Identiﬁcation and Characterisation of pST1023 A Mosaic, Multidrug-Resistant and Mobilisable IncR Plasmid

: We report the identiﬁcation and characterisation of a mosaic, multidrug-resistant and mobilisable IncR plasmid (pST1023) detected in Salmonella ST1023, a monophasic variant 4,[5],12:i: strain of widespread pandemic lineage, reported as a Southern European clone. pST1023 contains exogenous DNA regions, principally gained from pSLT-derivatives and IncI1 plasmids. Acquisition from IncI1 included oriT and nikAB and these conferred the ability to be mobilisable in the presence of a helper plasmid, as we demonstrated with the conjugative plasmids pST1007-1D (IncFII) or pVC1035 (IncC). A sul3 -associated class 1 integron, conferring resistance to aminoglycosides, chloramphenicol and trimethoprim-sulphonamides, was also embedded in the acquired IncI1 DNA segment. pST1023 also harboured an additional site-speciﬁc recombination system ( rfsF / rsdB ) and IS elements of the IS 1 , IS 5 (IS 903 group) and IS 6 families. Four of the six IS 26 elements present constituted two pseudo-compound-transposons, named PCT- sil and PCT-Tn 10 (identiﬁed here for the ﬁrst time). The study further highlighted the mosaic genetic architecture and the clinical importance of IncR plasmids. Moreover, it provides the ﬁrst experimental data on the ability of IncR plasmids to be mobilised and their potential role in the horizontal spread of antimicrobial-resistant genes.


Introduction
Plasmids are self-replicating, extra-chromosome genetic elements and are considered an important driving force of bacteria evolution as they contribute towards generating genetic variability and also simply provide selective advantages, such as antimicrobial resistance [1]. The latter is of great importance since resistance to most available antimicrobial classes has been recognised as an emerging health problem on a worldwide scale. Indeed, the global increase in multidrug-resistant (MDR) bacteria has drastically been reducing the range of antimicrobials available to treat bacterial infections. This has made antimicrobial resistance to bacteria the major cause of death worldwide, as recently reported by a comprehensive survey on this, covering over 204 countries and territories and published in The Lancet journal [2].
Antimicrobial resistance genes can be embedded within genetic elements, such as transposons, compound transposons and integrons (mainly of class 1), that are often carried by plasmids (particularly in Entebacterales), which, in turn, greatly contribute to antimicrobial resistance spreading and the insurgence of multidrug-resistant bacteria [3,4].
Indeed, many bacteria genomes contain multiple plasmids whose persistence is achieved by vertical transmission to daughter cells and (if conjugative) by transmission through cell-to-cell conjugation. The ability to be horizontally transferable represents an evolutive advantage in that it would further extend the host range, thus, increasing the general level of long-term persistence in bacteria population [5,6].
Plasmids can generally be classified into different types according to their replication (replicon type) or mobility (MOB typing) loci [7,8]. Plasmids sharing the same replication system are unable to co-exist stably within the same host cell and are clustered in the same replicon type (group of incompatibility or Inc group). MOB typing, based on relaxase protein phylogenies, allows classification of transmissible plasmids into MOB families. IncR is a relatively recent replicon type, first reported in 2009 and identified in pK245, a plasmid harboured by a multidrug-resistant Klebsiella pneumoniae strain [9]. IncR plasmids are not included in the MOB typing system as they do not contain a relaxase gene. Additionally, they do not possess conjugational transfer genes and, thus, are not conjugative. Since their first identification, IncR plasmids have been isolated the world over, mainly from clinical multidrug-resistant strains [3]. The IncR replication and maintenance systems are principally composed of repB (replication initiation) and its iterons, parAB (partition) and vagCD (toxin-antitoxin). In addition to their core backbone, IncR plasmids may carry various accessory modules, often conferring resistance to different classes of antimicrobials that extend the size of these up to 160 Kb [10].
In this study, we report the identification and characterisation of a mosaic, multidrugresistant and mobilisable IncR plasmid (pST1023) detected in the Salmonella enterica subsp. enterica serovar 4, [5],12:i: strain ST1023. ST1023 belongs to the widespread pandemic lineage reported as a Southern European clone [11] and the finding of a mobilisable IncR plasmid, experimentally demonstrated here for the first time, is of concern for its potential role in the spread of antimicrobial-resistance genes.

DNA Sequencing, Assembly and Annotation
Total genomic DNA was extracted by the cetyl trimethylammonium bromide method [16]. Plasmid DNA was isolated as described previously [17]. About 1 µg of DNA was fragmented by using the Ion Shear™ Plus Reagents Kit (Life Technologies, a part of Thermo Fisher Scientific Inc., Waltham, MA, USA), followed by barcoded adapter ligation using the Ion Xpress™ Barcode Adapters (Life Technologies, a part of Thermo Fisher Scientific Inc.) and Ion Plus Fragment Library Kit (Life Technologies, a part of Thermo Fisher Scientific Inc.) according to the manufacturer's protocol. The library size was selected (~400 bp) using E-Gel ® SizeSelect™ 2% Agarose Gel (Invitrogen, Carlsbad, CA, USA). Library concentrations were quantified using the Qubit dsDNA HS Assay Kit (Life Technologies, Waltham, MA, USA). Sequencing template was prepared by using the Ion 520 & 530 Kit-OT2 kit (Thermo Fisher Scientific Inc.) and then sequenced on an Ion 520 Chip using an Ion GeneStudio S5 System (Thermo Fisher Scientific Inc.). Raw data were quality filtered and assembled by using the SPAdes assembler version 3.15.4 [18] and the included pipeline plasmidSPAdes (-plasmid). pST1023 plasmid DNA was sequenced using MinION (Oxford Nanopore Technologies, Oxford, United Kingdom). Sequencing library was prepared using Rapid Sequencing Kit SQK-RAD004 (Life Technologies, a part of Thermo Fisher Scientific Inc) and 500ng of DNA following the manufacturer's instructions. The sequencing was performed using the R9.4.1 flongle flowcell FLO-FLG001 for 24h, according to the information provided by the manufacturer (https://store.nanoporetech.com/eu/flongle-flow-cell-pack.html, last access on 18 February 2022). The basecalling of the raw signals from the sequencing run was performed with Guppy v.5.0.11 (Oxford Nanopore Technologies) by the r9.4.1_450bps_hac model. Only the fastq files in the Guppy directory "pass", considered as high-quality reads, were used for the assembly. De novo genome assembly of basecalled reads was performed using Canu v.2.2 with default parameters [19]. The complete genome sequence was deposited in NCBI and annotated by the NCBI Prokaryotic Genome Annotation Pipeline (https://www.ncbi.nlm.nih.gov/genome/annotation_prok/, last access on 1 July 2022 [20]. The genome of ST1023 is publicly available under the Bioproject ID PRJNA854888 in GenBank. Plasmid sequences of pSLT (GenBank Acc. N • AE006471.2)) and pST1030-1A (GenBank Acc. N • MT507877) were used for comparison.

ST1023 is part of a collection of Salmonella MDR clinical strains isolated in Southern
Italy from 2006 to 2012 [12]. The genome of ST1023 was sequenced and analysis of the aroC, dnaN, hemD, hisD, purE, sucA and thrA housekeeping gene classified ST1023 in the MLST group ST19. Based on the absence of the fljB gene, detected by PCR, ST1023 was assigned to the monophasic variant 4, [5],12:i:- [26]. The genome sequence of Salmonella Typhimurium LT2 (GenBank Acc. N • AE006468.2) was used as reference for comparison with that of ST1023. A large chromosomal deletion of 70,456 bp was detected in ST1023: it started at 435 bp downstream of STM2692 and ended in the inverted repeat sequence hixR of the segment H. The segment H, required for phase variation, contains the promoter for fljB and the hin gene, encoding a DNA invertase necessary for inversion of the H segment. The chromosomal deletion included the prophage Fels-2 genome, the region from STM2741 to STM2769 and the fljAB operon, whose absence accounted for the monophasic variant 4, [5],12:i:-. In the place of the 70,456 bp deletion, there was a fragment of 5349 bp. It was composed of six Open Reading Frames (ORFs), named ORF1 to ORF6. ORF2 showed partial homologies with both STM1053 and STM1054 (genes of the prophage Gifsy-2). ORF3 (549 bp) was partially homologous with STM1997, a gene reported to encode for a component (UmuC) of the DNA polymerase V. ORF4 was homologous to STM2704. ORF5 and ORF6 showed partial homologies with STM2705 and STM2706, respectively. STM2704, STM2705 and STM2706 are part of the Fels-2 prophage genome. ORF1 had no homology with any LT2 genes.
ST1023 was resistant to Cm, Sm, Su, Tc and Tp encoded by cmlA1, (aadA1, aadA2), sul3, tetA(B) and dfrA12, respectively. aadA1, aadA2, cmlA1, dfrA12 and sul3 were part of a sul3-associate class1 integron, while tetA(B) was part of a pseudo-compound transposon (see below). The antimicrobial-resistance genes were harboured by an IncR plasmid, named pST1023. The genomic and phenotypic features detected for ST1023 were consistent with those characterising the pandemic serotype 4, [5],12:i:-lineage referred to as the Southern European clone, a monophasic variant of S. Typhimurium that has emerged as a major global cause of non-typhoidal disease in animals and humans [11,27,28]. pST1023 also harboured a second pseudo-compound transposon harbouring a sil operon (see below). However, ST1023 was found sensitive to AgNO 3 (MIC ranging from 4 to 8 µg/mL)

pST1023 Genetic Organisation
pST1023 consisted of 120,313 bp with an average G+C content of 51.4% ( Figure 1). The assembled sequence was confirmed by comparing the BamHI, ClaI, HindIII and KpnI restriction profiles generated in silico with those obtained from restrictions of plasmid DNA. Based on computational analysis for functional gene prediction, three major regions designated IncR backbone, I1 and SLT (the last two traceable to plasmids IncI and pSLT derivatives, respectively) were identified. pST1023 also harboured a Tn21-derived element lacking in the mer operon (termed Tn21-∆mer), two IS26-bound pseudo-compound transposons [29] carrying a Tn10-derived and a sil gene cluster, named PCT-Tn10 and PCT-sil, respectively, and IS elements (or their isoforms) of the families IS1, IS5 and IS6 (Table S1). The IncR backbone included repB (that encodes the replication initiation protein RepB) and its iterons (composed of 36 bp present in 10,4 copies), parAB (a partition system encoding the ATPase protein ParA and the centromere-binding protein ParB), umuCD (that encodes an error-prone DNA polymerase V, a key contributor in the SOS response), retA (a group IIB intron-encoding reverse transcriptase), rfsF-resD (site-specific recombination system) and vagCD (encoding a type II toxin-antitoxin (TA) system, consisting of The IncR backbone included repB (that encodes the replication initiation protein RepB) and its iterons (composed of 36 bp present in 10,4 copies), parAB (a partition system encoding the ATPase protein ParA and the centromere-binding protein ParB), umuCD (that encodes an error-prone DNA polymerase V, a key contributor in the SOS response), retA (a group IIB intron-encoding reverse transcriptase), rfsF-resD (site-specific recombination system) and vagCD (encoding a type II toxin-antitoxin (TA) system, consisting of VagC antitoxin and VagD toxin) [10,30]. vagCD was inversely oriented and separated from rfsF by region I1 (57,754 bp). Region I1 was composed of a remnant of fragment C (7.1 Kb, see below), a Tn21-∆mer (14.6 Kb) and a 32.5 Kb fragment (called I1-oriT); the last was linked to sequences from the leading region and part of the conjugative region of IncI1 plasmids [31]. The fragment C has mainly been identified in IncI1 plasmids, some of which had a Tn21-derived transposon (like the Tn21-∆mer but retaining the mer operon) inserted into ydfA [17]. The Tn21-∆mer also mapped next to 5 -∆ydfA and was composed of tnpA (transposase), tnpR (resolvase), tnpM (interrupted by a sul3-associated class 1 integron, carrying the array dfrA12-orfF-aadA2-cmlA1-aadA1-qacH) and mef (interrupted by the IS26 1 ). The fragment I1-oriT was composed of: I) the impCAB operon (impB was interrupted by the IS26 1 ) encoding (based on the homology to umuCD) an error-prone DNA repair system [32]; II) the psiAB operon (inhibition system of the SOS response that during conjugation prevents LexA autocleavage catalysed by RecA) [33]; III) the ardA gene, encoding an anti-restriction and anti-modification protein that prevents cleavage at foreign DNA entering a new bacteria host [34]; IV) the relaxation complex, composed of the oriT sequence (origin of transfer) and the nikAB gene cluster (encoding the oriT-specific DNA binding protein NikA and the relaxase NikB, respectively) [31]; V) the trbABC operon essential for conjugational transfer of IncI1 plasmids (TrbA and TrbC are key elements in delivery DNA molecules to be secreted across the T4SS) [35]; VI) the toxin/antitoxin system pndABC (Hok/Sok TA family), composed of a stable mRNA encoding a toxin (e.g., PndA), a more unstable antisense mRNA (e.g., PndB encoded mRNA) and pndC, which modulates pndA expression by promoting its translation [35]; VII) the entry exclusion system composed of excA and traY [36]; VIII) the traX gene encoding an inner membrane conjugal transfer pilus acetylation protein; and IX) the traW gene (encoding a lipoprotein) interrupted by the IS26 2 -v1 (classified by the recent proposed nomenclature as IS26-v1 in that differing from IS26 sequence for three nucleotides, of which two caused the single amino acid substitution G184N in the catalytic domain of Tnp26 [37]). Between pndABC and excA, there was an IS1N element flanked by an 8 bp (CGATAGCT) target site duplication (TSD). Between region I1 and vagCD, we mapped a multiple ∆IS locus (locus A) composed of a ∆IS102 element (IS5 family), truncated by an IS2 element (IS3 family), truncated by an ISEc15 (IS3 family), truncated by a Tn5393 (Tn3 family), interrupted by the IS26 2 -v1 ( Figure S1). Downstream of vagCD was the region SLT that included an rfsF-RsdB system, the type II TA system ccdAB [38,39] and the spvABCD operon (associated with strains that cause non-typhoid bacteraemia), with its positive regulator spvR [40]. A PCT-Tn10 (flanked by IS26) was inserted into spvC, splitting the SLT region into two fragments, of which one carried the 3 -∆spvC and one, inversely oriented, the 5 -∆spvC. TSD of 8 bp (CTTTAAAG) was detected downstream of both 3 -∆spvC and 5 -∆spvC. The PCT-Tn10 harboured three genes (jemA, jem B and jemC) unrelated to tetracycline resistance, genes encoding tetracycline resistance (tetA, tetC and tetD) and tetR encoding tetracycline transcriptional repressor [41]. tetD was interrupted by the IS26 4 -v1 that caused loss of IS10-R of the ancestral Tn10; of the IS10-L, only 66 bp were retained. Between the IS26 3 and ∆IS10-L being mapped were a locus named locus B, composed of a ∆TnEc1, a ∆IS1, a fragment of 68 bp and an IS903B element ( Figure S2). Following the SLT region mapped PCT-sil (flanked by IS26 4 -v1 and IS26 5 ), composed of the silE, silC, silF, silB, silA, silG and silP structural genes and the two-component silver-responsive transcriptional genes silR and silS [42]. The mechanism of silver resistance includes a cation sequestration in the periplasm (via SilE and SilF), an active silver efflux (via SilCBA efflux transporter and a putative P-type ATPase transporter SilP) and a signal transduction system, mediated by the sensor histidine kinase SilS and the response regulator protein SilR. The sil genes cluster was flanked by two ∆IS1 (referable in the IS finder database as either IS1A, IS1R or IS1S) interrupted by the IS26 4 -v1 and IS26 5 . Downstream of (554 bp) silP, there was an IS element (ISKpn74) of the IS5 family subgroup IS903 [43].
Between PCT-sil and the IncR backbone, there was a region of about 7227 bp where the 4651 bp just upstream of retA were found linked to many IncR plasmids (see below). This last region (termed N) was characterised by the two IS elements IS903B (subgroup IS903, family IS5 [44,45]) and IS1X3 (family IS1 [46], and a sequence of the IncN backbone composed of a remnant of repA (encoding the IncN replication initiation protein) and its iterons (composed of 37 bp present in 24.3 copies representing the RepA-binding site) [47,48]. The 2576 bp, spanning from PCT-sil to the N region, included an IS1X2 (family IS1) and an open reading frame of 741 bp, encoding for a putative tyrosine recombinase. Tyrosine recombinases constitute a large family of proteins involved in different biological processes (e.g., post-replicative segregation of plasmids) [49].
We also tried to track down IncR plasmids that shared markers with pST1023, whose common genetic organisation might mirror possible common evolutionary steps. Three DNA sequences were selected: that of region N (already reported in many IncR plasmids) and those of loci A and B (whose genetic structures were the output of precise events of insertions and/or recombination of IS and Tn elements) ( Figures S1 and S2, respectively). Seventy-four IncR sequences were found to harbour high-similar regions N (100% coverage and ≥99.9% identity) linked to the IncR backbone. Of the 74 retrieved sequences, 18 lacked vagCD (DNA homology ended to rfsF), 2 had vagCD separated from rfsF by DNA fragments different in size and genetic information and 48 retained vagCD, next to rfsF, linked to sequences (∆IS102-∆IS2 and ISEc15 or ∆ISEc15) related to locus A as follows: 1, pF18S020 (GenBank Acc. N • CP082454), held an intact ISEc15, 41 had ISEc15 disrupted (same nucleotide position) by an ISKpn60 and 6 the ISEc15 disrupted (same nucleotide position as in pST1023, see below) by a Tn5393 (GenBank Acc. N • CP057379, CP058068, CP064111, KM877517, LR890355 and LR890753). In pST01023, the ISEc15 (disrupted by Tn5393) was then interrupted by an IS26 2 -v1 element, thus, generating the structure of locus A. In pST1023, the region, including vagCD (3320 bp), was inversely oriented and separated from rfsF by the region I1.
In addition to locus A, the two plasmids with GenBank Acc. N • CP057379 and CP058068 (plasmids pRHB28-C14_2 and pRHB02-C19_6, respectively) also harboured an identical locus B, localised next to the remnant of gene brxL (nucleotide position 7467-8555 in GenBank Acc. N • CP058068). Locus B might have originated in a pKqq_18A069_2 (GenBank Acc. N • CP084820) -like plasmid, where a ∆IS1-TnEc1 was separated from the estP (nucleotide position 24.426-27.002) gene by a 55 bp fragment. Insertion of IS903B into estP (13 bp from its 3 end) generated the multi-IS locus B (IS903B separated from ∆IS1-∆TnEc1 by 68 bp that included the 13 bp 3 estP) found in the plasmid p1506-1 (GenBank Acc. N • CP059289). However, locus B in pRHB28-C14_2 and pRHB02-C19_6 was found next to the brxL gene or, as in pST1023, within the PCT-Tn10. The origin of this different genetic localisation may result from recombination events, possibly mediated by IS903B [50]. Plasmids pRHB28-C14_2 and pRHB02-C19_6 were virtually identical to each other (99.8% identity, coverage 100%), had a size (about 43 Kb) smaller than that of pST1023 and, interestingly, harboured a complete Tn10. These data together suggest that pRHB28-C14_2, pRHB02-C19_6 and pST1023 might have shared possible common evolutionary steps.

Discussion
Plasmids are genetic elements that may supply a valuable and variable gene pool. Moreover, features, such as maintenance over generations into daughter cells and ability to transfer from hosting to recipient cells, make plasmids a driving force of bacteria ecology and evolution [1]. Plasmids are subject to molecular evolution through genetic reassortments, mainly occurring among plasmids themselves or with other genetic elements, such as integrons, transposons, etc. In this respect, a set of plasmids, termed "mosaic", has recently been the topic of growing scientific interest, as shown by the number of studies published on this topic [52][53][54]. Identification and characterisation of mosaic plasmids can, indeed, help to better assess the extent of molecular dynamics (both known and yet to be discovered) on plasmid evolution and the role played by this set of plasmids in the spread of genes conferring selective advantages, including those encoding antimicrobial resistance [55,56].
pST1023 is a mosaic, multidrug-resistant and mobilisable IncR plasmid, harboured by the clinical strain ST1023, a monophasic variant 4, [5],12:i:-of the widespread pandemic lineage, reported as a Southern European clone. pST1023 acquired exogenous DNA fragments from other plasmids, mostly IncI1 and pSLT-derivatives. Some of these fragments conferred resistance to different classes of antimicrobials. One, probably acquired from an IncI1 plasmid, carried a sul3-associated class 1 integron (embedded into a Tn21-∆mer) that conferred resistance to aminoglycosides, chloramphenicol and trimethoprim-sulphonamides. The second, PCT-Tn10, conferred resistance to tetracycline. Other acquired fragments, also from IncI1 plasmids, carried oriT and nikAB that allowed pST1023 to be horizontally transferred by the copresence of a conjugative plasmid, as we demonstrated with the IncFII and IncC plasmids pST1007-1D and pVC1035, respectively.
Acquisition of some of pST1023 s exogenous DNA fragments could be the result of transposase activity of different IS elements, such as IS26 and IS903B [57]. The presence of six IS26 elements, of which four were organised into two pseudo-compound transposons (PCT-Tn10 and PCT-sil), might be consistent with that possibility and also reinforce the role played by this class of IS in shaping plasmids and spreading ARGs. In this respect, it is worth noting that two IS26s (IS26-2 and IS26-4) were classified as IS26-v1, an IS26 variant with enhanced activity [37], and that IS26-4 was part of PCT-Tn10. To the best of our knowledge, this is the first report on the detection of this pseudo-compound transposon and this finding poses concern regarding the diffusion of tetracycline resistance.
Silver ions and silver-based compounds (e.g., nanoparticles) are well-known antimicrobial agents used in clinical and medical practice, as well as in agricultural and industrial products [58,59]. As a result, isolation of silver-resistant bacteria has recently been increasing, making silver resistance an emerging problem [60,61]. Moreover, cryptic silver resistance that can be readily activated by single missense mutation in silS (encoding the sensor histidine kinase) has widely been reported for some genera (e.g., Enterobacter and Klebsiella) [62,63]. ST1023 is sensitive to AgNO 3 ; however, spontaneous ST1023 silverresistant mutants were selected on MHA plates added with AgNO 3 up to 256 µg/mL. Analysis of these mutants is in progress to assess the possible cryptic silver resistance of ST1023. Anyway, the finding of sil operon organised into an IS26-bound transposon (PCT-sil) is undoubtedly of concern, since IS26 elements are encountered with increasing frequency in genomes (especially plasmids) of clinically relevant bacteria [64].
Mechanisms of recombination, other than those mediated by transposases, for instance, site-specific recombination systems (rfsF-RsdB and rfsF-ResD), may also have contributed to the acquisition of the exogenous DNA fragments SLT and I1, respectively. In this regard, it is worth mentioning the finding of some IncR plasmids with vagCD separated from the core backbone through the insertion of different DNA fragments mapping just downstream rfsF. In pST1023, the rfsF-ResD system might have been involved in acquisition of the region I1 from IncI1 plasmids, such as pST1030-1A. Acquisition from IncI1 plasmids (like pST1030-1A) and involving the rfsF-ResD system has also been reported for the mosaic IncFII-plasmid pST1007-1A. It is noteworthy that pST1007-1A, pST1023 and pST1030-1A were all isolated from Salmonella clinical cases, which occurred in Apulia in the three-year period 2006-2008.
The broad host range of IncR plasmids has been linked to the possibility that these plasmids are mobilisable [3,65]. However, sequencing results showed that IncR plasmids did not possess relaxase genes nor origin of transfer sequences (oriT) and the mobilisation of IncR plasmids remains to be proven. In this study, we report the first experimental data on the mobilisation of a mosaic, multidrug-resistant IncR plasmid, pST1023, containing the IncI1 oriT-nikAB region. The study also attempts to outline possible evolutionary steps shared with other IncR plasmids.