Natural competence is active in very diverse species of the bacterial kingdom and probably participates in horizontal gene transfer. Recently, the genome sequence of various species, including Lactococcus lactis, revealed the presence of homologues of competence genes in bacteria, which were not previously identified as naturally transformable. We investigated the conservation among lactococcal strains of key components of the natural competence process in streptococci: (i) comX which encodes a sigma factor, allowing the expression of the late competence genes involved in DNA uptake, (ii) its recognition site, the cin-box and (iii) dprA which encodes a protein shown to determine the fate of incoming DNA. The comX and dprA genes and the cin-box appeared conserved among strains, although some L. lactis ssp. lactis strains presented an inactivated dprA gene. We established that ComX controls the expression of the late competence genes in L. lactis. In conclusion, our work strongly suggests that ComX has the same role in streptococci and L. lactis, i.e. the regulation of late competence genes. It also allowed the identification of a set of L. lactis strains and the construction of a comX overexpression system, which should facilitate the investigation of the natural competence activity in lactococci.
In streptococci, natural competence is controlled by a quorum-sensing system, which involves the production, the export-maturation and the detection by a two-component system (ComD, ComE) of a competence stimulating peptide (CSP) (Pestova et al., 1996). The phosphorylated response regulator, ComE∼P, stimulates the transcription of the genes responsible for CSP synthesis, export and detection, and also of the comX1 and comX2 genes that encode two homologous sigma factors (Lee & Morrison, 1999). Each of these sigma factors can initiate the next step of the competence process by switching on the expression of the late competence genes and stimulating the expression of competence-associated genes (Lee & Morrison, 1999). In order to stimulate gene expression, ComX interacts with the RNA polymerase core enzyme and the resulting holoenzyme recognizes a specific sequence, the cin-box located upstream of late competence genes (Lee & Morrison, 1999; Opdyke et al., 2001). In this system, ComX has a key role, linking the regulatory module constituted by the early competence genes to the DNA uptake module encoded by the late competence genes. The ComX-regulated genes are directly involved in the uptake of exogenous DNA (Claverys & Havarstein, 2002), in its protection from intracellular nucleases by the DprA protein (Berge et al., 2003), and in its incorporation in the chromosome by homologous recombination (Mortier-Barriere et al., 1998).
Lactococcus lactis is a gram-positive lactic acid bacterium (LAB), which is closely related to the streptococci (Stiles & Holzapfel, 1997). The two major L. lactis subspecies, L. lactis ssp. lactis and L. lactis ssp. cremoris, are industrially important as they are used as starters for the production of fermented dairy food (Stiles & Holzapfel, 1997). Lactococcus lactis was never identified as a species possessing a natural competence pathway and to the best of our knowledge, natural transformation was never observed in L. lactis strains. However, the analysis of the complete genome sequence of IL1403, an L. lactis ssp. lactis strain, revealed the presence of orthologs of several genes involved in the process of natural competence in various bacteria (Bolotin et al., 1999, 2001; Claverys & Martin, 2003). Based on sequence homology, a comX gene and all the late competence genes required for the formation of the DNA entry pore [comC (also referred to as CCL in S. pneumoniae), comEA, comEC, comFA, comFC and comGA to comGD] were identified (Bolotin et al., 1999, 2001). Comparison of the regions located upstream of these late competence genes allowed the identification of a conserved sequence which may constitute a recognition site of the ComX-RNA polymerase holoenzyme (Bolotin et al., 1999, 2001). These genetic features of IL1403 raise the question whether L. lactis can activate a natural competence system which may contribute to genetic transfer in dairy food ecosystems (Gasson, 2000; Jonas et al., 2001; Kharazmi et al., 2002; van den Eede et al., 2004).
Two observations questioned the functionality of natural competence in Lactococcus lactis. The first observation is that orthologs of the early competence genes (the ABC transporter encoding comA and comB genes, the two-component system encoding genes comD and comE, and the CSP encoding gene) responsible for the regulation by the CSP, were not clearly identified in IL1403 (Bolotin et al., 2001). The strain does, however, possess multigenic families encoding ABC transporters (at least 60 ABC transporters in IL1403) and two-component systems (seven complete systems and an additional response regulator) (Bolotin et al., 2001). In IL1403, Redon (2005) observed a slight progressive overexpression of comX during glucose starvation and proposed that specific starvation conditions may lead to the induction of natural competence in L. lactis.
The second observation is that the IL1403 dprA gene (also referred to as smf in several microorganisms as Bacillus subtilis) is inactivated by a frameshift resulting in a premature stop at nt 100 (protein of 33 amino acids). In several bacteria (Karudapuram et al., 1995; Ando et al., 1999; Smeets et al., 2000; Berka et al., 2002; Friedrich et al., 2002; Ogura et al., 2002), a determining role for DprA in the fate of the exogenous DNA taken up during competence was demonstrated. In an S. pneumoniae strain mutated for dprA, the incoming DNA is degraded and the nucleotides are randomly incorporated in the chromosome, while in a wild-type background the incoming DNA fragments are integrated in the chromosome by homologous recombination (Berge et al., 2003). Since DprA activity is crucial for the outcome of the natural transformation process and its contribution to horizontal gene transfer, we investigated whether L. lactis dprA+ strains could be identified in a collection of strains.
In order to evaluate if natural competence may allow the development of a general genetic tool for the modification of lactococcal strains, we investigated the conservation of two other key elements of the natural competence system, comX and the cin-box, in lactococci. Finally, as a prelude to a study on the functionality of natural competence in L. lactis, we overexpressed two alleles of the comX gene in IL1403 to test whether they regulate the expression of late competence genes.
Materials and methods
Bacterial strains and culture conditions
For this study, 31 lactococcal strains (Table 1) were chosen in the INRA national collection (INRA collection, URLGA, Jouy-en-Josas, France). These strains were classified in the different lactococcal subspecies (ssp. lactis, ssp. lactis biovar. diacetylactis and ssp. cremoris) according to their phenotypes (growth at 40°C or in the presence of 4% NaCl and arginine or citrate metabolism, J. C. Ogier, pers. comm.).
Groups 1, 2 and 3 correspond to the RAPD patterns the most frequently identified in the CNRZ collection of lactococci (Tailliez et al., 1998). Strains exhibiting specific RAPD patterns are referred to as RAPD individual patterns.
The 16S analysis did not discriminate between Lactococcus lactis ssp. lactis and L. lactis ssp. lactis biovar diacetylactis strains.
Strains indicated with were misclassified on the basis of their phenotypes.
The strain VI 7101 is a derivative of IL1403, which contains the nisR and nisK genes integrated in the chromosomal histidine biosynthesis operon (S. Calero and P. Renault, pers. comm.). Strain VI 7238 is a comX-deleted mutant of VI 7101 which resulted from a double crossing-over (DCO) using plasmid pVI6237 (carrying a 2 kb DNA insert constituted of the upstream (1 kb) and downstream (1 kb) regions of comX fused together). The DCO was performed as previously described for L. lactis (Biswas et al., 1993) and the mutated strain was identified among the ery-sensitive clones by PCR with the S99 and S102 primers (Table 2). The chromosomal structure of VI 7238 was then confirmed by Southern hybridization of digested chromosomal DNA (HindIII, XmnI and EcoRI) using a PCR fragment generated with primers S124 and S185 (Table 2) as a probe. The VI 7293, VI 7247 and VI 7298 strains resulted from the transformation of VI 7238 with plasmids pVI6253 (pnisA), pVI6213 (pnisA:comXIL) and pVI6255 (pnisA::comXMG), respectively. The pnisA promoter was induced by the addition of nisin (0.8 ng mL−1, Sigma, Lyon, France) in exponential phase cultures (OD600≤0.1) growing in SA medium (Jensen & Hammer, 1993) at 30°C. Otherwise, strains were propagated at 30°C in M17 (Terzaghi & Sandine, 1975) supplemented with 0.5% glucose. Erythromycin (ery) was added at 5 μg mL−1.
All the oligonucleotides were designed on the sequence of IL1403, except S183 and S184 which were designed on the sequence of MG1363.
DNA extraction and PCR amplification
Chromosomal DNA was extracted as previously described (Biswas et al., 1993). PCR DNA amplifications were performed with the ExTaq DNA polymerase (Takara, St Germain en laye, France) and the appropriate primers (Table 2) as recommended by the supplier. For the comX and dprA genes, primers homologous to the intergenic regions flanking the gene of interest in the IL1403 genome sequence were designed. These primers allowed to amplify fragments from all the strains of the lactis subspecies (as determined by the 16S analysis, Table 1) but not from the strains belonging to the cremoris ssp. For the cremoris ssp. strains (as determined with the 16S analysis), an internal fragment of the gene of interest was first amplified and sequenced (comX, a ∼300 bp fragment obtained with the primer pairs S141–S143 or S142–S143 and dprA, a 703 bp fragment obtained with S183–S184). Then, the upstream and downstream regions were amplified using primers homologous to this internal fragment (comX, S143 or S141 and dprA, S183 or S184) and the flanking genes (comX, S125, S126 or S127 and dprA, S98 or S186). The promoter region of the comG operon was amplified using primers homologous to the upstream (S149, S152 or S150) and downstream (S150, S154 or S155) sequences.
1 6S analysis
In order to characterize the 16S sequence of the 31 lactococcal strains selected from the INRA collection, an internal fragment (343 bp) of the 16S ribosomal RNA genes was amplified using the Y1 and Y2 primers described by Ward (1998) and chromosomal DNA as templates. The PCR products were then digested by CfoI and MboII to determine to which subspecies the strain belonged (Ward et al., 1998).
DNA sequencing and analysis
PCR generated fragments were sequenced using the Big Dye Terminator v3.1 (Applied Biosystems, Foster City, CA). Reaction products were analysed using a capillary sequencer (3700 DNA Analyser, ABI PRISM). Sequences were analysed using the GCG package (University of Wisconsin) and were aligned using CLUSTALW (Infobiogen, http://www.infobiogen.fr/) (Thompson et al., 1994), the percentage of divergence between the alleles was calculated by MEGA 2.1 (Kumar et al., 2001) and the data of the phylogenetic analysis were represented using TREEVIEW (Page, 1996). All sequences were deposited in the EMBL database under the following accession numbers: comX genes AJ890847, AJ890848, AJ890849, AJ890850, AJ890851, AJ890852, AJ890853, AJ890854, AJ890855, AJ890856, AJ890857, AJ890858, AJ890859, AJ890860, AJ890861, AJ890862, AJ890863, AJ890864, AJ890865, AJ890866, AJ890867, AJ890868, AJ890869, AJ890873, AJ890874, AJ890875, AJ890876, AJ890878, AJ890879, AJ890881, dprA genes AJ890918, AJ890919, AJ890920, AJ890921, AJ890922, AJ890923, AJ890924, AJ890925, AJ890926, AJ890927, AJ890928, AJ890929, AJ890930, AJ890932, AJ890933, AJ890934, AJ890935, AJ890936, AJ890937, AJ890938, AJ890944, AJ890945, AJ890946, AJ890947, AJ890948, AJ890949, AJ890950, AJ890951, AJ890952, AJ890953.
Construction of plasmids
pVI6237 is a derivative of the thermosensitive pG+host9 plasmid (Maguin et al., 1996) which carries a ∼2 kb SpeI–XhoI insert (a fusion of two 1 kb fragments corresponding to the upstream and downstream regions of comX and amplified with primers S99–S135 and S136–S102, respectively). pVI6213 is a derivative of the pJIM2246 vector (Renault et al., 1996) which contains between its XhoI and SacI sites a DNA insert containing 3 successive transcription terminators (T3) followed by a fusion between the inducible promoter pnisA and the comX gene of IL1403 (comXIL). This XhoI-SacI insert was generated as described below. (i) The pnisA::comXIL fusion was amplified by PCR using the S106 and S107 primers and as template the chromosomal DNA of VI 7145, which contains the pnisA::comXIL fusion. This PCR product was cloned in the pGEM-T vector (Promega, Charbonnieres, France) leading to pVI6205. (ii) The EcoRV-SacI fragment of pVI6205 (containing the fusion) was cloned in pVI6207 (pBluescriptSKII containing T3) leading to pVI6212 which contains the T3::pnisA::comXIL fusion on an XhoI–SacI fragment. In order to generate pVI6253 (pJIM2246::T3::pnisA), the comXIL gene was deleted from pVI6213 by SacI and NcoI digestions followed by a Klenow-T4 DNA polymerase treatment and self-ligation. Finally, plasmid pVI6255 (pJIM2246::T3::pnisA::comXMG) was obtained after SacI and NcoI digestions of pVI6253 and ligation with a SacI–NcoI PCR fragment containing comXMG (amplified from the MG1363 chromosomal DNA with primers S260 and S261).
Extraction of total RNAs
A volume of 100 mL of cultures at OD600≈0.1 were centrifuged (5 min, 9800 g at 4°C) and the cell pellet was washed with cold TE. The pellet was resuspended in 500 μL of cold water and transferred in a 2 mL screw-cap microcentrifuge tube containing 0.6 g of glass beads (Sigma), 200 μL of Macaloid (2%) (Sambrook et al., 1989), 500 μL of phenol–chloroform pH 4.7 and 25 μL of SDS (20%). Cells were disrupted by shaking in a Fastprep machine (BIO101) for 40 s at speed 5.5. After centrifugation at 12 000 g for 15 min (4°C), the aqueous supernatant, which contains the RNA, was treated with phenol-chloroform pH 4.7, precipitated with ethanol and resuspended in 30 μL of water.
Dnase treatment and RT-PCR
In all, 5 μg of total RNAs diluted in water (15 μL final) were incubated 3 min at 95°C and 10 min on ice. 3 μL of Dnase (2 units μL−1, Ambion, Huntington, UK) and 2 μL of buffer were added, and the mixture was incubated 1.5 h at 37°C. To synthesize cDNAs 1 μg of total RNA was incubated 10 min at 25°C and 5 min on ice with 1 μL of dNTPs (10 mM, Promega) and 1 μL of random primer (500 ng μL−1, NEB, Beverly, MA) in a final volume of 13 μL. A volume of 1 μL of reverse transcriptase (M-MLV Reverse transcriptase kit, Invitrogen Cergy-Pontoise, France), 4 μL of 5 × buffer, 2 μL of DTT (0.1 mM) were then added and the mixture was incubated 1 h at 37°C and 15 min at 75°C. The PCR amplification (final volume 50 μL) was performed from cDNA (1 μL) using appropriate primers (Table 2), Extaq polymerase (Takara), dNTPs and the reaction buffer as recommended by the supplier (Takara). The amplification was carried out as follows: 5 min at 94°C, (30 s at 94°C, 30 s at 57°C, 1 min at 72°C) 30 cycles. Note that the expression of ydbC was not monitored since the 2 pairs of oligonucleotides that we tested did not allow PCR amplification from a DNA template.
Characterization of the lactococcal strains
We selected 31 lactococcal strains including the two best characterized Lactococcus lactis strains, IL1403 (ssp. lactis) and MG1363 (ssp. cremoris) in the INRA national collection. These strains belong to the three major RAPD groups defined for the L. lactis strains (Tailliez et al., 1998) and according to their phenotypes, they belong to the two major L. lactis subspecies used in dairy industry (ssp. lactis and ssp. cremoris, Table 1).
In order to better characterize the chosen strains, their 16S sequences were analysed in a region which allowed to distinguish between the lactis and cremoris ssp. (Ward et al., 1998). Our data revealed that nine strains out of 31 (more than 29%) were misclassified on the basis of their phenotypes (Table 1). According to the 16S data, our set of strains comprises 20 strains of the lactis ssp. and 11 of the cremoris ssp. It is noteworthy that according to this new strain classification, our sample of the RAPD group 3 contained exclusively strains of cremoris spp.
Sequence analysis of the dprA gene
Sequence analysis revealed that in IL1403, the dprA gene contained a premature stop codon at position 34; the resulting truncated protein of 33 residues (instead of 282) is likely to be inactive. As DprA is expected to play a crucial role in natural transformation, its sequence was determined in all of the other 30 strains.
In all the strains belonging to the lactis ssp. according to the 16S analysis, two fragments (632 and 592 bp in IL1403), which together contained the dprA gene were amplified. For the remaining 11 strains of the cremoris ssp., unsuccessful amplification assays prompted us to change our strategy. An internal fragment (703 bp) of dprA was amplified and sequenced. The upstream and downstream regions were then amplified using oligonucleotides located in the internal sequenced region and in the dprA neighbouring genes.
The maximal dprA divergence was 21.7% and the sequence analyses revealed two allelic types, one detected in all the L. lactis ssp. lactis strains and the other in the ssp. cremoris strains; they are referred to as the lactis- and the cremoris-allelic types, respectively (Fig. 1a). The lactis- and cremoris-type exhibited an intratype variation of 3.1% and 3.7%, respectively. In the lactis-type, 10 genes (50%) presented a premature stop codon that would result in truncated proteins of 23 aa (A8), 33 aa (IL584, IL1403, DRC1), 167 aa (A7) or 193 aa (IL1306, A17, A15, AM2, IL1321). In all these cases, the resulting protein is expected to be inactive. For the cremoris-allelic type, all the predicted proteins had the full size of 282 aa.
Neighbour-joining unrooted phylogenetic trees inferred from the sequences of Lactococcus lactis genes. (a) dprA sequences. (b) comX sequences. The multiple nucleotide sequences were compared by CLUSTALW (Thompson et al., 1994). The tree representation was obtained with TREEVIEW (Page, 1996). The maximal divergence (max. div.) was calculated with MEGA 2.1 (Kumar et al., 2001). Asterisks (*) indicate the strains in which the gene is inactivated.
Complete dprA genes were identified in ∼68% of the tested strains, and in both the lactis and cremoris ssp. Since some strains contain an intact dprA gene, we were interested to see if the upstream regulatory elements were conserved; we therefore investigated the conservation of two other key elements for natural competence, comX and the cin-box, which regulate the expression of the late competence genes in S. pneumoniae.
Genetic variability of the comX gene among lactococci
As in the case of the dprA gene, the primers homologous to the intergenic regions flanking the comX gene in IL1403 led to a productive PCR only for the strains belonging to the lactis ssp. Other primers were designed for amplification from the cremoris strains (material and methods). The comX gene was detected in the sequence of all of the 31 fragments. The maximal divergence of the comX nucleotide sequences was 27.5% and all genes were intact. Further sequence comparison using CLUSTALW revealed two allelic types (Fig. 1b).
One allelic type was detected in all the lactis ssp. lactis strains (20 strains including IL1403); It exhibits a maximal intratype divergence of 4.5%. In total the 20 predicted proteins corresponding to this lactis-allelic type presented eight positions with substitution and one with an insertion (Fig. 2). However compared to the ComX of IL1403, the most divergent proteins (strains NCDO2091 and A7) only differed by four amino-acids: three substitutions (N38S, D114E, Q129E) and one insertion (D40) for NCDO2091 and four substitutions for A7 (N38S, L108S, Q129E, D145E). In the other strains presenting this comX allelic type, the deduced proteins only differed by one (six strains), two (four strains) or three (two strains) amino acids from the IL1403 ComX.
Amino-acid sequences of ComX. The black lines shown above the protein sequences indicate the putative functional domains (2.1, 2.2, 2.3, 2.4, 3.1, 4.1 and 4.2) as defined for other sigma factors (Lonetto et al., 1992). The first protein sequence corresponds to the ComX protein of the IL1403 strain. The lactis-T line shows all the amino-acid differences observed in the lactis-allelic type. The third sequence indicates the differences in the MG1363 ComX protein in comparison with the IL1403 protein; most of the substitutions indicated were found in all of the ssp. cremoris strains. The cremoris-T line shows amino-acid differences (other than that indicated for the MG1363 ComX protein) observed in the cremoris-allelic type. The numbers below the substitutions correspond to the number of strains in which the difference was observed. NCDO2091 is the strain exhibiting the aa (D) insertion in position 40 of ComX.
The second allelic type was identified in the eleven L. lactis ssp. cremoris strains (Fig. 1b); it has a maximal intratype divergence of 1.4%. Consequently, the 11 predicted proteins corresponding to this allelic type are very well conserved: only three positions with amino-acid substitutions (H54R, R67W and R130C) are observed (Fig. 2).
Comparison of the predicted proteins from the two comX allelic types (represented by the sequences from IL1403 and MG1363), revealed two clusters of amino-acid substitutions (Fig. 2 grey boxes) in regions of the sigma factor which are embedded in the functional domains 2.3 and 3.1 (Lonetto et al., 1992; Wosten, 1998; Gruber & Gross, 2003). Since both domains are proposed to be involved in the interaction with DNA (opening of the DNA duplex and DNA binding, respectively), we wondered whether this variability may be correlated to any divergence of the sigma recognition site in the promoters. We therefore determined the sequence of the putative cin-box in these strains.
Sequence of the cin-box in various lactococcal strains
In agreement with Bolotin (2001), MEME (Bailey & Elkan, 1994) generated a consensus sequence (GTTACAATN9TTTTCGTATA, Fig. 3) from the upstream region of the late competence operons comE, comF, comC, comG, several competence associated genes dprA, coiA, radA, recQ, ssbA and a gene of unknown function yqfG in IL1403. This sequence is proposed as the cin-box required for ComX recognition. Using PATSCAN (Dsouza et al., 1997) to search the IL1403 genome, a similar sequence was also identified upstream of the radC, recA, ydbC (encoding a conserved hypothetical protein, COG 4443) and ywgA (encoding a conserved hypothetical protein [COG 2137] which may belong to the RecX family [pfam02631]).
Comparison of the cin-box sequences. The upstream regions of the comC, comEA, comFA, comGA, dprA, coiA, ssbA, yqfG, radA, radC, recA, recQ, ydbC and ywgA genes of IL1403 and the consensus sequence referred to as cin-box obtained with MEME (Bailey & Elkan, 1994). Dark grey or pale grey highlight the bases found in a given position in at least 80% or at least 30% of the sequences. The sequences identified upstream of comGA in the other L. lactis strains studied are indicated below the cin-box. The G to A transition is only observed in A7.
In order to evaluate the conservation of this putative cin-box, the region corresponding to the 415 nucleotides upstream of the late competence comG operon in IL1403 was sequenced in the thirty lactococcal strains. Since the cin-box located upstream of ComGA slightly differs from the IL1403 consensus sequence, it seemed more likely to vary between strains. Comparison of these sequences revealed a conserved sequence identical to the IL1403 putative cin-box located upstream of the comG operon in all strains (Fig. 3). In conclusion, although two types of ComX proteins were detected in the lactococcal strains, an identical cin-box was revealed by this analysis.
Overexpression of the comXIL and comXMG genes
In order to test the functionality of two comX allelic types, comXIL and comXMG were cloned under the control of a nisin-inducible promoter (pnisA) and introduced in VI 7238, an IL1403 derived strain deleted of the comX gene (by double crossing-over) and carrying the nisR and nisK genes (required for the nisin-inducible expression system). The VI 7238 strain containing the pnisA-vector (pVI6253) was used as a control. After 30 min of incubation with nisin, total RNA was extracted from the various cultures and the expression of comX and late competence genes was monitored using RT-PCR (Fig. 4).
Expression of late competence genes without or with induction of comXIL or comXMG. Strains VI 7293 (pnisA), VI 7247 (pnisA::comXIL) and VI 7298 (pnisA::comXMG) were incubated 30 min with nisin. Total RNAs were then extracted and the expression of comXIL, comXMG and of most of the genes preceded by a putative cin-box were evaluated by RT-PCR. The expression of hu was used as a control and serial dilutions were also used to ensure that the amount of hu cDNA was similar in all samples (data not shown).
In the presence of nisin, the comXIL and comXMG transcripts were detected while they were absent in the control strain (VI 7238/pVI6253). To check the functionality of the comX gene products, the expression of all genes (except ydbC) exhibiting a putative cin-box in their upstream sequence was also monitored (Fig. 3). The expression of the comG, comE, comF, comC, dprA, coiA, ssbA and yqfG genes increased markedly after the over-expression of comXIL as well as that of the comXMG gene, whereas their expression remained barely detectable or low in the control strain treated with nisin. The radA, radC, recA and ywgA appeared already well expressed in the absence of ComX. Consequently, an accurate measurement of mRNA or cDNA amounts will be needed to determine if their expression increase in the presence of ComX. It is worth noting the similarity between the expression profiles after induction of ComXIL and ComXMG, which suggests that both are active.
Comparison of 16S analysis, dprA and comX sequences
The classification of the lactis strains used in this study on the basis of their 16S rRNA revealed that more than 29% of the selected strains had been misclassified on the basis of the phenotypic tests classically used to distinguish the L. lactis ssp. The comX and dprA sequences each revealed two allelic types, which perfectly matched the 16S classification of strains while only partially fitting the phenotypic or RAPD groups. The mean nucleic divergence observed between the lactis and cremoris alleles was 26.9% for comX and 20.9% for dprA. This observation strongly suggests that the natural competence genes were present in lactococci before the divergence between the lactis and cremoris ssp., i.e. around 17 million years ago (Bolotin et al., 2004) since the rate of divergence was estimated at ∼0.9% per million years (Ochman et al., 1999).
Conservation of key elements of the natural competence system
In all strains, intact comX genes were found and identical cin-boxes were identified upstream of the comG operon. Two types of ComX were distinguished, each being well conserved among the strains of one ssp. The observation that the same cin-box is conserved among all tested strains suggests that these two types of ComX recognize the same DNA sequence. We showed that in IL1403, both ComX variants were active and allowed the expression of all the late competence genes required for DNA uptake in other microorganisms. It establishes that comX is not degenerated in L. lactis, suggesting that having an active ComX might be an advantage for the strain for an as yet undiscovered reason. Contrary to comX, dprA was inactivated in about 32% of the strains all belonging to the ssp. lactis. Although the exact role of DprA remains to be established in L. lactis, this high occurrence of a mutated dprA gene may indicate that the natural competence process is used for another purpose than genetic diversity (i.e. natural transformation) such as a competitive advantage as previously shown in Escherichia coli (Finkel & Kolter, 2001). In a dprA background, the natural competence could provide DNA, which will be degraded leading to an increase of the intracellular nucleotide pool. As free bases are present in limited amounts in milk, one can propose that a dprA mutation might be advantageous. In addition to DNA uptake, the natural induction of competence may also allow the predation of non-competent cells or of other species via the expression of bacteriocin as being recently described for Streptococcus pneumoniae (Guiral et al., 2005) and S. mutans (Kreth et al., 2005).
For both the lactis and cremoris ssp., strains presenting a complete dprA gene were also found. The L. lactis strains presenting the full DprA (lactis-, cremoris-types) or a truncated form would be useful to test whether natural competence is functional in L. lactis and to investigate the role of this protein in lactococci. We demonstrated in this work that ComXMG (present in the cremoris ssp.) can stimulate the expression of the late competence genes in a ssp. lactis strain. Consequently a comX overexpression system could be introduced in the various strains exhibiting a complete or a truncated dprA gene (identified in this work) to test whether the upregulation of comX could lead to natural transformation. This system constitutes an interesting alternative to the laborious approach that is to seek conditions allowing the induction of competence in a given species by testing different media, starvation conditions, growth phases etc. This latter approach may be time consuming since the conditions of induction of natural competence vary between species (Dubnau, 1991).
In conclusion, this study revealed that the homologues of three key elements of the natural competence process are conserved among lactococci and it demonstrates for the first time in lactococci the activity of the ComX–cin-box regulatory circuit. Strains with different allelic types of the comX and dprA genes and the comX overexpression system constitute promising tools to test the functionality of natural transformation in this species.
We are grateful to D. Vailhen, A. Bolotin, M. van de Guchte, P. Serror for helpful discussion and suggestions. S. Wydau is the recipient of a fellowship from the Ministère de la Recherche.
(1994) Fitting a mixture model by expectation maximization to discover motifs in biopolymers. Proceedings of the Second International Conference on Intelligent Systems for Molecular Biology2: 28–36. AAAI Press, Menlo Park, CA.
(2002) Molecular analyses of the natural transformation machinery and identification of pilus structures in the extremely thermophilic bacterium Thermus thermophilus strain HB27. Appl Environ Microbiol 68: 745–755.
(1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22: 4673–4680.