A New Member of the Growing Family of Contact-Dependent Growth Inhibition Systems in Xenorhabdus doucetiae

Xenorhabdus is a bacterial symbiont of entomopathogenic Steinernema nematodes and is pathogenic for insects. Its life cycle involves a stage inside the insect cadaver, in which it competes for environmental resources with microorganisms from soil and the insect gut. Xenorhabdus is, thus, a useful model for identifying new interbacterial competition systems. For the first time, in an entomopathogenic bacterium, Xenorhabdus doucetiae strain FRM16, we identified a cdi-like locus. The cdi loci encode contact-dependent inhibition (CDI) systems composed of proteins from the two–partner secretion (TPS) family. CdiB is the outer membrane protein and CdiA is the toxic exoprotein. An immunity protein, CdiI, protects bacteria against inhibition. We describe here the growth inhibition effect of the toxic C-terminus of CdiA from X. doucetiae FRM16, CdiA-CTFRM16, following its production in closely and distantly related enterobacterial species. CdiA-CTFRM16 displayed Mg2+-dependent DNase activity, in vitro. CdiA-CTFRM16-mediated growth inhibition was specifically neutralized by CdiIFRM16. Moreover, the cdi FRM16 locus encodes an ortholog of toxin-activating proteins C that we named CdiCFRM16. In addition to E. coli, the cdiBCAI-type locus was found to be widespread in environmental bacteria interacting with insects, plants, rhizospheres and soils. Phylogenetic tree comparisons for CdiB, CdiA and CdiC suggested that the genes encoding these proteins had co-evolved. By contrast, the considerable variability of CdiI protein sequences suggests that the cdiI gene is an independent evolutionary unit. These findings further characterize the sparsely described cdiBCAI-type locus.


Introduction
Xenorhabdus, a member of the Enterobacteriaceae family, is a natural symbiont of entomopathogenic Steinernema spp. nematodes living in the soil [1,2]. The bacterium-nematode symbiosis is used for the biological control of various groups of insects [3], but the genus Xenorhabdus is also a pertinent and tractable model for investigating processes of antagonism between bacteria [4]. Indeed, the complex lifecycle of this bacterium involves an alternation between ecological niches: the nematode gut, the hemocoel of the living insect and the insect cadaver. Xenorhabdus proliferation in the insect cadaver is dependent on the ability of the the corresponding sites of pGJ907 to yield pGJ907_CdiA-CT FRM16 . Constructs were checked by sequencing (MWG-Eurofins, Germany). Plasmids pGJ907 and pGJ907_ CdiA-CT FRM16 were introduced into E. coli strain EPI400 (Epicentre, France) or E. coli strain WM3064 by transformation. The WM3064 transformants were used to transfer plasmids pGJ907 and pGJ907_ CdiA-CT FRM16 inside Xenorhabdus bovienii strain CS03 by conjugative mating (according to the protocol described by [34]) to construct the recombinant strain Xb_CS03_ PGJ907 and Xb_CS03_ PGJ907_ CdiA-CT FRM16 .

Plasmid construction for protein overproduction in E. coli BL21
The cdiA-CT FRM16 _XDD1_1120 fragment and the XDD1_1120 gene were amplified by PCR with the primer pairs F_EcorI_Cter_tpsA_III/ XD_1120_sal1R_bis and XD1120_EcoR1_F/ XD_1120_sal1R_bis, respectively. The PCR products were digested with EcoRI and SalI and inserted into the corresponding sites of pET28b to yield pET28b_CdiA-CT FRM16 _CdiI-His6 and pET28b_CdiI-His6.
Toxicity assays with CdiA-CT FRM16 E. coli strain EPI400 cells carrying pGJ907 (empty vector control), pGJ907_CdiA-CT FRM16 were cultured overnight in LB medium supplemented with kanamycin. The cultures were then diluted 1/500 in fresh LB supplemented with kanamycin and 200 μL of the resulting suspensions was used to inoculate 96-well plates (Greiner). The plates were incubated at 28˚C, with orbital shaking, in an Infinite M200 microplate reader (Tecan). Absorbance at 600 nm was measured every 30 minutes. When the OD 600 reached 0.15 to 0.2, aTc was (final concentration = 100 ng.mL -1 ) or was not (controls) added to the medium to induce CdiA-CT FRM16 expression under the control of the P tet promoter. Growth was evaluated for~10 hours. All experiments were performed in triplicate. We used the same protocol for Xenorhabdus growth inhibition assays.

Modulation of growth inhibition
Recombinant EPI400 E. coli_ PGJ907_ CdiA-CT FRM16 / pUC18, E. coli_ PGJ907_ CdiA-CT FRM16 / pUC18_XD118 and E. coli_ PGJ907_ CdiA-CT FRM16 / pUC18_XD1120 cells were used to inoculate 96-well plates in LB medium supplemented with IPTG (0.2 mM final), and were incubated at 28˚C, as described above. We added aTc to the medium when the OD 600 reached 0.15 to 0.2, to induce CdiA-CT FRM16 expression under the control of the P tet promoter. Growth was assessed for~10 hours. E. coli strain EPI400 cells carrying pGJ907 and pUC18 were used as empty vectors control. All experiments were performed in triplicate.
As a control, purified CdiI FRM16 -His6 alone was obtained from E. coli strain BL21 carrying pET28b_cdiI FRM16 and purified by Ni 2+ -affinity chromatography, as described above (without the denaturation step).
Purified proteins were analyzed by SDS-PAGE on a 15% polyacrylamide gel in the presence of SDS as previously described [35]. The samples, 5 μg of purified fraction, were run under reducing conditions, reduction being achieved by treating (3 min, 100˚C) the samples with a solution containing β-2-mercaptoethanol (0.5% final concentration). The molecular mass was determined by the use of Page Ruler TM plus prestained protein ladder (Thermo Scientific, USA). Gels were stained with Coomassie brilliant blue.

Pull down experiment
Purified CdiA-CT FRM16 (5 μM) was incubated with CdiI FRM16 -His6 (5 μM) for 30 min at room temperature and an aliquot was removed for analysis by SDS-PAGE. 500 μL of Ni 2+ -NTA resin, previously washed four times with binding buffer (20 mM sodium phosphate pH 7.0 containing 150 mM sodium chloride and 10 mM β-mercaptoethanol), was then added and incubated for 1 h 30 at 4˚C. The resin was then collected by centrifugation and the supernatant (unbound proteins) was collected for analysis by SDS-PAGE. Ni 2+ -NTA was washed with binding buffer three times, and the bound proteins were eluted in elution buffer (20 mM sodium phosphate pH 7.0 containing 150 mM sodium chloride, 10 mM β-mercaptoethanol and 250 mM imidazole). Samples were finally analysed by SDS-PAGE and the gel was stained with Coomassie blue as described above.

Nuclease assays
The activity of purified CdiA-CT FRM16 was assayed in vitro on genomic DNA from Xd_FRM16. CdiA-CT FRM16 (final concentration of 1 μM) was incubated with 2 μg of DNA in 50 μL of sterilized water supplemented with 2 mM MgCl 2 or CaCl 2 for 1 h at 37˚C. Where indicated, purified CdiI FRM16 -His6 protein from pET28b_cdiI FRM16 (immunity protein purified alone without CdiA-CT) was included at different final concentrations and allowed to bind CdiA-CT FRM16 for 30 minutes at room temperature before the addition of substrate DNA. Reactions were quenched with EDTA, and the reaction mixture was subjected to electrophoresis in a 1% agarose gel stained with ethidium bromide.

RNA isolation and RT-PCR analysis
X. doucetiae FRM16 in LB broth was incubated at 28˚C (100 mL) with horizontal shaking. Three samples were collected at different incubation times: i) early exponential growth phase (OD at 600 nm = 0.75); ii) end of the exponential growth phase (OD at 600 nm = 1.70); iii) stationary phase (OD at 600 nm = 3.70).
At each sampling time, total RNA was extracted with the RNeasy Protect Bacteria miniprep kit (from Qiagen), including incubation with DNase I, according to the manufacturer's instructions. RNA concentration was determined by measuring absorbance at 260 nm. For each RNA preparation, we assessed DNA contamination by carrying out a control PCR targeting the 16S ribosomal RNA gene, by using Xenorhabdus-specific 16S primers (see S1 Table).
The quantity and quality of total and messenger RNA, respectively, were assessed with a Nano-Drop 2000 spectrophotometer (Thermo Scientific) and an Agilent 2100 Bioanalyzer with the RNA 6000 Nano LabChip kit (Agilent). Total RNA (2 μg) was reverse-transcribed with Super Script IV Reverse Transcriptase (Invitrogen) and random hexamers (100 ng.μL −1 , Applied Biosystems) according to the manufacturer's instructions. The primers used to amplify cdi genes and cdi gene junctions are listed in S1 Table. All PCRs on cDNA were performed with GoTaq DNA polymerase (Promega) or iProof DNA Polymerase (BioRad), in accordance with the manufacturer's recommendations. The final PCR products were subjected to electrophoresis in 1% agarose gels in TAE, alongside the 1 kb DNA Ladder Plus (Euromedex).

Genomic analysis
We used the Genoscope microscopy platform (http://www.genoscope.cns.fr/agc/microscope/ home/) to identify the tpsAB loci present in available Xenorhabdus and Photorhabdus complete genome sequences, by searching for protein sequences with the conserved NPNGI and NPNL motifs; the N-terminal TPS domain is a hallmark of TpsA proteins.
We used BlastP with default parameters to search for orthologs of the CdiB/CdiC/CdiA/ CdiI proteins in other bacteria present in the NCBI non-redundant protein database.
The strains used in phylogenetic analysis and their accession numbers are shown in S2 Table. Results

Xd FRM16 harbors a new cdi locus
A putative cdi locus was identified in the genome of Xenorhabdus doucetiae strain FRM16 (Xd_FRM16). It contains four ORFs, XDD1_1117 to XDD1_1120 ( Fig 1A). XDD1_1117 encodes a homolog of the outer membrane protein CdiB (29% and 36% identity to the CdiB of E. coli 536 and Burkholderia thailandensis E264, respectively). XDD1_1118 encodes a homolog of acyltransferase proteins generally described as toxin-activating protein C (34% identity to HlyC of E. coli strain PM152 [36]). For this reason, we named the ORF XDD1_1118 CdiC FRM16 . XDD1_1119 is similar to CdiA (36% and 40% identity to the sequences of CdiA of E. coli 536 and B. thailandensis E264, respectively) and encodes the VENN motif separating the conserved N-terminus (~4 100 aa) from the variable C-terminus (~300 aa) in many CdiA proteins ( Fig 1A). This carboxy-terminal region is referred to below as CdiA-CT FRM16 . Finally, XDD1_1120 encodes a putative protein of unknown function. Its location just after the cdiA gene suggests that it may encode the CdiI immunity protein.
The cdi locus of Xd_FRM16 is located within a previously described integrative and conjugative element (ICE) [23] (Fig 1B) harbouring the remnant of a pilus synthesis locus (Fig 1B), but it contains the essential machinery for mobilization [23]. The ICE is embedded in a large genomic island (GI) inserted between the fba and mltC genes of the core genome (Fig 1B and S3  Table for genetic content). This GI contains several genes or gene clusters potentially involved in host interactions: i) the mcf gene, which encodes a toxin active against caterpillars [37], ii) a paa-like locus encoding enzymes of the phenylacetic acid catabolic pathway required for the oral pathogenicity of Burkholderia cenocepacia in Caenorhabditis elegans [38], iii) an iol-like locus encoding enzymes involved in myoinositol catabolism, the products of which have been implicated in the pathogenicity of the human fungal pathogen Cryptococcus neoformans [39].

Growth inhibition function of the CdiA-CT FRM16 fragment and identification of the CdiI protein
The region encoding the CdiA-CT FRM16 polypeptide (residues 4167 to 4436) was inserted into plasmid pGJ907 under the control of the P tet promoter, which is inducible by adding aTc (anhydrotetracycline) to the culture medium [41]. When introduced into Xenorhabdus bovienii strain CS03, which does not possess the cdi locus [42], the resulting plasmid, pGJ907_cdiA-CT FRM16 , conferred a growth inhibition phenotype to the bacterium cultured in LB broth, upon aTc induction (Fig 2A). Similar results were obtained with Escherichia coli EPI400 ( Fig  2B). The CdiA-CT FRM16 fragment therefore confers a capacity to inhibit the growth of both closely related (X. bovienii) and more distantly related (E. coli) species.
This growth inhibition was completely blocked in E. coli/pGJ907_cdiA-CT FRM16 following co-expression of the XDD1_1120 gene within pUC18, under the control of the P lac promoter (Fig 2C). These results confirm that XDD1_1120 confers immunity to CdiA-CT FRM16 mediated growth inhibition. We therefore named the XDD1_1120-encoded protein CdiI FRM16 . By The cdi locus is shown in blue. The conserved genes of the ICE, as defined in a previous study [40], are highlighted in green. Notable cargo genes (e.g. potentially involved in host interactions) are highlighted in red. The label numbers of the genes are indicated above the locus. The ICE is embedded in a large genomic island inserted between the fba and mltC genes (black boxes). The genetic content of the genomic island is described in S3 Table. doi: 10 contrast, the expression of cdiC FRM16 gene (XDD1_1118) does not display significant consequence on growth inhibition caused by cdiA-CT FRM16 (Fig 2C).

CdiA-CT FRM16 displays DNAse activity
The most closely related ortholog of CdiA-CT FRM16 is CdiA-CT from Dickeya dadantii 3937 (67% amino-acid identity), an Mg 2+ -dependent DNase [27]. We assessed the DNase activity of CdiA-CT FRM16 , by using the pET28b vector to overproduce (i) the CdiA-CT FRM16 fragment CdiA-CT FRM16 -mediated growth inhibition in E. coli EPI400. E. coli EPI400 carrying pGJ907 (black lines) or pGJ907_cdiA-CT FRM16 (red lines) was grown at 37˚C in LB broth supplemented with kanamycin (dotted curves), or LB broth supplemented with kanamycin plus aTc (continuous curves). C. XDD1-1120 confers immunity to CdiA-CT FRM16 toxicity in E. coli EPI400. E. coli carrying pGJ907 and pUC18 (black line), pGJ907_cdiA-CT FRM16 and pUC18 (red line), pGJ907_cdiA-CT FRM16 and pUC18_ XD1120 (blue line), pGJ907_cdiA-CT FRM16 and pUC18_ XD1118 (green line) were grown at 37˚C in LB broth supplemented with kanamycin. The times when IPTG and aTc were added at the culture are indicated by an arrow. In each panel, optical density at 600 nm was recorded every 30 minutes. When required, aTc was added 2 hours after the start of culture (OD 600 nm~0 .15). The results shown are the mean and standard deviation of three experiments. and the immunity protein CdiI FRM16 (Fig 3A) or (ii) the immunity protein CdiI FRM16 alone. His6-tagged CdiI FRM16 was purified by Ni-NTA chromatography under denaturing conditions. The CdiA-CT FRM16 toxin was purified together with its cognate His6-tagged CdiI FRM16 immunity protein, to prevent autotoxicity in the E. coli strain overproducing it. It was then separated from the immunity protein by SDS-PAGE under denaturing conditions. The purified CdiA-CT FRM16 and CdiI FRM16 polypeptides had observed sizes of 30 and 18 kDa, respectively (Fig 3A), consistent with the predicted sizes of the recombinant products. Purified recombinant protein yield was~500 mg.L -1 for CdiA-CT FRM16 and~100 mg.L -1 for CdiI FRM16 . We underwent pull-down experiment to confirm that the over-produced CdiI FRM16 immunity protein purified alone binds specifically to cognate over-produced CdiA-CT FRM16 protein (Fig 3B).
We then assessed the nuclease activity of the purified CdiA-CT FRM16 fragment in vitro ( Fig  3C). Xd_FRM16 genomic DNA was completely degraded following incubation with 0.5 μg of purified CdiA-CT FRM16 , in the presence of the bivalent cations Mg 2+ and Ca 2+ , or Mg 2+ alone (lanes 1 and 4, Fig 3C). CdiA-CT FRM16 also displayed DNase activity against genomic DNA from different Xenorhabdus and Photorhabdus species, supercoiled plasmid DNA and eukaryotic genomic DNA (data not shown). In the absence of Mg 2+ ions, little or no nuclease activity was observed in the absence or presence of Ca 2+ cations (lanes 2 and 3, Fig 3C). The CdiI FRM16 polypeptide had no DNase activity (lane 5, Fig 3C). We then evaluated the impact on CdiA-CT FRM16 DNase activity of increasing concentrations of CdiI FRM16 by pre-incubation of the two polypeptides. Protection against DNase activity was observed for a CdiA-CT FRM16 / CdiI FRM16 ratio greater than two (lanes 6 to lanes 9, Fig 3C). Thus, CdiA-CT FRM16 displays Mg 2+ -dependent DNase activity that is neutralized by its cognate CdiI FRM16 immunity protein.

cdiB, cdiC, cdiA and cdiI belong to the same transcriptional unit
We investigated whether the cdi locus could be considered to function as an operon, by mapping the RNAs of the cdi genes by PCR (see Fig 4A for location of the primers). We checked that RNA is free of DNA by absence of 16S rDNA gene amplification (data not shown). We then showed that the four genes were transcribed at different time points during bacterial growth in LB (Fig 4B). We finally mapped the cdi FRM16 cDNA with primers allowing the amplification of nucleotides 400 of cdiB through 2779 of cdiA (4641-bp fragment flanked by the primers 5F and 5R) or of nucleotides 400 of cdiB through 6568 of cdiA (8430-bp fragment flanked by the primers 5F and 6R), and the amplification of nucleotides 10184 of cdiA through 287 of cdiI (3430-bp fragment flanked by the primers 6F and 7R) or of nucleotides 5625 of cdiA through 287 of cdiI (7988-bp fragment flanked by the primers 7F and 7R) (Fig 4C). The cdiB-cdiC-cdiA and cdiA-cdiI genes were respectively co-transcribed, consistent with an operon structure for the cdi FRM16 locus.

The cdiBCAI locus is widespread in environmental bacteria interacting with insects, plants, rhizospheres and soils
We investigated the frequency of occurrence of the cdiBCAI-type locus in entomopathogenic bacteria, by searching for orthologs of genes encoding the CdiA, CdiB, CdiC and CdiI proteins in 41 Xenorhabdus and Photorhabdus available genomes. cdi-like loci are frequently redundant in all Photorhabdus genomes. However, a cdiBCAI-type locus was detected only in the P. luminescens BA1 genome. By contrast, cdi-like loci are present in only~30% of Xenorhabdus genomes, but a majority of Xenorhabdus cdi-like loci were of cdiBCAI type. The cdiBCAI-type loci encountered in the genomes of Xenorhabdus and P. luminescens BA1 are described in Table 1.  4 and 5, respectively). The bound protein fraction (lane 6) was eluted using imidazole. All fractions were run on SDS-polyacrylamide gels and stained with Coomassie blue. M, protein marker (PageRuler plus prestained protein ladder, Thermo scientific). C. CdiA-CT FRM16 has DNAse activity CdiA-CT FRM16 was assayed for DNase activity with bacterial genomic DNA from Xd_FRM16. DNA was visualized by electrophoresis in a 0.7% agarose We then expanded our biodiversity survey by searching for proteins orthologous to CdiC FRM16 , the hallmark of the cdiBCAI-type locus, in the NCBI public database. CdiC FRM16 orthologs are systematically encoded inside cdi-like loci (S4 Table). Most of the CdiC orthologs were associated with (i) entomopathogenic bacteria, such as Xenorhabdus and Serratia, (ii) bacteria interacting with plants, such as Pseudomonas syringae and D. dadantii or (iii) soil and rhizosphere bacteria, such as Pseudomonas fluorescens, iv) and numerous pathogenic enterobacteria, e.g. E. coli strains.
Phylogenetic analysis of a representative set of bacterial species clustered the CdiC sequences in two clades (Fig 5). Clade I contained CdiC sequences from Enterobacteriaceae, including environmental strains such as insects pathogens (Xenorhabdus strains, P. luminescens strain BA1 and Serratia) and plant pathogens (Dickeya, Pantoea and Pectobacterium), whereas clade II contained CdiC sequences from environmental proteobacteria of the Pseudomonadaceae and Burkholderiaceae families. Some features suggested probable horizontal genetic transfers of the cdiC gene between closely related species. For example, the CdiC P. luminescens BA1 sequence clustered with CdiC FRM16 (Fig 5). Moreover, the CdiC proteins of X. doucetiae and X. cabanillasii (locus 1) were 100% identical, despite the distant relationship of these two species within the genus Xenorhabdus.

Evolutionary history of cdiBCAI genes
We analyzed the co-evolution of the Cdi proteins by comparing their phylogeny. As previously described for CdiC, the sequences of the CdiB, CdiA and CdiI proteins were retrieved by BlastP searches of the NCBI public database with CdiA FRM16 , CdiB FRM16 and CdiI FRM16 as queries (S4 Table). The phylogenetic trees for the CdiB, CdiA and CdiC proteins were congruent, suggesting co-evolution of the three genes (S1 Fig). The CdiB and CdiA sequences from "E. coli-type" and "Burkholderia-type" cdi loci, used as references, branched outside these trees, confirming that the cdiBCAI-type locus was phylogenetically distinct from these previously characterized CDI systems. By contrast, the phylogenetic tree of CdiI (Fig 6) differed from those for CdiB, CdiA and CdiC and from bacterial species phylogeny.
Finally, we found no genes encoding orthologs of CdiI FRM16 in the other Xenorhabdus genomes. The closest CdiI FRM16 orthologs were encoded by genes in the Photorhabdus genus (P. asymbiotica ATCC43949, P. heterorhabditis VMG, Photorhabdus temperata M1021 and numerous strains of P. luminescens), in Pseudomonas syringae strains, and, more surprisingly, in a few strains of Gilliamella apicola, a gut symbiont of honey bees (Fig 6).

Discussion
CDI systems may play a key role in competition strategies, by delivering toxins that kill neighbouring bacteria, thereby eliminating bacterial competitors. Known CDI systems are encoded by two different classes of genomic loci: those of the ''E. coli-type" generally found in Enterobacteriaceae [27], and those of the ''Burkholderia-type" found in the genera Burkholderia, Cupriavidus, Ralstonia [32,33]. In this study, we used Xenorhabdus genomic resources to identify a new type of cdi locus, the cdiBCAI-type locus, in the Xenorhabdus doucetiae FRM16 gel. Lanes 1 to 4: The purified CdiA-CT FRM16 protein and genomic DNA were incubated for 1 hour at 37˚C with or without CaCl 2 (2 mM) and MgCl 2 (2 mM). Lane 5: The CdiI FRM16 protein purified alone and genomic DNA were incubated for 1 hour at 37˚C with CaCl 2 (2 mM) and MgCl 2 (2 mM). Lanes 6 to 9: The purified CdiA-CT FRM16 and CdiI FRM16 proteins were first incubated for 30 minutes at 37˚C in the presence of various concentrations of CdiI FRM16 . Genomic DNA was then added and the mixture was incubated for 1 hour at 37˚C with CaCl 2 (2 mM) and MgCl 2 (2 mM). Lane M: 1-kb DNA ladder (Eurogentec).
doi:10.1371/journal.pone.0167443.g003 Xenorhabdus  Fig 4. Xd_FRM16 cdiC, cdiB and cdiA are transcribed as a single transcription unit. A. Position of the primers used in RT-PCR analyses. The cdi genes are indicated by large arrows, the size of which is proportional to gene length. The positions of the primers are indicated by thin black arrows for internal PCRs and thin red arrows for junction PCRs (see S1 Table for a description of the primers). B. cdiB, cdiC, cdiA and cdiI were transcribed at different time points during bacterial growth in LB broth. cDNAs were synthesized from total RNA extracted from cultures of X. doucetiae FRM16 in LB broth at different incubation times: 1 (OD at 600 nm = 0.75), 2 (OD at 600 nm = 1.70), 3 (OD at 600 nm = 3.70). 50 ng of cDNA were then amplified with specific primers targeting the four genes of the X. doucetiae FRM16 cdi locus (see panel A and S1 Table). +,~70 ng of genomic DNA from X. doucetiae FRM16; -, 1 μL of water; M, 5 μL of 1 kb DNA Ladder Plus (Euromedex). The PCR products were subjected to electrophoresis in 1% agarose gels. C. cdiB, cdiC, cdiA and cdiI genes form a single operon. Specific primers allowing the mapping of the cdiB-cdiC-cdiA and the cdiA-cdiI junctions were used (see panel A and S1 Table).~50 ng of cDNA of the condition 1 (panel B) were used. +,~70 ng of genomic DNA from X. doucetiae FRM16; -, 1 μL of water; M, 5 μL of 1 kb DNA Ladder Plus (Euromedex). genome. This locus is characterized by the presence of an additional ORF, located between cdiB and cdiA that we named cdiC, due to the orthology of its product with toxin-activating proteins C. Although the presence of such an accessory gene in cdi loci has been reported in some Gram-negative strains [43], it was never exhaustively investigated. A more exhaustive search of cdiBCAI-type loci in a bacterial database showed that the cdiBCAI type locus was frequently present in Xenorhabdus genomes (~30% of the sequenced genomes) and many environmental bacteria interacting with insects, plants, rhizosphere and soil. Previous studies have suggested that cdi loci probably undergo horizontal gene transfer (HGT) between bacteria, because they are often present in genomic and pathogenicity islands [26,28]. The Xd_FRM16 cdi locus is the first example found to be located within an ICE, a class of mobile genetic elements known to mediate HGT [44,45]. The CdiA toxin module of X. doucetiae FRM16, CdiA-CT FRM16 , has many features in common with other described CdiA-CT. It is flanked at its C-terminal end by a VENN motif, and it is toxic when produced within the cells of closely or more distantly related bacteria [26]. CdiA-CT FRM16 is an Mg 2+ -dependent DNase, as shown for its ortholog CdiA-CT in D. dadantii strain 3937 [27]. Several other CdiA-CTs display nuclease activity in vitro, probably killing their target bacteria [27,33,46]. The CdiA-CT of E. coli EC869 has DNase activity, but its amino-acid sequence is very different from that of CdiA-CT FRM16 (30% identity) and its nuclease activity is Zn 2+ -dependent [47,48]. This finding may reflect highly convergent evolution for the nuclease activity of CdiA-CT fragments.

Contact-Dependent Inhibition in
We found that cdiB FRM16 , cdiC FRM16 , cdiA FRM16 and cdiI FRM16 were cotranscribed as an operon in both the exponential and the stationary phases of growth in LB culture. Moreover, we performed phylogenetic analysis, which displays that cdiB FRM16 , cdiC FRM16 and cdiA FRM16 genes co-evolved. We therefore hypothesize that the product of the cdiC gene is involved in the functional CDI system. Due to the similarity of the CdiC protein with HlyC [36], CdiC FRM16 might play a role in activation of CDI through lysine acetylation. In our growth inhibition assay in E. coli, we did not observe any effect of the cdiC FRM16 expression on CdiA-CT FRM16 activity. Some authors suggest that CdiC activity may promote CdiA or CdiB association with membrane [49]. Indeed, CdiC FRM16 may play an important role in membrane-associated CDI functions, such as CdiBA biogenesis, CdiA target cell binding or toxin translocation through the cell envelope.
We identified and characterized the function of CdiI FRM16 in protecting the bacteria producing it from the toxic activity of CdiA-CT FRM16 . Our phylogenetic analysis suggested that cdiBCA and cdiI were different evolutionary units. Ruhe and coworkers previously observed that, in many CDI toxin-immunity protein pairs, the CdiI proteins display much greater sequence diversity than the CdiA-CT toxins [26]. They suggest that cdiI evolution is rapid as long as CdiI maintains sufficient affinity for CdiA-CT to provide immunity [26]. Investigation of the genetic context of cdiI FRM16 orthologs in available complete genomes showed that they are either orphan genes or participate in orphan cdiA-CT-cdiI pairs (J-C. Ogier, unpublished data). These orphan cdiI genes or cdiA-CT-cdiI pairs may be horizontally exchanged between bacteria, potentially enabling bacteria to protect themselves against neighbouring bacteria that are producing and delivering CdiA-CT toxins. Interestingly, no cdiI FRM16 orthologs were found in Xenorhabdus genomes, and the closest cdiI FRM16 orthologs, associated or unassociated with cdiA-CT, were mostly found in Photorhabdus strains, a genus having a life cycle similar to that of Xenorhabdus, including a major stage in the insect cadaver [22]. below the phylogenetic tree reflects the numbers of amino-acid substitutions per site. The protein sequences are split into three clades, I, II and III, indicated at the right of the tree. Accession numbers of the sequences are indicated in S2 Table. doi:10.1371/journal.pone.0167443.g006

Contact-Dependent Inhibition in Xenorhabdus
Several attempts for assessing role of the in vivo activity of the cdiBCAI locus were conducted between X. doucetiae FRM16 and strains of Xenorhabdus that naturally do not harbor any cdi loci. However, no clear effect could be detected (data not shown). This is likely due to the highly redundant arsenal of Xenorhabdus factors involved in antagonism between bacteria [4]. For example, a locus encoding a tail-phage bacteriocin is responsible for eliminating phylogenetically close bacteria from the insect cadaver [15].
In conclusion, Xenorhabdus genus is a pertinent resource for identifying genomic loci potentially involved in interbacterial competition systems, such as cdi loci. In order to determine the specificities of the cdiBCAI-type locus, further studies should be conducted.
Supporting Information S1  The phylogenetic trees were built by the maximum likelihood (ML) method, and branch support values (estimated by the aLRT(SH-like) method) are indicated at the nodes. The branch length scale bar below the phylogenetic tree reflects the number of amino-acid substitutions per site. The Xd_FRM16 sequences are highlighted in blue. For some taxa, orthologs are highlighted in red when the topology is not congruent with the species tree. The sequences from "E. coli-type" and "Burkholderia-type" cdi loci are used as outgroups, and are highlighted in bold. Accession numbers of the sequences are indicated in S2 Table. (PPTX)