Development of an Efficient In Vivo System (P_junc-TpaseIS ₁₂₂₃ ) for Random Transposon Mutagenesis of Lactobacillus casei

Plasmid pVI116 was constructed as described in Fig. 1 and its legend to provide the transposase of IS1223 expressed under the control of the L. delbrueckii subsp. bulgaricus P_hlbA promoter (9). Plasmid pVI115 was constructed from pVI162 (see Table 1 for details of construction) to provide the transposition substrate corresponding to an abutted IRR-IRL junction of IS1223 separated by 3 base pairs, named P_junc (Fig. 1A). It replicates only in the TG1 RepA strain of Escherichia coli (18). Plasmid pVI116 (P_hlbA-Tpase_IS1223) and the control plasmids, pGB2 and pVI113 (Tpase_IS1223), were electroporated into E. coli TG1 as previously described (12). The resulting strains were electroporated with identical amounts (100 ng) of pVI115, as a nonreplicative source of the P_junc, or pVI119 (Table 1), as a P_junc-less nonreplicative control, and pGEMT, as a replicative plasmid. Cells were directly plated on LB agar plates supplemented with chloramphenicol (10 μg/ml) or ampicillin (50 μg/ml). Plates were incubated for 20 h at 37°C, and colonies were counted to score for integration or transformation. Since pVI115 cannot replicate in these E. coli strains, the resulting Cm^r transformants were considered pVI115 chromosomal integrants. The integration efficiency obtained with pVI115 in the absence of Tpase_IS1223 was very low (∼10⁻⁸) compared to that of the strain carrying a Tpase_IS1223 without a cloned promoter (∼10⁻⁶), as well as that obtained with Tpase_IS1223 fused to the P_hlbA promoter (up to 10⁻³). In all strains, the integration efficiency of pVI119 was very close to the background level observed in the absence of identified promoter (10⁻⁸ to 10⁻⁷). These results clearly show that Tpase_IS1223 triggers pVI115 integration using the P_junc substrate in trans and that the P_hlbA promoter drastically enhances the expression of Tpase_IS1223 in E. coli. The transposon as developed in this work mimics the double-strand DNA intermediate and integrates as a nonreplicative element in the target sequence (Fig. 1C). The target sites of 19 integrants were determined by direct sequencing of genomic DNA (GATC Biotech) using the primer OLB215 (Table 2), which targets one transposon extremity (Fig. 1C). Fourteen insertions had occurred in different putative open reading frames; three were located in noncoding regions and two in repetitive extragenic palindromic (REP) sequences (data not shown). To confirm randomness of integration and saturation of the chromosome by pVI115, a pVI115-mutagenized Lac-positive (Lac⁺) E. coli strain culture was diluted and spread onto LB medium with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) to screen Lac-negative (Lac⁻) mutants. Of the 10 Lac⁻ clones analyzed, none was redundant, strongly supporting the randomness of integration and saturation of the E. coli chromosome by pVI115. Altogether, these results validated the fact that Tpase_IS1223 is active in E. coli and efficiently recognizes P_junc as a substrate leading to random transposition.

The highly efficient P_junc-Tpase_IS1223 transposition system was adapted to LAB, namely, L. delbrueckii subsp. bulgaricus and L. casei. P_hlbA-Tpase_IS1223 was cloned in the E. coli–Gram-positive bacterium shuttle vector, pVI1056, to give pVI129 (Fig. 1B and Table 1), a plasmid providing the Tpase_IS1223. Plasmid pVI129 possesses the pIP501 replication origin, which is thermosensitive in several Gram-positive bacteria (4, 15, 21, 26, 30), including L. delbrueckii subsp. bulgaricus (31). This property allows the efficient elimination of the transposase-expressing plasmid so as to avoid further transposition events. The P_junc sequence was combined with the erythromycin resistance cassette ermB to generate pVI110 (Fig. 1A), the suicide transposon plasmid.

L. delbrueckii subsp. bulgaricus VI104 and L. casei LC334 cells were first transformed with pVI129 as previously described (2, 32), and the resulting strains, LBpVI129 and LCpVI129, respectively, were then electroporated with 4 μg of pVI110, an optimal amount determined by preliminary assays with different amounts of plasmid DNA (data not shown). The cells were directly plated on MRS agar plates supplemented with erythromycin (5 μg/ml), and the plates were incubated for 2 days at 42°C or 37°C for VI104 and LC334 strains, respectively, under static anaerobic growth conditions. The Em^r colonies obtained after transformation with suicide transposon pVI110 were considered genomic (chromosome or indigenous plasmid) integrants. The transposition efficiency was determined by the number of Em^r colonies obtained for 50 μl of electrocompetent cells with 1 μg of pVI110 plasmid. With LBpVI129, the number of integrants was estimated between 300 and 1,500 for ∼2 × 10⁸ viable cells, while with LCpVI129, this number reached between 2,500 and 7,500 for ∼10⁹ viable cells in more than 10 independent experiments. These results demonstrate that P_junc-TpaseIS₁₂₂₃ is functional and efficient in the two species. Negative controls were made using two other combinations of plasmids introduced successively: (i) pVI1056, as a Tpase_IS1223-nonexpressing vector, and pVI110 and (ii) pVI129 and pVI137, a P_junc-less plasmid. For the two strains, less than 10 Em^r colonies were obtained using these plasmid combinations. These last data show that the P_junc is not the substrate of a genomic indigenous putative transposase produced by L. delbrueckii subsp. bulgaricus or L. casei and that P_junc is strictly required for pVI110 integration in these two species.

Preliminary results (data not shown) revealed that the growth of L. casei was seriously affected at temperatures above 40°C, making the elimination of pVI129 at a high temperature undesirable. The segregational stability of pVI129 in L. casei at 37°C was estimated at 86% per generation as described by Heap et al. (20). Thus, the inherent pVI129 instability and the resulting loss of Tpase_IS1223 in L. casei mutants considerably limit the risk of genomic instability of the mutants.

pVI110 insertion mutants of L. delbrueckii subsp. bulgaricus and L. casei were randomly selected. Mutant genomic DNA digested by NgoMIV for L. delbrueckii subsp. bulgaricus and by HindIII for L. casei was analyzed by Southern hybridization with a pVI110-specific probe generated by PCR amplification with primers ERYF and ERYR (Table 2). Plasmid pVI110 integrated in a single locus in each mutant, except for mutant 2 of L. casei, which presented two bands of different intensity, suggesting two distinct mutants in the sample (Fig. 2). Overall, the diversity of fragment sizes among the tested clones indicated that pVI110 had inserted randomly into both the L. delbrueckii subsp. bulgaricus and L. casei genomes. L. casei strain LC334 carries pLSEIA (GenBank accession number NC_008502.1), a 29-kbp indigenous plasmid. Since indigenous plasmids are often targets of preferential insertion, leading to a reduction in efficiency of random transposon mutagenesis in chromosomal targets (28), we determined the plasmid copy number (PCN) of pLSEIA by quantitative PCR (qPCR). Real-time PCRs were performed as previously described (25) from whole DNA of L. casei with primer pairs LSEI0004F-LSEI0004R and LSEI0145F-LSEI0145R (for the chromosome) and LSEIA04F-LSEIA04R and LSEIA13F-LSEIA13R (for pLSEIA) (Table 2). The PCN was determined using the following equation: PCN = (Ec)^C_Tc/(Ep)^C_Tp, considering different amplification efficiencies [E = 10^(−1/slope)] and cycle threshold (C_T) values for the two amplicons (chromosome, c, and plasmid, p) (33). The PCN of pLSEIA was estimated at 2.8 ± 1.4 (mean ± standard deviation) plasmid copies per chromosome (from 3 independent DNA extracts). Taking into account the respective sizes of pLSEIA (29 kbp) and the L. casei chromosome (2.9 Mbp), the theoretical percentage of pVI110 nonpreferential integration in pLSEIA should be between 1 and 5%.

To confirm the diversity of mutants and to identify the nature of the target sequences of the pVI110 transposon, the randomly selected mutants were identified by genomic DNA sequencing with primers OLB215 and OLB221 (Table 2), which target the transposon sequence extremities (Fig. 1C and Table 3). In regard to L. delbrueckii subsp. bulgaricus, more than 80% (n = 17) of sequenced targets were located in intergenic regions (IGR). Although four mutants were obtained in the IGR Ldb2086/Ldb2087, the target sequences of the pVI110 insertions were different, suggesting that this region is most likely not a hot spot of integration. Noticeably, target sites are surrounded by inverted repeats predicted to form hairpins with ΔG < −9 Kcal (calculated with Oligo Analyser freeware). Alignment of pVI110 target sequences revealed a preferential insertion in A/T-rich regions, as seen for other mobile elements, like Tn1545 in Clostridium and Listeria (5, 8), and several insertion sequences (11, 28). The nucleotide sequences of pVI110-target junctions in L. delbrueckii subsp. bulgaricus also revealed a 3-bp (occasionally 4-bp) duplication generated upon integration. Analysis of the target sequences suggests that triplets C/A A/T T/A may be preferential target sites for pVI110 in L. delbrueckii subsp. bulgaricus. Of the 20 random pVI110 transposon targets sequenced for L. casei (Table 3), 50% (n = 10) were located in intergenic regions, while the L. casei genome contains about 20% noncoding regions (7). Ten percent (n = 2) of mutants correspond to two different integration sites of pVI110 in pLSEIA, which represents only twice the maximal theoretical rate. This result reveals that pLSEIA is not a significant preferential host for pVI110 integration, indicating that the presence of the pLSEIA plasmid in LC334 is not an obstacle to obtaining saturated mutagenesis libraries. In contrast to the results for L. delbrueckii subsp. bulgaricus, only 20% of the L. casei target sites were located in inverted repeats predicted to form hairpins. Moreover, no preferential insertion in A/T-rich regions was observed. Despite the general, presumed random insertion of most transposons, many of them show a target preference (for reviews, see references 11 and 28). This targeting could in fact be a result of selective means to avoid affecting host fitness and, consequently, to promote transposon dissemination. Since L. delbrueckii subsp. bulgaricus is closely related to L. johnsonii, the original host of IS1223 (39), IS1223 is likely to preferentially target noncoding sequences to preserve its host genome. Interestingly, this bias is reduced in L. casei, which is phylogenetically more distant from L. johnsonii (3, 40), and is reduced even further in E. coli, suggesting that insertion sequences may display a more random integration in phylogenetically distant bacteria.

In conclusion, this work describes the use of an IS3-like transposition mechanism to engineer a novel transposition system based on the P_junc-TpaseIS₁₂₂₃ two-plasmid system for Gram-positive bacteria. Our results demonstrate that this system is functional in L. delbrueckii subsp. bulgaricus and L. casei and produces a high rate of stable integrants (at least 10,000 mutants per transformation for L. casei) despite the relatively poor transformation rate of lactobacilli. This system presents the advantage of promoting transposition of a suicide plasmid which contains P_junc (pVI110) provided in trans with a helper plasmid (pVI129) supplying TpaseIS₁₂₂₃. Thanks to this design, no sibling clones from early transposition events (31) can appear, and as pVI110 is stably produced in E. coli, it can be easily manipulated by inserting a reporter gene or used for signature-tagged mutagenesis. In view of the efficient transposition activity observed in the species tested (e.g., Bacillus subtilis, Lactococcus lactis, Lactobacillus plantarum, and Enterococcus faecalis; unpublished results), this transposition system may have a broad application in Gram-positive bacteria, particularly in LAB.