Surface exclusion of IncC conjugative plasmids and their relatives

The phenomenon of exclusion allows conjugative plasmids to selectively impede the entry of identical or related elements into their host cell to prevent the resulting instability. Entry exclusion blocks DNA translocation into the recipient cell, whereas surface exclusion destabilizes the mating pair. IncC conjugative plasmids largely contribute to the dissemination of antibiotic-resistance genes in Gammaproteobacteria. IncC plasmids are known to exert exclusion against their relatives, including IncC and IncA plasmids, yet the entry exclusion factor eexC alone does not account for the totality of the exclusion phenotype. In this study, a transposon-directed insertion sequencing approach identified sfx as necessary and sufficient for the remaining exclusion phenotype. Sfx is an exclusion factor unrelated to the ones described to date. A cell fractionation assay localized Sfx in the outer membrane. Reverse transcription PCR and beta-galactosidase experiments showed that sfx is expressed constitutively at a higher level than eexC. A search in Gammaproteobacteria genomes identified Sfx homologs encoded by IncC, IncA and related, untyped conjugative plasmids and an uncharacterized family of integrative and mobilizable elements that likely rely on IncC plasmids for their mobility. Mating assays demonstrated that sfx is not required in the donor for exclusion, ruling out Sfx as the exclusion target. Instead, complementation assays revealed that the putative adhesin TraN in the donor mediates the specificity of surface exclusion. Mating assays with TraN homologs from related untyped plasmids from Aeromonas spp. and Photobacterium damselae identified two surface exclusion groups, with each Sfx being specific of TraN homologs from the same group. Together, these results allow us to better understand the apparent incompatibility between IncA and IncC plasmids and to propose a mechanistic model for surface exclusion mediated by Sfx in IncC plasmids and related elements, with implications for the rampant dissemination of antibiotic resistance. Author summary Bacterial conjugation plays a pivotal role in the evolution of bacterial populations. The circulation of drug resistance genes bolsters the emergence of multidrug-resistant pathogens, with which contemporary medicine struggles to cope. Exclusion is a natural process preventing the redundant acquisition of a plasmid via conjugation by a host harbouring an identical or similar plasmid. Although exclusion has been known for the past half-century, the mechanisms involved remain poorly understood. This study describes an exclusion factor, Sfx, encoded by IncC, IncA and related conjugative plasmids and by unrelated integrative and mobilizable elements. We report that Sfx is a lipoprotein of the recipient that selectively inhibits conjugation based on the adhesin TraN expressed at the surface of the donor. We propose a mechanistic model for Sfx-mediated exclusion. Ultimately, a better understanding of exclusion could facilitate the design of conjugation inhibitors targeting mating pair formation to curb the circulation of drug-resistance genes in healthcare settings, agriculture, animal husbandry and food and drug production.


Introduction
IncC plasmids are large conjugative plasmids frequently associated with multidrug resistance phenotypes in a broad range of Gammaproteobacteria species [1].IncC plasmids' ability to mobilize non-autonomous, unrelated integrative and mobilizable elements (IMEs) of the Salmonella Genomic Island 1 (SGI1) and MGIVchHai6 families exacerbates their importance in antibiotic resistance dissemination [2][3][4][5].SGI1 and its multiple variants are important vehicles of antibiotic resistance genes and are frequently found in Salmonella enterica and a broad range of Gammaproteobacteria [6].MGIVchHai6 conferred multidrug resistance to Vibrio cholerae non-O1/non O139 infecting cholera patients during the 2010 cholera outbreak in Haiti [4].IncC, IncA and untyped IncA/C-like (ACL) plasmids (e.g., pAsa4c, pAhD4-1, pAQU1) share a syntenic set of genes involved in conjugation, DNA repair, and regulation.All seem regulated by closely related homologs of the master activator of transfer AcaCD [7][8][9].IncC plasmid-encoded AcaCD promotes the excision and mobilization of MGIVchHai6-like IMEs [4,5].AcaCD also stimulates the excision and replication of SGI1 and the expression of its traN, traG and traH genes, which modify the mating apparatus of IncC plasmids to enhance its dissemination [7,10,11].
Large conjugative plasmids are autonomous replicons that usually remain at a low-copy number.
The entry into the same cell of a plasmid sharing similar replication or partitioning determinants promotes plasmid destabilization and loss, a phenomenon known as incompatibility [12].
Exclusion refers to the mechanisms employed by conjugative elements to preclude instability resulting from incompatibility by hindering redundant transfer into the cells they reside [13].
While the definite nature of exclusion mechanisms remains elusive, they can be categorized as entry and surface exclusion by whether they interfere with DNA transfer or mating pair stabilization [13,14].Exclusion seems to be a staple of the conjugative plasmid lifestyle.
Exclusion is not specific to the transferred DNA [25], and proteins in the donor have been identified as targets of exclusion factors in several systems.EexA and EexB of IncHI1 plasmids and TrbK of pKPK_UVA01 appear to act concurrently as exclusion factors and exclusion targets, as they are necessary in both the donor and recipient for exclusion to take place [15,21].Other exclusion factors target different plasmid-encoded proteins.The IncI ExcA targets TraY, which is found in the cytoplasmic membrane of the donor [16].Similarly, the IncF TraS and SXT/R391 Eex of the recipient seem to interact directly with the VirB6 homolog TraG of the donor, and the regions involved in both proteins are cytoplasmic [23,26].In the cases above, the genes encoding the exclusion factor and its target are adjacent.In IncP, IncW and IncN plasmids, for which the exclusion target remains unknown, the exclusion gene abuts a gene coding for a VirB6 homolog, suggesting a widespread usage of this target beyond the IncF and SXT/R391 families [13].
The surface exclusion factor TraT of IncF plasmids reduces the percentage of mating aggregates, and five contiguous residues determine specificity between exclusion groups [14,27].Harrison et al. proposed that TraT destabilizes the mating pair by interacting with the pilus tip; however, it was eventually shown that the pilin does not define exclusion specificity [27,28].The discovery of interactions between TraT and the outer membrane protein OmpA in the recipient and between OmpA in the recipient and the putative adhesin TraN in the donor suggested surface exclusion could instead result from TraT preventing docking of TraN to OmpA, thus hindering mating pair stabilization [29][30][31].The identification of TraN receptor specificity groups across IncF plasmids makes this hypothesis extremely appealing, yet traN substitutions failed to flip exclusion specificity between F and R100-1 [30,32,33].
IncC plasmids share with IncA plasmids the same entry exclusion system, eexC/traG, and inhibit each other's entry into the cell in which they reside [9,34].During the characterization of eexC/traG, we reported that knocking-out eexC of the IncC plasmid pVCR94 in the recipient only reduced the exclusion phenotype in mating assays using an IncC + donor.Likewise, the deletion of a 45.6-kbfragment encompassing 33 open reading frames located between traN and traF led to a statistically significant reduction of exclusion.Hence, another mechanism likely prevents redundant transfer between cells containing identical plasmids.
In this report, we investigate the cause of the incomplete abolition of exclusion by a recipient strain bearing an IncC plasmid lacking the entry exclusion gene eexC.Using a transposondirected insertion sequencing (TraDIS) approach, we identified IncC-borne candidate genes whose disruption in recipient cells facilitated the acquisition of an incoming mobilizable plasmid from an IncC + donor.Further analyses revealed that a single open reading frame, hereafter

vcrx085 impedes conjugative transfer between IncC + cells
To identify additional factors responsible for the inhibition of transfer between IncC + cells, we used transposon-directed insertion sequencing (TraDIS) aimed at mapping Tn5 insertions that enhance the entry of the mobilizable plasmid pClo transferred from an IncC + donor strain into a Tn5 + IncC + recipient library (Fig 1A).pClo was used as a proxy for conjugative transfer to bypass IncC plasmid incompatibility.Briefly, we constructed a high-density mini-Tn5 (Sp R ) insertion library in E. coli GG56 carrying pVCR94 Kn ΔeexC::cat, a kanamycin-resistant variant of pVCR94 unable to exert entry exclusion in recipient cells [9].This set of mutants formed the input library containing Tn5 insertions in the chromosome or pVCR94 that allowed plasmid replication and maintenance.The input library was then used as recipient in a mating assay with a donor strain carrying pVCR94 Kn and the mobilizable plasmid pClo.GG56 transconjugant colonies that acquired pClo made up the output library.We expected that genes or sequences of pVCR94 Kn ΔeexC::cat more frequently disrupted in the output than in the input (high insertion index ratio) were those impeding the mobilization of pClo between IncC + cells.Since our assays monitored pClo entry only, we did not consider the role of chromosomal genes in the recipient.The tracks plot the number of reads from two independent replicates as a function of position in pVCR94 Kn ΔeexC::cat for both the input and output libraries.ORFs with similar functions are colour-coded as indicated in the panel.This panel was created using the UCSC Genome Browser (http://genome.ucsc.edu).(C) Nine loci of unknown function show a mini-Tn5 insertion index ratio greater than 5 (dashed line), calculated as the ratio of insertion counts between the output and input libraries.ORFs with similar functions are colour-coded as indicated in Fig S1 .(D) sfx (vcrx085) is the only functional exclusion factor among candidates identified in (B) and suffices alongside eexC (eex) to account for the totality of IncC exclusion.E. coli VB112 (Rf r ) containing pVCR94 Kn and pClo (Ap r ) served as the donor strain, and E. coli GG56 (Nx r ) containing the specified derivatives of pVCR94 Sp served as the recipient strains.When indicated, eexC or sfx were expressed in the recipients from single-copy, chromosomally integrated pAH56 (Kn r ) in the presence of IPTG or together from pBeloBAC11 (Cm r ) under the control of their respective native promoters.Transconjugants containing pClo were selected as the Nx r Ap r colonies.Crossbars show the mean and standard error of the mean of three independent experiments.One-way ANOVA (p=1.6e-25) with a Tukey-Kramer post-test was used on the We identified nine candidate genes in pVCR94 Kn ΔeexC::cat with insertion index ratios above the arbitrary threshold of 5 (Table 1, Fig 1B and 1C, S1 Fig).Among these genes, only acr2 (vcrx150) is a known repressor of conjugative transfer [7].To validate the ability of the candidates to inhibit incoming conjugative transfer, we tested mobilization of pClo into the corresponding deletion mutants of pVCR94 Sp used as recipients.All but one had no impact on transfer of pClo (Fig 1D).The deletion of vcrx085 (hereafter referred to as sfx for surface exclusion) improved conjugation to the same level similar as the deletion of the entry exclusion factor eexC.Like eexC, we could complement the sfx deletion in trans (Fig 1D , pAH56).Both mutations combined completely abolished exclusion, allowing conjugation to a level comparable to an empty recipient.Conversely, the standalone expression of sfx and eexC from their native promoter in the absence of an IncC plasmid in the recipient matched the full exclusion phenotype of the wildtype IncC plasmid (Fig 1D , pBeloBAC11).

Sfx is a constitutively expressed outer membrane protein
The SignalP 6.0 server predicts a lipoprotein signal peptide in the translation product of sfx (S3 suggests that the high expression level of sfx 3xFLAG from P tac on a multi-copy vector saturates the Lpt trafficking pathway.We could not undoubtedly assign the processed and unprocessed forms in the inner membrane fraction, and two additional products appeared between 20 and 25 kDa on the western blot.These weaker bands are presumably degradation products of an unknown nature.

Surface exclusion specificity is determined by TraN
To test the role of Sfx in the donor, we monitored the transfer of pClo from donors carrying the ΔeexC and Δsfx deletion mutants of pVCR94 Sp toward an IncC + strain (pVCR94 Kn ).No alleviation of exclusion, i.e., an increase in transfer frequency, was observed (Fig 4A).On the contrary, a modest yet significant decrease was observed for the Δsfx and ΔeexC Δsfx mutants, suggesting that augmented redundant transfer between donors was detrimental to transfer toward recipients.Hence, sfx is not self-targeting to promote surface exclusion.
Next, we focused on TraN as a potential target for surface exclusion, as the gene traN encodes a putative adhesin thought to stabilize the mating pair, and it is adjacent to sfx in most conjugative plasmids found but pAQU1 [30,31,39]

Discussion
All known conjugative plasmids encode at least one exclusion factor [13].An entry exclusion system involving the entry exclusion factor EexC and the VirB6-homolog TraG was previously identified in IncA and IncC conjugative plasmids [9].Here, we identified a second exclusion factor within a region suspected by Humbert et al. to be responsible for additional exclusion activity [9].A TraDIS experiment pointed to nine candidate genes likely to enhance incoming transfer once disrupted.sfx was the only one whose activity in exclusion could be and IncC plasmids are expected to exert surface exclusion against each other, hence contributing alongside entry exclusion to the previously observed exclusion phenotype between the two incompatibility groups [9,34] We showed that sfx Aut1 of IEVchAut1 is functional (Fig 4D).Since these IMEs lack a traN gene, we propose they engage in surface exclusion as a defensive mechanism, forcing their host to favour mating partners that could enhance both their stability and transmissibility.Hence, these IMEs could modulate the circulation of conjugative elements in bacterial populations without affecting whether surface exclusion from a potential recipient will target transfer mediated by their helper element.Such behaviour would contrast with the way SGI1 manipulates the exclusion system of its helper IncC plasmid.SGI1, which is not known to exert exclusion, evades IncC entry exclusion by substituting TraG in the mating apparatus with a distant VirB6 homolog [11].Manipulation of exclusion by mobilizable elements is not well known.While homologous ColE1 mobilizable plasmids were reported to inhibit each other's transfer mildly, this phenotype was attributable to the accessory mob gene mbeD with no identified target in the donor [20,46,47].It thus remains unclear whether exclusion is at play in that case.The putative IMEs we reported here may be the first instance of exclusion exerted by mobilizable elements.its gene content and the presence of AcaCD binding motifs suggest mechanistic parallels with that of SGI1, albeit using totally unrelated mobilization factors.In ACL + cells, AcaCD likely promotes IEVchAut1's excision through the expression of rdfM [38].A predicted homolog of the replication protein P from bacteriophage lambda suggests the excised IME is replicative [48].

The close relationship between Sfx proteins encoded by
IME DNA transfer seems to rely on the IME-encoded TraI, a predicted relaxase of the MOB F family [49].While SGI1 encodes its own replication and DNA processing functions [10,50,51], IEVchAut1 additionally encodes a type IV coupling protein, presumably making it independent from the IncC-encoded TraD and DtrJ (also known as TraJ) [5,52].
IncHI1 EexB and IncF TraT are two other known surface exclusion factors.While EexB seems to have a dual role as an exclusion factor and exclusion target, TraT's mechanism remains elusive as exclusion specificity could not be attributed to direct interaction with pilin or indirect interaction with the putative adhesin TraN through OmpA [15,27,28,30] Although a direct interaction between Eex and TraG has been suggested in SXT/R391 ICEs and likely extends to other VirB6-targeting entry exclusion factors [26], direct contact between Sfx and TraN is unlikely for surface exclusion.In fact, the binding of outer membrane proteins on donor and recipient cells is hard to reconcile with the idea that surface exclusion physically prevents the formation or disrupts mating aggregates [14].Instead, surface exclusion factors have been proposed to act by concealing outer membrane receptors involved in mating aggregation [29].According to Riede and Eschbach, TraT of F directly interacts with the major outer membrane protein OmpA as TraT inhibits the OmpA-specific phage K3 [29].Yet, despite ample evidence that TraN F interacts with OmpA during conjugation, it does not seem to determine exclusion specificity as TraN F substitution with TraN R100-1 does not counter exclusion by TraT of F [30][31][32].
We found that sfx 94 is constitutively expressed independently of AcaCD at a much higher level recognize distinct receptors on the recipient surface and that Sfx prevents mating pair stabilization by specific binding to the receptor of its cognate TraN.Further investigation will be required to identify the receptors involved.

Plasmid and strain construction
Plasmid DNA was prepared using the QIAprep Spin Miniprep kit (Qiagen), and genomic DNA was isolated with the QIAamp DNA mini kit (Qiagen) as recommended by the manufacturer.
PCR products were purified using the PCR purification kit (Qiagen).All molecular biology manipulations were carried out by standard procedures following the Current Protocols in Molecular Biology [70].The oligonucleotides used in this study are described in S1 Table.
Amplicons were digested with NdeI and SalI and cloned into NdeI/SalI-digested pAH56.psfx 3xFlag was constructed from psfx 94 using primer pair pAH56_3xFlag.F/94sfx_3xFlag.R and the Q5 Site-Directed Mutagenesis Kit (New England Biolabs) according to the manufacturer's instructions.
The resulting constructs were single-copy integrated into the attB λ chromosomal site of CAG18439 using pINT-ts.
To assemble a DNA fragment encompassing sfx 94 and eexC alongside their native promoters, the region spanning from acaB to traG in pVCR94 Kn was deleted using the one-step chromosomal gene inactivation technique with primer pair 94del86acaB.for/94del144traG.rev and pKD3 as the template.After the excision of the resistance cassette using pCP20, the region of interest was amplified using primer pair 94eexBamHI.for/94sfxHindIII.rev.were calculated as the ratio of an element's transfer frequency toward an empty recipient to its transfer frequency toward the tested recipient.

Transposon-directed insertion sequencing (TraDIS)
A conjugation-assisted random transposon mutagenesis experiment was performed on E. coli GG56 bearing pVCR94 Kn ΔeexC::cat. to select for mini-Tn5 (Sp) insertions that allowed the entry and replication of pClo into the recipients.After overnight incubation at 37°C, Nx, Kn, Sp, Cm, and Ap-resistant colonies were collected and subsequently resuspended in LB broth, washed and resuspended in 4.5 ml of LB broth and cryopreserved.These samples were designated as 'output libraries'.The total DNA of a 1.5 ml aliquot of the output libraries was extracted and used for sequencing.

Preparation of TraDIS libraries and Illumina sequencing
For each library, a 1.5 ml frozen stock aliquot was thawed on ice for 15 min and used to prepare sequencing libraries as described previously [71].Mutant libraries were then pooled and sequenced by Illumina using the NextSeq® 500/550 High Output Kit v2 at the RNomics platform of the Laboratoire de Génomique Fonctionnelle de l'Université de Sherbrooke (https://rnomics.med.usherbrooke.ca)(Sherbrooke, QC, Canada).The transposon data analysis was carried out as described previously [71].

Fractionation of cellular proteins
Cell fractionation was carried out using a protocol derived from Sandrini et al. [72].Briefly, E. coli DH5α λpir bearing pacaDC 3×Flag or psfx 3xFlag was grown overnight with or without 0.1 mM IPTG.
The cells were pelleted at 3,900 g for 20 min at 4°C. Cell pellets were washed twice with 10 mM Tris buffer (pH 7.5), resuspended in 10 ml of the same buffer, and frozen for 2 h at -80°C.
Samples were lysed by ultrasonication (Qsonica Q125) in an ice bath using five cycles of 30 s each at 70% amplitude, followed by 30 s of cooling.The lysates were treated with DNAse I (1 mg/ml).Cell debris were removed by centrifugation (3,900 g for 30 min at 4°C).The proteins in the supernatant were separated by ultracentrifugation (185,000 g for 30 min at 4°C).The supernatant (cytoplasmic proteins) was stored at -20°C.The pellet (total membrane proteins) was washed three times with 10 mM Tris buffer (pH 7.5), resuspended in 10 mM Tris buffer with 2% (v/v) Triton X-100, and incubated for 30 min at room temperature.The mixture was centrifuged at 185,000 g for 30 min at 4°C.The supernatant (inner membrane proteins) was concentrated using Pierce™ Protein Concentrators PES 3K, following the manufacturer's instructions (Thermo Scientific, cat.# 88515) and stored at -20°C.The pellet (outer membrane proteins) was washed three times with 10 mM Tris buffer (pH 7.5) and stored at -20°C.

Molecular biology
Plasmid DNA was extracted using the EZ-10 Spin Column Plasmid DNA Minipreps Kit (Bio Basic) following the manufacturer's instructions.Enzymes used in this study were purchased from New England Biolabs.PCR assays were performed with primers listed in Table S1.PCR conditions were as follows: (i) 3 min at 94°C; (ii) 30 cycles of 30 s at 94°C, 30 s at the appropriate annealing temperature, and 1 minute/kb at 68°C; and (iii) 5 min at 68°C.When required, the resulting products were purified using the EZ-10 Spin Column PCR Products Purification Kit (Bio Basic) following the manufacturer's instructions.E. coli strains were transformed by electroporation as described previously [73] in a Bio-Rad Gene Pulser Xcell device set at 25 μF, 200 V and 1.8 kV using 1-mm gap electroporation cuvettes.

RNA extraction and cDNA synthesis
RNA extractions were performed as follows.E. coli GG56 containing pVCR94 Kn with or without pacaCD was grown at 37°C for 16 h in LB broth containing the appropriate antibiotics.The cultures were diluted 1:200 in fresh medium containing the appropriate antibiotics and grown to an OD 600 of 0.2 before being diluted 1:10 again in fresh medium containing the appropriate antibiotics and supplemented with 0.02% arabinose when needed.After a 2-h incubation period, Details of statistical tests are provided in S2 File.Graphics were rendered via the ggplot2 R package (v3.3.6)[77].All figures were prepared using Inkscape 0.92 (https://inkscape.org/).

Phylogenetic analyses
The primary sequence of Sfx 94 and TraN 94 homologs were obtained using the NCBI blastp algorithm [78] against the nr/nt database restricted to Gammaproteobacteria (taxid: 1236).
Primary sequences sharing less than 45% identity and under 85% minimum coverage were filtered out of subsequent analyses.The distantly related Sfx AQU1 from pAQU1 was added manually to the Sfx dataset as an outgroup.The datasets were clustered with CD-HIT [79] to the best cluster that met the 0.90 identity cut-off before alignment.Where applicable, representative sequences ("seeds") were manually substituted with sequences of interest within the same clusters for downstream analyses.The resulting datasets were then aligned with MUSCLE [80], and poorly aligned regions were discarded from the resulting amino acid alignments using TrimAl v1.3 software with the automated heuristic approach [81].Evolutionary analyses were performed within MEGA11 (v 11.0.13)[82] using the Maximum Likelihood method (PhyML) [83] and the Le and Gascuel matrix-based model [84] with gamma distribution (LG + G).Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model and then selecting the topology with superior log likelihood value.Percent identity values were retrieved from the MUSCLE output [80].

Bioinformatic predictions
The prediction of signal peptides and cleavage sites in Sfx 94 and TraN 94 was performed using SignalP 6.0 with the slow model mode against other organisms [85].Terminators were predicted on both strands using ARNold [86,87].oriT loci and mobilization proteins in putative IMEs were predicted using oriTfinder [88].
named sfx, flanked by traN and acaB, accounts for the residual exclusion phenotype.Deletion of sfx with eexC completely abolished exclusion.Signal peptide prediction and a subcellular localization assay suggest the Sfx protein is an outer-membrane lipoprotein and support a role as a surface exclusion factor.Sfx belongs to an uncharacterized family of broadly distributed surface exclusion proteins encoded by IncC plasmids, their relatives and several putative IMEs.Surface exclusion assays designed to test sfx/traN gene pairs from ACL plasmids revealed that sfx and traN are part of a surface exclusion system, enabling us to propose a model of surface 135 exclusion of ACL plasmids.136

log 10 -
transformed values to compare the means.Statistical significance (S2 File) is shown as a compact letter display for pairwise comparisons where means grouped under identical letters are not statistically different.Exclusion indices, shown at the bottom of each crossbar, are calculated as the frequency of transfer into the empty recipient divided by the frequency of transfer into the indicated mutant.

Fig),
Fig), suggesting Sfx is displayed on the cell surface [35].A C-terminally 3xFLAG-tagged version of sfx was cloned under the control of the inducible promoter P tac .The Sfx 3xFLAG polypeptide has a predicted size of 33.4 kDa before signal peptide cleavage and 31.7 kDa afterwards.A cell fractionation assay confirmed the presence of Sfx 3xFLAG in the inner and outer membrane

Fig 2 .
Fig 2. Expression of IncC surface exclusion is independent of the master activator of

Fig 3 .
Fig 3. Sfx homologs are found across a wide range of Gammaproteobacteria. Maximum . A search for homologs of TraN 94 yielded 412 homologous proteins (S3 File), most encoded by ACL plasmids.Others belonged to IncCmobilizable IMEs of the SGI1 family integrated into trmE or dusA, with no clear relationship between TraN phylogeny and insertion site (Fig 3B).The cluster represented by pAQU1 differs significantly from the close ACL plasmids as the dsbC-traC-trhF-traWUN and acaB/bet-exo regions are not adjacent, and sfx lies elsewhere.

Fig 4 .
Fig 4. TraN directs surface exclusion specificity.(A) The absence of eexC and sfx in the donor decreases the frequency of transconjugant formation.pClo (Ap r ) was mobilized from E. coli VB112 (Rf r ) bearing the indicated deletion mutants of pVCR94 Sp (donor, D) into E. coli GG56 (Nx r ) bearing pVCR94 Kn (recipient).Transconjugants containing pClo were selected as the Nx r Ap r colonies.Crossbars show the mean and standard error of the mean of three experimentally validated.Together, eexC and sfx account for the totality of the exclusion phenotype(Fig 1D), akin to the exclusion mediated by traS and traT of the F plasmid[43].We could not detect any impact of abolishing exclusion on cell viability (S2 Fig).This result strongly contrasts with reports on F plasmid exclusion for which double mutants of traS and traT could not be isolated[43].This discrepancy could be due to the strong repression of the conjugative machinery of IncC plasmids (Fig 2D), whereas F transfer is constitutive due to the inactivation of F finO by IS3[7,44].The translation product of sfx is a predicted lipoprotein, and we detected Sfx in the outer membrane (Fig 2A), a trait shared with the surface exclusion factor TraT of the F plasmid.In addition, we showed that exclusion mediated by sfx is inhibited or abolished by the substitution of the putative adhesin TraN (Fig 4C), suggesting that exclusion specificity is determined by sfx/traN pairs.Together, these observations support a mechanism of surface exclusion[14,45].Our search in the Genbank database using Sfx 94 failed to reveal any known entry or surface exclusion factor, including F plasmid TraS or TraT, establishing Sfx of IncA and IncC plasmids as an unrelated surface exclusion factor (S3 File).As the ACL plasmids pAsa4c and pAhD4-1 belong to the same surface exclusion group as pVCR94 despite their divergence (Fig 4C), IncA IEAveTha1 and IEVchAut1 with those tested in this study (Sfx 94 , Sfx AhD4-1 , and Sfx Asa4c ) suggests they belong to the same exclusion group (Fig 3, S4 Fig).This activity has been corroborated by experimental evidence for Sfx Aut1 (Fig 4C, S5 Fig), establishing surface exclusion as an actual barrier to the circulation of mobile genetic elements on par with other bacterial defense mechanisms instead of being limited to preventing redundant transfer and, potentially, mediating mating pair separation [13,39].Surprisingly, none of the IMEs identified by the TraN search, including SGI1, appears in the Sfx tree, suggesting they do not encode a surface exclusion factor related to Sfx 94 .Conversely, IMEs identified by the Sfx search are ostensibly absent from the TraN tree as they lack the components of the conjugative apparatus.This observation suggests different survival strategies.Whereas SGI1 maximizes its own dissemination by remodelling the mating apparatus to improve transfer and bypass entry exclusion [11], IEAveTha1 and IEVchAut1 may ensure persistence by hindering via surface exclusion the entry of destabilizing, excisioninducing ACL plasmids.Although the mobilization mechanism of IEVchAut1 remains unknown, than the entry exclusion factor eexC or the adhesin gene traN when donor cells are in pure culture (Fig 2C and 2D).The high expression of sfx relative to eexC, despite similar levels of exclusion exerted by the two genes (Fig 1D, Fig 2D), is consistent with an interaction with a highly abundant surface protein.While few EexC proteins might suffice to target TraG in one mating pore during conjugation, an abundance of Sfx proteins is likely required to conceal most, if not all, TraN receptors on the recipient surface at any time.In pure cultures, expression of traN remained low, suggesting that the level of Sfx could be sufficient to quench spurious donordonor mating (Fig 2D).Unlike TraT of F, we have identified TraN as the specificity determinant for Sfx-mediated exclusion between ACL plasmids.Thus, we propose that TraN 94 and TraN AQU1

Table 1 . Candidate genes identified by TraDIS. ORF Fold change Predicted size (aa) Predicted product
[38]e).A second group represented by IEVchAut1 comprises six IMEs integrated at the 3' end of yicC (putative RNase adaptor protein YicC) in environmental and clinical Vibrio strains.In addition to a putative recombination directionality factor related to RdfM of MGIVflInd1[38], these IMEs carry a set of conserved genes, including a predicted type IV coupling protein (traD) and a putative relaxase of the MOB F family (traI).Finally, two IMEs coding for an identical Sfx are integrated at the 3' end of tRNA- TRP and the 5' end of rlmD (23S rRNA uracil methyltransferase) in environmental isolates of Shewanella chilikensis and Shewanella putrefaciens, respectively.Despite sharing a conserved core set of genes with IEVchAut1, the Sfx protein of these two IMEs resembles and clusters with Sfx 94 (Fig3A).Although we can predict AcaCD binding sites across these IMEs (S2 Table), no other conserved features of the SGI1 and MGIVchHai6 families was found, suggesting they belong to an unrelated IME family mobilizable by IncC plasmids.
. Among the Sfx homologs tested in this study, only Sfx AQU1 , which failed detection by the NCBI blastp search due to low similarity with Sfx 94 , belongs to a like elements constitute a new family of IMEs likely to be mobilizable by ACL plasmids (Fig 3C).
. Our results demonstrate that Sfx is only required in the recipient to enable exclusion (Fig 4A), ruling out the self-targeting of Sfx.Instead, we found that Sfx exclusion can be neutralized by substituting

Table 2 . Strains and plasmids used in this study Strain, plasmid or element Relevant genotype or phenotype Reference
The amplicon was digested with BamHI and HindIII and cloned into a BamHI/HindIII-digested pBeloBAC11.traNAsa4c , traN AhD4-1 , traN AQU1 were amplified using primer pairs Asa4traNEcoRI.f/Asa4traNSalI.r,AhD4traNEcoRI.for/AhD4traNEcoRI.revand AQU1traNEcoRI.for/AQU1traNEcoRI.rev,respectively.Amplicons were digested with EcoRI or EcoRI/SalI and cloned into EcoRI-or EcoRI/SalI-digested pBAD30.All constructs were verified by PCR and DNA sequencing at the Plateforme de Séquençage et de Génotypage du Centre de Recherche du CHUL (Québec, QC, Canada).Bacteria were grown for 16 h in LB broth with the appropriate antibiotics.Mating assays were carried out by mixing 100 µl of donor and recipient cells.Cells were pelleted by centrifugation, then washed once in 1 volume of LB broth and resuspended in 1/20 volume of LB broth.Bacterial mixtures were incubated for 6 h on LB agar plates at 37°C to allow conjugation.Serial dilutions were then plated on selective LB-agar plates with appropriate antibiotics to discriminate between donor, recipient and transconjugant CFUs.Transfer frequencies were calculated by dividing the number of transconjugant CFUs by the number of donor CFUs.Exclusion indices The transposition system was composed of E. coli MFDpir The TraDIS experiment was performed in several successive steps.First, pFG051 was transferred by conjugation from MFDpir to GG56 bearing pVCR94 Kn ΔeexC::cat in a 2-h mating experiment at 30°C on LB agar plates supplemented with DAP in duplicates.Once in the recipient strain that lacks cI, the constitutively expressed Tn5 machinery of pFG051 mediates random mini-Tn5 (Sp) insertions in the genome.The mating mixture was then entirely spread onto 40 large LB agar plates (150 mm) supplemented with Cm, Kn, Nx, and Sp.Plates were incubated until near confluence (40k to 60k CFUs) to select clones carrying mini-Tn5 (Sp) insertions.After overnight incubation at 37°C, Cm Nx Kn Sp-resistant colonies were collected using a cell scraper and resuspended in LB broth.The collected sample, designated as the 'input library' was washed, then resuspended in 4.5 ml of LB broth and cryopreserved.The total DNA of a 1.5 ml aliquot of the input library was extracted for sequencing.Another 1 ml aliquot of input library was used to inoculate 50 ml of LB broth supplemented with Cm, Kn, Nx and Sp, which was then grown overnight at 37°C.The resulting culture was used as the recipient in a mating assay, mixed in equal volumes with E. coli KH95 carrying pVCR94 Kn and pClo used as the donor.After 2-h incubation at 37°C, mating mixtures were spread onto 20 large LB agar plates (150 mm) supplemented with Nx, Kn, Sp, Cm, and Ap