In particular, an important application of the Cre-lox system is selectable-marker excision in gene replacements. Commonly used gene replacement strategies result in the introduction of a selectable marker into the genome, facilitating the selection of gene mutations that might cause growth retardation. However, the expression of the marker may result in polar effects on the expression of genes located upstream and downstream. Selectable-marker removal from the genome by Cre-lox recombination is an elegant and efficient way of circumventing this issue and has therefore been used frequently, for example, for plants (20), mouse cell lines (2), and yeast (22).

Notably, use of native loxP sites for consecutive rounds of gene replacement and subsequent selectable-marker removal would lead to the integration into the genome of multiple loxP sites that can still be recognized by Cre. To minimize genetic instability, lox sites containing mutations within the inverted repeats (lox66 and lox71 [Fig. 1 ]) have been used for plants (4) and chicken cell lines (5). Recombination of lox66 and lox71 results in a lox72 site that shows strongly reduced binding affinity for Cre, allowing for repeated gene deletion in a single genetic background.

FIG. 1.

Schematic representation of mutant lox66 and lox71 sites, which after Cre recombination result in a double-mutant lox72 site. Boldfaced sequences are mutated compared to the native loxP site (shown as a reference).

In contrast to the many eukaryotic examples, the Cre-lox system has been used much less frequently for prokaryotic organisms. For example, mechanistic studies on Cre recombination have been performed on Escherichia coli (3, 44). The Cre-lox system was used for conditional gene deletions in Lactobacillus plantarum in the murine gastrointestinal tract (10). In addition, Cre-lox-mediated selectable-marker removal in gene replacements has been used for the gram-negative bacteria Methylobacterium extorquens, Burkholderia fungorum, Escherichia coli, and Pseudomonas aeruginosa (41, 46, 49). However, these experiments used native loxP sites for multiple gene replacements. Previously, it was shown for Lactococcus lactis, Corynebacterium glutamicum, and Salmonella enterica serovar Typhimurium that Cre readily excises or inverts large fragments of DNA flanked by loxP sites in prokaryotes (14, 15, 57, 62), thereby leading to genomic instability.

Here we describe the construction of a Cre-lox-based toolbox for multiple gene deletions in a single genetic background in gram-positive bacteria, a system that combines the advantages of selectable gene replacement and a marker-free, in-frame gene deletion in the final strain. For this purpose, a mutagenesis vector was constructed and used for classic double-crossover replacement of target genes by the selectable-marker cassette lox66-P₃₂-cat-lox71, which can be recombined from the chromosome into a double-mutant lox72 site by transient Cre recombinase expression from a second plasmid.

To validate the effectiveness of the system described here, the mutagenesis targets melA, bsh1, and lacS2 were selected in the model organism L. plantarum WCFS1, whose genome has been sequenced (28). The melA and bsh1 genes are genetically unlinked loci, whereas the melA and lacS2 loci are closely linked genetically. Although this system has been experimentally tested only for this bacterium, the simple, functional implementation of the same basic characteristics for other bacterial hosts will be discussed.

MATERIALS AND METHODS

Bacterial strains, plasmids, and primers.

The bacterial strains, plasmids, and primers used in this study and their relevant features are listed in Tables 1 and 2.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant feature(s)a	Reference(s)
Strains
L. plantarum
WCFS1	Single-colony isolate of L. plantarum NCIMB8826	28
NZ5304	Derivative of WCFS1 containing a lox66-P32-cat-lox71 replacement of bsh1 (bsh1:: lox66-P32-cat-lox71)	This work
NZ5305	Derivative of NZ5304 containing a lox72 replacement of bsh1 (Δbsh1)	This work
NZ5334	Derivative of WCFS1 containing a lox66-P32-cat-lox71 replacement of melA (melA:: lox66-P32-cat-lox71)	This work
NZ5335	Derivative of NZ5334 containing a lox72 replacement of melA (ΔmelA)	This work
NZ5337	Derivative of NZ5335 containing a lox72 replacement of melA and bsh1 (ΔmelA Δbsh1)	This work
NZ5338	Derivative of NZ5335 containing a lox66-P32-cat-lox71 replacement of part of lacS2 (ΔmelA lacS2::lox66-P32-cat-lox71)	This work
NZ5339	Derivative of NZ5338 containing a lox72 replacement of melA and part of lacS2 (ΔmelA ΔlacS2)	This work
E. coli
DH5α	Cloning host; F′ φ80dlacZΔM15 endA1 recA1 hsdR17 (rK− mK+) supE44 thi-1 gyrA96 relA1 Δ(lacZYA-argF)U169 deoR λ−	63
MC1061	Cloning host; F−araD139 Δ(ara-leu)7696 galE15 galK16 Δ(lac)X74 rpsL (Strr) hsdR2 (rK− mK+) mcrA mcrB1	16, 61
L. lactis MG1363	Plasmid-free derivative of NCDO 712	19
Plasmids
pCR-Blunt	Kanr; cloning vector for blunt-end PCR products	Invitrogen
pGID023	Emr; pJDC9 derivative containing the pE194 replication functions; unstable in lactobacilli	24
pGIZ850	Cmr Emr Apr; pUC18 derivative containing a P32-cat cassette that is selectable at the single-copy level	21
pNZ273	Cmr; pNZ124 carrying the promoterless gusA gene from E. coli	47
pNZ5315	Cmr Emr; pNZ7101 derivative containing a lox66 site	This work
pNZ5317	Cmr Emr; pNZ5315 derivative containing a lox71 site	This work
pNZ5318	Cmr Emr; pNZ5317 derivative for multiple gene replacements containing las and pepN terminators	This work
pNZ5319	Cmr Emr; pNZ5318 derivative for multiple gene replacements in gram-positive bacteria	This work
pNZ5340	Cmr Emr; pNZ5319 derivative containing homologous regions up- and downstream of WCFS1 melA	This work
pNZ5325	Cmr Emr; pNZ5319 derivative containing homologous regions up- and downstream of WCFS1 bsh1	This work
pNZ5344	Cmr Emr; pNZ5319 derivative containing homologous regions up- and downstream of lacS2 in the melA::lox72 strain NZ5335	This work
pNZ5346	Cmr; pNZ273 derivative containing the promoter region and ribosome binding site of lp_1144 of WCFS1	This work
pNZ5347	Cmr; pNZ7110 derivative containing the promoter region and ribosome binding site of lp_1144, driving expression of cre	This work
pNZ5348	Emr; pGID023 derivative containing cre under the control of the lp_1144 promoter	This work
pNZ7101	Cmr Emr; pACYC184 derivative for gene replacements	8
pNZ7110	Apr; pUC18 derivative containing the cre gene	9
pNZ84	Cmr; pACYC184 derivative	59

^aKan^r, kanamycin resistant; Cm^r, chloramphenicol resistant; Em^r, erythromycin resistant; Ap^r, ampicillin resistant.

TABLE 2.

Primers used in this study

Primer	Sequence (5′ to 3′)
Bsh1fr1F	GATTAAGTTTGCAGGACATGGAG
Bsh1R	GCCAGCCATTGGAACTTACTCTG
Cat96F	TCAAATACAGCTTTTAGAACTGG
Cat97R	ACCATCAAAAATTGTATAAAGTGGC
CreF	CTAACTCGAGTGATCACCAATTC
CreR	GGCTATCAATCAAAGCAACACG
EryintF	CGATACCGTTTACGAAATTGG
EryintR	CTTGCTCATAAGTAACGGTAC
Lastermi_F	ACGTCCGCGGGGACAATATGGGGTAAGCG
Lastermi_R	AAGAAGATCTCTAAAGCTGACGGGGTAAAC
MellacF	ACCGTTAAGATGCGTGGGACTGG
MellacR	CATAGTAAAATCTTCCCTTCGCTA
pNZ84F	CGGGATCCCAACAGTACTGCGATGAG
pNZ84R	GGGGTACCATCCAGTGATTTTTTTCTCC
80	CGTTTAAACAATTTAAATCTACCGTTCGTATAATGTATGCTATACGAAGTTATGACA
81	TTGTCATAACTTCGTATAGCATACATTATACGAACGGTAGATTTAAATTGTTTAAACGTGCT
82	GACCCATAACTTCGTATAATGTATGCTATACGAACGGTACAGCCCGGGCATGAGCTCCGAT
83	CGGAGCTCATGCCCGGGCTGTACCGTTCGTATAGCATACATTATACGAAGTTATGG
85	GTTTTTTTCTAGTCCAAGCTCACA
87	GCCGACTGTACTTTCGGATCCT
90	CATAGTAAAATCTTCCCTTCGC
91	GTCGTAAAGTGTTCTTCTTAGC
92	GCTAAGGACTAAGCTCAGCC
93	GAGTTTAGGACTACAGGGGGC
95	AATATGTGTACAGGCTGAGCTTAGTCCTTAGCC
101	GATTGCGATTGATATCGATGGC
102	TATGGCAGTACACATAACTAGTAATCCTCC
103	TACTATGCAGTTAACTAAAAGCC
104	CTTACCAATCATGCGTCCCG
106a	GTTCGATAAAGAATGAGGATGGC
107a	TTATCGCAAGTATCTCAAATTGCG
108	TATTGGCCTTCCCACCATTAGC
109	CACGTTATTTACGGCGACGGG
111	CGTGTTGCTTTGATTGATAGCC
113	GCTCGAGCGCGTTATCGGTCCTTTAATTGG
124	GCGCCTCGAGCCGCAATCGCTTTGATTTCTGCC
125	TCCGATTGCATGGCGAGTCGG
126	TACCGTTAAGATGCGTGGG
127	GGCGGGATCCTTATACGGTGACAGCAGACGG
128	CATTAAACCAACGGATGGTCGCG
130	CGTGGTTGGATGGCATTTGGG
137	CTTCTACCCATTATTACAGCA
1144F	GGATCCGTCGACCGCGATTTTTGTATGAGATG
1144R	GGATCCGCTGTTCGCCACCCTTTCTA

As a model strain for gram-positive bacteria, L. plantarum WCFS1 (28) was used. L. plantarum was grown at 37°C in MRS broth (Difco, West Molesey, United Kingdom) without aeration. Escherichia coli strains DH5α (63) and MC1061 (16, 61) were used as intermediate cloning hosts and were grown at 37°C in TY broth (27) with aeration (53). When appropriate, antibiotics were added to the media. For L. plantarum, 10 μg/ml chloramphenicol and 10 μg/ml or (for replica plating) 30 μg/ml erythromycin were used. For E. coli, 10 μg/ml chloramphenicol and 250 μg/ml erythromycin were used.

DNA manipulations.

Plasmid DNA was isolated from E. coli on a small scale using the alkaline lysis method (7). Large-scale plasmid DNA isolations were performed using Jetstar columns as recommended by the manufacturer (Genomed GmbH, Bad Oberhausen, Germany). For DNA manipulations in E. coli, standard procedures were used (53).

L. plantarum DNA was isolated and transformed as described previously (25), with slight modifications. For DNA isolation, an overnight culture of L. plantarum WCFS1 was diluted 20 times in 50 ml of fresh MRS medium and cells were grown to an optical density at 600 nm (OD₆₀₀) of 1. Cells were pelleted by centrifugation for 10 min at 4,500 rpm (Megafuge 1.0R; Heraeus, Hanau, Germany), resuspended in 2.5 ml of THMS buffer (30 mM Tris-HCl [pH 8], 3 mM MgCl₂, 0.73 M sucrose) containing 50 mg/ml lysozyme, and incubated for 2 h at 37°C. Cells were pelleted by centrifugation and resuspended in 2.5 ml Tris-EDTA containing RNase. Subsequently, 125 μl of 10% sodium dodecyl sulfate was added, and cells were incubated for 15 min at 37°C. Then 25 μl of 20-mg/ml proteinase K was added, and the solution was subjected to phenol-chloroform extraction three times. The total DNA was precipitated with isopropanol, washed with 70% ethanol, dried, and taken up in water. For transformation of L. plantarum WCFS1, a preculture in MRS broth was diluted in MRS broth containing 1% glycine and cells were grown to an OD₆₀₀ of 1. Cells were kept on ice for 10 min and pelleted by centrifugation for 10 min at 4,000 rpm (Megafuge 1.0R; Heraeus, Hanau, Germany). Cells were then resuspended in ice-cold 30% polyethylene glycol 1450 and kept on ice for 10 min. Finally, cells were pelleted by centrifugation for 10 min at 4,000 rpm and concentrated 100-fold into ice-cold 30% polyethylene glycol 1450. Subsequently, 40 μl of the cell suspension and a maximum of 5 μl of the plasmid DNA solution were electroporated using a GenePulser Xcell electroporator (Bio-Rad, Veenendaal, The Netherlands) in cuvettes with a 2-mm electroporation gap at 1.5 kV, 25 μF capacitance, and 400 Ω parallel resistance.

Restriction endonucleases, Taq, Pfx, and Pwo DNA polymerases, T4 DNA ligase, and Klenow enzyme were used as specified by the manufacturers (Promega, Leiden, The Netherlands; Boehringer, Mannheim, Germany). Primers were obtained from Genset Oligos (Paris, France).

Mutagenesis vector construction.

To facilitate construction of chromosomal gene replacements, the universal mutagenesis vector pNZ5319 was constructed (Fig. 2A). For this purpose, the pACYC184-derived origin of replication was amplified by PCR (using Pfx polymerase, primers pNZ84F and pNZ84R, and pNZ84 [59] as template DNA) and cloned into the NaeI restriction site of pGIZ850 (21), resulting in pNZ7101 (8). To introduce lox66 and blunt-end restriction sites SwaI and PmeI, 80 and 81 linkers (Table 2) were annealed and cloned into the Bsp1286I and Tth111I restriction sites upstream of the P₃₂-cat cassette of pNZ7101 (8), yielding pNZ5315 (Table 1). Subsequently, 82 and 83 linkers (Table 2), which contained lox71 and blunt-end restriction sites Ecl136II and SrfI, were annealed and cloned into the PpuMI and PvuI restriction sites downstream of the P₃₂-cat cassette of pNZ5315, yielding pNZ5317 (Table 1). Furthermore, both the las operon and pepN terminator regions (38, 39) were amplified by PCR from L. lactis MG1363 (19) using primers LasTermi_F and LasTermi_R and primers 111 and 113, respectively (Table 2). The PCR fragment containing the las terminator and the BglII restriction site was digested with Ecl136II (to avoid introduction of an extra Ecl136II restriction site into the mutagenesis vector) and cloned into pNZ5317, which had been digested with AflIII and treated with Klenow enzyme to generate blunt ends. Subsequently, the PCR fragment containing the pepN terminator and the XhoI restriction site was cloned into the PvuII restriction site of the targeting vector, yielding pNZ5318 (Table 1). Residual and nonfunctional DNA sequences were removed from the pNZ5318 mutagenesis vector by BbsI and SalI digestion, treatment with Klenow enzyme to generate blunt ends, and self-ligation, yielding pNZ5319 (Table 1).

FIG. 2.

(A) Schematic representation of the construction of mutagenesis vector pNZ5319. The 80 and 81 linkers were annealed (Table 2) and cloned into pNZ7101 digested with Bsp1286I and Tth111I, yielding pNZ5315 (Table 1). Subsequently, the 82 and 83 linkers (Table 2) were annealed and cloned into pNZ5315 digested with PpuMI and PvuI, yielding pNZ5317 (Table 1). Then the las terminator (38, 39) and pepN terminator (58) were amplified by PCR from L. lactis MG1363 (19) using primers LasTermi_F and LasTermi_R and primers 111 and 113, respectively (Table 2). The PCR fragment containing the las terminator and the BglII restriction site was digested with Ecl136II, cloned into pNZ5317 digested with AflIII, and treated with Klenow enzyme. Subsequently, the PCR fragment containing the pepN terminator and the XhoI restriction site was cloned into the PvuII site of the targeting vector, yielding pNZ5318 (Table 1). Finally, nonfunctional DNA sequences were removed by BbsI and SalI digestion, treatment with Klenow enzyme, and self-ligation, resulting in pNZ5319 (Table 1). (B) Schematic representation of mutagenesis vector pNZ5319. Indicated are the pACYC184-derived origin of replication (ori), the erythromycin resistance gene (ery), the chloramphenicol resistance gene under the control of the P₃₂ promoter (P₃₂-cat), flanked by lox66 and lox71 sites, and the lactococcal T_las and T_pepN terminators. The presence of rare-cutting blunt-end restriction sites SwaI, PmeI, SrfI, and Ecl136II allows direct cloning of blunt-end PCR products of the flanking regions of the target locus. As an example, the regions used for PCR amplification and cloning into pNZ5319 for construction of a L. plantarum melA mutant are indicated. In addition, the sticky-end restriction sites XhoI and BglII can be used in combination with the blunt-end restriction sites. The presence of two selectable-marker gene cassettes (P₃₂-cat and ery) on the mutagenesis vector allows direct selection of double-crossover integrants based on their antibiotic resistance (Cm^r) and sensitivity (Em^s) phenotype. Black arrowheads indicate primers used in this study. (C) Schematic representation of the in-frame insertion that is left in the genome after lox72 replacement of the target gene. Depending on the restriction sites used for cloning of the homologous DNA fragments (as indicated in the figure) that encompass a whole number of codons of the target gene, the number of foreign nucleotides left in the genome is 45 (cloning using SwaI and SrfI restriction sites), 54 (PmeI/SrfI or SwaI/Ecl136II), or 63 (PmeI/Ecl136II), thereby creating an in-frame deletion of the target gene.

Construction of gene-specific mutagenesis vectors.

For construction of gene-specific mutagenesis vectors, a standard cloning procedure was used (Fig. 2B). Typically, a 1-kb fragment of the upstream sequence and a 1-kb fragment of the downstream sequence of the target locus were amplified by PCR using a proofreading polymerase and cloned into the SwaI or PmeI and Ecl136II or SrfI blunt-end restriction sites of pNZ5319, respectively. When desired, XhoI or BglII sticky-end restriction sites can be used in combination with the blunt-end restriction sites. To ultimately obtain in-frame gene deletions (following Cre recombination [see below and Fig. 2C]), primers were designed in such a way that the 5′- and 3′-flanking regions of the target gene encompassed the first and last five codons of that gene, respectively. The efficiency of cloning of the PCR amplification products of the 5′- and 3′-flanking regions of the target gene into the mutagenesis vector was enhanced by removal of the self-ligation vector in the ligation mixture by digestion with SwaI, PmeI, Ecl136II, or SrfI, depending on the restriction site used for cloning. Colonies harboring the anticipated insert in the desired orientation could be identified effectively by colony PCR using a vector-specific primer (annealing to the P₃₂-cat region; reverse primer 85 for cloning of 5′ sequences into the SwaI or PmeI restriction site and forward primer 87 for cloning of 3′ sequences into the Ecl136II or SrfI restriction site), combined with an insert-specific primer (see also below) (Table 2).

By following the strategy described above, the melA mutagenesis vector pNZ5340 was constructed by successive cloning of the 5′- and 3′-flanking regions of melA (lp_3485) (28) (Fig. 2B) into the SwaI and Ecl136II restriction sites of pNZ5319 (amplified by PCR using Pfx, L. plantarum WCFS1 genomic DNA as a template, and primer sets 91-90 and 92-93, respectively [Table 2]). Clones that harbored the anticipated inserts were identified by PCR using primer sets 91-85 and 87-93, respectively (Table 2; Fig. 2B).

Likewise, the bsh1 replacement vector pNZ5325 was constructed by the successive cloning of PCR products of the 5′- and 3′-flanking regions of WCFS1 bsh1 (lp_3536) (28) (amplified using Pfx polymerase, WCFS1 genomic DNA as a template, and primer sets 101-102 and 103-104, respectively [Table 2]) into pNZ5319 digested with SwaI and Ecl136II, respectively. Clones that harbored the correct inserts were identified by PCR using primer sets 101-85 and 87-104, respectively (Table 2).

The third locus, lacS2 (lp_3486) (28), was targeted for mutagenesis in a L. plantarum WCFS1 ΔmelA background (NZ5335 [Table 1]). For construction of the corresponding mutagenesis vector pNZ5344, PCR products of 5′ and 3′ regions of lacS2 (amplified using Pfx polymerase, ΔmelA [NZ5335] template DNA, and primer sets 124-125 and 126-127, respectively [Table 2]) were digested with XhoI and BamHI, respectively, and sequentially cloned into XhoI- and SwaI-digested and BglII- and Ecl136II-digested pNZ5319. Clones harboring the correct insert were identified using primer sets 124-85 and 87-127, respectively (Table 2).

Mutant construction.

In order to engineer lox66-P₃₂-cat-lox71 gene replacements, 4 μg of the appropriate mutagenesis vector was transformed into L. plantarum WCFS1 by electroporation as described previously (25). Chloramphenicol-resistant (Cm^r) transformants were selected and replica plated to check for an erythromycin-sensitive (Em^s) phenotype. Candidate double-crossover clones (Cm^r Em^s) were analyzed by PCR amplification of the cat and ery genes using primers cat96F-cat97R and eryintF-eryintR, respectively (Table 2). Correct integration of the lox66-P₃₂-cat-lox71 cassette into the genome (for melA replacement, see Fig. 3A) was confirmed by PCR amplification of the flanking regions of the integrated lox66-P₃₂-cat-lox71 cassette using primers annealing uniquely to genomic sequences (for melA, primer 108 for amplification of the 5′ region and primer 109 for amplification of the 3′ region) combined with the mutagenesis vector-specific primers 85 and 87, respectively, which annealed to the P₃₂-cat region (Table 2; Fig. 3A). Likewise, for analysis of the flanking regions of the lox66-P₃₂-cat-lox71 replacement of bsh1, primers 106a-85 and 87-107a were used; for lacS2 replacement, primers 130-85 and 87-109 were used (Table 2; Fig. 3A).

FIG. 3.

(A) Strategy for construction of double-crossover mutants and subsequent Cre-lox-mediated selectable-marker removal (melA and subsequent lacS2 mutagenesis are shown as an example). After transformation of the gene-specific mutagenesis vector, the target gene (melA) is replaced by a lox66-P₃₂-cat-lox71 cassette by homologous recombination. After selection of the double-crossover mutant, the lox66-P₃₂-cat-lox71 cassette is resolved to a single double-mutant lox72 site by transient Cre expression from a curable plasmid. Subsequently, the next round of gene replacement (lacS2) can be performed. Black arrowheads indicate primers that were used for amplification of the homologous DNA fragments used for targeting cloning in the pNZ5319 mutagenesis vector (primers 91 and 90 and primers 92 and 93 for melA targeting; primers 124 and 125 and primers 126 and 127 for lacS2 targeting), confirmation of correct integration of the mutagenesis vector into the genome (primers 108 and 85 and primers 87 and 109 for melA targeting; primers 130 and 85 and primers 87 and 109 for lacS2 targeting), and confirmation of Cre resolution of the lox66-P₃₂-cat-lox71 cassette (primers 108 and 109 for melA targeting). (B) Possible products of Cre recombination during multiple gene replacement in a single genetic background, as exemplified by recombination of lox66-P₃₂-cat-lox72 at the lacS2 locus in a ΔmelA background. Black arrowheads indicate primers that were used to distinguish the four possible products of lacS2::lox66-P₃₂-cat-lox72 recombination in a ΔmelA background by PCR (primers 128, 137, and 95 in one reaction mixture). The corresponding PCR product sizes are given on the right. (Diagram 1) No recombination occurred. (Diagram 2) Correct recombination occurred, removing the P₃₂-cat selectable marker cassette. (Diagram 3) Incorrect recombination between lox66 and lox72 occurred, resulting in deletion of the intermediate region. (Diagram 4) Incorrect recombination between lox71 and lox72 occurred, resulting in deletion of the intermediate region.

Transient Cre expression vector.

For expression of Cre, a 100-bp region upstream of the L. plantarum WCFS1 gene lp_1144 (28), containing the functional promoter P₁₁₄₄ (11), was amplified by proofreading PCR using primers 1144F and 1144R and then cloned into pCR-Blunt (Invitrogen, Breda, The Netherlands) (Tables 1 and 2). To assess the promoter activity of the P₁₁₄₄ fragment, it was digested from the pCR-Blunt vector with BamHI and cloned upstream of the gusA gene into BglII-digested pNZ273 (47), resulting in pNZ5346. Quantitative β-glucuronidase activity measurements were performed as described previously (18). To determine cre expression, the promoter fragment was digested from pNZ5346 with SalI and BamHI and then cloned upstream of cre into SalI- and BamHI-digested pNZ7110 (9), yielding pNZ5347. Finally, the P₁₁₄₄ promoter-cre cassette was digested from pNZ5347 using KpnI and HindIII and then cloned into correspondingly digested pGID023, yielding pNZ5348, which is unstable in lactobacilli (24).

The stability of the pGID023 replicon in L. plantarum was determined in duplicate by culturing for 10 generations without selection pressure. Subsequently, cells were plated with and without selection pressure, and the CFU count per milliliter was determined. The presence of pGID023 in L. plantarum was verified by colony PCR using primers eryintF and eryintR (Table 2).

Cre-mediated mutant locus resolution.

To excise the P₃₂-cat selectable-marker cassette from the chromosome, 4 μg of the transient erythromycin-selectable cre expression plasmid pNZ5348 was transformed into lox66-P₃₂-cat-lox71 gene replacement mutants. After 48 to 72 h of growth, Em^r colonies were checked by PCR for the presence of cells that had undergone Cre-mediated recombination, using primers spanning the recombination locus (specifically, primers 108 and 109 for melA [Fig. 3A], bsh1fr1F and bsh1R for bsh1, and 128, 137, and 95 in one PCR for lacS2 [Fig. 3B]) (Table 2). The pNZ5348 vector was cured from appropriate colonies of L. plantarum mutants by growth without erythromycin selection pressure for 10 generations. To obtain clonal strains, single-colony isolates were selected for which curing of the Cre expression vector was confirmed by the absence of PCR amplification of ery (using primers eryintF and eryintR) and cre (using primers creF and creR) (Table 2), and Cre-mediated recombination was confirmed by PCR amplification as described above. Additionally, the presence of a correctly resolved lox72 site (Fig. 2C) was confirmed by sequencing (Baseclear, Leiden, The Netherlands) using primers 95, bsh1fr1F, and 95 for melA, bsh1, and lacS2 replacement, respectively.

Southern blot analysis.

To confirm the genotype of WCFS1 melA::lox66-P32-cat-lox71 (NZ5334), ΔmelA (NZ5335), ΔmelA lacS2::lox66-P32-cat-lox71 (NZ5338), and ΔmelA ΔlacS2 (NZ5339) mutant derivatives (Table 1), Southern blot analysis was performed as described previously (53) using AvaI and DraI digests of total DNA. As a probe, a PCR amplification product of the intergenic region of melA and lacS2 (amplified with Taq polymerase, WCFS1 total DNA, and primers mellacF and mellacR) was used.

HPLC assay of bile salt hydrolase activity.

To determine the bile salt hydrolase activity of L. plantarum, an overnight culture was inoculated 1:10 into fresh MRS medium and cells were grown to an OD₆₀₀ of 5. Cells were pelleted by centrifugation for 10 min at 4,500 rpm (Megafuge 1.0R; Heraeus, Hanau, Germany) and resuspended in MRS medium to an OD₆₀₀ of 100. For determination of bile salt hydrolase activity, wild-type L. plantarum cells were diluted in MRS to an OD₆₀₀ of 10, whereas cells of bsh1 deletion strains were used undiluted. Conversion of the bile salt glycocholic acid (Sigma, Zwijndrecht, The Netherlands) was determined by high-performance liquid chromatography (HPLC) as described previously (17). Separations were carried out with a reversed-phase resin-based column (PLRP-S; 5-μm particles; 300-Å pore size; 250-mm column length; 4.6-mm inner diameter; Polymer Laboratories, Shropshire, United Kingdom) and a matching precolumn. Bile salts were detected using a pulsed amperometric detector (EG&G Princeton Applied Research, Princeton, NJ) equipped with a gold working electrode and a reference electrode (Ag/AgCl).

Nucleotide sequence accession number.

The sequence of the pNZ5319 mutagenesis vector is available in the GenBank database under accession number DQ104847. The sequence of the P₁₁₄₄-cre cassette of the cre expression plasmid pNZ5348 is available in the GenBank database under accession number DQ340306.

RESULTS

Strategy of gene replacement and selectable-marker removal.

For generation of gene deletions, the mutagenesis vector pNZ5319 (Fig. 2B) was constructed and implemented in the gram-positive model organism Lactobacillus plantarum WCFS1 (28). This medium-copy-number E. coli cloning vector contains a PACYC184 origin of replication, which is suitable for “suicide” mutagenesis in lactic acid bacteria, as described previously (18, 59).

Cloning of PCR-amplified homologous DNA fragments upstream and downstream of the mutagenesis locus (necessary for targeting of the mutagenesis vector to the genomic locus) in pNZ5319 is facilitated by the presence of PmeI, SwaI, SrfI, and Ecl136II rare-cutting blunt-end restriction sites in the vector, flanked by the lactococcal las (38, 39) and pepN (58) terminators (Fig. 2B).

Furthermore, the mutagenesis vector pNZ5319 contains both a chloramphenicol (lox66-P₃₂-cat-lox71) and an erythromycin (ery) resistance cassette that can be selected at the single-copy chromosomal level. Following transformation of the mutagenesis vector to the target organism, the antibiotic resistance cassettes allow for direct selection of double-crossover mutants on the basis of their antibiotic resistance and sensitivity phenotype. In the resulting double-crossover mutant strains, the target gene is replaced by a lox66-P₃₂-cat-lox71 cassette. The presence of the lox sites renders the P₃₂-cat cassette excisable from the genome of the double-crossover mutant strain by Cre recombinase. Using native loxP sites, multiple gene deletions would lead to the integration into the genome of multiple loxP sites that can cause genomic instability in the presence of Cre (14, 15, 57). Therefore, loxP sites containing mutations within the inverted repeats (lox66 and lox71) (4) were used (Fig. 1); after Cre recombination, this strategy results in a double-mutant loxP site (lox72), which shows strongly lowered affinity for Cre.

Transient Cre expression was driven by P₁₁₄₄ (upstream of the lp_1144 gene) from the pNZ5348 vector, which contains a replicon that is unstable in lactobacilli (24). Although this plasmid can be introduced into L. plantarum when selective (Em^r) conditions are maintained, its intrinsic instability in this host resulted in rapid curing of the plasmid when selective pressure was relieved. In our experiments, a ca. 1,000-fold reduction in plasmid retention was obtained by culturing for 10 generations in the absence of selection pressure, as determined by antibiotic resistance profiling and PCR (data not shown). Finally, the P₁₁₄₄ promoter is predicted to drive constitutive, moderate levels of transcription of the downstream gene (pcrA), encoding a DNA helicase. Previous experiments in our laboratory using gusA (which encodes β-glucuronidase) as a promoter probe confirmed the prediction of the characteristics of the P₁₁₄₄ promoter (data not shown).

Notably, in our design, the PCR-amplified homologous DNA fragments used for locus targeting include a number of complete codons of the 5′ and 3′ ends of the gene targeted for mutagenesis. Thus, depending on the restriction sites in pNZ5319 that were used for cloning of the 5′ and 3′ homologous regions, the total number of foreign nucleotides left in the genome after Cre-mediated excision of the selectable-marker cassette is 45 (using SrfI and SwaI), 54 (using SrfI and PmeI or Ecl136II and SwaI), or 63 (using Ecl136II and PmeI), thereby generating an in-frame deletion (Fig. 2C).

Single-locus mutagenesis.

To validate the gene deletion system, the melA gene (lp_3485) and the bsh1 gene (lp_3536) of L. plantarum WCFS1 (28) were chosen as target genes for single-locus mutagenesis. The melA gene encodes an α-galactosidase (EC 3.2.1.22), which is predicted to be involved in hydrolysis of the sugar melibiose into galactose and glucose. In L. plantarum, the melA gene is induced by melibiose and repressed by glucose (56). The melA gene is a convenient target for mutagenesis, since it is predicted to encode a nonredundant function in L. plantarum WCFS1 and its phenotype is likely to be measurable both quantitatively (by hydrolysis of a chromogenic substrate) and qualitatively (by the absence of growth on melibiose as a sole carbon source). The bsh1 gene of L. plantarum WCFS1 is predicted to encode a bile salt hydrolase (Bsh; EC 3.5.1.24). Bile salt hydrolases catalyze cleavage of the amino acid moiety from the steroid nucleus of conjugated bile salts. The DNA sequence of L. plantarum WCFS1 bsh1 is 99% identical to the sequence of the bsh gene of L. plantarum LP80, for which a previously constructed bsh mutant derivative was shown to be deficient in bile salt hydrolase activity (35). However, the L. plantarum WCFS1 genome appears to be fourfold redundant for this function, containing genes annotated as bsh1 to bsh4 (28). The L. plantarum bsh1 gene was selected as a target for mutagenesis in order to investigate whether it is the sole bile salt hydrolase-encoding gene in this strain.

For construction of melA and bsh1 mutant strains, gene-specific mutagenesis vectors pNZ5340 and pNZ5325 were constructed by direct cloning of PCR-amplified homologous DNA fragments upstream and downstream of melA or bsh1, respectively, and transformed to L. plantarum WCFS1. The genetic events involved are schematically illustrated for the melA locus in Fig. 3A. During primary selection of melA double-crossover mutants, 34 Cm^r colonies were found, 9 of which appeared to display the Em^s phenotype correlating to the melA::lox66-P₃₂-cat-lox71 genotype. A single-colony isolate was selected and designated NZ5334 (Table 1). For bsh1 mutagenesis, a slightly lower number of Cm^r colonies, 12, was found during primary selection of double-crossover mutants; the proportion of colonies displaying an Em^s phenotype (correlating to a bsh1::lox66-P₃₂-cat-lox71 genotype), 3 of 12, appeared to be similar to that observed for the melA locus. A single-colony isolate was selected and designated NZ5304 (Table 1). Correct integration of the lox66-P₃₂-cat-lox71 cassette into the genome was confirmed by PCR and by Southern blot analysis (Fig. 4).

FIG. 4.

Southern blot analysis of WCFS1 and its melA::lox66-P₃₂-cat-lox71 (NZ5334), ΔmelA lacS2::lox66-P₃₂-cat-lox71 (NZ5338), ΔmelA ΔlacS2 (NZ5339), and ΔmelA (NZ5335) derivatives. The restriction enzymes (DraI and AvaI, respectively) used for restriction of genomic DNA of the strains were chosen in such a way that all strains could be distinguished based on the predicted sizes of the hybridizing bands (given below the blot in base pairs). The sizes of the hybridizing bands of the various strains were as expected (compare the predicted sizes with the positions of the bands relative to the DNA ladder on the right). DraI, DraI digestion of total DNA; AvaI, AvaI digestion of total DNA.

To resolve the lox66-P₃₂-cat-lox71 cassette at the mutation locus to a single in-frame lox72 site, the erythromycin-selectable Cre expression plasmid pNZ5348 was transformed to the NZ5334 and NZ5304 double-crossover mutant strains. After 48 h of incubation, 22 Em^s colonies of each strain were analyzed, of which 9 (for the melA-mutagenized strain) and 15 (for the bsh1-mutagenized strain) colonies were found to contain cells that had undergone Cre recombination, as determined by PCR. Subsequently, the Cre expression vector was allowed to be lost from the cells by culturing without selective pressure. Single-colony isolates (designated NZ5335 [ΔmelA] and NZ5305 [Δbsh1] [Table 1]) were selected and analyzed for loss of the Cre expression vector and correct Cre-mediated recombination of the target loci, as determined by an Em^s phenotype, PCR (Fig. 3A), DNA sequencing of the mutated locus, and Southern blot analysis (Fig. 4).

To further confirm the genotypes of these mutants, the anticipated physiological consequences were analyzed. As expected, growth of the L. plantarum WCFS1 melA deletion mutant NZ5335 on MRS plates in the presence of glucose as a carbon and energy source was unaffected compared to growth of the wild type. However, in the presence of melibiose as the sole carbon and energy source, growth of the L. plantarum WCFS1 melA deletion mutant NZ5335 on MRS plates was completely abolished, whereas growth of the wild-type strain was normal (data not shown). This confirms the lack of functional redundancy for this gene in the L. plantarum WCFS1 genome and indicates that melA is essential for growth on melibiose. In analogy, quantitative determination of the bile salt hydrolase activity of the bsh1 deletion mutant NZ5305 revealed that bsh1 deficiency leads to a 90% reduction in bile salt hydrolase activity in L. plantarum WCFS1 (Fig. 5).

FIG. 5.

Bsh activities of Δbsh1 (NZ5303), ΔmelA (NZ5335), and ΔmelA Δbsh1 (NZ5337) strains relative to that of WCFS1 (taken as 100%). Deletion of bsh1 in WCFS1 or NZ5335 results in a 90% reduction in Bsh activity.

Mutagenesis of genetically unlinked loci in a single genetic background.

To establish that the mutagenesis system constructed was indeed suitable for efficient construction of successive selectable-marker-free gene deletions, the bsh1 gene was mutated in the ΔmelA derivative (NZ5335) of L. plantarum WCFS1.

After transformation of the bsh1 mutagenesis vector pNZ5325 into NZ5335 (Table 1), 38 Cm^r colonies were found during primary selection of double-crossover integrants, and 4 of these colonies displayed an Em^s phenotype, correlating to a bsh1::lox66-P₃₂-cat-lox71 genotype (as confirmed by PCR). Subsequently, the lox66-P₃₂-cat-lox71 cassette in the ΔmelA, bsh1::lox66-P₃₂-cat-lox71 strain was efficiently resolved to a single in-frame lox72 site by transient Cre expression from the erythromycin-selectable vector pNZ5348. After 72 h of incubation, all of the 15 Em^r colonies that were analyzed contained cells that had undergone Cre-mediated recombination as confirmed by PCR. Following removal of the Cre expression vector by culturing without selection pressure, single-colony isolates were taken and designated NZ5337 (L. plantarum ΔmelA Δbsh1). Loss of the Cre expression vector and correct recombination of the target locus were confirmed by PCR and DNA sequencing of the mutated locus. Furthermore, the anticipated double-mutant (ΔmelA Δbsh1) phenotype could be confirmed, i.e., by lack of growth on melibiose as a sole carbon and energy source and strongly reduced Bsh activity (Fig. 5).

Mutagenesis of genetically linked loci in a single genetic background.

The approaches described above showed that mutagenesis of two genetically unlinked loci (melA and bsh1) could be performed effectively. However, secondary-gene replacement at a locus that is closely linked to the initially mutated locus leads to close proximity of the lox66-P₃₂-cat-lox71 cassette and the lox72 site of the primarily targeted locus (Fig. 3), which could lead to incorrect resolution by the Cre enzyme. Provided that Cre recognizes all available lox sites with a certain affinity, four different modes of resolution of the locus could occur, involving the lox66 and lox71 sites alone or involving the lox66 and/or lox71 site in combination with the lox72 site. These modes of resolution can be distinguished by PCR analysis of resulting individual clones (Fig. 3B). To evaluate the frequency of these potential artifact resolutions, a small part of the WCFS1 lacS2 gene (lp_3486) (28), which is located directly upstream of melA (lp_3485) and in the same orientation as melA, was chosen as a secondary target for mutagenesis in the WCFS1 ΔmelA strain NZ5335. In the presence of glucose as a carbon and energy source, mutation of lacS2 was not expected to affect growth. The lacS2-specific mutagenesis vector pNZ5344 was constructed by cloning of PCR-amplified homologous DNA fragments into the mutagenesis vector pNZ5319. Following transformation of pNZ5344 into the ΔmelA strain, double-crossover integrants were selected based on their Cm^r and Em^s phenotype. Out of 22 Cm^r transformants, 12 clones displayed an Em^s phenotype and the corresponding ΔmelA lacS2::lox66-P₃₂-cat-lox71 genotype, as could be confirmed by PCR and Southern blot analysis (Fig. 4). After Cre-mediated resolution, recombination patterns could be distinguished by PCR using primers 128, 137, and 95 in a single PCR (Table 2; Fig. 3B). Of the 192 colonies analyzed, 179 gave a PCR product, the vast majority of which (126 colonies [70.4%]) appeared to have undergone correct Cre recombination (ΔmelA ΔlacS2), while almost all of the residual colonies (52 colonies [29.1%]) appeared to contain a mixed population of correctly resolved (ΔmelA ΔlacS2) and unresolved (ΔmelA lacS2::lox66-P₃₂-cat-lox71) cells. Only a single colony (0.6%) that had undergone incorrect recombination between the lox66 site at the lacS2 locus and the lox72 site at the melA locus was detected, while no colonies that had undergone incorrect recombination between the lox71 site at the lacS2 locus and the lox72 site at the melA locus were detected. To further establish correct Cre-mediated resolution, the genotype of a single ΔmelA ΔlacS2 colony (designated NZ5339) was confirmed by DNA sequencing of the mutated locus and Southern blot analysis (Fig. 4). Taken together, these experiments support the robustness and selectivity of Cre-mediated resolution of the mutant loxP sites introduced into the genome and exemplify the suitability of the mutagenesis system presented here for the construction of multilocus mutants in a single genetic background.

DISCUSSION

Here we describe the construction of an effective Cre-lox-based toolbox for multiple gene replacements in a single genetic background. The procedure consists of three steps. In the first step, a gene-specific mutagenesis vector is constructed using standard cloning procedures. In the second step, the target gene is replaced by a lox66-P₃₂-cat-lox71 cassette by double-crossover recombination. In the third step, the lox66-P₃₂-cat-lox71 cassette introduced at the target locus is resolved by transient Cre expression, resulting in an in-frame lox72 site in the genome (Fig. 1; 2C).

The mutagenesis vector presented facilitates efficient direct cloning of blunt-end PCR amplification products that represent homologous 5′- and 3′-flanking regions of any desired target locus in rare-cutting blunt-end restriction sites in the medium-copy-number E. coli cloning vector. The presence of two selectable-marker gene cassettes on the mutagenesis vector enables direct selection of double-crossover integrants. Direct selection of mutants provides a major advantage in procedures aiming to generate gene mutations that might result in growth retardation, where mutants may not be obtained when a method that implements a single selectable marker on the mutagenesis vector (i.e., pUC18 or pG⁺host) is used (12, 64). However, the presence of a selectable marker in the chromosome hampers multiple gene replacements and may influence the expression of surrounding genes. As opposed to most mutagenesis systems that enable direct selection of double-crossover mutants by implementation of two selectable markers on the mutagenesis vector (such as pUC18ery) (60), the method described here adds the option of selectable-marker removal from the genome by transient Cre expression in the gene replacement background using an unstable and easily curable Cre-expressing vector that ensures the removal of Cre activity before the introduction of additional mutations. Thus, our system provides an important advantage over mutagenesis systems for gram-positive bacteria that have been described to date.

Targeting of the melA and bsh1 locus of the gram-positive model organism L. plantarum WCFS1 (28) showed the effectiveness of our system for construction of single-locus double-crossover mutants and subsequent marker removal. The melA gene was shown to be essential for the growth of L. plantarum WCFS1 on melibiose as a sole carbon and energy source. Although predicted to be fourfold redundant, the bsh1 gene appeared to be the major bile salt hydrolase-encoding gene in L. plantarum WCFS1. This is in good agreement with the finding that L. plantarum WCFS1 bsh1 is almost identical to the bsh gene of L. plantarum LP80, for which a previously constructed mutant was shown to be bile salt hydrolase deficient (35). However, to determine whether bsh2, bsh3, and/or bsh4 plays a role in the remaining Bsh activity in the Δbsh1 derivative of strain WCFS1, further investigation is required.

Furthermore, successful generation of multiple gene deletions by Cre-lox recombination depends on correct resolution events even in the presence of previously integrated lox72 sites. In the work presented here, the efficiency of Cre-lox-based removal of the P₃₂-cat selectable-marker cassette in multiple gene replacements was determined by successive mutagenesis of genetically unlinked and closely linked chromosomal loci. Thus, bsh1was effectively deleted in an L. plantarum WCFS1 ΔmelA strain, resulting in a ΔmelA Δbsh1 double-mutant strain, which displayed the anticipated combination of the single-locus mutant phenotypes. During successive mutagenesis of the genetically closely linked loci melA and lacS2 in L. plantarum WCFS1 (Fig. 3), Cre-mediated resolution of lox66 and lox71 in close proximity to a lox72 site was shown to occur correctly in 99.4% of the colonies that were successfully analyzed. Moreover, no colonies were found to have undergone incorrect resolution of lox71 and lox72. However, the latter type of recombination (lox71 and lox72) results in a lox71 site, which in turn can recombine with the still remaining lox66 site, thereby generating a single lox72 site (Fig. 3B). This situation is indistinguishable from direct recombination of lox66 and lox72, which was found to occur in only a single colony tested (0.6%). The fact that correct Cre resolution of lox66 and lox71 occurs almost exclusively even in close proximity to a lox72 site emphasizes the selectivity of the Cre enzyme and underlines the advantage of this system relative to mutagenesis systems that employ native loxP sites, which have been used for some gram-negative bacteria (41, 46, 49). Especially for prokaryotes, where only one chromosome is present, the use of loxP sites to resolve the selectable-marker cassette in multiple gene deletions is highly undesirable, because it can lead to (large) genomic inversions and/or rearrangements (14, 15, 57).

Although the Cre-lox system as described here was used for L. plantarum WCFS1, it can easily be adapted to other gram-positive bacteria, including other lactic acid bacteria such as Lactococcus lactis and Streptococcus thermophilus. The only prerequisite for the use of this system is the availability of a method to establish transient Cre expression. For example, a temperature-sensitive vector such as pG⁺host (40) can be used for this purpose by subcloning of the P₁₁₄₄-Cre expression cassette present in pNZ5348. The applicability of this approach for L. lactis has already been established in our laboratory (R. Brooijmans, personal communication). More-advanced alternative possibilities could also be employed, including, for example, strictly controlled Cre expression from a permanently present plasmid or from a chromosomal locus, or expression of Cre from a plasmid in which Cre or the origin of replication is flanked by lox sites to ensure cessation of Cre activity before construction of additional mutations in the same genetic background. Analogously, the pNZ5319 mutagenesis vector could be modified by replacement of the pACYC184 replicon with a temperature-sensitive replicon (pG⁺host derived) in order to facilitate mutagenesis in bacterial species with low transformation frequencies that preclude the suicide mutagenesis strategy employed here. These relatively simple modifications allow functional implementation of the Cre-lox-based mutagenesis system with a range of other gram-positive or gram-negative bacteria and exemplify the broad applicability of the system presented here.

In conclusion, the multiple gene deletion system for gram-positive bacteria presented here allows for effective, standardized construction of double-crossover mutants that can be introduced in a single genetic background by a simple repetitive procedure using mutant lox sites for Cre-mediated selectable-marker removal.