Conjugation mediates transfer of the Ll.LtrB group II intron between different bacterial species


  • Kamila Belhocine,

    1. Department of Microbiology and Immunology, McGill University, Montréal, Québec, Canada H3A 2B4.
    Search for more papers by this author
  • Isabelle Plante,

    1. Department of Microbiology and Immunology, McGill University, Montréal, Québec, Canada H3A 2B4.
    Search for more papers by this author
  • Benoit Cousineau

    Corresponding author
    1. Department of Microbiology and Immunology, McGill University, Montréal, Québec, Canada H3A 2B4.
      E-mail; Tel. (+1) 514 398 8929; Fax (+1) 514 398 7052.
    Search for more papers by this author

E-mail; Tel. (+1) 514 398 8929; Fax (+1) 514 398 7052.


Some self-splicing group II introns (ribozymes) are mobile retroelements. These retroelements, which can insert themselves into cognate intronless alleles or ectopic sites by reverse splicing, are thought to be the evolutionary progenitors of the widely distributed eukaryotic spliceosomal introns. Lateral or horizontal transmission of introns (i.e. between species), although never experimentally demonstrated, is a well-accepted model for intron dispersal and evolution. Horizontal transfer of the ancestral bacterial group II introns may have contributed to the dispersal and wide distribution of spliceosomal introns present in modern eukaryotic genomes. Here, the Ll.LtrB group II intron from the Gram-positive bacterium Lactococcus lactis was used as a model system to address the dissemination of introns in the bacterial kingdom. We report the first experimental demonstration of horizontal transfer of a group II intron. We show that the Ll.LtrB group II intron, originally discovered on an L. lactis conjugative plasmid (pRS01) and within a chromosomally located sex factor in L. lactis 712, invades new sites using both retrohoming and retrotransposition pathways after its transfer by conjugation. Ll.LtrB lateral transfer is shown among different L. lactis strains (intraspecies) (retrohoming and retrotransposition) and between L. lactis and Enterococcus faecalis (interspecies) (retrohoming). These results shed light on long-standing questions about intron evolution and propagation, and demonstrate that conjugation is one of the mechanisms by which group II introns are, and probably were, broadly disseminated between widely diverged organisms.


Self-splicing group II introns are large autocatalytic RNAs (ribozymes) (Lambowitz and Belfort, 1993; Saldanha et al., 1993; Michel and Ferat, 1995; Belfort et al., 2002). Some group II introns harbouring an intron-encoded protein (IEP) are also mobile elements. These retroelements are capable of inserting themselves into both cognate intronless alleles (homing sites; HSs) and ectopic sites (non-homologous sites) by retrohoming and retrotransposition respectively (Belfort et al., 2002). The Ll.LtrB group II intron was initially discovered in the industrially important Gram-positive bacterium, Lactococcus lactis (Mills et al., 1996; Shearman et al., 1996). Ll.LtrB is the first bacterial group II intron that was shown to splice and to be mobile in vivo (Mills et al., 1996; 1997; Shearman et al., 1996). The mobility pathways of Ll.LtrB for both retrohoming (Cousineau et al., 1998) and retrotransposition (Cousineau et al., 2000; Ichiyanagi et al., 2002) were studied using genetic systems in both L. lactis and Escherichia coli. Retrohoming is a very efficient process compared with the insertion of the intron at ectopic sites through the retrotransposition pathway (Cousineau et al., 1998; 2000; Ichiyanagi et al., 2002).

The Ll.LtrB group II intron (2.5 kb) encodes, within the loop region of domain IV, a protein of 599 amino acids called LtrA (1.8 kb) (Mills et al., 1996). The LtrA protein carries three functional domains (reverse transcriptase, maturase, endonuclease) essential in promoting Ll.LtrB mobility via retrohoming (Matsuura et al., 1997; Cousineau et al., 1998). The retrohoming and retrotransposition pathways of the Ll.LtrB group II intron are duplicative processes that, similarly to other retroelements (e.g. retrotransposons and retroviruses), proceed through an RNA intermediate (Cousineau et al., 1998; 2000; Ichiyanagi et al., 2002). Retrohoming occurs through a target DNA-primed reverse transcription mechanism (TPRT). The first step in retrohoming is splicing of the intron from the pre-mRNA, which absolutely requires the maturase activity of its intron-associated protein LtrA in vivo. Active ribonucleoprotein particles (RNPs; intron RNA + LtrA) are liberated from the pre-mRNA after splicing of the intron and ligation of the two flanking exons. These RNPs then identify uninterrupted homing sites, which the intron invades by complete reverse splicing into the sense strand (mRNA-like strand) of double-stranded DNA. After insertion of the intron RNA, the endonuclease activity of LtrA cuts the bottom strand 9 nucleotides (nt) upstream of the intron insertion site to complete the staggered double-strand cut. The liberated 3′-OH of the DNA antisense strand is then recognized by the RT domain of LtrA and is used as a primer to synthesize a full cDNA copy of the intron. The final steps of the retrohoming pathway are mainly supported by the host DNA repair mechanisms (Cousineau et al., 1998). Even though host DNA repair seems to be involved in establishing intron insertion, the retrohoming pathway of Ll.LtrB was shown to be completely independent of the major RecA-dependent homologous recombination pathway (Mills et al., 1997; Cousineau et al., 1998).

From an evolutionary perspective, group II introns are also very fascinating. They are considered to be the ancestors of the spliceosome-dependent eukaryotic nuclear introns (Sharp, 1991). Indeed, group II and spliceosomal introns share numerous striking similarities. Both intron types are excised as lariats and use the same splicing mechanism (Saldanha et al., 1993; Michel and Ferat, 1995). Moreover, the small nuclear RNAs (snRNAs) that are part of the spliceosome machinery are proposed to be structurally similar to specific domains of group II introns (Sharp, 1991). This theory suggests that fragmentation and degeneracy of ancestral group II introns gave rise to the current eukaryotic spliceosome machinery (Sharp, 1991). On the other hand, the phylogenetic distribution and relative abundance of group II versus spliceosomal introns are quite different. The non-mobile and highly abundant spliceosomal introns (>16% of the human genome) are found exclusively in the nucleus of eukaryotes. Group II introns, some of which are mobile elements, are found in bacteria and in eukaryotic organelles derived from bacteria, such as fungal and plant mitochondria, as well as plant chloroplasts (Lambowitz and Belfort, 1993; Saldanha et al., 1993; Michel and Ferat, 1995; Belfort et al., 2002). A better understanding of the pathways supporting group II intron propagation, within and between species, should help to explain the idiosyncratic distribution of these two types of introns and the prominence of the spliceosomal introns in eukaryotes. Interspecies dissemination of mobile group II introns, through lateral transfer, may have contributed to the wide distribution of eukaryotic introns. Horizontal transfer of group II introns between organisms, although never experimentally demonstrated, is a well-accepted model of intron evolution. Horizontal transfer was proposed to explain the presence of closely related introns at different locations (genes or species) and to rationalize why, in specific cases, introns are more conserved than their flanking exons (Lambowitz and Belfort, 1993). The recent study of group II intron distribution in bacterial genomes also suggests substantial horizontal transfer of these introns within the bacterial kingdom (Zimmerly et al., 2001).

Here, we describe L. lactis genetic assays for lateral transfer of group II introns between bacteria using a biologically relevant and genetically tractable experimental setting. The Ll.LtrB group II intron was originally found to interrupt a relaxase gene (ltrB) within two very similar L. lactis mobilizable elements: a conjugative plasmid (pRS01) (Mills et al., 1996) and a sex factor embedded within the chromosome of the L. lactis 712 strain (Shearman et al., 1996). We thus explored the possibility that conjugation is a basis for the lateral transfer of group II introns promoting their dissemination to new homologous (retrohoming) and non-homologous (retrotransposition) sites both within and between bacterial species.


Conjugation/mobility assay in L. lactis

To determine whether the Ll.LtrB group II intron could be transferred after conjugation of its host elements between different L. lactis strains, we built a conjugation/mobility assay (Fig. 1). We used the pLE12 plasmid (16 kb) (Mills et al., 1996; Cousineau et al., 1998), which is smaller and easier to handle than either the original Ll.LtrB-containing conjugative plasmid pRS01 (48.4 kb) (Mills et al., 1994; 1996) or the chromosomally located sex factor (50 kb) (Shearman et al., 1996). The pLE12 plasmid contains only two (Tra1–2) of the four conjugative transfer regions (Tra1–Tra4) identified in pRS01 (Mills et al., 1994) (Fig. 1A). The Ll.LtrB intron interrupts a relaxase gene (ltrB; Fig. 1A) absolutely required for pRS01 conjugation; hence, splicing of the intron is essential for plasmid transfer (Mills et al., 1996; Zhou et al., 2000). The Tra1–2 region from pRS01 also contains, between ltrD and ltrE, a conjugation transfer origin (oriT) recognized by the relaxase to initiate conjugative transfer (Fig. 1A) (Mills et al., 1998).

Figure 1.

Conjugation and mobility assays.
A. Schematic of the Tra1–2 region from the L. lactis conjugative plasmid pRS01 (adapted from Mills et al., 1996). The PstI fragment (7.5 kb, Tra1–2 region) from pRS01 present in the pLE12 vector and the Ll.LtrB group II intron are shown. The conjugative origin of transfer (oriT) is also represented (black circle).
B. Conjugation/retrohoming assay between two L. lactis strains. The first step (1) represents the transfer of pLE12 by conjugation from the donor to the recipient strain. The second step (2) shows the invasion (retrohoming) of the homing site (E1/E2) present on the recipient plasmid (pMNHS) by the Ll.LtrB intron expressed from the donor plasmid (pLE12), thus forming a mobility product (pMNHS+).
C. Conjugation/retrotransposition assay between two L. lactis strains. The first step (1) represents the transfer of pGNIK by conjugation from the donor to the recipient strain. The second step (2) illustrates Ll.LtrB invasion of the recipient strain's chromosome by retrotransposition (group I). The loss of the intron-carrying plasmid [pGNIK, temperature sensitive (Ts)] from transconjugants, following a temperature shift, allowed the selection of Ll.LtrB chromosomal insertions (KanR). The chromosome is only depicted for the recipient strain.
D. Scoring of intron mobility efficiency (patch hybridization) after conjugation. The plasmid mix (donor, recipient and mobility products) from 10 independent transconjugants are recovered by plasmid preparation (1) and retransformed in E. coli DH5α cells (LB/Spc) (2). One hundred individual colonies, containing either mobility events or the uninterrupted recipient plasmid, are patched on LB/Spc plates (3) and transferred onto a nylon membrane (4). The nylon membrane is then hybridized with an intron-specific probe to calculate the ratio of recipient plasmids that received the intron. Ll.LtrB group II intron, grey; E1 and E2, Ll.LtrB exons.

We first analysed the conjugation efficiency of pLE12 between two L. lactis strains. The donor strain (LM0230) contained the pLE12 vector, and the recipient strain (MMS372) contained the pMNHS plasmid, harbouring the intron homing site (HS) (LM0230/pLE12 X MMS372/pMNHS). Using milk-plate conjugation assays, we noticed, as expected from previous studies (Mills et al., 1994), that the Tra1–2 region poorly promotes conjugation of its host plasmid. Only a small increase in conjugation was observed between the backbone plasmid pLE1 and pLE12 (Table 1; 3.3-fold). More-over, the overall conjugation efficiencies were relatively low compared with what was previously observed for pRS01 transfer (10−6 versus 10−3) (Mills et al., 1994), suggesting that some conjugative factors, probably those expressed from the Tra3 and Tra4 regions, were missing.

Table 1. . Conjugation efficiency of Ll.LtrB-carrying plasmids.
LM0230 to
MMS372 (10−6)
NZΔltrB a to
MMS372 (10−4)
NZ9800 to
LM0230 (10−3)
NZ9800 to
MMS372 (10−3)
NZ9800 to
JH2-2 (10−7)
  • Efficiencies are an average of three independent conjugation assays. All the recipient cells contained the pMNHS (SpcR) recipient plasmid.

  • a. NZΔltrB = NZ9800ΔltrB::TetR.

  • b. The conjugation efficiency is 10−8.

  • c

    . DNase I/RNase conjugation control assays.

  • d. For these assays, the pMNHS recipient plasmid is SpcS-ErmR.

  • NZΔltrB, NZ9800, LM0230(recA+) and MMS372(recA) are L. lactis strains, and JH2-2 is an E. faecalis strain.

  • ND, not detected; no transconjugants (three assays) (<10−8).

pLE11.6 ± 0.3ND4.9 ± 1.2bNDND
pLE125.2 ± 0.32.2 ± 0.62.3 ± 0.62.3 ± 0.51.6 ± 0.3
pLE12c  2.7 ± 0.6 6.0 ± 4.2
pLE12I  3.7 ± 0.84.8 ± 1.1 
pLE12IKd  3.7 ± 0.1 5.5 ± 0.8
pLE12RT  1.4 ± 0.52.5 ± 0.5 
pLE12Mat 7.7 ± 2.9b2.7 ± 0.45.5 ± 0.7 
pLE12Endo  2.4 ± 0.91.9 ± 0.1 
pLE12ΔORF  4.3 ± 0.4  
pLE12ΔD5  3.9 ± 1.0  

In order to study Ll.LtrB transfer from pLE12, but in the context of its complete conjugative element (pRS01 or sex factor), we analysed pLE12 transfer from L. lactis donor strains harbouring a chromosomal copy of the sex factor. These strains provided, by complementation, all the conjugation machinery necessary for pLE12 conjugative transfer. The L. lactis NZ9800 strain was derived from MG1363, a strain harbouring a chromosomal copy of the sex factor (Shearman et al., 1996). We confirmed the presence of the sex factor within the chromosome of the NZ9800 strain and its absence from the LM0230 and MMS372 strains by polymerase chain reaction (PCR) and by Southern blots (data not shown). Using NZ9800 as the donor strain and either LM0230 or MMS372 as the recipient strain, we observed, in both cases, a significant increase in conjugation efficiencies between pLE1 (10−8) and pLE12 (10−3) (Table 1). These experiments showed that the sex factor within the NZ9800 chromosome complements pLE12 transfer by conjugation and that the transfer of pLE1 is significantly increased by the presence of the Tra1–2 region from pRS01 (100 000-fold, 10−8−10−3) (Table 1), possibly by the recognition of the transfer origin (oriT). Accordingly, we saw a dramatic drop in the conjugation efficiency of the pLE12 derivative harbouring a splicing-deficient intron [pLE12Mat; no relaxase (LtrB)] from the NZ9800ΔltrB::TetR strain (10−8) but not from NZ9800 (10−3) (Table 1). NZ9800 thus provides the relaxase from its chromosome and drives the conjugation of the pLE12Mat plasmid that is unable to produce its own relaxase. In fact, all the intron mutants, including the non-splicing variants (Mat, ΔORF, ΔD5), show efficient conjugation levels from the NZ9800 strain (Table 1). A modest but reproducible drop (10-fold) in conjugation efficiency was also noticed for pLE12 when only the plasmid-encoded relaxase gene is present (NZ9800ΔltrB) (Table 1).

To verify further that the recipient cells harbouring both plasmids (transconjugants) acquired pLE12 by conjugation, we performed six independent conjugation assays (NZ9800/pLE12 × LM0230/pMNHS) (Fig. 1B). For half the assays, a cocktail of DNase I and RNase was applied onto the mating plates (Trieu-Cuot et al., 1998). Conjugation efficiencies were identical when supplemented or non-supplemented plates were used (Table 1), confirming that pLE12 acquisition by the recipient cells is achieved through conjugation.

Transfer of the Ll.LtrB group II intron between L. lactis strains

To study whether the newly transferred Ll.LtrB intron was still expressed, active and proficient in invading its homing site after its conjugative transfer, we recovered and analysed the plasmid mix from 10 independent transconjugants (Fig. 1B). The presence of the homing site-containing plasmid pMNHS within the recipient cells gave us the opportunity to assess directly Ll.LtrB mobility from pLE12 after its mobilization by conjugation. These plasmids were first analysed by agarose gel electrophoresis, where both donor (D) and recipient (R) plasmids could be seen and distinguished (Fig. 2A), confirming that the CamR/SpcR recipient cells were indeed transconjugants. We also noticed the presence of an additional characteristic band, corresponding to an intron-interrupted recipient plasmid or mobility event (pMNHS + Ll.LtrB) (Fig. 2A) (Cousineau et al., 1998). This additional band and the donor plasmid pLE12 were both absent from recipient cells carrying only the recipient plasmid (LM0230/pMNHS) (Fig. 2A). We confirmed that the additional band corresponds to Ll.LtrB mobility events by Southern blot, using a Ll.LtrB-specific probe (Fig. 2B). The 32P-labelled intron probe annealed to the intron-containing recipient plasmid (M) and to both the donor plasmid (D; pLE12) and the positive control (M; pMNHS + Ll.LtrB) and did not hybridize to the uninterrupted recipient plasmid (R; pMNHS).

Figure 2.

Lateral transfer of Ll.LtrB.
A. Agarose gel (0.5%) containing the plasmid mix (undigested) recovered from 10 recipients (SpcR) and 10 transconjugants (SpcR/CamR) after pLE12 conjugation (NZ9800/pLE12 × LM0230/pMNHS, Fig. 1B). The presence of the donor (D), recipient (R) and mobility product (M) is indicated.
B. Southern blot of the agarose gel shown in (A) (Cousineau et al., 1998).

Using patch hybridization assays (Fig. 1D) (Cousineau et al., 1998), we scored the mobility efficiency of the Ll.LtrB intron for the 10 independent mobility assays and calculated an average efficiency of 10.5% (Table 2). The mobility efficiency of the Ll.LtrB intron after conjugation of pLE12 is slightly higher (2.4-fold) than previously observed when the pLE12 and pMNHS plasmids were co-transformed in the same strain (LM0230) (Cousineau et al., 1998).

Table 2. . Efficiency of Ll.LtrB mobility from pLE12 derivatives after their transfer by conjugation.
 NZ9800/plasmid × LM0230(recA+)/pMNHS
gpI SJgpIIgpI lossgpI SJgpIIgpI loss
 1101 110055100 0 0 2 0 0
 2104 410011100 2 0 0 0 0
 3107 710022100 0 0 0 0 0
 4112 210022100 0 0 1 0 0
 5 82 210044100 0 0 2 04
 6 82 210044100 0 0 0 0 0
 7104 41005510055 096 0 0
 8133 310011100 0 010 0 0
 9163 310044100 1 0 0 0 0
10 94 410011100 0 1 447 0
Mean10.5%  3.2%  2.9% <1%<1% 1.1%<1%<1%
SEM±0.8 ±0.5  ±0.5   ±0.5  
 NZ9800/plasmid × MMS372(recA+)/pMNHS
gpI SJgpIIgpI loss
  1. The mobility efficiencies of the Ll.LtrB intron variants from the donor plasmids (Fig. 3) were calculated by patch hybridization assays (Experimental procedures) analysing 100 colonies for each of the 10 independent transconjugants (nos 1–10).

  2. For conjugation assays with pLE12IK, the pMNHS used was Spcs-ErmR.

  3. Mobility efficiencies from supermobiles are underlined and not included in the means.

  4. gpI SJ, number of colonies revealed with the td group I intron splice junction probe (ligated exons); gpII, number of colonies revealed with the group II intron probe; gpI loss, percentage of mobility events lacking the group I intron.

 11001460  –  044 1
 2 14191910086 0 4
 3 502020100100 0 1
 4 352122 95  091 4
 5 261617 94  0 0 1
 6 191515100  0 0 3
 7941313100  0 0 0
 8 422424100  0 028
 9 221985  –  0 0 1
10 13151510010024 4
Mean 27.6% 18.1%  <1%<1% 2.1%
SEM ±4.8 ±1.3   ±0.5

Retrohoming of Ll.LtrB after conjugation

To determine whether Ll.LtrB invades its homing site via retrohoming, we constructed a series of Ll.LtrB mutants (Fig. 3) (Cousineau et al., 1998; 2000) in pLE12 and calculated the resulting intron mobility efficiencies after conjugation from NZ9800 to LM0230/pMNHS recipients (Table 2). Two of these constructs were artificial twintrons (Fig. 3; pLE12I, pLE12IK) in which the phage T4 group I td intron was present within domain IV of Ll.LtrB, downstream of LtrA. This autocatalytic group I intron splices only at the RNA level. The ratio of mobility products compared with the total amount of recipient plasmids was calculated by patch hybridization assays for the different Ll.LtrB intron variants. Mobility products were identified using an intron-specific probe (gpII), whereas a group I splice junction probe (gpI SJ) was used to assess whether the mobility products had lost the group I intron (gpI loss). As observed previously (Cousineau et al., 1998), the twintron constructs, carrying additional sequences, are not as proficient as the wild-type intron (Table 2). However, mobility events obtained after conjugation of the Ll.LtrB twintron constructs (pLE12I, pLE12IK) showed conclusively that Ll.LtrB invades its homing site using an RNA intermediate, as all the mobility events had lost the retromobility indicator td intron (Table 2, columns gpI SJ and gpI loss) (Cousineau et al., 1998; 2000; Ichiyanagi et al., 2002). If Ll.LtrB mobility had proceeded through a DNA-based pathway, the group I intron would still have been present within Ll.LtrB after its insertion. Specific mutants of the intron-encoded protein (IEP) [reverse transcriptase (RT), maturase (Mat) and LtrA (ΔORF)] and of the autocatalytic core of the intron (ΔD5) (Matsurra et al., 1997) were also analysed (Fig. 3). These mutations reduced the mobility efficiency of Ll.LtrB below our detection limit for this assay (>10 fold; <10−2) (Table 2). However, we detected some mobility of the endonuclease mutant (Endo). Its efficiency was reduced 9.5-fold compared with the wild-type Ll.LtrB. Interestingly, this Ll.LtrB variant, which is also known to be deficient in RT activity (San Filippo and Lambowitz, 2002), was shown previously to be virtually immobile (59-fold lower) in Escherichia coli (Cousineau et al., 1998). The analysis of this Ll.LtrB mutant in its original host environment (L. lactis versus E. coli) during and/or after conjugation of its carrying plasmid could account for the difference observed in mobility efficiencies. Nevertheless, this result is not completely unexpected, as some bacterial group II introns harbour IEPs lacking an endonuclease domain (Martinez-Abarca and Toro, 2000a; Dai and Zimmerly, 2002), one of which, the RmInt1 intron from Sinorhizobium meliloti, has already been shown to be mobile (Martinez-Abarca et al., 2000; Martinez-Abarca and Toro, 2000b).

Figure 3.

Different constructs of the Ll.LtrB group II intron. Ll.LtrB group II intron, grey; phage T4 group I td intron, black; Ll.LtrB exons, E1 and E2; wild type Ll.LtrB group II intron, WT.

We also evaluated the mobility efficiency of some of the Ll.LtrB constructs after their transfer by conjugation from NZ9800 to MMS372, a recA isogenic strain of LM0230. We noticed that, although the conjugation levels are comparable (Table 1), the Ll.LtrB retrohoming efficiency was higher when using MMS372 as the recipient strain (Table 2, pLE12 (WT) and pLE12Endo). A similar increase in Ll.LtrB mobility in MMS372 (recA) cells compared with LM0230 (recA+) was observed previously when pLE12 and pMNHS were co-transformed in these same strains (Mills et al., 1997; Cousineau et al., 1998). These data show that the homologous recombination system is not involved in Ll.LtrB mobility during and/or after conjugation. We also noticed that the mobility efficiency of the RT (RT) and maturase (Mat) mutants of LtrA did not vary between these strains.

While scoring the mobility efficiency of the different Ll.LtrB constructs after conjugation, we observed unexpectedly high mobility levels for some of the 10 independent events examined (Table 2, underlined values). These events, often much more proficient than the wild-type intron, were observed in different conditions for all constructs studied (Fig. 3 and Table 2). The detailed mechanism promoting these mobility events is currently under investigation.

Taken together, these results show that Ll.LtrB is retrohoming to its new location during and/or after conjugation of its host plasmid, and that the three functions of LtrA (RT, maturase, endonuclease) are involved in its mobility pathway. The complementation/conjugation system in which the relaxase enzyme (LtrB) was expressed from the chromosome of the donor strain gave us the opportunity to study the transfer of different intron mutants, between L. lactis strains, after conjugation of their host plasmid. The level of LtrB produced from the chromosome is sufficient to sustain the transfer of our pLE12 variants to wild-type conjugation levels (Table 1; 10−3).

Ll.LtrB retrotransposition after its transfer by conjugation

In order to determine whether the Ll.LtrB group II intron was able to invade ectopic sites after its transfer by conjugation, we performed conjugation assays between L. lactis strains (NZ9800/pGNIK × NZ9800ΔltrB::CamR) in which the recipient strain was plasmid free (Fig. 1C). We then looked, after conjugation of the donor plasmid, for chromosomal insertions of Ll.LtrB in recipient cells.

In this conjugation/retrotransposition assay (Fig. 1C), the donor plasmid pGNIK (pG + host5 based) contains a temperature-sensitive origin of replication (Ts). This plasmid replicates normally in cells grown at the permissive temperature (30°C) but cannot replicate and gets diluted out of cultures grown at 37°C. This feature allowed us to select for retrotransposition events (KanR) following the loss of the intron-carrying plasmid. The Ll.LtrB variant we used for this assay harboured the td group I intron and the KanR gene, and was only flanked by small portions of both exons (Fig. 3, pGNIK). Although pGNIK is a non-conjugative plasmid, we were nevertheless able to obtain transconjugants (NZ9800ΔltrB/pGNIK, CamR/ErmR). The conjugation frequencies were comparable, when the mating plates were supplemented (3.3 ± 0.6 × 10−8) or not (2.6 ± 0.4 × 10−8) with the DNase I/RNase cocktail (three assays each). Despite the low conjugation efficiencies, the DNase I/RNase control suggests that pGNIK is indeed transferred between the L. lactis strains by conjugation.

We selected for Ll.LtrB insertions within the L. lactis chromosome of transconjugants after loss of the donor plasmid upon temperature shift. Kanamycin-resistant colonies were obtained even if intron expression was not induced with nisin. Using Southern blots (Ll.LtrB probe), we showed that these independently isolated recipient strains contained only one Ll.LtrB insertion per genome at five different sites. Moreover, using the td group I intron splice junction probe, we found that the td intron was precisely spliced out in all cases (Fig. 4A). The absence of the td intron from these chromosomal insertions suggests that Ll.LtrB invaded the chromosome of the recipient cell by retrotransposition (ectopic sites, RNA intermediate). Taking advantage of the kanamycin gene present within the newly inserted introns, we cloned and sequenced the five independent Ll.LtrB chromosomal insertion sites (Fig. 4B). The sequences confirmed that these Ll.LtrB insertions were retrotransposition events and that they were new insertion sites never observed in previous retrotransposition studies (Cousineau et al., 2000; Ichiyanagi et al., 2002). The homology between these sites and the wild-type homing site are, as expected, confined to a short 13–17 bp region spanning the intron insertion sites. This suggests that the intron invaded these target sites through reverse splicing (Cousineau et al., 2000; Ichiyanagi et al., 2002). As anticipated from previous retrotransposition studies using this Ll.LtrB variant (pGNIK) (Cousineau et al., 2000; Ichiyanagi et al., 2002), the intron was inserted, in all five cases, in the same orientation as the interrupted genes.

Figure 4.

Retrotransposition sites.
A. Southern blot on digested genomic DNA (NcoI) from five independent KanR/pGNIK NZ9800ΔltrB::CamR colonies (1–5) and from NZ9800ΔltrB::CamR (NZΔL, negative control). The 32P-labelled td splice junction probe revealed only one signal per lane all at different positions (1–5).
B. Sequences of five independent Ll.LtrB retrotransposition sites isolated from the chromosome of the NZ9800ΔltrB::CamR strain are shown and compared with the WT retrohoming site (HS). White nucleotides on a black background represent identity with HS or potential G-U basepairs (asterisk) with intron RNA. The arrowhead indicates the Ll.LtrB insertion site. The interrupted genes and the potential interactions between the intron RNA and its substrate at the insertion sites are illustrated (IBS1/EBS1, IBS2/EBS2, δ/δ′). Intron binding sites, IBS1 and IBS2; exon binding sites, EBS1 and EBS2.

These data demonstrate that the Ll.LtrB group II intron can be widely spread, after its transfer by conjugation, invading non-homologous sites within the chromosome of the recipient strain by retrotransposition. Ll.LtrB is thus not restrained to move exclusively by retrohoming to a very specific site but can also invade a multitude of sites within its new host's chromosome. Moreover, we showed that Ll.LtrB can be transferred even when present on a non-conjugative plasmid. This situation is probably restricted to donor cells housing some minimal transfer functions allowing cell–cell contacts and the formation of a mating channel.

Lateral transfer of the Ll.LtrB intron between Gram-positive bacteria

To demonstrate further the biological relevance of Ll.LtrB lateral transfer, we asked whether pLE12 could relocate from L. lactis to other bacterial species by conjugation. We thus looked at the conjugation of pLE12 from L. lactis to Enterococcus faecalis, another low-GC Gram-positive bacterium, by filter mating assays (Sasaki et al., 1998). Using the JH2-2 E. faecalis strain as the recipient strain (NZ9800/pLE12 × JH2-2/pMNHS), we obtained transconjugants at a frequency of 10−7 (Table 1). Although the conjugation frequencies from L. lactis to E. faecalis (10−7) are much lower than those observed between L. lactis strains (10−3), the DNase I/RNase control assay confirmed that pLE12 was transferred to E. faecalis by conjugation (Table 1).

Using the pLE12 and pLE12IK constructs, we detected lateral transfer of the Ll.LtrB group II intron by retrohoming with efficiencies of 6 ± 1% and 1.2 ± 0.2% respectively. Again, the mobility efficiency of the group I/KanR-containing construct is lower than the wild-type intron, and they are both less efficient in E. faecalis than in L. lactis. The presence of L. lactis promoters could explain the mobility difference in the two cell backgrounds, with lower expression of the ltrB and/or ltrA gene leading to lower levels of active RNPs in E. faecalis. However, the absence of the group I intron from the great majority of the mobility events (67%) demonstrates that the intron is invading its homing site via the retrohoming pathway in E. faecalis. These results confirm that plasmid conjugation can support the lateral or horizontal transfer of Ll.LtrB in the Gram-positive branch of the bacterial kingdom.


In this study, we developed conjugation/retrohoming and conjugation/retrotransposition assays to analyse the lateral transfer of the Ll.LtrB group II intron from L. lactis. We have presented the first experimental demonstration that bacterial group II introns can be laterally transferred in the bacterial kingdom using both retrohoming (L. lactis to L. lactis and L. lactis to E. faecalis) and retrotransposition (L. lactis to L. lactis) pathways during and/or after their transfer by conjugation. This work demonstrates that group II introns can invade either resident plasmids or the chromosome of the recipient strain after their transfer by conjugation. We also showed that group II introns can be laterally transferred by conjugation even when present on non-conjugative plasmids if the donor cells house some minimal transfer functions allowing cell–cell contacts and the formation of a mating channel.

The data presented are biologically relevant for the following reasons. First, Ll.LtrB was originally discovered in L. lactis on a natural conjugative plasmid (pRS01) and within a chromosomally located sex factor (Mills et al., 1996; Shearman et al., 1996). These two elements were shown to be actively transferred by conjugation between L. lactis strains and Gram-positive bacteria (Mills et al., 1996; Shearman et al., 1996). Moreover, it was proposed that pRS01 can be transferred by conjugation from L. lactis to other Gram-positive genera such as streptococci and lactobacilli (Mills et al., 1996), indicating a possible vast dissemination of Ll.LtrB by conjugation within the Gram-positive bacterial branch. Secondly, in our conjugation/mobility assays, natural levels of active RNPs were produced as the expression of the ltrB gene, interrupted by the intron, and ltrA present within the intron were under the control of their native promoters (not overexpressed). Thirdly, Ll.LtrB can be efficiently disseminated after its transfer by conjugation, through retrohoming, not only between closely related L. lactis strains, but also from L. lactis to other Gram-positive bacteria (E. faecalis). Fourthly, Ll.LtrB can also invade multiple chromosomal sites by retrotransposition into a recipient cell after its transfer by conjugation. Fifthly, Ll.LtrB splicing directly controls the expression level of the relaxase gene (ltrB), essential for conjugative transfer, thus controlling the conjugation of its host element and at the same time its own dissemination and survival.

Taken together, these results suggest that conjugation of these broad-host-range mobile elements (pRS01 and sex factor) should happen in nature and may support a wider dissemination of Ll.LtrB than we demonstrated using pLE12 in a constrained laboratory setting. Plasmid pRS01 and the L. lactis sex factor can thus be considered as infectious elements driving the horizontal transfer and spread of their associated mobile intron.

As conjugation is the most efficient way to transfer genetic information between widely diverged bacterial species and even across phyla (Lambowitz and Belfort, 1993), conjugation may have been and probably still is an important means of intron dispersal.

Interestingly, the great majority of group II introns found in bacteria are associated with various mobile elements, such as insertion sequence (IS) elements, transposons, conjugative plasmids, pathogenicity islands and virulence plasmids (Martinez-Abarca and Toro, 2000a; Dai and Zimmerly, 2002). The results presented in this study suggest that these introns could also be disseminated and spread by horizontal transfer in the bacterial kingdom, following the transfer of their host mobile elements. The association of some group II introns with other mobile elements may have been a means of survival for these introns and could explain why they are still present and unexpectedly highly represented within contemporary bacterial genomes.

Experimental procedures

Strains and plasmids

The strains and plasmids used in this study are described in Table 3. L. lactis and E. faecalis strains were grown without shaking in M17 supplemented with 0.5% glucose (GM17) (30°C) and brain–heart infusion (BHI) media (37°C) respectively. E. coli strains (DH5α, DH10β), used for cloning and mobility scoring, were grown with shaking at 37°C in LB broth. The milk plates used in conjugation assays were made of 5% dry milk (Carnation), 1% dextrose and 1.5% agar. The pLE12 plasmid consists of the Tra1–2 region from pRS01 (7.5 kb, PstI, Fig. 1A) cloned into the pLE1 vector at its unique PstI site. The pMNHS plasmid (pMN1343) contains a 271 bp homing site (exon 1: 179 bp; exon 2: 92 bp) (HindIII) inserted at the unique HindIII restriction site in the pDL278 vector. The pLE12-based constructs harbouring different Ll.LtrB mutants were made by replacement cloning using either BsrGI and BsiWI or BsrGI and KpnI restriction enzymes. Ll.LtrB mutants are as follows: RT, reverse transcriptase mutant (YADD to YAAA) (Matsuura et al., 1997); Mat, maturase defective (SC463 to LA) (Matsuura et al., 1997; Cousineau et al., 1998); Endo, endonuclease domain deletion (amino acids 543–599) (Matsuura et al., 1997); ΔORF, amino acids 40–572 in LtrA were replaced by TR (RT, Mat, Endo) (Matsuura et al., 1997); ΔD5, domain V of the intron was deleted (non-splicing) (Matsuura et al., 1997). The pGNIK plasmid is a pG+host5-based construct (Ts) carrying the Ll.LtrB intron (HindIII, td intron, KanR) under the control of the nisin-inducible promoter. Selective media contained the following concentrations of antibiotics: chloramphenicol, 5 or 10 µg ml−1; spectinomycin, 300 µg ml−1; erythromycin, 300 µg ml−1; kanamycin, 20 µg ml−1; fusidic acid 25 µg ml−1; tetracycline, 3 µg ml−1.

Table 3. . Bacterial strains and plasmids.
Strain or plasmidRelevant characteristicsaSource, description and/or reference
L. lactis strains
 LM0230Plasmid free, Rec+Donor and recipient strain for conjugation assaysb
 MMS372Plasmid free, RecRecipient strain for conjugation assays, isogenic to LM0230c
 NZ9800Plasmid free, Rec+Donor strain for conjugation assays, chromosomal sex factor (Ll.LtrB intron)d
 NZ9800ΔltrBCamR, plasmid free, Rec+The Ll.LtrB intron was deleted with parts of its flanking exons (HindIII)e
 NZ9800ΔltrBTetR, plasmid free, Rec+The Ll.LtrB intron was deleted with parts of its flanking exons (HindIII)f
E. faecalis strain
 JH2-2Plasmid free, RifR, FusRRecipient strain for conjugation assays (L. lactis to E. faecalis)g
 pLE1CamR, 8.7 kbBackbone shuttle plasmid (Gram+/Gram–)h
 pLE12CamR, 16.2 kbpLE1 containing the Tra1–2 region from pRS01 (Ll.LtrB; 7.5 kb)I
 pLE12ICamR, 16.6 kbLl.LtrB contains the td group I intron in domain IVj
 pLE12IKCamR, 17.7 kbLl.LtrB contains the td group I intron and KanR gene in domain IV
 pLE12RTCamR, 16.2 kbLl.LtrB carrying LtrA with a mutated reverse transcriptase domain
 pLE12 MatCamR, 16.2 kbLl.LtrB carrying LtrA with a mutated maturase domain
 pLE12 EndoCamR., 16.0 kbLl.LtrB carrying LtrA missing the endonuclease domain
 pLE12ΔORFCamR, 14.6 kbLl.LtrB carrying a large deletion within LtrA (RT, Mat, Endo)
 pLE12ΔD5CamR, 16.2 kbLl.LtrB missing domain V, the LtrA protein is intact
 pMNHSSpcR, 6.9 kbShuttle plasmid pDL278 containing the Ll.LtrB homing site (HindIII, 271 bp)k
 pMNHS/ErmRErmR, SpcS, 7.7 kb The Erm gene (PCR) was inserted in the middle of the SpcR gene (KpnI)
 pGNIKErmR, KanR, 11.2 kbTs plasmid, Ll.LtrB intron (td intron, KanR) is flanked by short exons (HindIII),  intron is located downstream from the nisin-inducible promotere,f

Conjugation assays

Lactococcus lactis strains (donor, recipient) were diluted (0.4 or 0.8 ml in 10 ml) from overnight saturated cultures and grown for 7 h at 30°C with the appropriate antibiotics. Cells were recovered by centrifugation, the pellets were mixed (1:1), spread on milk plates and incubated at 30°C for 12 h. The cell mixtures were recovered with 1× PBS (1 ml), and serial dilutions were made to score donor (CamR), recipient (SpcR) and transconjugant (CamR/SpcR) cells (Mills et al., 1996).

Conjugation assays from L. lactis (NZ9800) to E. faecalis (JH2-2) were done on GM17 plates (filter mating) at 37°C (Sasaki et al., 1998). The identity of the recipient strain (JH2-2) containing both donor and recipient plasmids was confirmed by its resistance to fusidic acid (25 µg ml−1). The conjugation efficiencies (three assays) were calculated as the number of transconjugants/donor cells (CamR-SpcR/CamR).

In order to perform conjugation controls, a cocktail of DNase I and RNase (100 U ml−1 each) (Trieu-Cuot et al., 1998) was applied onto the mating plates, and the conjugation efficiencies were calculated as described previously. The typical conjugation control experiment uses only DNase I but, as the Ll.LtrB intron is a retroelement, we also controlled for the possible non-conjugative uptake of active RNP particles [intron RNA + intron-encoded protein (IEP)].

Mobility assay (patch hybridization)

The plasmid mixtures (donor, recipient and mobility products) from 10 independent transconjugants were prepared (same conjugation assay), retransformed into E. coli (DH5α) and plated on LB/Spc plates to select for recipient plasmids interrupted or not by the intron (Fig. 1C). In order to calculate the percentage of recipient plasmids that received the intron (mobility products), 100 isolated colonies (SpcR) were patched for each independent assay (LB/Spc plates). The patches were lifted on nylon membranes and hybridized with the appropriate 5′-32P end-labelled probes for homing [Ll.LtrB probe, 5′-GTATGGCTATGCCCGGAATAC-3′ (3′ end of the intron) or 5′-CCGTGCTCTGTTCCCGTATCAGC-3′ (5′ end of the intron)] and retrohoming (td intron splice junction probe, 5′-ATTAAACGGTAGACCCAAGAAAAC-3′) (Cousineau et al., 1998). The td intron splice junction probe recognizes the two td ligated exons (12 nt each exon) flanking it and gives a positive signal only upon group I intron loss in mobility products (Belfort et al., 1990).

Retrotransposition assay

The conjugative transfer of pGNIK from NZ9800 to NZ9800ΔltrB::CamR was performed as described above. After conjugation, a transconjugant colony (NZ9800ΔltrB/pGNIK, CamR/ErmR) was grown overnight in GM17 at 37°C to prevent replication of the pGNIK Ts vector. Two successive overnight cultures at 37°C (1:1000) were done followed by a final overnight culture in GM17/Kan at 37°C (1:1000) to select for Ll.LtrB chromosomal insertions [retrotransposition events (RTP)]. The cells were then diluted, plated on GM17/Kan at 30°C, and isolated colonies were picked and grown overnight (GM17/Kan, 30°C). These cultures were split for plasmid and genomic DNA isolation. Plasmid preparations confirmed that pGNIK was lost during growth at 37°C. Genomic DNAs were digested (SpeI or NcoI) and hybridized with the td splice junction probe and showed only one signal at different positions (Fig. 4A). This result was confirmed, again by Southern blot (SpeI), using a Ll.LtrB-specific probe (5′-CCGTGCTCT GTTCCCGTATCAGC-3′). To isolate and characterize the RTP events, genomic DNA was digested with SpeI, HindIII or PvuII, cloned in the pBS vector and selected for on LB/Kan plates (DH5α or DH10β). Both 5′ and 3′ junctions between Ll.LtrB and its five independent insertion sites were obtained by sequencing using intron-specific primers (5′-CCGTGCTCTGTTCCCGTATCAGC-3′, 5′-CAGAGCCG TATACTCCGAG-3′) (Fig. 4B). Genomic DNAs from the RTP events were also digested with PstI and hybridized with an exon 2 probe (5′-GTGAGAGTTACCTGGAGACT-3′) to confirm the identity of the recipient genomic DNA (NZ9800ΔltrB::CamR) (data not shown).


We thank M. Belfort for kindly providing the NZ9800ΔltrB::TetR strain. We also thank M. Belfort, K. Fiola, R. Lease, J. Wyse and S. Zimmerly for providing comments on the manuscript, and A. Villeneuve for technical assistance. B.C. is a CIHR New Investigator Scholar and a McGill William Dawson Scholar. This work was supported by an NSERC grant to B.C.