Mobilization of IMEs Integrated in the oriT of ICEs Involves Their Own Relaxase Belonging to the Rep-Trans Family of Proteins

Integrative mobilizable elements (IMEs) are widespread but very poorly studied integrated elements that can excise and hijack the transfer apparatus of co-resident conjugative elements to promote their own spreading. Sixty-four putative IMEs, harboring closely related mobilization and recombination modules, were found in 14 Streptococcus species and in Staphylococcus aureus. Fifty-three are integrated into the origin of transfer (oriT) of a host integrative conjugative element (ICE), encoding a MobT relaxase and belonging to three distant families: ICESt3, Tn916, and ICE6013. The others are integrated into an unrelated IME or in chromosomal sites. After labeling by an antibiotic resistance gene, the conjugative transfer of one of these IMEs (named IME_oriTs) and its host ICE was measured. Although the IME is integrated in an ICE, it does not transfer as a part of the host ICE (no cis-mobilization). The IME excises and transfers separately from the ICE (without impacting its transfer rate) using its own relaxase, distantly related to all known MobT relaxases, and integrates in the oriT of the ICE after transfer. Overall, IME_oriTs use MobT-encoding ICEs both as hosts and as helpers for conjugative transfer. As half of them carry lsa(C), they actively participate in the dissemination of lincosamide–streptogramin A–pleuromutilin resistance among Firmicutes.


Introduction
Integrative conjugative elements (ICEs) are mobile genetic elements (MGEs) integrated in bacterial chromosomes (or plasmids) that are able to excise as a circular form and transfer autonomously by conjugation (see [1,2] for reviews). These elements are widespread in bacterial genomes, even more than conjugative plasmids in almost all bacterial clades [3]. Therefore, they probably play a key role in horizontal transfer of genes, in particular, in virulence and antibiotic resistance spread [4,5]. ICE transfer is thought to be similar to transfer mediated by conjugative plasmids. The mechanistic aspects of conjugative transfer of plasmids are quite well characterized in Gram-negative bacteria and, more particularly, Proteobacteria, but conjugative transfer of ICE is less well known. Initiation of conjugative transfer requires nicking of DNA at a sequence called origin of transfer (oriT) by a protein called relaxase (or Mob for mobilization protein). The relaxase binds covalently to DNA and, by interacting with a so-called coupling protein (CP), brings one of the strands of DNA to an active transport system encoded by the conjugative element. This system of mating pair formation a tRNALys gene). Transfer of the ICE hosting IME_oriT (specifically integrated in rpsI) in the donor strain was not observed [17]. This raises interesting evolutionary questions about the interactions between IMEs and their helper and/or host conjugative element. If IME integration disrupts an essential sequence or gene of the ICE, does it impair ICE transfer or does the IME excise when the transfer system of the ICE is activated (as previously suggested by Coluzzi et al. [13])? Is the IME only mobilized in trans, thus enabling its co-transfer with the ICE? Does the use of the ICE T4SS by the IME interfere with ICE transfer? Overall, is this IME a harmful pirate or a parasite of the ICE that hampers ICE transfer or a gentle hitchhiker that pacifically co-exists with the ICE?
In this study, an exhaustive search of this family of IMEs that target oriT of ICEs revealed 64 elements, not only in many species of streptococci but also in one strain of Staphylococcus aureus.
In order to answer the questions raised above, we studied excision and transfer by conjugation of an IME_oriT. For these purposes, we made genetic constructs to create adequate donor and recipient strains that were then used in various conjugation experiments. These mating assays demonstrated that this IME_oriT can be mobilized in trans but not in cis by the ICE, and does not interfere with ICE transfer.

In Silico Search and Analysis of IMEs Encoding a Relaxase or an Integrase Related to the One of IME_SagLMG15084_oriT
The relaxase (GenBank ID: EPW57073.1) and integrase (GenBank ID: EPW57075.1) of IME_SagLMG15084_oriT described previously [15] were used as queries to find homologs by BlastP and tBlastN analyses (with default parameters on non-redundant protein sequences and nucleotide databases, respectively; last NCBI database interrogation on 2020/02/11). Hits with more than 40% of amino acid identity with the whole query sequence were further analyzed. Sequences adjacent to the identified genes were extracted from genomes or contigs and then manually explored using Geneious Prime (version 2020.0.5, Biomatters Ltd., Auckland, New Zealand to identify the boundaries of the elements (search of direct repeats). IME_oriTs were also identified using a previously described method [13] on 21 genomes of Streptococcus salivarius sequenced in our laboratory (see more details in the accompanying paper [14]) and were added to the set of IMEs. Denominations of the IMEs include letters for the species (Sag for S. agalactiae, San for Streptococcus anginosus, Saur for S. aureus, Saus for Streptococcus australis, Sco for Streptococcus constellatus, Smi for S. mitis, Smu for S. mutans, Sora for Streptococcus oralis, Spn for Streptococcus pneumoniae, Spse for Streptococcus pseudopneumoniae, Ssal for S. salivarius, Ssan for Streptococcus sanguinis, Ssp for unspecified Streptococcus species, and Sth for Streptococcus thermophilus) and the name of the bacterial strain that hosts the IME. Denominations also include the name of the target sequence (oriT or oriT* if the IME is integrated in a secondary site). A multiple alignment of the IME sequences was made using Clustal Omega 1.2.2 in order to construct a full distance matrix. This enabled the selection of representative sequences for further analyses. Based on the distance matrix obtained for IME sequences, sequences were aligned two-by-two using BlastN (with default parameters). These comparison files were then used to draw alignment graphs using Artemis Comparison Tool (ACT) provided by the Sanger Centre [18]. Manual editing of the comparison figure was performed using Inkscape.
Cargo genes of the putative IMEs were analyzed using the Conserved Domain search tool on NCBI website (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi).

Construction of Phylogenetic Trees
Sequences of the integrase genes, relaxase proteins and Lsa(C) proteins encoded by the putative IMEs were extracted using Geneious Prime (version 2020.0.5, Biomatters Ltd.) and aligned using ClustalW. Phylogenetic trees were built with MEGA7 [19] using the maximum likelihood method based on the Tamura-Nei model [20] without outgroup. Initial tree(s) for the heuristic search were obtained automatically by applying neighbor-joining and BioNJ algorithms to a matrix of pairwise distances estimated using the maximum composite likelihood approach and then selecting the topology with superior log likelihood value.

Bacterial Strains and Growth Conditions
The bacterial strains used in this study are indicated in Table 1. Strain deficient in exopolysaccharide production, Ery R and carrying ICESt3 Spc R with its integrase gene replaced by a Cm R gene This work 1 As described in the material and methods, Integrative Conjugative Elements (ICEs) and Integrative Mobilizable Elements (IMEs) are named according to the bacterial species and strain in which they were identified and their site of insertion. Deletion of genes introduced into the element are also mentioned (int for integrase, rel for relaxase, and espE for exopolysaccharides production locus). For more clarity, antibiotic resistance genes used to label the elements are not included in the element name but are indicated in the genotype column; Spc R , spectinomycin resistance; Kan R , kanamycin resistance; Cm R , chloramphenicol resistance; Ery R , erythromycin resistance.

IME and ICE Tagging with a Resistance Gene and Deletion of Integrase and Relaxase Genes of the Integrative Elements
Overlap PCR was used to insert an antibiotic resistance gene in integrative elements of S. salivarius L25 as described previously [21]. Deletion of the integrase genes (of ICE SsalF1-4_fda in S. salivarius F1-4 and ICESt3 of S. thermophilus in LMG18311) was done by insertion of a chloramphenicol resistance cassette as described previously [24]. IME_SsalL25_oriT and 1-kb-upstream and downstream flanking regions were PCR-amplified before natural transformation of S. thermophilus LMG18311(ICESt3) as described by Gardan et al. [25]. The deletion strategy, used to construct IME_SsalL25_oriT∆rel to avoid introducing an antibiotic resistance gene and polar effect, was similar to the strategy used previously [23]. Sequence of the primers used is indicated in Table S1.

Excision Tests and Mating Experiments
Excision tests were done as described previously [21]. Sequence of the primers used is indicated in Table S1. Relevant amplicons, attI and attB were sequenced (Eurogentec, Angers, France).
Mating experiments using S. salivarius and S. thermophilus strains were made as described previously [21,24]. All the experiments done are listed in Table S3. At least three technical replicates of three independent biological replicates were done. Data are expressed as means ± SD. Statistical analysis was performed using the paired Students t-test. A p-value < 0.05 was considered as significant.

Distribution of IME_oriTs in Firmicutes
All the IMEs integrated in an oriT of an ICE that were previously described [15,17] encode very closely related putative relaxases and integrases (92 to 100% of amino acid identity). We chose to use the sequence of the putative relaxase and the integrase of IME_SagLMG15084_oriT to search for homologs in bacterial genomes available on the NCBI website and in 21 genomes of S. salivarius sequenced in our laboratory [14]. Sixty-nine proteins very closely related (91 to 100% of amino acid identity) to the relaxase of IME_SagLMG15084_oriT were found in 14 different Streptococcus species and in one strain of S. aureus. Analysis of these putative relaxases indicate that they all exhibit a PF2486 domain characteristic of Rep_trans_proteins. This family includes the MobT relaxases of ICEs belonging to the Tn916 superfamily (Tn916, ICEBs1, ICESt3) and the RCR initiators involved in the maintenance of many small plasmids of Firmicutes (for example, plasmids of the pT181 family like RepC of pT181 and RepD of pC221 [26]). Even though the three groups of Rep-trans proteins are only very distantly related (less than 20% of amino acid identity except for RepSTK1, see the phylogenetic tree presented in Supplementary Figure S3), sequence alignment indicate that they all display the five previously described characteristic motifs that include three acidic residues involved in cation coordination and a catalytic tyrosine [23,26]. Sequences adjacent to the putative relaxase and integrase genes were analyzed in order to identify and delineate putative IMEs. All the relaxase genes were found associated with direct repeats and an integrase gene, two characteristics of integrative elements. In five of these elements (elements found in two strains of S. thermophilus, in two strains of S. anginosus, and in one strain of Streptococcus equinus), the integrase gene is truncated (pseudogene). Thus, these elements were considered as defective and were not included in Table S2 that synthetizes results of the in silico search. Among the 64 detected putative IMEs, several almost identical elements were found in various species (Table S2): (i) IME_SagGB00984_oriT in 10 other strains of S. agalactiae, one strain of Streptococcus cristatus (98% of nucleotide identity) and one strain of S. mitis (99% of nucleotide identity); (ii) IME_SagUCN70 carrying lsa(C) found in another strain of S. agalactiae, four strains of S. salivarius, four strains of Streptococcus parasanguinis, two strains of S. pseudopneumoniae, and two strains of Streptococcus sp. (98 to 99% of nucleotide identity), and (iii) IME_SaurER01570.3_oriT carrying lsa(C) found in one strain of S. aureus, one strain of S. oralis, and two strains of Streptococcus sp. (100% of nucleotide identity).

Chromosomal Integration Sites of IME_oriTs
In most of the cases (n = 53), integration of the IME occurred upstream the gene encoding the MobT relaxase: in Tn916-related ICEs (16 ICEs closely related to Tn916 that integrate in low G+C regions, one Tn5801 integrated in the 3 end of guaA), in ICESt3-related ICEs (15 integrated in the 3 end of a tRNALys gene, six in the 3 end of fda, 7 in the 3 end of rpsI, four in the 5 end of a ebfC and three in the 3 end of rpmG), and in one ICE related to ICE6013 (that integrates at many different sites [27]) in S. aureus (Table S2). For ICEs of the Tn916 superfamily, this region includes the origin of transfer (oriT) of the ICE [23,28]. As described previously [15], the IME more precisely integrates in a stem-loop sequence that likely includes the nicking site of the relaxase (by homology with the nic sequence demonstrated for ICEBs1 by Lee and Grossman [29]). Interestingly, an IME_oriT was also detected inside an unrelated IME (integrated in rpmG) between the gene encoding a putative relaxase (that belongs to the Rep_trans family, is devoid of HTH domain and is distantly related to the known MobT relaxases) and the gene encoding its putative TcpA CP (Table S2). In addition, eight of these putative IMEs were found integrated elsewhere in the chromosome (not inside another MGE). The integrases of these eight IMEs do not cluster together on a phylogenetic tree (see sequences indicated by a star in Figure S1). In the same way, the integrases of IMEs targeting different subfamilies of ICEs (Tn916-related elements, ICESt3 elements or ICE6013) or different integration sites (in particular those targeting fda, rpsI, or rpmG) do not group together on the phylogenetic tree ( Figure S1). Elements with identical integrases were even found integrated in different subfamilies of ICEs.

Cargo Genes of IME_oriTs
In addition to the relaxase and recombination genes, required for the mobility of the IME and a regulator gene (found in all IME_oriTs, except two), all the IME_oriTs detected by the in silico analysis carry diverse cargo genes (Table S2 and Figure 1). The sole exception is IME_Smu11VS1_oriT that contains only the three mobility genes ( Figure 1). This diversity of cargo genes has an impact on the length of the IMEs that expands from 2.7 to 10.9 kb (IME_Smu11VS1 and IME_SspA12_oriT, respectively, Figure 1).
Most of these cargo genes encode putative hypothetical proteins but some of them encode a protein with a specific domain that could give information about the possible function of the protein (Table S2).
Several encode proteins that could confer resistance against antibiotics, bacteriocins, or heavy metals (Table S2). In particular, half (n = 34) of the detected IMEs carry a lsa(C) resistance gene that is known to confer cross-resistance to lincosamides, streptogramin A, and pleuromutilins in S. agalactiae [16]. Douarre et al. [17] described in S. agalactiae a variant of Lsa(C) that shares 93% of amino acid identity with Lsa(C) found initially in S. agalactiae strain UCN70 [16]. This variant was found in 11 strains of S. agalactiae (including S. agalactiae GB00984, Figure 1), in one strain of S. mitis, and in one strain of S. cristatus (Table S2). These IMEs are integrated in different sites (inside a Tn916-related element, inside an ICESt3 element integrated in a tRNALys gene, and in the other locations indicated as oriT*). A second variant of Lsa(C) showing 96% of amino acid identity with Lsa(C) of S. agalactiae strain UCN70 was discovered in two strains of S. pseudopneumoniae (Table S2). In addition, Streptococcus sp. AM28-20 could encode an ABC-type multidrug transport system. Strain A12 also likely encodes an ABC transporter that could confer protection against lantibiotics (proteins with TIGR03732, TIGR03733, and TIGR03740 domain, respectively, Table S2). Lastly, a heavy metal resistance gene (Hg ++ resistance operon) was found on the IMEs hosted by S. agalactiae LMG15084, 18RS21, and NCTC11079.
In silico analysis indicated that IME_oriTs also convey genes that are not involved in resistance: a putative alkylation repair protein encoded by an IME found in S. anginosus (strain NCTC11064), a putative regulator (N terminal domain of TfoX competence regulator in 6 strains of S. agalactiae, S. salivarius L11, and S. cristatus BCA6), a two-component system (in strain A12), and a protein with a domain found in chromosomal segregation proteins (in strains A12 and S. salivarius 39-07) ( Figure 1 and Table S2).  Table S2 for more details and the name of the strain is indicated in the name of the IME. When the IME is integrated in the origin of transfer of a mobile genetic element (MGE), the targeted ICE/IME is shown in brackets. oriT* is used for IMEs integrated elsewhere in the chromosome and NA when information is not available. Direct repeats (DRs) delimiting IMEs are shown as black bars. Coding DNA sequences (CDSs) appear as arrows (truncated genes are indicated by Δ). The recombination genes (encoding the integrase (int) and the recombination directionality factor (RDF) also known as excisionase (xis)) appear in red, the relaxase gene (rel) in blue, and the regulation gene (reg) in green. Genes encoding a hypothetical protein with unknown function appear in white. Cargo genes encoding proteins with a putative function inferred from an in silico analysis are indicated in color: in purple for a putative alkylation repair protein, in yellow for resistance operon/genes, in light green for a putative regulator (with a TfoX_N domain), in brown for a two-component system, in orange for ABC transporters, and in pink for a putative chromosome segregation protein (SMC) and its adjacent gene (encoding a protein with a DUF4297 Figure 1. Diversity of the IMEs carrying a relaxase related to the one of IME_SagLMG15084_oriT. IMEs exhibiting more than 85% of nucleic acid identity on the whole sequence with another one are not indicated on the figure (see Table S2 for more details Ssp for unspecified Streptococcus species, and Sth for S. thermophilus) and the name of the strain is indicated in the name of the IME. When the IME is integrated in the origin of transfer of a mobile genetic element (MGE), the targeted ICE/IME is shown in brackets. oriT* is used for IMEs integrated elsewhere in the chromosome and NA when information is not available. Direct repeats (DRs) delimiting IMEs are shown as black bars. Coding DNA sequences (CDSs) appear as arrows (truncated genes are indicated by ∆). The recombination genes (encoding the integrase (int) and the recombination directionality factor (RDF) also known as excisionase (xis)) appear in red, the relaxase gene (rel) in blue, and the regulation gene (reg) in green. Genes encoding a hypothetical protein with unknown function appear in white. Cargo genes encoding proteins with a putative function inferred from an in silico analysis are indicated in color: in purple for a putative alkylation repair protein, in yellow for resistance operon/genes, in light green for a putative regulator (with a TfoX_N domain), in brown for a two-component system, in orange for ABC transporters, and in pink for a putative chromosome segregation protein (SMC) and its adjacent gene (encoding a protein with a DUF4297 domain). A nucleic acid sequence identity between sequences higher than 75% is indicated in light grey and in dark grey when higher than 90%.

IME_oriTs Identified in S. salivarius Strains L11, L25 and L45 Are Able to Excise from the Chromosome
In both L11 and L25 S. salivarius strains, an IME_oriT is integrated into oriT of an ICE of the ICESt3 family, which is integrated in the 3 end of fda. In S. salivarius L45, another IME_oriT is integrated in oriT of an ICESt3-related ICE integrated in the 3 end of rpsI. Excision of these three IMEs and of their host ICEs was tested by PCR for wild type strains. All the six integrative elements (IMEs and ICEs) were found to excise (Figure 2, data shown for elements of strain L25).
In silico analysis indicated that IME_oriTs also convey genes that are not involved in resistance: a putative alkylation repair protein encoded by an IME found in S. anginosus (strain NCTC11064), a putative regulator (N terminal domain of TfoX competence regulator in 6 strains of S. agalactiae, S. salivarius L11, and S. cristatus BCA6), a two-component system (in strain A12), and a protein with a domain found in chromosomal segregation proteins (in strains A12 and S. salivarius 39-07) ( Figure  1 and Table S2).

IME_oriTs Identified in S. salivarius Strains L11, L25 and L45 Are Able to Excise from the Chromosome
In both L11 and L25 S. salivarius strains, an IME_oriT is integrated into oriT of an ICE of the ICESt3 family, which is integrated in the 3' end of fda. In S. salivarius L45, another IME_oriT is integrated in oriT of an ICESt3-related ICE integrated in the 3' end of rpsI. Excision of these three IMEs and of their host ICEs was tested by PCR for wild type strains. All the six integrative elements (IMEs and ICEs) were found to excise (Figure 2, data shown for elements of strain L25).  Table S1. 3.3. IME_oriT of S. salivarius Strain L25 (IME_SsalL25_oriT) Can Be Transferred by Conjugation to Another S. salivarius Strain and to S. thermophilus IME_SsalL25_oriT and the ICE hosting this IME (ICE_SsalL25_fda) were labeled by kanamycin and spectinomycin resistance genes, respectively, to study their transfer by conjugation. The recipient strain used in the mating pair was S. salivarius strain F1-4 that hosts an ICE of the ICESt3 family integrated in fda. To immobilize this ICE in the recipient strain, the integrase gene of this ICE was deleted and replaced by a chloramphenicol resistance gene (ICE_SsalF1-4_fda∆int, Table 1). No excision of this ICE was detected (data not shown). Transfer was observed for both ICE_SsalL25_fda and IME_SsalL25_oriT (Figure 3). When transconjugants (TC) harbor the three elements, the sequences of oriTs of the two ICEs are too similar to determine whether the IME_SsalL25_oriT has a preferred insertion site. Some of these transconjugants carried only IME_SsalL25_oriT and ICE_SsalF1-4_fda∆int (Figure 3). This does not exclude a co-transfer of ICE_SsalL25_fda in these transconjugants since the helper ICE could have been lost in the recipient strain after transfer due to the occupation of the fda site by ICE_SsalF1-4_fda∆int in the recipient strain. and IME_SsalL25_oriT (Figure 3). When transconjugants (TC) harbor the three elements, the sequences of oriTs of the two ICEs are too similar to determine whether the IME_SsalL25_oriT has a preferred insertion site. Some of these transconjugants carried only IME_SsalL25_oriT and ICE_SsalF1-4_fdaΔint (Figure 3). This does not exclude a co-transfer of ICE_SsalL25_fda in these transconjugants since the helper ICE could have been lost in the recipient strain after transfer due to the occupation of the fda site by ICE_SsalF1-4_fdaΔint in the recipient strain. Only a few transconjugants were obtained using a S. salivarius strain as recipient. This is in agreement with previous results indicating poor conjugative transfer in the S. salivarius species in laboratory conditions [21]. IME_SsalL25_oriT was thus transferred (by transformation) to a S. thermophilus strain (LMG18311) that proved proficient for ICE transfer [22,30] and carries ICESt3 (labeled by a chloramphenicol resistance gene) [22]. ICE and IME present in this LMG18311 strain are able to excise ( Figure S2). Further conjugation experiments were performed using this strain (LMG18311(ICESt3::IME_SsalL25_oriT)) as donor in the mating pairs. To increase the transfer frequency, we used S. thermophilus LMG18311ΔepsE mutant, as recipient strain [24]. Using the mating pair LMG18311(ICESt3::IME_SsalL25_oriT) (donor)/LMG18311ΔepsE (recipient), the frequencies of transfer of the ICE and of the IME were 3.3 × 10 −4 ± 1.3 × 10 −4 and 3.1 × 10 −6 ± 1.5 × 10 −6 TC/donor cell, respectively (Table S3). IME_SsalL25_oriT transfer was thus observed at a 100-fold lower frequency than ICE transfer. When the recipient strain already carries an ICESt3 element (that offers an integration site for the IME_oriT) the frequency of TC obtention was 3-fold higher (9.3 × 10 −6 ± 6.5 × 10 −6 TC/donor cell) (Table S3). . PCR analysis of four transconjugants (TC1 to TC4) obtained using the mating pair S. salivarius L25(ICE_SsalL25_fda::IME_SsalL25_oriT) (donor D)/S. salivarius F1-4(ICE_SsalF1-4_fda∆int) (recipient R). Specific regions were amplified to discriminate the elements attR for ICE_SsalF1-4_fda∆int and IME_SsalL25_oriT and the region of ICE_SsalL25_fda that includes accessory genes (different from the one of ICE_SsalF1-4_fda), indicated as var in the figure. Resistance phenotypes of the transconjugants are indicated above the wells: Cm R , chloramphenicol resistance (conferred by ICE_ SsalF1-4_fda∆int); Kan R , kanamycin resistance (conferred by IME_SsalL35_oriT); and Spc R , spectinomycin resistance (conferred by ICE_SsalL25_fda).
Only a few transconjugants were obtained using a S. salivarius strain as recipient. This is in agreement with previous results indicating poor conjugative transfer in the S. salivarius species in laboratory conditions [21]. IME_SsalL25_oriT was thus transferred (by transformation) to a S. thermophilus strain (LMG18311) that proved proficient for ICE transfer [22,30] and carries ICESt3 (labeled by a chloramphenicol resistance gene) [22]. ICE and IME present in this LMG18311 strain are able to excise ( Figure S2). Further conjugation experiments were performed using this strain (LMG18311(ICESt3::IME_SsalL25_oriT)) as donor in the mating pairs. To increase the transfer frequency, we used S. thermophilus LMG18311∆epsE mutant, as recipient strain [24]. Using the mating pair LMG18311(ICESt3::IME_SsalL25_oriT) (donor)/LMG18311∆epsE (recipient), the frequencies of transfer of the ICE and of the IME were 3.3 × 10 −4 ± 1.3 × 10 −4 and 3.1 × 10 −6 ± 1.5 × 10 −6 TC/donor cell, respectively (Table S3). IME_SsalL25_oriT transfer was thus observed at a 100-fold lower frequency than ICE transfer. When the recipient strain already carries an ICESt3 element (that offers an integration site for the IME_oriT) the frequency of TC obtention was 3-fold higher (9.3 × 10 −6 ± 6.5 × 10 −6 TC/donor cell) (Table S3).

IME_SsalL25_oriT Can Be Mobilized in Trans
To check if IME_SsalL25_oriT can be mobilized in trans, other strains were constructed. The integrase gene of the ICE hosted by strain LMG18311(ICESt3) was deleted, resulting in LMG18311(ICESt3∆int). Excision and conjugative transfer of this element were not detected. IME_SsalL25_oriT was transferred (by conjugation) in this strain to obtain LMG18311(ICESt3∆int:: IME_SsalL25_oriT). Then, epsE was also deleted in the LMG18311(ICESt3∆int) strain to obtain LMG18311∆epsE(ICESt3∆int) strain. In conjugation experiments using the mating pair LMG18311(ICESt3∆int::IME_SsalL25_oriT) as donor and LMG18311∆epsE(ICESt3∆int) as recipient, transconjugants were obtained at a frequency of 4.8 × 10 −6 ± 4.0 × 10 −6 TC/donor cell (Table S3). In each experiment, four transconjugants were analyzed by PCR and all of them exhibited the same profile ( Figure 4). Since the ICE hosting the IME was immobilized in the donor strain, this indicated that IME_ SsalL25_oriT was mobilized in trans, i.e., excised and hijacked the conjugation machinery of the ICE to transfer by conjugation. A control conjugation experiment using the mating pair LMG18311(ICESt3∆int::IME_SsalL25_oriT) (donor)/LMG18311∆epsE (recipient without integration site for the IME) gave no transconjugant, indicating that the IME_ SsalL25_oriT integrates efficiently only in an oriT (Table S3).
( Figure 4). Since the ICE hosting the IME was immobilized in the donor strain, this indicated that IME_ SsalL25_oriT was mobilized in trans, i.e. excised and hijacked the conjugation machinery of the ICE to transfer by conjugation. A control conjugation experiment using the mating pair LMG18311(ICESt3Δint::IME_SsalL25_oriT) (donor)/LMG18311ΔepsE (recipient without integration site for the IME) gave no transconjugant, indicating that the IME_ SsalL25_oriT integrates efficiently only in an oriT (Table S3).  Table S1.

The Relaxase of IME_SsalL25_oriT Is Needed for IME Mobilization and Does Not Contribute to the Transfer of the Helper ICE
The IME_oriTs encode a protein belonging to the Rep_trans family that is very distantly related to known MobT relaxases such as the one of ICESt3 [23] and could be the relaxase of the IME. The gene of IME_SsalL25_oriT encoding this Rep_trans protein was deleted in strain LMG18311(ICESt3::IME_SsalL25_oriT) and conjugation experiments were performed using this strain (LMG18311(ICESt3::IME_SsalL25_oriTΔrel)) as donor and the LMG18311ΔepsE (ICESt3Δint) strain as recipient. No transfer of the IME was observed using this mating pair (Table S3). Thus, this protein is essential to its mobilization and cannot be complemented by the relaxase of the ICE. Therefore, it is the relaxase of the IME. Additional conjugation experiments were performed using the LMG18311ΔepsE strain as recipient. Despite a frequency of ICE transfer reaching 3.5 × 10 −4 ± 2.0 × 10 −4 TC/donor cell, no co-transfer of the IME was observed (Table S3). Therefore, no cis-mobilization (i.e., passive transfer as a passenger of the ICE) of IME_oriT could be detected.
IME_SsalL25_oriT was transferred by conjugation into a strain deleted for the relaxase gene of ICESt3 to obtain LMG18311(ICESt3Δrel::IME_SsalL25_oriT) used as donor in conjugation experiments with LMG18311ΔepsE strain as recipient. This ICE was still able to excise (data not shown) but no transconjugant was obtained (Table S3). Therefore, the relaxase of the ICE cannot be complemented (or at a very low efficiency that was undetectable in the laboratory conditions used) by that of the IME. Specific regions were amplified to discriminate the recipient strain, ermB for ∆epsE, and attL for IME_SsalL25_oriT. The integrase deletion (done by insertion of a chloramphenicol resistance cassette) was detected with the amplification of the attR site of ICESt3. The sizes of the PCR fragments were confirmed by parallel migration of a DNA ladder (M for marker on the figure). The primers pairs used for these amplifications and expected sequence lengths are listed in Table S1.

The Relaxase of IME_SsalL25_oriT Is Needed for IME Mobilization and Does Not Contribute to the Transfer of the Helper ICE
The IME_oriTs encode a protein belonging to the Rep_trans family that is very distantly related to known MobT relaxases such as the one of ICESt3 [23] and could be the relaxase of the IME. The gene of IME_SsalL25_oriT encoding this Rep_trans protein was deleted in strain LMG18311(ICESt3::IME_SsalL25_oriT) and conjugation experiments were performed using this strain (LMG18311(ICESt3::IME_SsalL25_oriT∆rel)) as donor and the LMG18311∆epsE (ICESt3∆int) strain as recipient. No transfer of the IME was observed using this mating pair (Table S3). Thus, this protein is essential to its mobilization and cannot be complemented by the relaxase of the ICE. Therefore, it is the relaxase of the IME. Additional conjugation experiments were performed using the LMG18311∆epsE strain as recipient. Despite a frequency of ICE transfer reaching 3.5 × 10 −4 ± 2.0 × 10 −4 TC/donor cell, no co-transfer of the IME was observed (Table S3). Therefore, no cis-mobilization (i.e., passive transfer as a passenger of the ICE) of IME_oriT could be detected.
IME_SsalL25_oriT was transferred by conjugation into a strain deleted for the relaxase gene of ICESt3 to obtain LMG18311(ICESt3∆rel::IME_SsalL25_oriT) used as donor in conjugation experiments with LMG18311∆epsE strain as recipient. This ICE was still able to excise (data not shown) but no transconjugant was obtained (Table S3). Therefore, the relaxase of the ICE cannot be complemented (or at a very low efficiency that was undetectable in the laboratory conditions used) by that of the IME.

IME_SsalL25_oriT and Helper ICE Do Not Compete for the Conjugation Machinery
Since IME_SsalL25_oriT needs to use the conjugation machinery of its helper ICE, we wondered whether it is at the expense of the ICE transfer rate. Thus, ICESt3 transfer was measured for three distinct structures: ICESt3 devoid of IME_SsalL25_oriT, ICESt3 hosting an IME_SsalL25_oriT and ICESt3 hosting an IME deleted for its relaxase gene (and thus unable to transfer [23]). Frequencies of ICESt3 transfer were similar (Table S3). This indicates that, in these laboratory mating assay conditions, IME mobilization does not impair the ICE transfer rate ( Figure 5).
Conversely, immobilization of the helper ICE in the donor cell (using ∆int mutant of ICESt3) does not have any impact on the IME transfer rate in the tested laboratory conditions (Table S3). This indicates that use of the conjugation machinery is not a limiting factor for ICE and IME transfer in the tested laboratory conditions. Since IME_SsalL25_oriT needs to use the conjugation machinery of its helper ICE, we wondered whether it is at the expense of the ICE transfer rate. Thus, ICESt3 transfer was measured for three distinct structures: ICESt3 devoid of IME_SsalL25_oriT, ICESt3 hosting an IME_SsalL25_oriT and ICESt3 hosting an IME deleted for its relaxase gene (and thus unable to transfer [23]). Frequencies of ICESt3 transfer were similar (Table S3). This indicates that, in these laboratory mating assay conditions, IME mobilization does not impair the ICE transfer rate ( Figure 5). Conversely, immobilization of the helper ICE in the donor cell (using Δint mutant of ICESt3) does not have any impact on the IME transfer rate in the tested laboratory conditions (Table S3). This indicates that use of the conjugation machinery is not a limiting factor for ICE and IME transfer in the tested laboratory conditions.

Discussion and Conclusions
MobT relaxases from ICEs of the Tn916 superfamily (Tn916, ICEBs1, and ICESt3) belong to a larger family of proteins, Rep_trans, that are classically considered as RCR initiators mainly involved in plasmid maintenance [23,31]. None has ever been shown to be involved in the transfer of a mobilizable element. However, we recently found that many integrative elements from streptococci, which do not encode any MPF protein, encode Rep_trans "RCR initiators" that are very distantly related to genuine MobTs from ICEs [12][13][14][15]. We previously proposed that the "Rep-trans" proteins of these elements are MobT relaxases, enlarging this family, and that these elements are IMEs [13].
Here, we showed for the first time that (i) the Rep_trans "RCR initiators" encoded by one of these types of elements (IME_oriTs) are actually relaxases involved in their conjugative transfer; and (ii) these elements are IMEs that use the MPFFA and TcpA of helper ICEs to promote their own transfer. This suggests that the tridimensional structures of the proteins are conserved despite a low conservation of the primary sequence, as previously described for the Rep proteins RepSTK1 of Geobacillus stearothermophilus and staphylococcal Rep proteins [26]. This works substantiates the very high prevalence of IMEs encoding relaxases related to RCR initiators involved in plasmid and virus replication [13]. It remains to be examined if plasmids harboring Rep-trans proteins can be mobilized by ICEs encoding a MobT relaxase. Such mobilization (by ICEBs1) has been described for plasmids that encode RCR Initiators of the Rep_1 family [32]. In addition to their function in conjugative transfer, the MobT/Rep_Trans proteins of Tn916 and ICEBs1 also catalyze the initiation of RCR needed for the maintenance of the excised ICEs [28,33]. As suggested previously, this could also be the case for the IMEs [11].
We previously described twelve members of this IME_oriT family [15]. In this study, the extensive search of related IMEs in available bacterial genomes led to the identification of 52 new

Discussion and Conclusions
MobT relaxases from ICEs of the Tn916 superfamily (Tn916, ICEBs1, and ICESt3) belong to a larger family of proteins, Rep_trans, that are classically considered as RCR initiators mainly involved in plasmid maintenance [23,31]. None has ever been shown to be involved in the transfer of a mobilizable element. However, we recently found that many integrative elements from streptococci, which do not encode any MPF protein, encode Rep_trans "RCR initiators" that are very distantly related to genuine MobTs from ICEs [12][13][14][15]. We previously proposed that the "Rep-trans" proteins of these elements are MobT relaxases, enlarging this family, and that these elements are IMEs [13]. Here, we showed for the first time that (i) the Rep_trans "RCR initiators" encoded by one of these types of elements (IME_oriTs) are actually relaxases involved in their conjugative transfer; and (ii) these elements are IMEs that use the MPF FA and TcpA of helper ICEs to promote their own transfer. This suggests that the tridimensional structures of the proteins are conserved despite a low conservation of the primary sequence, as previously described for the Rep proteins RepSTK1 of Geobacillus stearothermophilus and staphylococcal Rep proteins [26]. This works substantiates the very high prevalence of IMEs encoding relaxases related to RCR initiators involved in plasmid and virus replication [13]. It remains to be examined if plasmids harboring Rep-trans proteins can be mobilized by ICEs encoding a MobT relaxase. Such mobilization (by ICEBs1) has been described for plasmids that encode RCR Initiators of the Rep_1 family [32]. In addition to their function in conjugative transfer, the MobT/Rep_Trans proteins of Tn916 and ICEBs1 also catalyze the initiation of RCR needed for the maintenance of the excised ICEs [28,33]. As suggested previously, this could also be the case for the IMEs [11].
We previously described twelve members of this IME_oriT family [15]. In this study, the extensive search of related IMEs in available bacterial genomes led to the identification of 52 new IME_oriT. These 64 elements are distributed in 14 streptococcal species and in S. aureus. Almost identical elements were found in distant species, indicating a very recent horizontal transfer of these elements. These elements target the oriT of ICEs belonging to three distantly related families (Tn916, ICESt3 and ICE6013) of the Tn916 superfamily that all exhibit a MobT relaxase. More surprisingly, one IME_oriT was found inside an IME integrated in rpmG. It is worth mentioning that this IME also encodes a relaxase belonging to the enlarged MobT family. Therefore, this suggests that the IME_oriT recognizes specifically the oriT associated to MobT relaxases and more precisely a very short conserved sequence that includes the site that is nicked by the relaxase of a very large array of integrated elements encoding very distant relaxases. However, some of IME_oriT elements were also found integrated in other chromosomal sites (neither in ICEs nor in IMEs) even though they do harbor an integrase closely related to the one of the IMEs integrated in oriT. These sites could be secondary ones that present some similarities with the site targeted by the integrase.
Diverse cargo genes were found on the IMEs of the IME_oriT family. Half of these elements carry lsa(C) that confers cross-resistance to lincosamides, streptogramin A, and pleuromutilins. Thus, these IMEs play a major role in the dissemination of this particular cross-resistance. Interestingly, one of these IMEs carrying lsa(C) is integrated in an ICE6013. This is only the second description of an ICE6013 carrying an antibiotic resistance gene. The first description is an ICE6013 carrying a penicillin resistance gene through insertion of Tn552 but, in this case, the transposon disrupts the DDE transposase gene of the ICE and thus impairs its excision and transfer [27]. In the present example, the ICE likely remains active, provided that IME first excises. We recently reported the contribution of different families of IMEs to the dissemination of antibiotic resistance in Streptococcus suis [12]. IMEs have been very poorly studied until now because they are difficult to detect in bacterial genomes due to their small size and high diversity. However, their role in antibiotic resistance dissemination and, in particular, in streptococci, should be better considered. In addition, three IMEs carry merA that encodes a mercuric reductase conferring mercury resistance. This gene is frequently found on MGEs, in particular, in oral bacteria belonging to the Firmicutes group [34] due to the use of mercury in dental care in the past. Furthermore, one IME could encode an ABC transporter protecting its bacterial host from lantibiotics. Bacteriocins are ribosomally synthesized, antimicrobial substances produced by many bacteria to survive in competitive environments [35]. Bacteriocins from Firmicutes are classified into two main classes: (i) post-translationally modified bacteriocins (lantibiotics) and (ii) non-modified bacteriocins. Different mechanisms of immunity against lantibiotics have been described, including ABC transporters that export lantibiotics into the surrounding medium [36]. This family of IMEs could also contribute to the fitness of the host by providing an additional alkylation DNA repair protein.
Alkylating agents, which are widespread in the environment, also occur endogenously in bacteria as primary and secondary metabolites. Such compounds have intrinsically extreme cytotoxic and frequent mutagenic effects, to which organisms have developed resistance by evolving multiple repair mechanisms to protect cellular DNA [37].
Although the elements are integrated in ICEs, the experimental characterization of the mobility of an IME_oriT failed to find any mobilization in cis by its host ICE. This suggests that disruption of the origin of transfer of the ICE by integration of the IME impairs ICE transfer and that prior excision of the IME is required to visualize ICE transfer. Furthermore, conjugative transfer of the IME by mobilization in trans was observed, albeit at a 100-fold lower frequency than that of ICE. This transfer does not impair the ICE transfer rate. This indicates that IME_oriT benefits from the DNA transport system encoded by their helper ICE but does not impair ICE transfer. Similar interactions could also occur between the many IMEs that integrate inside the genes encoding conserved or essential proteins (such as VirD4 CP) of ICEs of another superfamily widespread in Firmicutes (Tn5252 superfamily) and their ICE host [11,12]. The impact of an IME on the transfer and/or stability of their helper co-resident element has been studied in depth only for IMEs of the SGI-1 family from the proteobacterium Salmonella [38][39][40][41][42]. Although these IMEs are mobilized only by conjugative plasmids of the IncA/C incompatibility group, SGI-elements and IncA/C plasmids are never found together in clinical multidrug resistant isolates. Firstly, the conjugative plasmids transfer from the donor cell to a recipient cell strain harboring the quiescent IME, inducing the excision of the IME. Secondly, the IME subverts the conjugation apparatus of the plasmid to invade the donor cells harboring IncA/C plasmids by retrotransfer and then probably invades all the cells of the population that harbor the plasmids. Thirdly, it suppresses plasmid transfer and leads to plasmid loss in the cell by interfering with plasmid regulation. Therefore, in contrast to the IME_oriT studied here, these IMEs are harmful pirates of conjugative IncA/C plasmids that await the arrival of the conjugative plasmid and then are probably transferred to all the cells harboring the plasmids and then eliminate these plasmids [41,42].
In conclusion, IMEs that integrate inside the origin of transfer of both ICEs and IMEs are widespread in streptococci and are also found in other Firmicutes. They vehicle cargo genes, particularly antibiotic resistance genes, and can thus participate in the dissemination of these threatening genes. They can also offer other helpful traits to their bacterial host (protection against bacteriocins or repair of alkylation damage), which help them persist in bacterial communities. The IME_oriTs are the first elements that are clearly shown to use a conjugative element both as a host for their integration and as helper supplying their transfer engine (CP and MPF). Some IMEs, like SGI1, can act as harmful pirates of their host/helper but some (such as IME_oriTs) can also be harmless hitchhikers of the ICEs. Other unrelated IMEs in Firmicutes may have similar capacities. Further studies are needed in order to decipher the mechanistic aspects of their mobilization.