Genomic and functional characterisation of IncX3 plasmids encoding blaSHV-12 in Escherichia coli from human and animal origin

The blaSHV-12 β-lactamase gene is one of the most prevalent genes conferring resistance to extended-spectrum β-lactams in Enterobacteriaceae disseminating within and between reservoirs, mostly via plasmid-mediated horizontal gene transfer. Yet, studies regarding the biology of plasmids encoding blaSHV-12 are very limited. In this study, we revealed the emergence of IncX3 plasmids alongside IncI1α/γ in blaSHV-12 in animal-related Escherichia coli isolates. Four representative blaSHV-12-encoding IncX3 plasmids were selected for genome sequencing and further genetic and functional characterization. We report here the first complete sequences of IncX3 plasmids of animal origin and show that IncX3 plasmids exhibit remarkable synteny in their backbone, while the major differences lie in their blaSHV-12-flanking region. Our findings indicate that plasmids of this subgroup are conjugative and highly stable, while they exert no fitness cost on their bacterial host. These favourable features might have contributed to the emergence of IncX3 amongst SHV-12-producing E. coli in the Netherlands, highlighting the epidemic potential of these plasmids.

IncX3 plasmid backbone is highly syntenic and conserved. Comparison between whole sequences of the four IncX3 plasmids from this study and twenty IncX3 plasmids available in GenBank revealed a highly conserved plasmid backbone and their organization into a number of distinct clades (Fig. 2). Plasmids of animal origin pEC-125 and pEC-243 were closely clustered together (MUMi distance 0.018) and grouped with the human-derived plasmid pEC-NRS18 that showed distance from 0.085 (pEC-125) to 0.093 (pEC-243). The animal-derived plasmid pEC-393 (turkey meat) clustered with a Klebsiella pneumoniae-encoded pIncX-SHV from human source in Italy (MUMi distance 0.002). IncX3 plasmids recovered in the Netherlands clustered closely with pOXA181 (China) and pKS22 (Switzerland) encoding bla OXA-181 and qnrS1, respectively, with MUMi distances varying from 0.140 (pOXA181 with pEC-125) to 0.179 (pKS22with pEC-393).
The variable region of pEC-NRS18 contained bla TEM-1 embedded in a Tn3 transposon, as well as genes bla SHV-12 and qnrS1 associated with the upstream presence of IS26 in the opposite and same orientation, respectively. Similarly, pEC-125 and pEC-243 contained both bla SHV-12 and qnrS1 genes, whereas pEC-393 encoded only bla SHV-12 associated to IS26. The genetic environment surrounding bla SHV-12 was characterized by two flanking copies of IS26 distributed in opposite orientation to form a composite IS26-IS26 transposon (3,633 bp); this structure was conserved in three of the four IncX3 plasmids (pEC-NRS18, pEC-125 and pEC-243; Fig. 3). BLAST analysis revealed that this composite transposon is 100% identical to previously described transposons located on plasmids of K. pneumoniae (IncFIB; GenBank accession no. CP019048.1) and Aeromonas veronii (IncA/C2; GenBank accession no. CP014775.1), as well as into the genome of Pseudomonas aeruginosa (GenBank accession no. GU592828.1). The flanking region of bla SHV-12 on pEC-393 showed a partial 2,484-bp overlap with corresponding regions of pEC-NRS18, pEC-125 and pEC-243 (Fig. 3) and encoded genes deoR, ygbJ and truncated ygbK also present on several K. pneumoniae chromosomes (GenBank accession no. CP000647, CP002910 and CP008831). This 4,783-bp region exhibits 100% identity to a fragment of a bla SHV -harboring IncR plasmid of K. pneumoniae (GenBank accession no. KF954150.1).
IncX3 plasmids do not contribute to bacterial pathogenicity. Annotation of the four IncX3 plasmids revealed the presence of a Type 4 Secretion System (pilX1-11) and several ORFs with unknown function that could potentially act as virulence effectors (Fig. 3). The Galleria mellonella in vivo infection model was employed to evaluate the impact of harbouring an IncX3 plasmid on bacterial pathogenicity. The LD 50 value after 24 h was determined to be 10 7 CFU/larvae and survival curves were compared between the isogenic control E. coli DH10b strain and DH10b transformed strains harbouring each of the IncX3 plasmids (Fig. 6). All four transformed strains carrying IncX3 plasmids displayed comparable virulence to the plasmid-free control strain (mortality = 40-86%), with no significant difference in the 96 h survival curves (Fig. 6). In both control groups all larvae survived.

Discussion
Our results confirmed that IncI1α/γ plasmids are the major facilitators of the bla SHV-12 diffusion in E. coli of human and animal origin and mirrored the global plasmid repertoire associated with bla SHV-12 37,39 . However, a gradual decrease in the prevalence of IncI1α/γ and a parallel increase in IncX3 plasmids encoding bla SHV-12 was documented mostly in animal-related commensal E. coli. Previously, the IncX3 plasmid subgroup was only incidentally associated with bla SHV-12 9,16-19 and/or qnrS1 2,9,12-15 among clinically recovered E. coli isolates, and very recently it was identified among poultry isolates in Germany 40    Nevertheless, no association between IncX3 plasmids and other resistance genes (apart from bla SHV-12, bla TEM-1 and qnrS1) in Enterobacteriaceae was found in the Netherlands (data not shown).
The high degree of synteny and conservation in the backbone of IncX3 plasmids among E. coli isolates of both human and animal origin reflects the ecological success of this plasmid subgroup 2 . In addition to encoding genes essential for their maintenance and dissemination, IncX3 plasmids contained a variable region encoding resistance to clinically important antimicrobial agents (fluoroquinolones and/or extended spectrum cephalosporins). Our findings confirm the potential of this subgroup for accumulation of resistance genes via IS-mediated transposition, with the likely consequence of limiting effective treatment options for possible human infections 9,13,15,16,18,21 . As previously described 41 , the presence of IS26 linked to bla SHV-12 and other co-linear genes originating from the chromosome of K. pneumoniae was documented within the IncX3 variable region, confirming the hypothesis of IS26-mediated mobilization of a bla SHV ancestor gene from the chromosome of K. pneumoniae 42 . Although we cannot prove how bla SHV-12 was integrated into IncX3 plasmids, the documented ability of IS26 to participate in both replicative transposition and self-targeted transposition creating IS26-bounded transposons 43 could have facilitated the formation of the composite bla SHV-12 -encoding IS26-IS26 transposon seen in the majority of the IncX3 plasmids studied here. The presence of identical composite transposons, mostly on plasmids of diverse replicon types, indicates its mobilization ability preferentially onto plasmids rather than the chromosome, as previously suggested 44 . In addition to the contribution of IS26 to the mobilization on conjugative plasmids and the subsequent dissemination of bla SHV-12 , it has been shown that IS26 supplies a promoter −35 box that can be coupled with a −10 box in the adjacent DNA 45 , possibly also contributing to the expression of this resistance gene.
IncX3 plasmids, as well as the archetypal R6K plasmid of the IncX family, have been investigated for their ability to conjugate 13,15,16,18,19,46 . We documented the temperature-dependent transfer of animal-derived IncX3 plasmids with frequencies higher at 30 °C, suggesting a more efficient transfer in the environment than in the animal gut. Similarly to other conjugative plasmids, IncX3 plasmids hold a gene encoding a H-NS-like protein 2,[47][48][49][50][51] . Several studies demonstrated the inhibitory role of H-NS-like proteins on gene thermoregulation owing to their ability to polymerize along and bridge adjacent DNA regions at 37 °C and to derepress H-NS-regulated genes at lower temperatures 47,48,52 . The involvement of these proteins in the temperature-dependent conjugative transfer of  IncHI1 plasmids suggests a similar role for IncX3 plasmids that can only be speculated here. The ability of IncX3 plasmids to replicate and be stably maintained in α-, βand γ-Proteobacteria 53 , in combination with a higher conjugation frequency at 30 °C, underscore a potential mesophilic Proteobacteria reservoir of this plasmid subgroup.
In contrast with studies describing that plasmids impose fitness cost on their bacterial hosts 54 , growth kinetics obtained over 24 h showed no evidence of fitness cost on the bacterial E. coli host. It has been demonstrated that H-NS-like proteins are able to silence newly introduced foreign sequences (including plasmids), based on increased adenine and thymine (AT) content in comparison with the chromosome 51,55,56 . Taking into consideration the high AT content of the IncX3 plasmids, we hypothesize that the H-NS gene present on these plasmids allows them to invade bacterial hosts with a minimal impact on their fitness, ensuring the future competitiveness of the new plasmid-host combination even without the presence of antibiotic selective pressure. The significant IncX3 plasmid-mediated fitness enhancement of E. coli under antibiotic selective pressure highlights the ecological advantage and subsequent successful proliferation of these plasmids in antibiotic-rich reservoirs.
All IncX3 plasmids encoded the widespread partitioning system ParAB ensuring the correct inheritance of these plasmids to the daughter cells 57 . The observed high stability of IncX3 plasmids is potentially due to their high conjugation frequency and absence of a fitness burden, as well as to a low rate of segregational loss.
Our data show that IncX3 plasmids encode a type IV secretion system (T4SS, pilX1-11), typically used for the exchange of genetic material within bacteria, toxin secretion and the translocation of virulent effector proteins into eukaryotic host cells 58,59 . Yet, a virulence potential of IncX3-carrying E. coli DH10b was not observed, suggesting that T4SS does not play a role in the virulence of E. coli, at least in the G. mellonella model, conversely from other Gram-negative pathogens 58 .
In conclusion, we report the first genetic characterisation of IncX3 plasmids of animal origin, as well as the first functional analysis of human-and animal-derived plasmids of this subgroup, including their conjugation frequencies, stability, fitness cost and virulence potential. IncX3 plasmids are highly conserved, syntenic, conjugative and highly stable, while they exert no fitness cost on their bacterial host, independent of their origin. Although clonal expansion of E. coli strains could also play a role as suggested by our finding in The Netherlands of E. coli from the same clonal complex and carrying the same IncX3 plasmid (data not shown), the favourable plasmid functional features potentially contributed to their emergence amongst SHV-12-producing E. coli in the Netherlands, highlighting the epidemic potential of this plasmid subgroup.

Materials and Methods
Bacterial strains, transformants and plasmids. A total of 129 non duplicate bla SHV-12 encoding E. coli, consecutively recovered during national antimicrobial resistance monitoring programmes or national projects between 2009 and 2014 were included in the study 60 . Identification of the isolates was performed by MALDI-TOF Mass spectrometry (Brucker, Coventry, UK). Genes conferring an ESC R phenotype were sought by microarray analysis followed by PCR amplification and sequencing 61 . Plasmid location of bla SHV-12 was determined using a transformation-based approach. Briefly, plasmids encoding bla SHV-12 were extracted from the parental strain using a miniprep method and transformed into E. coli DH10b cells (Invitrogen, Van Allen Way, CA USA) by electroporation under the following conditions: 1.25 kV/cm, 200 Ω, 25 μFar, as previously described 61 . Transformants were selected on Luria-Bertani (LB) agar plates supplemented with cefotaxime (1 mg/L) and confirmed for the presence of bla SHV-12 gene. The presence of a single plasmid in the transformants was confirmed by S1-PFGE on the transformants, followed by Southern blot hybridization using DIG-labelled probe (DIG DNA Labeling and Detection Kit, Roche, Mannheim, Germany) targeting the bla SHV-12 gene, as previously described 62 . Plasmid typing was confirmed by PCR-based replicon typing (PBRT KIT, DIATHEVA, Fano, Italy) on the transformants. Host E. coli sequence type were assigned by MLST based on the allelic profiles of seven housekeeping genes (adk, fumC, gyrB, icd, mdh, purA and recA) 63 .
Plasmid sequencing, assembly and analysis. Four genetically and epidemiologically unrelated IncX3 plasmids encoding bla SHV-12 were randomly selected for further analysis from E. coli isolates belonging to the single ST of human origin (ST69) and diverse STs of animal origin, including the predominant animal-related ST (ST117). The relevant characteristics of the selected plasmids are specified in Table 2. Plasmid DNA from transformants was isolated using the QIAfilter Plasmid Midi Kit (QIAGEN, Hilden, Germany) according to the manufacturer's recommendations. Deep sequencing of the plasmid genomes was performed using 300-bp paired-end sequencing libraries (Nextera TAG-mentation sequencing kits [Epicentre]) on an Illumina MiSeq sequencer. High-quality filtered reads were subsequently assembled de novo using SPAdes algorithm (SPAdes version 3.7.1) for Illumina-derived reads and then manually curated to close the gaps. Putative open reading frames (ORFs) were identified by RAST version 2.0 and manually curated when necessary 64 . BLASTP analyses of the putative ORFs against the NCBI non-redundant proteins (NR) database, Pfam, and Interpro scan were used to assess their putative functions by identification of structural features and motifs 65,66 . ResFinder (version 2.1), PlasmidFinder (version 1.3) and ISfinder were used to determine the presence of resistance genes, replicon types and insertion sequences, respectively [67][68][69] . Plasmid sequences were hierarchically clustered and displayed as a phenogram using the BioNJ algorithm, where the underlying distance matrix was calculated from the pairwise non-overlapping maximal unique matches (MUMs) using Nucmer version 3.07 70,71 . Relative pairwise distances were obtained by dividing the pairwise MUMs' sum by the average genome size of the two paired genomes (MUMi genomic distance) 72 . BioNJ trees were generated from the MUMi distance matrix using SplitsTree4 73 . BLAST analysis was used to assess sequence identity between the bla SHV-12 -surrounding region and nucleotide sequences deposited to NCBI 74 .
Mating assays. Plasmid  with donor E. coli DH10b transformed strains carrying the different IncX3 plasmids, as previously described 75 . Overnight cultures of recipient and donor strains in mid-exponential phase were co-incubated (100 μl each) onto sterile nitrocellulose filters of 0.45 μm pore size (Schleicher and Schuell GmbH, Dassel, Germany) for 4 h at 25 °C, 30 °C and 37 °C. Transconjugants were selected on LB agar supplemented with chloramphenicol (32 mg/L) and cefotaxime (1 mg/L). Positive transconjugants were confirmed by PCR amplification for the resistance and yfp genes. Conjugation frequency was calculated as the number of transconjugants per donor cell, with the absence of transconjugants suggesting either non-conjugative plasmids or conjugation frequencies below the detection limit (≤1 × 10 −9 ). For statistical analysis, conjugation frequencies were transformed to log10 values, the differences between the temperatures and plasmids were tested using a non-parametric Kruskal-Wallis test, and a p value < 0.05 was considered to be statistically significant. All analysis was performed using R and RStudio (version 1.0.143) 76,77 . Fitness cost assays. Liquid cultures of E. coli DH10b transformed strains carrying different IncX3 plasmids were incubated overnight in 3 mL LB medium at 37 °C with 180 rpm shaking. Cultures were then diluted 100-fold into 3 mL of fresh pre-warmed LB medium with and without antibiotic (1 mg/L of cefotaxime) and incubated under the same conditions until mid-exponential phase (OD 600 of ≈0.5). 200 µL of each culture were loaded in triplicate in wells of a 100-well honeycomb plate and incubated at 37 °C with shaking for 24 h. Growth rates were obtained by measuring optical density at 600 nm every 30 min by using a Bioscreen C Reader (Oy Growth Curves, Helsinki, Finland). Assays were performed in triplicate. Relative growth rates were calculated by dividing the generation time of each DH10b transformed strain by the generation time of the wild-type DH10b strain which was included in each individual assay 78 . Growth rates between strains were compared using the Wilcoxon rank sum test with a Bonferroni adjustment for multiple comparisons. All statistical analysis were performed in R studio (version 1.0.143) 76 .
Stability assays. E. coli DH10b transformants carrying different IncX3 plasmids were propagated in antibiotic-free LB medium at 37 °C with 180 rpm shaking for 10 days (∼180 generations) to determine their stability in an E. coli population. Cultures of each strain were daily diluted 1000-fold into 3 mL of fresh pre-warmed LB medium without antibiotics. On day 10, cultures were plated onto antibiotic-free LB agar and 100 randomly chosen colonies of each evolved line were replica-plated onto antibiotic-free and antibiotic-containing (1 mg/L of cefotaxime) LB agar plates. Plasmid presence was confirmed by colony PCR targeting the taxC gene of the IncX3 plasmids 2 . Colony growth on antibiotic-free but not on antibiotic-containing plates indicated the proportion of bacteria that lost the plasmid. Assays were performed in triplicate. The chance of E. coli DH10b keeping the plasmid was estimated for every plasmid using @Risk 6.3.1 (Palisade Corporation, Newfield, NY, USA), and the proportions of plasmid-harbouring colonies for each plasmid were compared.
Galleria mellonella survival assays. G. mellonella caterpillars in the final-instar larval stage were obtained in bulk from Livefood UK Limited (Rooks Bridge, Somerset, United Kingdom) and stored at 15 °C in the dark on wood shavings prior to use. Ten randomly chosen larvae weighing between 250 mg and 350 mg were employed for each group of an experiment. Strains included in the assay were grown overnight in LB broth and washed twice in sterile phosphate-buffered saline (PBS). The optimal bacterial inoculum was determined by injecting 10 larvae with 10 μl of bacterial suspensions containing 10 4 to 10 7 CFU/larva of organism in PBS. Bacterial inoculum concentration was determined by viable bacterial count on LB agar identifying the inoculum which killed 50% of larvae after 24 hours incubation at 37 °C (LD 50 ). The optimal inoculum was then injected into the hemocoels of the caterpillars via a left proleg using 25-μl Hamilton syringes (Cole-Parmer, London, United Kingdom). Following injection, larvae were incubated in petri dishes lined with filter paper at 37 °C for 96 h and scored for survival by 2 independent observers daily. Larvae were considered dead when they displayed no movement in response to touch. Two control groups were used per experiment, including larvae that were inoculated with PBS to control for any lethal effects of the injection process and larvae that received no injection. All G. mellonella survival assays were performed in triplicate using different batches of larvae. Survival curves were plotted using the Kaplan-Meier method and differences in survival were calculated by the log-rank test using R studio (version 1.0.143) 76 .
Data availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information files). Accession codes. The reported plasmid sequences are deposited in GenBank under the following accession numbers: KX618696 (pEC-NRS18), KX618697 (pEC-393), KX618698 (pEC-243) and KX618703 (pEC-125).