Conjugative Transfer of the Integrative Conjugative Elements ICESt1 and ICESt3 from Streptococcus thermophilus

Abstract

Integrative and conjugative elements (ICEs), also called conjugative transposons, are genomic islands that excise, self-transfer by conjugation, and integrate in the genome of the recipient bacterium. The current investigation shows the intraspecies conjugative transfer of the first described ICEs in Streptococcus thermophilus, ICESt1 and ICESt3. Mitomycin C, a DNA-damaging agent, derepresses ICESt3 conjugative transfer almost 25-fold. The ICESt3 host range was determined using various members of the Firmicutes as recipients. Whereas numerous ICESt3 transconjugants of Streptococcus pyogenes and Enterococcus faecalis were recovered, only one transconjugant of Lactococcus lactis was obtained. The newly incoming ICEs, except the one from L. lactis, are site-specifically integrated into the 3′ end of the fda gene and are still able to excise in these transconjugants. Furthermore, ICESt3 was retransferred from E. faecalis to S. thermophilus. Recombinant plasmids carrying different parts of the ICESt1 recombination module were used to show that the integrase gene is required for the site-specific integration and excision of the ICEs, whereas the excisionase gene is required for the site-specific excision only.

Horizontal transfer of genomic islands plays a key role in bacterial evolution, leading to the acquisition of advantageous functions or to major modifications of the bacterial way of life (37, 38). However, the mechanism of their transfer and integration remains generally unknown. Recent in silico analyses revealed that numerous genomic islands could be integrative and conjugative elements (ICEs) or elements deriving from them (18, 30, 62, 67) The ICEs, also called conjugative transposons, were defined as elements which encode their own excision, their transfer by conjugation, and their integration in a replicon of the recipient cell, whatever their specificity and their mechanism of integration and conjugation (18). Most of the ICEs encode a putative integrase belonging to the tyrosine recombinase family. This enzyme, generally associated with an excisionase, catalyzes the recombination between identical sequences carried by the attL and attR recombination sites flanking the element. This event leads to the excision of a circular form of the ICE harboring an attI site, resulting from the recombination between attL and attR sites, and to a chromosome carrying an empty attB site. After conjugative transfer, most of the ICEs integrate by site-specific recombination between identical sequences carried by the attI site and the attB site. The attB sites are generally located in the 3′ end of a tRNA gene or a specific open reading frame (ORF). However, some ICEs, such as Tn916, can integrate in numerous sites (57). Host ranges can be very different in different ICEs. While Tn916, from the firmicute Enterococcus faecalis, has a very broad host range (7, 58), the sex factor from the firmicute Lactococcus lactis can self-transfer only between strains of this species (32).

The ICEs, like the other transferable elements such as prophages and plasmids, have a modular structure and evolution (18, 19, 70). Each module is composed of all the sequences and genes involved in a biological function, such as site-specific recombination, conjugation, or antibiotic resistance, and can be exchanged between transferable elements.

Stimuli inducing the conjugative transfer of ICEs have been identified in a small number of elements, including Tn916, clc, and ICEs from Bacteroides (11, 24, 59, 61). Recently, DNA-damaging agents were found to derepress the excision and transfer of ICEBs1 from Bacillus subtilis (2) and SXT from Vibrio cholerae (3). This is similar to the derepression of numerous prophages by DNA damage that leads to activation of the lytic pathway (29, 45).

Genomic islands related to each other were found to be integrated in the 3′ end of fda from Streptococcus thermophilus, an ORF encoding a putative fructose-1,6-diphosphate aldolase. They carry sequences almost identical to L. lactis sequences, including insertion sequences (ISs), a restriction-modification module, and a cadmium resistance module (19, 20, 54). Two of these elements, ICESt1 and ICESt3, harbor very closely related recombination modules that encode a tyrosine integrase and an excisionase. They excise by site-specific recombination between two 27-bp direct repeats included in their two attachment sites, attL and attR (20, 54). A DNA-damaging agent, mitomycin C (MC), derepresses the site-specific excision of ICESt1 and ICESt3 (4). These elements also carry an almost identical putative conjugative module distantly related to those of two ICEs, Tn916 and ICEBs1, suggesting that these genomic islands are ICEs or derived from ICEs (19, 54). In the present report, we show that these elements are able to transfer by conjugation to other strains of S. thermophilus and we characterize the ICESt3 host range. In the vast majority of transconjugants, the S. thermophilus ICEs were found to be integrated by site-specific recombination into the 3′ end of the fda gene. Moreover, the transfer of this ICE is derepressed by MC-induced DNA damage. ICESt3 is still active in the transconjugants and can transfer again to a new recipient. Finally, recombinant plasmids carrying different parts of the ICESt1 recombination module were used to determine the role of the integrase, the excisionase, and the attI site in the site-specific recombination events.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

The original strains and plasmids used for electroporation to obtain recipient or donor strains are listed in Table 1. Strains which were modified or mutated were named according to the modification. For example, Streptococcus pyogenes ATCC 12202 Rif^r corresponds to a mutated derivative of S. pyogenes ATCC 12202 resistant to rifampin, and S. thermophilus LMG18311/pMG36e (Erm^r) corresponds to a derivative of S. thermophilus LMG18311 carrying the plasmid pMG36e encoding resistance to erythromycin.

TABLE 1.

Strains and plasmids used in this work

Strain or plasmid	Relevant phenotype or genotypea	Source or reference
Strains
S. thermophilus
LMG18311	Wild-type strain, no element integrated in the fda gene	BCCM/LMG
CNRZ368	Wild-type strain carrying ICESt1	INRA-CNRZ
CNRZ385	Wild-type strain carrying ICESt3	INRA-CNRZ
CNRZ368 ICESt1cat	S. thermophilus CNRZ368 carrying ICESt1 tagged with the cat gene inserted in a noncoding region flanked by fragments of IS1192 and of IS981, Cmr	This work
CNRZ368 ICESt1spc	S. thermophilus CNRZ368 carrying ICESt1 tagged with the spc gene inserted in a noncoding region flanked by fragments of IS1192 and of IS981, Spcr	This work
CNRZ385 ICESt3cat	S. thermophilus CNRZ385 carrying ICESt3 tagged with the cat gene inserted in the pseudogene Ψorf385J, Cmr	This work
CNRZ385 ICESt3spc	S. thermophilus CNRZ385 carrying ICESt3 tagged with the spc gene inserted in the pseudogene Ψorf385J, Spcr	This work
S. pyogenes ATCC 12202	Wild-type strain, no element integrated in the fda gene	ATCC
S. agalactiae
COH1	Wild-type strain, no element integrated in the fda gene	74
539-22	Wild-type strain, no element integrated in the fda gene	16
L. lactis subsp. cremoris MG1363	Wild-type strain, no element integrated in the fda gene	31
L. lactis subsp. lactis IL1403	Plasmid-free strain derived from L. lactis subsp lactis IL594, no element integrated in the fda gene	22
E. faecalis JH2-2	No element integrated in the fda gene, Rifr Fusr	40
E. casseliflavus 664.1H1	Wild-type strain, oral isolate from monkey with amalgam fillings, presence of an element integrated in the fda gene not determined, Tcr	A.O. Summers (Athens, GA)
Lactobacillus delbrueckii subsp. bulgaricus VI104	Wild-type strain, presence of an element integrated in the fda gene not determined	Collection Génétique Microbienne INRA
B. subtilis
168	Wild-type strain, no element integrated in the fda gene	43
CU2189	Wild-type strain, presence of an element integrated in the fda gene not determined	23
Staphylococcus aureus RN4220	Restriction-deficient derivative of phage group III strain 8325-4, presence of an element integrated in the fda gene not determined	42
C. difficile
CD37	Wild-type strain, presence of an element integrated in the fda gene not determined, Tcr	63
R20291	Wild-type strain, presence of an element integrated in the fda gene not determined	Anaerobe Reference Laboratory, Cardiff, United Kingdom
E. coli
EC101	supE hsd-5 thi (lac-proAB) F (traD6 proAB lacIqlacZ M15) repA, derivative of strain TG1 (56)	44
DH5α	supE44 lacU169 (φ80 lacZ M15) hsdR17 endA1 gyrA96 thi-1 relA1	56
HB101	supE44 hsdS20 recA13 ara-14 proA2 lacY1 galK2 rpsL20 xyl-5 mtl-1, no element integrated in the fda gene	47
Plasmids
pG+host9	3.8 kb, pWV01-type thermosensitive replication origin from pVE6002, Ermr	46
pSET4s	4.5 kb, pWV01-type thermosensitive replication origin from pVE6002, lacZ′, Spcr	66
pSET5s	4.4 kb, pWV01-type thermosensitive replication origin from pVE6002, LacZ′, Cmr	66
pMG36e	3.4 kb, replication origin from pWV01, Ermr	72
pTRK398/θH	10.8 kb, P6 constitutive promoter from Lactobacillus acidophilus ATCC 4356 upstream of the llaIR restriction cassette, Ermr	27
pNST260+ and pNST260−	6.2 kb, plasmid derived from pG+host9 carrying the int gene and the attI site from ICESt1 cloned downstream from the P6 constitutive promoter from pTRK398/θH; + and − indicate that the int gene is in forward or reverse orientation compared to the promoter, respectively	This work
pNST261+ and pNST261−	6.5 kb, plasmid derived from pG+host9 carrying the xis and the int genes and the attI site from ICESt1 cloned downstream from the P6 constitutive promoter from pTRK398/θH; + and − indicate that the int gene is in forward or reverse orientation compared to the promoter, respectively	This work
pNST262+ and pNST262−	6.1 kb, plasmid derived from pG+host9 carrying the xis and int genes and the attR site from ICESt1 cloned downstream from P6 constitutive promoter from pTRK398/θH; + and − indicate that the int gene is in forward or reverse orientation compared to the promoter, respectively	This work

^aAbbreviations: Cm^r, chloramphenicol resistance; Ery^r, erythromycin resistance; Fus^r, fusidic acid resistance; Rif^r, rifampin resistance, Spc^r, spectinomycin resistance; Tc^r, tetracycline resistance.

S. thermophilus strains were grown in reconstituted skim milk (10%, wt/vol), M17 broth supplemented with 0.5% lactose (LM17) (Oxoid), Hogg-Jago broth supplemented with 1% lactose (HJL) (64), or Belliker broth (53) at 42°C under anaerobic conditions (GENbox Anaer atmosphere generators and incubation jars from bioMérieux, France). L. lactis strains were grown in M17 broth supplemented with 0.5% glucose (GM17) at 30°C under anaerobic conditions. Lactobacillus delbrueckii subsp. bulgaricus was grown in MRS broth at 37°C under anaerobic conditions. S. pyogenes, E. faecalis, Enterococcus casseliflavus, and Clostridium difficile strains were grown in brain hearth infusion (Difco) broth at 37°C without shaking, whereas Streptococcus agalactiae, Bacillus subtilis, and Staphylococcus aureus strains were grown in the same broth and at the same temperature with shaking. These species were spread on brain hearth infusion plates supplemented with defibrinated horse blood (5%). C. difficile plates were also supplemented with Clostridium difficile selective supplement (Oxoid). Cultures were supplemented with the following antibiotic when required: chloramphenicol, 4 μg · ml^−1; erythromycin, 5 μg · ml⁻¹; fusidic acid, 25 μg · ml⁻¹; rifampin, 25 μg · ml⁻¹; spectinomycin, 50 μg · ml⁻¹; or tetracycline, 10 μg · ml⁻¹.

Escherichia coli strains were grown in Luria-Bertani (LB) broth at 37°C in aerobiosis (56). Recombinant plasmids derived from pG+host9 were transformed into E. coli EC101, a strain containing a chromosomal copy of the pWV01 repA gene (17), and selected at 37°C on LB containing 150 μg of erythromycin ml⁻¹.

DNA manipulations and recombinant DNA.

Preparation of chromosomal and plasmid DNAs and Southern analysis were performed according to standard protocols (56). Primers used in this study were purchased from Eurogentec (Angers, France) and are listed in Table S1 in the supplemental material. Their location and orientation are indicated in Fig. 1. In the figures and in Table S1 in the supplemental material, each letter (A to H) indicates a class of primers hybridizing on the same locus in different species. For example, the primers O132.3, Lla fba2 MG1363, Lla fba2 IL1403, Efa fba2, Bsu fba2, and Sau fba2, belonging to class A, hybridize on the noncoding region flanking the stop codon of the fda ORF in S. thermophilus, L. lactis MG1363, L. lactis IL1403, E. faecalis, B. subtilis, and Staphylococcus aureus, respectively. Most of the primers used are strain specific, species specific, or ICE specific. PCRs and high-fidelity PCRs were carried out according to the instructions with the ThermoPol PCR kit (New England Biolabs, United Kingdom) and with the Triple Master PCR system (Eppendorf, France), respectively. The transformants and the integrants that were characterized by PCR amplification were previously purified by two successive streakings on agar plates. The restriction and modifying enzymes were purchased from New England Biolabs, United Kingdom.

FIG. 1.

Schematic representation of the locations and orientations of the primers used in this work to analyze the transconjugants. The locations of the recombination, conjugation, or regulation modules of the S. thermophilus ICEs are indicated by gray and white rectangles. Recombination sites are drawn as vertical rectangles: black, sequence identical in attL, attR, attI, and attB sites; checkerboards, arm of attR sites and related arm of attI sites; hatched boxes, arm of attL sites and related arm of attI sites. The recombination sites are magnified. The locations and orientations of the primers hybridizing into the ICEs (ICESt1 or ICESt3) and into the flanking sequences are indicated by black arrowheads. Each letter (A to H) indicates a class of primers hybridizing on the same locus in different species (see Table S1 in the supplemental material). The primer pairs A-B, E-F, A-F, and E-B allow the amplification of fragments carrying the attL, the attR, the attB, and the attI sites, respectively. The primer pair C-D allows the amplification of an internal fragment of the regulation module of ICESt1 or ICESt3. The amplification of an internal fragment of the fda gene using the species-specific primers G and H allows the identification of the species. The spectinomycin (spc) or the chloramphenicol (cat) resistance genes introduced in order to tag the ICEs are indicated by a large black rectangle. The sequences between the regulation module and the attL site are specific to each S. thermophilus ICE. The accession number of the untagged ICESt1 and ICESt3 nucleotide sequences are AJ278471 and AJ586568, respectively.

(i) ICE tagging.

S. thermophilus strains carrying a tagged ICE were constructed as previously described by Thibessard et al. (68) with the following modifications. Briefly, the two regions flanking the locus to be tagged were independently amplified by PCR using primers carrying restriction enzymes sites, digested by these enzymes, and joined by ligation together and with pG+host9 digested by the appropriate enzymes. Thereafter, restriction fragments carrying the spectinomycin (spc) or the chloramphenicol (cat) resistance genes from the plasmids pSET4s and pSET5s were ligated within the digested restriction site located between the two PCR fragments. After introduction of the recombinant plasmid into S. thermophilus, two crossovers, upstream and downstream of the tagged region, were selected (68). Thus, the locus of interest is replaced by the tagged locus carried by the plasmid.

(ii) Construction of plasmids carrying fragments of the ICESt1 recombination module.

Fragments of the ICESt1 recombination module, i.e., the int and xis genes and the attI site, were amplified by high-fidelity nested PCRs using primers carrying a SalI or a XhoI restriction enzymes site and then digested by these enzymes. The amplicons were introduced into the plasmid pTRK398/θH digested by the same enzymes. SalI and XhoI produce compatible cohesive ends that allow the selection of the ligation of the ICESt1 integration module in both orientations relative to the P6 constitutive promoter from Lactobacillus acidophilus ATCC 4356 carried by this plasmid (sense and antisense transcriptions). Thereafter, the P6 constitutive promoter and the integration module fragments in either the sense or antisense orientation were amplified by PCR using primers carrying a PstI or KpnI restriction enzymes site. These two amplicons were digested by PstI or KpnI and joined by ligation with the thermosensitive replication vector pG+host9 digested by the same restriction enzymes (Fig. 2). The different cloned amplicons carry the int gene and the attI site (plasmids pNST260+ and pNST260−); the int gene, the xis gene, and the attI site (pNST261+ and pNST261−); or the int gene, the xis gene, and the attR site (pNST262+ and pNST262−).

FIG. 2.

Map of the thermosensitive replication plasmids carrying the P6 constitutive promoter and the various fragments of the ICESt1 recombination module. The location of the thermosensitive replication origin [Ori(pWV01^TS)] is indicated by a white rectangle. The location and the orientation of the ermB gene, encoding resistance to erythromycin, is indicated by a gray arrow. The PstI and KpnI restriction sites used for molecular cloning are indicated. The location and the orientation of the P6 constitutive promoter from Lactobacillus acidophilus ATCC 4356 are indicated by a right-angle arrow. Black arrows indicate the locations and orientations of the int and xis genes, encoding the integrase and the excisionase, respectively. The location and the orientation of a fragment corresponding to the last 60 nucleotides of the fda gene is indicated by a black triangle. Recombination sites are drawn as rectangles: black, sequence identical in attL, attR, and attI sites; checkerboards, arm of attL sites and related arm of attI sites; hatched boxes, arm of attR sites and related arm of attI sites. The recombination sites are magnified. The putative rho-independent terminator downstream from the int gene is indicated by a lollipop. The locations and orientations of the primers AttI2 and AttI3 are indicated by white arrowheads (see Table S1 in the supplemental material).

Electroporation procedure.

S. thermophilus, L. lactis, Lactobacillus delbrueckii subsp. bulgaricus, and B. subtilis strains were electroporated as previously described by Tyvaert et al. (71), O'Sullivan and Klaenhammer (52), Serror et al. (60), and Matsuno et al. (48), respectively. E. faecalis and Staphylococcus aureus strains were electroporated as described by Okamoto et al. (50). The introduced plasmids carry the thermosensitive replication origin from plasmid pVE6002 (46). Plasmids harboring this replication origin can replicate in numerous gram-positive bacteria, including lactococci (8), streptococci (55, 69), enterococci (25, 49), staphylococci (39, 65), and bacilli (51).

Filter matings.

The filter mating protocol was modified from that described by Wang et al. (73) as follows. Both donor and recipient strains were grown for 15 h. A 15-ml culture of the relevant broth was inoculated with 150 μl of overnight culture of the recipient or the donor strain. The cultures were grown at the relevant temperature until mid-exponential phase (optical density at 600 nm of 0.4). Cultures of the donor and recipient were mixed and centrifuged for 15 min in a prewarmed centrifuge at 4,500 × g to form a cell pellet. The pellet was resuspended in 1 ml of LM17 broth with or without 50 μg/ml DNase I (23), and 150-μl aliquots were spread on 0.45-μm-pore-size nitrocellulose filters (Sartorius) on LM17 agar plates which were incubated for 20 h at 42°C (intraspecies transfer) or at 37°C (interspecies transfer). The filters were removed from the agar plates, placed in 50-ml bottles containing 10 ml of sterile LM17 broth, and vortexed for 30 s. Various dilutions were spread on agar plates supplemented with the appropriate antibiotics, and the plates were incubated for 24 h to 48 h in order to count CFU of the donor, the recipient, and the transconjugants.

Integration frequencies of the recombinant plasmids.

The recombinant plasmids created in this work carry a thermosensitive replication origin. The permissive temperature of the plasmids carrying this replication origin is 30°C, whereas temperatures greater than or equal to 35°C are restrictive (46). The plasmids were introduced by electroporation into the bacteria of interest. The transformants were cultivated at 30°C, and various dilutions were spread on plates both with and without erythromycin and incubated at a restrictive temperature. The integration frequency was calculated as the number of CFU resistant to erythromycin divided by the total number of CFU. The integration of the plasmids was verified by PCR amplifications of the resulting attachment sites.

Induction of ICE transfer by MC.

S. thermophilus LMG18311/pMG36e was used as the recipient in the following filter mating experiments. The S. thermophilus donor strains carrying ICESt1 or ICESt3 were grown in LM17 at 42°C to an optical density at 600 nm of 0.4. They were then diluted 10-fold in 10 ml of prewarmed LM17 containing MC at a concentration equal to one-half or one-quarter of the MIC, i.e., 0.2 and 0.1 μg/ml in LM17. A 10-fold dilution without MC was used as a control. After 1 hour of culture, the cells were harvested by centrifugation in a prewarmed centrifuge and washed once with 10 ml of warmed LM17. The cells treated with MC were mixed with exponentially growing recipients (optical density at 600 nm of 0.4) and centrifuged for 15 min in a prewarmed centrifuge to form a cell pellet. The pellet was resuspended in 1 ml of LM17 broth, and 150 μl was spread on 0.45-μm-pore-size nitrocellulose filters on LM17 agar plates, which were incubated for 3 h at 42°C in anaerobiosis. The filters were then treated as in other filter matings.

Statistical analysis.

Statistical analysis was performed as described by Georgin and Mouet (33) and Cumming et al. (26). The means and standard errors from at least five independent experiments are indicated.

RESULTS

ICESt1 and ICESt3 intraspecies transfer.

Genes encoding resistance to spectinomycin (spc) or chloramphenicol (cat) were introduced into loci that should not be involved in the ICE excision or transfer, i.e., the noncoding sequence flanked by IS1192 and IS981 fragments in ICESt1 and a pseudogene (Ψorf385J) in ICESt3 (Fig. 1 and Table 1).

Reconstituted skim milk and the rich media LM17, HJL, and Belliker were used during the development of a mating protocol for S. thermophilus ICEs. The mating attempts were performed using liquid cultures or nitrocellulose filters placed on agar plates. The only conditions which allowed putative transconjugants to be recovered were filter matings on skim milk or LM17 agar plates. The numbers of putative transconjugants per donor were similar with both media whether the filter matings were performed in aerobiosis or in anaerobiosis (data not shown). All of the following filter matings were performed on LM17 agar plates. When the cell-cell contacts between the donor and the recipient strains were prevented by use of a filter, no putative transconjugants were recovered in spite of satisfactory growth. Moreover, the numbers of putative transconjugants per donor were similar with or without DNase I (data not shown). Together, these data strongly suggest that the putative transconjugants were obtained following conjugation events.

S. thermophilus LMG18311/pMG36e, a strain devoid of an element integrated in the 3′ end of the fda gene and carrying the plasmid pMG36e, encoding erythromycin resistance, was used as the recipient in the intraspecies filter mating experiments. ICESt3 tagged with a chloramphenicol resistance gene (ICESt3cat) was transferred from strain S. thermophilus CNRZ385 ICESt3cat. The number of putative transconjugants per donor resistant to erythromycin and chloramphenicol is 3.4 × 10⁻⁶ ± 0.5 × 10⁻⁶. The frequencies of spontaneous mutations leading to the appearance of S. thermophilus strains resistant to erythromycin or chloramphenicol are less than 10⁻⁹ and thus do not interfere with the calculation of the transfer frequencies. Putative transconjugants were isolated and tested by PCR. The strain identity was confirmed by the amplification of the central part of the cse gene (Fig. 3A). This gene, which is specific to S. thermophilus, carries a central part that is highly variable in size and sequence between S. thermophilus strains (12; S. Layec, personal communication). Each pair of donor and recipient strains used have different cse alleles allowing their identification. These amplifications have shown that all of the putative transconjugants analyzed (>100) were derived from the recipient strain and carry ICESt3. For all of these clones, the ICE detection was confirmed by the amplification of an internal fragment of the ICESt3 regulation module (Fig. 3A). Therefore, ICESt3 is a functional S. thermophilus ICE.

FIG. 3.

Characterization by PCR of an intraspecies ICESt3 transconjugant. CNRZ385 ICESt3cat and LMG18311/pMG36e were used as the donor and the recipient, respectively. Don, Rec, and Tc indicate donor, recipient, and transconjugant, respectively. (A) ICESt3, amplification of a fragment of the regulation module. cse, amplification of the central part of the cse gene. (B and C) attR, attL, attB, and attI, amplifications of fragments carrying these attachment sites. The sizes of the PCR fragments are deduced from sequence analysis and confirmed by comigration with a DNA ladder. The classes of primer pairs used for these amplifications are indicated in parentheses, and their localizations are indicated in Fig. 1.

Fragments carrying the attL and the attR sites resulting from the ICESt3 site-specific integration into the fda 3′ region were amplified in all 50 transconjugants analyzed, showing that ICESt3 integrates in the fda 3′ end with a high specificity in S. thermophilus (Fig. 3B). Moreover, fragments carrying the attB and the attI sites resulting from the site-specific excision of ICESt3 were also detected in these transconjugants (Fig. 3C). Therefore, this element excises in the S. thermophilus transconjugants, suggesting that it is capable of transfer.

The strain S. thermophilus CNRZ368 carrying ICESt1 tagged with the cat gene was used as the donor in intraspecies filter mating. Only one putative ICESt1 transconjugant resistant to erythromycin and chloramphenicol was obtained. This clone was confirmed by PCRs as being a transconjugant, i.e. deriving from the recipient S. thermophilus LMG18311/pMG36e and carrying ICESt1 (data not shown). Moreover, the ICE was found to be integrated in the fda locus and to be still able to excise. This suggests that ICESt1 is, like ICESt3, an S. thermophilus ICE but transfers at a very low frequency (<10⁻⁹) to other S. thermophilus strains. ICESt3 and ICESt1 are the two first conjugative elements (ICEs or conjugative plasmids) identified and characterized in S. thermophilus. Moreover, ICESt1 and ICES3 integrate with a high specificity in the fda 3′ end in S. thermophilus and are still able to excise in the transconjugants.

ICESt3 interspecies transfer.

Various Firmicutes that are close to, distant from, or very distant from S. thermophilus and a gammaproteobacterium, Escherichia coli, were used as recipients in order to determine the ICESt3 host range (Table 1). The following species were used as recipients: Streptococcus pyogenes, Streptococcus agalactiae (two strains), Lactococcus lactis (two strains), Enterococcus faecalis, Enterococcus casseliflavus, Lactobacillus delbrueckii subsp. bulgaricus, Bacillus subtilis (two strains), Staphylococcus aureus, Clostridium difficile (two strains), and E. coli (strain HB101). S. thermophilus strain CNRZ385 carrying ICESt3 tagged with the spc gene was used as the donor in all the interspecies filter matings. The frequencies of spontaneous mutations leading to the appearance of donor or recipient strains resistant to the antibiotics used are less than 10⁻⁹ and thus do not interfere with the calculation of the transfer frequencies. Some of these filter matings were done in aerobiosis, but the intraspecies ICESt3 transfer frequencies resulting from intraspecies filter matings in aerobiosis and anaerobiosis were found to be similar (data not shown).

Numerous putative transconjugants of S. pyogenes ATCC 12202 Rif^r and E. faecalis JH2-2 Rif^r Fus^r were obtained (Fig. 4). Some putative transconjugants were purified and analyzed by PCR. All the primers hybridizing to the chromosomes of the recipient strains (primers belonging to the classes A, F, G, and H) hybridize to a species-specific sequence (Fig. 1; see Table S1 in the supplemental material). Therefore, PCR products obtained with these primers are also species specific. The amplification of an fda fragment specific to each species and of an internal fragment of the ICE regulation module has shown that the putative ICESt3 transconjugants of E. faecalis JH2-2 were derived from the recipient and that they carried the ICE (Fig. 1 and 4A). Similar amplifications by PCR have confirmed that transconjugants of S. pyogenes were obtained (data not shown). The hemolytic property of S. pyogenes was also used to differentiate S. pyogenes from S. thermophilus. The numbers of transconjugants per donor for S. pyogenes and E. faecalis were 3.0 × 10⁻⁶ ± 1.5 × 10⁻⁶ and 3.9 × 10⁻⁷ ± 0.9 × 10⁻⁷, respectively. Therefore, ICESt3 transfers to various Firmicutes.

FIG. 4.

Characterization of an ICESt3 transconjugant of E. faecalis by PCR. S. thermophilus CNRZ385 ICESt3spc and E. faecalis JH2-2 Rif^r Fus^r were used as the donor and the recipient, respectively. Don, Rec, and Tc indicated donor, recipient, and transconjugant, respectively. (A) E. faecalis and ICESt3, amplification of an internal fda fragment specific to E. faecalis and amplification of a fragment of the ICESt3 regulation module, respectively. (B and C) attR, attL, attB, and attI, amplifications of fragments carrying these attachment sites. The sizes of the PCR fragments are deduced from sequence analysis and confirmed by comigration with a DNA ladder. The classes of primer pairs used for these amplifications are indicated in parentheses, and their localizations are indicated in he Fig. 1.

Fragments carrying the attL and the attR sites resulting from the site-specific recombination between the fda 3′ end from E. faecalis and the ICESt3 attI site were amplified from 40 transconjugants (Fig. 1 and 4B). These sites were also amplified in 25 transconjugants of S. pyogenes ATCC 12202 (data not shown). Therefore, ICESt3 integrates in the fda 3′ end with a high specificity in these species. The attB site and the attI site resulting from the site-specific excision of ICESt3 were amplified from 40 E. faecalis transconjugants (Fig. 4C) and from 25 S. pyogenes transconjugants (data not shown), indicating that this element is still able to excise in the transconjugants of these species.

One putative ICESt3 transconjugant of L. lactis MG1363/pMG36e was also recovered. This clone was analyzed by PCR like the previous interspecies transconjugants. These amplifications confirmed that this clone is L. lactis MG1363 carrying ICESt3. This shows that ICESt3 transfers to this species at low frequency (<10⁻⁹). However, in this transconjugant, ICESt3 did not enter the fda locus. Additionally, the detection of a chromosomal attI site in the transconjugant strongly suggests that this site was not used as a substrate for the integration of the element in a putative attB′ site (data not shown).

After the incubation of S. thermophilus carrying ICESt3 and one of the S. agalactiae recipient strains on a filter, no viable S. thermophilus colonies were recovered. Therefore, the conditions allowing the coculture of these two species on a filter could not be determined.

In spite of satisfactory growth conditions during cocultures, no transconjugants were recovered after at least three filter matings using recipients from the other species (Lactobacillus delbrueckii subsp. bulgaricus, B. subtilis, Staphylococcus aureus, C. difficile, and E. coli). This suggests that ICESt3 is poorly able to establish a functional mating apparatus with these species and/or is poorly able to integrate into the genomes of these strains.

ICESt3 transfers from E. faecalis.

The ability of the newly incoming ICESt3 to transfer from a transconjugant was analyzed. E. faecalis JH2-2 carrying ICESt3spc and S. thermophilus LMG18311/pMG36e were used in filter mating as donor and recipient, respectively. Some putative transconjugants were purified and analyzed by PCR as described above. Thirteen ICESt3-containing transconjugants of S. thermophilus were obtained, at a frequency of 3.2 × 10⁻⁹ ± 0.8 × 10⁻⁹ per donor. PCR amplification of a fragment of the ICESt3 regulation module and an fda fragment showed that all 13 transconjugants were derived from the recipient and carried the ICE (data not shown). The PCR amplification of fragments harboring the specific attR and attL sites showed that ICESt3 is integrated in the fda 3′ ends of all the transconjugants. Moreover, the amplification of fragments carrying the S. thermophilus attB site and the ICESt3 attI site in the transconjugants showed the site-specific excision of the ICE in the transconjugants. Therefore, ICESt3 is still able to transfer from E. faecalis transconjugants. The amplification of the attB and the attI sites in the transconjugants of S. pyogenes and S. thermophilus strongly suggests that ICESt3 could also transfer again from these strains.

Integration and excision of recombinant plasmids harboring the ICESt1 recombination module.

All the attachment sites of ICESt1 and ICESt3 contain, in S. thermophilus, a 27-bp sequence identical to the last 20 nucleotides of the S. thermophilus fda gene and the 7 nucleotides downstream the stop codon of this gene. Only the 20 nucleotides belonging to the fda ORF are similar in members of the Firmicutes (Fig. 5). Moreover, the S. thermophilus ICEs were shown in this work to integrate overwhelmingly in the fda 3′ end. Therefore, this sequence is the putative attB site of the S. thermophilus ICEs.

FIG. 5.

Sequences of the fda 3′ ends of the strains used as recipients and integration frequencies of pNST260+ in some of these strains. (A) Cladogram of the species used as recipients in this work. (B) The stop codon of the fda ORFs is enclosed by a rectangle. The bold and underlined nucleotides indicate the differences from the S. thermophilus sequence. The sequences in this figure correspond to those found in the sequenced strains. Only one of the two Staphylococcus aureus alleles, the most common in databases, is indicated. The second Staphylococcus aureus allele is more different from the S. thermophilus sequence: the adenine at the position 8 is replaced by a thymine. The fda 3′ end sequences available in the databases are identical for all the strains of the other species. The sequence of the E. casseliflavus fda 3′ end is not available in databases. The fda 3′ end sequences from Lactobacillus delbrueckii subsp. bulgaricus and E. coli do not display a significant identity with that of S. thermophilus. (C) pNST260+ integration frequencies in five strains used as recipients in intra- and interspecies filter mating. S. thermophilus LMG18311 and B. subtilis 168 were used in this experiment. ND, not determined.

Recombinant thermosensitive plasmids were created in order to investigate the ability of the S. thermophilus ICEs to integrate in the attB sites of various Firmicutes and to excise. Different parts of the ICESt1 recombination module, i.e., the integrase gene (int), the excisionase gene (xis), and the attI site of this element, were cloned in both orientations downstream from the P6 constitutive promoter, which is active in L. lactis, E. faecalis, and B. subtilis (27), in the vector pG+host9 (Fig. 2). In some plasmid constructs, the promoter fusions are expected to drive expression of the cloned gene(s), and in others there should be no expression (recombinant plasmid names end with a + or −, respectively).

The plasmids pNST260+ and pNST260− carrying the int gene and the attI site were introduced in S. thermophilus LGM18311, L. lactis IL1403, E. faecalis JH2-2, B. subtilis 168, and Staphylococcus aureus RN4220. The integration frequencies of pNST260+ in S. thermophilus LGM18311 and L. lactis IL1403 are 9.7 × 10⁻¹ ± 0.6 × 10⁻¹ and 9.5 × 10⁻¹ ± 0.6 × 10⁻¹ (Fig. 6), respectively. Therefore, pNST260+ has a very high ability to integrate in the chromosomes of S. thermophilus and L. lactis, whereas no ICESt3 transconjugant of L. lactis integrated in the fda locus was recovered. The integration frequencies of pNST260+ in E. faecalis JH2-2, B. subtilis 168, and Staphylococcus aureus RN4220 are 5.0 × 10⁻¹ ± 0.2 × 10⁻¹, 5.5 × 10⁻² ± 1.2 × 10⁻², and 6.3 × 10⁻² ± 2.8 × 10⁻², respectively (Fig. 5). Hence, pNST260+ integrates in the E. faecalis genome with an integration efficiency approximately two times lower than that in S. thermophilus or in L. lactis, whereas it does not integrate efficiently in B. subtilis 168 and Staphylococcus aureus RN4220. The integration frequencies of pNST260− in all these strains are less than 10⁻³, strongly suggesting that the integration of pNST260+ was achieved through the expression of the int gene (data not shown).

FIG. 6.

Schematic representation of the site-specific integration of plasmid pNST260+ into the fda 3′ end and characterization of an E. faecalis transformant carrying this plasmid. (A) The locations and orientations of primers hybridizing to the chromosome and to the plasmid are indicated by black and white arrowheads, respectively. The recombination sites are magnified. The primer pairs A-AttI3, AttI2-F, A-F, and AttI3-AttI2 allow the amplification of fragments carrying the attL, the attR, the attB, and the attI sites, respectively (see Table S1 in the supplemental material). The classes of primer pairs used for these amplifications are indicated. The symbols used to represent plasmid pNST260+ are identical to those used in Fig. 3. (B) Amplifications by PCR using the DNA of E. faecalis JH2-2 and a clone of E. faecalis JH2-2 carrying plasmid pNST260+. attRç attL, and attB, amplifications of fragments carrying these attachment sites.

Four purified E. faecalis transformants carrying pNST260+ were characterized by PCR using species-specific and ICE-specific primers. The amplification of the attR and attL sites resulting from the site-specific recombination between the attI site carried by pNST260+ and the attB site was obtained in the four integrants (Fig. 6). Thus, the plasmid pNST260+ integrates by site-specific recombination in the fda 3′ end in E. faecalis. Moreover, a fragment carrying the attB site could be amplified in the E. faecalis wild-type strain but not in the integrant, showing that the plasmid cannot excise (Fig. 6). This strongly suggests that the integrase alone cannot catalyze the excision of the plasmid.

Four purified pNST260+ transformants of S. thermophilus, L. lactis, B. subtilis, and Staphylococcus aureus were also characterized by PCRs using species-specific and ICE-specific primers. Transformants of S. thermophilus and L. lactis were similar to those of E. faecalis. However, PCRs failed to detect the integration of pNST260+ in the fda locus in B. subtilis and in Staphylococcus aureus (data not shown).

The plasmids pNST261+ and pNST261−, carrying the int and the xis genes and the attI site, were introduced in S. thermophilus LGM18311. Whereas pNST261− integrants were not recovered, integrants of pNST261+ were isolated at a frequency of 4.6 × 10⁻² ± 2.2 × 10⁻². The attR and the attL sites resulting from the site-specific integration of pNST261+ into the fda 3′ end were detected by PCR (data not shown). Moreover, the amplification of the attI site of the plasmid and the chromosomal attB site from a purified pNST261+ integrant showed that this plasmid is also able to excise by site-specific recombination. Therefore, while the expression of the integrase gene is necessary and sufficient to obtain site-specific integration, the expression of the excisionase is required for site-specific excision. Previous results showed that the integrase gene is involved in the ICESt1 site-specific excision (20). Together, these data strongly suggest that the expression of the integrase and the excisionase is required for the pNST261+ site-specific excision.

The plasmids pNST262+ and pNST262−, carrying the int and the xis genes and the attR site, were also introduced in S. thermophilus LGM18311; however, no integrants were isolated (Fig. 2). Therefore, while the integrase and the excisionase are expressed in pNST262+ (data not shown), the plasmid does not integrate in the fda 3′ end. Thus, the site-specific recombination between the attB and the attR sites cannot occur, suggesting that the presence of attI is also required for the site-specific integration of a circular molecule in the attB site.

The amino acid sequences of the proteins encoded by the int and the xis genes from ICESt1 and ICESt3 exhibit a high degree of identity of over 99% of their length. The 27-bp sequences of the attI sites from the S. thermophilus ICEs into which the site-specific recombinations occurred are identical, whereas the nucleotide sequences carrying these sites (285 bp) show more than 98% identity. These data strongly suggest that the site-specific recombinations of ICESt1 and ICESt3 are nearly identical.

ICESt3 transfer to recipients carrying pNST260+.

S. thermophilus LGM18311, L. lactis IL1403, and L. lactis MG1363 harboring pNST260+ integrated in the fda locus were used as recipients in filter mating in order to determine if the overexpression of the int gene in these strains could have an impact on the ICESt3 transfer frequency. The recovering of transconjugants was confirmed by PCRs as described above (data not shown). The number of ICESt3 transconjugants of LGM18311/pNST260+ per donor, 3.8 × 10⁻⁶ ± 1.5 × 10⁻⁶, is similar to that obtained with LGM18311/pMG36e (3.4 × 10⁻⁶ ± 0.5 × 10⁻⁶); therefore, the overexpression of the int gene in the recipient strain LGM18311 does not have an obvious impact on ICESt3 transfer frequency. No ICESt3 transconjugants of L. lactis IL1403 or L. lactis MG1363 carrying pNST260+ was obtained. Hence, the overexpression of the int gene in this species cannot complement a hypothetical incoming ICESt3 that could not express the ICE integrase.

Derepression of the conjugative transfer.

Previous results have shown that MC, a DNA-damaging agent, derepresses the excision of ICESt1 and ICESt3 at least 10-fold (4). The MC concentration providing the highest derepression factor was equal to half of the MIC. S. thermophilus strains harboring ICESt1 or ICESt3 treated with MC and S. thermophilus LGM18311/pMG36e were used in filter matings as donor and recipient, respectively. The donor cells were treated with various concentrations of MC. The purified transconjugants were characterized by PCR as described above (data not shown). The transfer frequencies are expressed as the number of transconjugants per recipient instead of the number of transconjugants per donor because of the mortality of the donor caused by MC (2, 3). When the donor cells were not treated with MC, the number of ICESt3 transconjugants per recipient was 3.8 × 10⁻⁷ ± 1.3 × 10⁻⁷. The numbers of transconjugants per recipient were 1.0 × 10⁻⁵ ± 0.4 × 10⁻⁵ and 7.6 × 10⁻⁶ ± 2.6 × 10⁻⁶ after treatment with MC concentrations equal to one-fourth and one-half of the MIC, respectively. Therefore, MC treatment of cells carrying ICESt3 leads to a derepression of the conjugative transfer of this ICE by almost 25-fold. Whereas MC derepresses the excision of ICESt1 and ICESt3, no ICESt1 transconjugants of LMG18311 were characterized after MC treatment of the donor cells carrying this element, suggesting that MC does not have a detectable impact on ICESt1 transfer frequency.

DISCUSSION

Many ICESt3 transconjugants of S. thermophilus, S. pyogenes, and E. faecalis and one ICESt1 transconjugant of S. thermophilus were isolated and characterized in this work. Furthermore, the presence of DNase does not alter the transfer frequency, and ICESt3 transfer is not observed if the donor and recipient strains are separated by a filter. Therefore, ICESt3 is a functional conjugative element, the first one proven and characterized in S. thermophilus. Moreover, its retransfer from E. faecalis to S. thermophilus has shown that this element is able to transfer from different species. The observation of a single ICESt1 transfer event can leave doubt as to the functionality of this ICE and its ability to transfer in the environment.

The late competence genes were found to be functional in some strains of S. thermophilus, S. pyogenes, and E. faecalis, and sequence data suggest that intergenomic recombinations have occurred. However, natural transformation was nevertheless never reported for these species (6, 9, 10, 14, 41). Furthermore, the chromosomal sequences near the attL site from the S. thermophilus ICEs, i.e., the sequence downstream from the fda gene in an empty attB site, are completely different in the various species used as recipient or donor. Therefore, the acquisition of an S. thermophilus ICE by natural transformation is very unlikely in interspecies matings, since no RecA-dependent recombination could occur in the chromosomal sequence near attL. During intraspecies matings, the donor and the recipient are two S. thermophilus strains. Therefore, the integration by homologous recombination of a linear DNA fragment carrying an ICE could happen if this transforming fragment is internalized in the recipient. Nevertheless, no ICE transfer was recovered in liquid cultures, which are the conditions used for the artificial induction of S. thermophilus competence (9). Therefore, the acquisition of an S. thermophilus ICE by natural transformation is also very unlikely in intraspecies matings.

The nucleotide sequences of the recombination and the conjugation modules of ICESt1 and ICESt3 are almost identical (95%) (54). Moreover, in spite of the high identity between the recombination modules of the S. thermophilus ICEs, the excision frequencies of ICESt1 and ICESt3 are very different: 1 × 10⁻⁶ and 0.9 × 10⁻³, respectively (54). The regulation modules are more distantly related than the recombination and the conjugation modules. They shared three genes whose nucleotide sequences are 82 to 85% identical and which encode two putative regulatory pathways (5, 54). These pathways involved two ORFs named arp1 and arp2, belonging to the cI and the cI-like repressor families, respectively. However, numerous particularities such as specific putative regulatory genes or differences in the sequences of the promoters and ribosome binding sites have been previously described in the regulation modules (5). We have proposed that the differences between the regulation modules of ICESt1 and ICESt3 could be responsible for the dissimilar excision frequencies of these elements. Together, these data suggest that these differences could also be involved in the very distinct transfer frequencies of the S. thermophilus ICEs.

In this work, the derepression by MC of the ICESt3 conjugative transfer by a factor of 25 was demonstrated. Previous results have shown that MC treatment of cells harboring ICESt1 or ICESt3 led to a derepression of the excision of these elements by at least a factor of 10 (4). Thus, although the excision frequencies of ICESt1 and ICESt3 are very different and these elements seem to be differently regulated, MC treatments cause similar derepression of their excision frequencies. Nevertheless, no ICESt1 transconjugant was recovered after such treatments.

While the role of the arp1 gene, encoding a cI homolog, has been previously determined in ICESt1, the role of the arp2 gene, encoding a cI-like homolog, has not yet been investigated (4). Furthermore, some recent results strongly suggest that the arp1 gene is also functional in ICESt3 (unpublished results). The role of the arp2 gene encoding a cI-like homolog is also undetermined in this ICE. To our knowledge, the S. thermophilus ICEs are the only integrative elements (i.e., phages, ICEs, and related elements) carrying both a cI repressor and a cI-like repressor. Although these two repressor families are unrelated, they are both inactivated in response to DNA damage, which leads to a derepression of the target genes of the repressors (1, 13, 29, 45). MC-induced DNA damage is also known to derepress the conjugative transfer of others ICEs, such as SXT from Vibrio cholerae and ICEBs1 from B. subtilis, which carry a cI homolog and a cI-like homolog, respectively (2, 3).

ICESt3 from S. thermophilus was characterized in this work as able to transfer to the other Firmicutes S. pyogenes and E. faecalis, whereas no ICESt3 transconjugants of S. agalactiae, E. casseliflavus, Lactobacillus delbrueckii subsp. bulgaricus, B. subtilis, Staphylococcus aureus, C. difficile, and E. coli were recovered. If optimal coculture conditions of the donor and the recipient strains were used, some traits specific to one of these strains were able to interfere with the ability of ICESt3 to transfer. For example, the coculture of S. thermophilus and each of the two S. agalactiae recipient strains led to the death of the donor cells. This impossible coculture could be due to the synthesis by these S. agalactiae recipients of a compound such as a bacteriocin that is active against S. thermophilus.

Many studies with att site mutations have demonstrated that efficient recombination occurs only when the two attachment sites have very closed direct repeats. The plasmid pNST260+ was created in order to investigate the ability of the integrase to catalyze site-specific integrations into different fda 3′ ends, i.e., the putative ICESt3 attB site of the recipient strains. This plasmid carrying the int gene under the control of the P6 promoter and the attI site integrates efficiently in the fda 3′ ends from S. thermophilus, L. lactis, and E. faecalis but does not in B. subtilis and Staphylococcus aureus. The activity of the P6 promoter carried by the recombinant plasmids was shown in B. subtilis (27). Therefore, this strongly suggests that the integrase is expressed from pNST260+ but cannot catalyze the site-specific integration of the plasmid in this species. B. subtilis and Staphylococcus aureus are related species, and therefore we hypothesize that pNST260+ should have behaved the same in these two species. The integration range of plasmid pNST260+ and the ICESt3 host range are similar except for L. lactis (Fig. 5). Together these data suggest that although the promoters controlling the expression of the integrase in the plasmid and the ICE are not the same, a defective ICESt3 integration similar to that observed for pNST260+ would be a limiting step of their transfer and another part of the explanation of their host range.

L. lactis is a species close to S. thermophilus and carries a putative attB site identical or almost identical to the attB site of S. thermophilus (Fig. 5). ICESt3 transfers easily to E. faecalis, a species more distant from S. thermophilus than L. lactis and harboring the same putative attB site as L. lactis IL1403. Thus, the recovery of numerous ICESt3 transconjugants of L. lactis integrated in the fda 3′ end was expected. Surprisingly, only one transconjugant of L. lactis was isolated and characterized. Analyses by PCR have confirmed that this clone derives from the recipient strain and that it carries the ICE, i.e., that this clone is an ICESt3 transconjugant. Nonetheless, the amplification of the attL and the attR sites resulting from the site-specific integration of the ICE in the fda 3′ end of the recipient cell was not obtained. Together these data suggest that ICESt3, which cannot replicate itself and does not share any sequence identity with the L. lactis genome, was transferred and integrated in the chromosome of this transconjugant by a mechanism of illegitimate recombination. This integration in a locus other than fda could result from a defective integration process. Nevertheless, pNST260+, which overexpresses the int gene, integrates with a high efficiency in L. lactis. Differences in the integrase transcription from the P6 promoter of pNST260+ and from the natural integrase promoter in ICESt3 could lead to this defective integration. However, no ICESt3 transconjugant of L. lactis was recovered when a strain carrying pNST260+ was used as the recipient. Thus, pNST260+ cannot complement a hypothetical incoming ICESt3 that could not express the ICE integrase. This suggests that ICESt3 is poorly able to establish a functional mating apparatus with L. lactis.

The lactic acid bacterium S. thermophilus is extensively used as a starter in the manufacture of dairy products with other lactic acid bacteria, such as Lactococcus lactis. Sequence comparisons and hybridizations have revealed horizontal gene transfers between a large array of lactic acid bacteria. The quasi-identities of the IS905 sequence of L. lactis (28) and the IS1191 sequence of S. thermophilus and the distribution of these ISs suggested that IS905/IS1191 were horizontally transferred from S. thermophilus to L. lactis (35, 36). In the same way, horizontal transfers of IS981 and ISS1 from L. lactis to S. thermophilus have recently occurred, probably in cocultures used for cheese manufacture (15, 35, 36). The sequencing of S. thermophilus genomes has shown that this species also acquired exogenic DNA from Lactobacillus delbrueckii subsp. bulgaricus and has suggested DNA exchanges with commensal streptococci (10). However, conjugative plasmids that might be involved in these transfers were never identified in S. thermophilus.

The S. thermophilus ICEs and the related genomic islands contain many examples of genetic exchange. They contain different types of enterococcal and lactococcal ISs and IS fragments (21, 54, 75). The genomic island ΔCIME308, which is related to ICESt1 and ICESt3, carries a 10-kb sequence almost identical to the sequence of the plasmid pAH82 from L. lactis. Moreover, ICESt1 and ICESt3 are closely related to the putative ICE RD2 of S. pyogenes (34), to the putative ICE SmuE of Streptococcus mutans (18), and to six putative ICEs and genomic islands from S. agalactiae (16). Finally, one of the genomic islands related to the S. thermophilus ICEs carries a cadmium resistance module closely related to plasmids from L. lactis, Listeria innocua, Staphylococcus aureus, Staphylococcus saprophyticus; to the putative ICEs ICELm1 from Listeria monocytogenes, ICE_2603_rpiL from S. agalactiae, and ICESde3396 from S. dysgalactiae subsp. equisimilis; and to SCCmec genomic islands from staphylococci (15, 50; unpublished data).

The presence of closely related genes in the S. thermophilus ICEs and the related genomic islands in various species strongly suggests that all these elements transfer in the environment and that they evolve by cointegration, deletion, and module exchanges. The genetic mixing between ICEs and other elements seems to be an important mechanism allowing the spread of exchanged genes into different ecological niches.