Emergence of IncHI2 Plasmids With Mobilized Colistin Resistance (mcr)-9 Gene in ESBL-Producing, Multidrug-Resistant Salmonella Typhimurium and Its Monophasic Variant ST34 From Food-Producing Animals in Italy

A collection of 177 genomes of Salmonella Typhimurium and its monophasic variant isolated in 2014–2019 from Italian poultry/livestock (n = 165) and foodstuff (n = 12), previously screened for antimicrobial susceptibility and assigned to ST34 and single-locus variants, were studied in-depth to check the presence of the novel mcr-9 gene and to investigate their genetic relatedness by whole genome sequencing (WGS). The study of accessory resistance genes revealed the presence of mcr-9.1 in 11 ST34 isolates, displaying elevated colistin minimum inhibitory concentration values up to 2 mg/L and also a multidrug-resistant (MDR) profile toward up to seven antimicrobial classes. Five of them were also extended-spectrum beta-lactamases producers (blaSHV–12 type), mediated by the corresponding antimicrobial resistance (AMR) accessory genes. All mcr-9-positive isolates harbored IncHI2-ST1 plasmids. From the results of the Mash analysis performed on all 177 genomes, the 11 mcr-9-positive isolates fell together in the same subcluster and were all closely related. This subcluster included also two mcr-9-negative isolates, and other eight mcr-9-negative ST34 isolates were present within the same parental branch. All the 21 isolates within this branch presented an IncHI2/2A plasmid and a similar MDR gene pattern. In three representative mcr-9-positive isolates, mcr-9 was demonstrated to be located on different IncHI2/IncHI2A large-size (∼277–297 kb) plasmids, using a combined Illumina–Oxford Nanopore WGS approach. These plasmids were also compared by BLAST analysis with publicly available IncHI2 plasmid sequences harboring mcr-9. In our plasmids, mcr-9 was located in a ∼30-kb region lacking different genetic elements of the typical core structure of mcr-9 cassettes. In this region were also identified different genes involved in heavy metal metabolism. Our results underline how genomics and WGS-based surveillance are increasingly indispensable to achieve better insights into the genetic environment and features of plasmid-mediated AMR, as in the case of such IncHI2 plasmids harboring other MDR genes beside mcr-9, that can be transferred horizontally also to other major Salmonella serovars spreading along the food chain.

A collection of 177 genomes of Salmonella Typhimurium and its monophasic variant isolated in 2014-2019 from Italian poultry/livestock (n = 165) and foodstuff (n = 12), previously screened for antimicrobial susceptibility and assigned to ST34 and singlelocus variants, were studied in-depth to check the presence of the novel mcr-9 gene and to investigate their genetic relatedness by whole genome sequencing (WGS). The study of accessory resistance genes revealed the presence of mcr-9.1 in 11 ST34 isolates, displaying elevated colistin minimum inhibitory concentration values up to 2 mg/L and also a multidrug-resistant (MDR) profile toward up to seven antimicrobial classes. Five of them were also extended-spectrum beta-lactamases producers (bla SHV −12 type), mediated by the corresponding antimicrobial resistance (AMR) accessory genes. All mcr-9-positive isolates harbored IncHI2-ST1 plasmids. From the results of the Mash analysis performed on all 177 genomes, the 11 mcr-9-positive isolates fell together in the same subcluster and were all closely related. This subcluster included also two mcr-9-negative isolates, and other eight mcr-9-negative ST34 isolates were present within the same parental branch. All the 21 isolates within this branch presented an IncHI2/2A plasmid and a similar MDR gene pattern. In three representative mcr-9-positive isolates, mcr-9 was demonstrated to be located on different IncHI2/IncHI2A large-size (∼277-297 kb) plasmids, using a combined Illumina-Oxford Nanopore WGS approach. These plasmids were also compared by BLAST analysis with publicly available IncHI2 plasmid sequences harboring mcr-9. In our plasmids, mcr-9 was located in a ∼30-kb region lacking different genetic elements of the typical core structure of mcr-9 cassettes. In this region were also identified different genes involved in heavy metal metabolism. Our results underline how genomics and WGS-based surveillance are increasingly INTRODUCTION Salmonellosis, among the most important foodborne zoonoses worldwide, is the second most commonly reported gastrointestinal infection in humans after campylobacteriosis in the European Union/European Economic Area, mainly caused by the consumption of contaminated food (European Food Safety Authority [EFSA] and European Centre for Disease Prevention and Control [ECDC], 2021). In Europe, among all confirmed salmonellosis cases, the three most commonly reported Salmonella serovars in 2019 were Salmonella Enteritidis (50.3%), Salmonella Typhimurium (11.9%), and monophasic Salmonella Typhimurium (1,4,[5],12:i:-) (8.2%) (European Food Safety Authority [EFSA] and European Centre for Disease Prevention and Control [ECDC], 2021). Sequence types (STs) 19, 34, 313, and 213 represent the most frequently reported STs for S. Typhimurium. Starting from the past two decades, the global pandemic S. Typhimurium ST34 clone has been increasingly reported, replacing to the "traditional" clone ST19 (Biswas et al., 2019). The efficacy of antimicrobial therapy for salmonellosis may be impaired by the spread of antimicrobial-resistant isolates (Michael and Schwarz, 2016), particularly to last-resort drugs used to treat severe infections in humans.
mcr-9 is an inducible plasmid-borne gene mainly associated with the IncHI2 replicon type , encoding an acquired phosphoethanolamine transferase not conferring by itself clinical resistance to colistin. However, subinhibitory concentrations of colistin could induce its expression, which, in E. coli, is mediated by a two-component regulatory system encoded by the qseC and qseB genes, located downstream of mcr-9 (Kieffer et al., 2019).
The aims of this work were to (i) reanalyze by whole genome sequencing (WGS) a historical collection of S. Typhimurium and its monophasic variant within ST34 (along with its single-locus variants) to screen for the presence of mcr-9 and to investigate the relatedness of their genomes and (ii) determine the genetic environment of the mcr-9 gene and fully reconstruct the mcr-9-harboring plasmids by using a combined Illumina-Oxford Nanopore WGS approach.

Salmonella Isolates
A collection of 177 genomes of S. Typhimurium (n = 39) and its monophasic variant (n = 138) assigned to ST34 and single-locus variants by WGS (Supplementary Table 1) was studied in-depth. These isolates were obtained from animal samples (n = 165), collected from Italian poultry (n = 1 broiler chicken, n = 2 laying hens, and n = 14 fattening turkeys) and livestock [n = 106 fattening pigs and n = 42 bovine animals <12 months (veal calves)] and also foodstuff (n = 12), in the frame of antimicrobial resistance (AMR)-monitoring activities (according to Decision 2013/652/EU 1 ) conducted from 2014 to 2019 by the National Reference Laboratory for Antimicrobial Resistance (NRL-AR) and previously screened for antimicrobial susceptibility.

Antimicrobial Susceptibility Testing (AST)
AST was performed as minimum inhibitory concentration (MIC) determination by broth microdilution, using the EU consensus 96-well microtiter plates (Trek Diagnostic Systems, Westlake, OH, United States). The results were interpreted according to epidemiological cutoffs (ECOFFs) included in the Annex A of the EU Decision 2013/652/EU, and for sulfamethoxazole the tentative ECOFF of >256 mg/L according to the EURL-AR protocol for antimicrobial susceptibility testing of Escherichia coli, Salmonella, and Campylobacter 2 . E. coli ATCC 25922 was used as quality control strain.
novo assembly of raw reads were performed according to Alba et al. (2020).
Molecular characterization was performed on all the assembled genomes with the ABRicate tool 3 using the Genomic Epidemiology (CGE) databases of ResFinder 4 and PlasmidFinder 5 , and the MLST 6 and SeqSero2  tools for the detection of the genetic basis of AMR and plasmid replicon types and to confirm the STs and the serotype in silico, respectively.
The pMLST 2.0 online version 7 was also used for pMLST analysis on mcr-9-positive isolates and in representative mcr-9negative isolates. The presence of the two-component regulatory system of mcr-9 inducible expression encoded by the qseC and qseB genes (Kieffer et al., 2019), was also searched by BLAST analysis.
In order to determine the genetic relatedness of the whole set of Salmonella genomes analyzed, all raw reads from the 177 genomes were also compared and clustered by using the Mash algorithm (Katz et al., 2019).

Oxford Nanopore Technologies (ONT) Long-Read Sequencing and Bioinformatics Analysis
In order to resolve the complete sequence, fully reconstruct the mcr-9-harboring plasmids, and precisely identify and locate mcr-9-harboring regions, three selected mcr-9-positive (IDs: 19063952, 15060500, and 19093665) and one mcr-9-negative (ID: 17021625; one of the most closely related to positive ones selected for comparison purposes) isolates were also sequenced using the nanopore-based MinION device (ONT) with the rapid barcoding kit (SQK-RBK004). A hybrid (Illumina-Oxford Nanopore) assembly was performed using the Unicycler pipeline (Branton et al., 2008) with the default parameters.
The assembly obtained was annotated using the RAST Server (Aziz et al., 2008). Additionally, a manual curation for the obtained annotation was performed, especially for the Insertion Sequences (ISs) by using the ISfinder database (Siguier et al., 2006). The presence of virulence genes was also determined on the obtained plasmid sequences using the VF database 8 .

AST
The AMR phenotypes of all Salmonella isolates are reported in Supplementary Table 1. As for the 11 mcr-9-positive isolates, beside the AMR gene profiles, also the MIC values are reported in detail in Table 1.
For isolates reported in Table 1, the MIC values of highest priority critically important antimicrobials (HPCIAs) such as colistin, cefotaxime, ceftazidime, and ciprofloxacin were interpreted also according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) 9 clinical breakpoints. In synthesis, out of 11 isolates, 6 showed microbiological and clinical resistance to cefotaxime (FOT) and ceftazidime (TAZ) (MIC range 4-8 mg/L for FOT and 16 mg/L for TAZ). Five showed microbiological resistance (MIC range 0.25-0.5 mg/L), and one also clinical resistance (MIC = 2 mg/L) to ciprofloxacin (CIP). The 11 mcr-9-positive isolates displayed elevated colistin MIC values: 4 of them had a MIC of 2 mg/L, on the epidemiological cutoff and clinical breakpoint, and 7 had a MIC of 1 mg/L.
Frontiers in Microbiology | www.frontiersin.org FIGURE 1 | Circular representation of the Mash analysis of the 177 Salmonella genomes analyzed. Branch length is not proportional to the genetic distance. The red branches indicate the mcr-9-positive isolates harboring IncHI2 plasmids, and the green branches indicate the mcr-9-negative isolates harboring IncHI2 plasmids.
qnrB2 or qnrS1 (four isolates) genes, encoding, respectively, sulfamethoxazole, tetracycline, trimethoprim, all beta-lactams, phenicol, and fluoroquinolone resistance. The abovementioned AMR gene profiles were confirmed by the corresponding AMR phenotypes ( Table 1). No mcr-9-positive isolates presented known chromosomal point mutations associated with AMR phenotypes, and all tested negative for the presence of qseC and qseB regulatory genes. All mcr-9-positive isolates harbored an IncHI2/2A, two an IncI1 and one an IncQ replicon type. Moreover, the pMLST analysis revealed that all of them harbored an IncHI2-ST1 plasmid, while one representative mcr-9-negative isolate harbored an IncHI2-ST4 replicon ( Table 1).
The results of the Mash analysis are represented in Figure 1. In the Mash tree, three main clusters were identified, with two of them consisting of 7 and 29 ST34 monophasic variant isolates, and the third one, being the most complex, grouped all the remaining isolates in several subclusters. No clear clustering based on year of sampling, sample type, or origin of the isolates was observed. The 11 mcr-9-positive isolates fell together in one of the subclusters within the most complex cluster and were all closely related (marked in red, Figure 1). This subcluster included also two mcr-9-negative isolates (17050877 and 17082797 marked in green, Figure 1). Additionally, other eight mcr-9-negative ST34 isolates were present within the same parental branch (marked in green, Figure 1).
Only the 21 isolates within this branch (both the 11 mcr-9 positive and the 10 negative ones) presented an IncHI2/2A plasmid (Figure 2A and Supplementary Figure 1A) and a similar MDR gene pattern, with three of them being also positive for mcr-1.1, associated with colistin resistance (Figure 2B and Supplementary Figure 1B).
Annotation of plasmid sequences identified four main genetic regions in the four IncHI2/IncHI2A plasmids. The plasmid FIGURE 2 | (A) Presence or absence of plasmid replicons identified by Mash analysis in the mcr-9-positive and related mcr-9-negative isolates. (B) Presence or absence of AMR genes identified by Mash analysis in the mcr-9-positive and related mcr-9-negative isolates.
backbone (located in the region 1..40.000 nucleotides (nt); ∼13% of the total plasmid size) included a replication region containing the repA gene encoding a replication initiation protein, a stability region, and the conjugative and transfer genes (Figure 3).
Comparison of our three mcr-9-carrying plasmids revealed that with a coverage range of 94-98%, they shared an identity of 99.9%, while the mcr-9-negative pMOL625 presented a lower coverage range (71-75%) and an identity of 98.7% with the abovementioned plasmids. Besides mcr-9, all the four IncHI2 plasmids harbored the majority of the AMR genes detected encoding resistance to all beta-lactams, aminoglycosides, trimethoprim, tetracyclines, sulfamethoxazole, and phenicols ( Table 1). In particular, pMOL-952 and pMOL-665 shared a variable region (240,020..250,837 nt; Figure 3) of a class 1 integron carrying several AMR genes and qacE 1 (quaternary ammonium compound resistance gene). The studied plasmids also harbored genes involved in the metabolism of heavy metals as arsenic, nickel, and mercury. In this regard, a mercury-resistance (mer) operon (101,550..105,526 nt; Figure 3) was shared by the mcr-9-positive pMOL-952, pMOL-665, and pMOL-500 plasmids and was absent in the mcr-9-negative pMOL625.
No virulence genes were detected in the four IncHI2 plasmids. In these plasmids, the mcr-9 gene was located on an almost conserved region in the three mcr-9-harboring plasmids, of approximately 24,548 bp. mcr-9 was located adjacent to a cupin fold metalloprotein gene (wbuC) with the insertion sequences IS903 upstream and the IS26 downstream of both genes. Additionally, located upstream of this structure were genes associated with heavy metal metabolism (Figures 3, 4). No other AMR genes have been detected in this region.
The results obtained with BLAST analysis with the two previously published IncHI2 mcr-9-carrying plasmids revealed that our three IncHI2 mcr-9-carrying plasmids were closely related (88% coverage and 99.9% identity) to the IncHI2 ∼293 kb plasmid (p3846) described in Italy by Marchetti et al. (2021) in E. cloacae. As for the mcr-9/bla VIM −1 -carrying IncHI2 plasmid (pRH-R27, size ∼299 kb) described in Germany in S. Infantis from a fattening pig farm, it shared with our three mcr-9-carrying plasmids a coverage range of 80-84% and an identity of 99.9%.
The specific region where mcr-9 was located in pMOL952, pMOL665, and pMOL500 was almost identical to the corresponding region of p3846 and pRH-R27 with an identity of 99% including the wbuC gene. The structural analysis revealed that pMOL952, pMOL665, and pMOL500 presented a mobile element (IS903) that was not identified in p3846 or pRH-R27 (Figures 3, 4).

DISCUSSION
In this study, we report for the first time in Italy the presence and genomic features of 11 MDR, mcr-9-positive S. Typhimurium, and single-locus variant isolates from food-producing animals (fattening pigs and veal calves) belonging to the emerging ST34 clone, with some of them also displaying resistance to highest priority critically important antimicrobials (HPCIAs), as third-and fourth-generation cephalosporins. Indeed, plasmidborne mcr genes pose a significant threat to public health at international level, not only because of the colistin-resistant phenotype they can induce but also because they can be transferred horizontally to foodborne pathogens in combination with other resistance genes, with the potential to be transmitted to humans and impair the treatment options.
From the Mash analysis, we identified 21 ST34 isolates within the same parental branch (both the 11 mcr-9 positive and the 10 negative ones) harboring large-size (∼277-297 kb) IncHI2/2A plasmids and a similar MDR gene pattern (Figure 2 and Supplementary Figure 1). Interestingly, only these isolates shared the presence of these plasmid types and presented a greater abundance of AMR genes compared to the rest of the Salmonella genomes analyzed. Because of the intrinsic characteristic of the Mash approach, the existence of a large-size IncHI2/2A plasmid harboring several similar AMR genes could have determined the high genetic relatedness of these isolates represented on the Mash tree.
the two-component regulatory system qseBC. Therefore, mcr-9 may silently spread and remain undetected, unless induced by subinhibitory concentrations of colistin (Kieffer et al., 2019). However, the effectiveness of this regulatory system may be dependent on the genetic context (e.g., mcr-9 could be located on a plasmid or integrated into the chromosome) or may differ in relation to different strain backgrounds (e.g., serotypedependent colistin susceptibility observed in Salmonella isolates) (Tyson et al., 2020), or other still unknown mechanisms could be involved. Additionally, the mcr gene family may undergo further genetic microevolution (and possibly to changes in the susceptibility pattern to colistin) due to selection pressure in the food-producing animal industry.
However, only very few complete IncHI2 plasmid sequences carrying mcr-9 from food-producing animals and related foodstuff are available in public repositories for comparison, so far. In this regard, we fully reconstructed two mcr-9positive IncHI2 plasmids from the monophasic variant of S. Typhimurium isolated from pig sources and one from the cecal content of a veal calf. To the best of the authors' knowledge, this represents the first report of MDR, mcr-9positive Enterobacteriaceae detected in bovine animals.
All 11 IncHI2 plasmids harboring mcr-9 were assigned to ST1 by pMLST analysis. IncHI2-ST1 plasmids have been the most frequently reported IncHI2 ST, representing a major vehicle in mediating mcr-9 and AMR gene dissemination also in Enterobacteriaceae isolates from clinical settings . Interestingly, the mcr-9-negative IncHI2 plasmid we sequenced with ONT was assigned to another ST, ST4.
IncHI2 plasmids are known carriers of resistance determinants not only to antibiotics but also to heavy metals such as silver, mercury, arsenic, copper, tellurium, and others with the potential to coselect for the concomitant presence of mcr-9 (Tyson et al., 2020).
Moreover, pigs are considered one of the most significant vectors for the monophasic variant of S. Typhimurium ST34, and pork meat represents one of the main infection sources for humans (Biswas et al., 2019). As copper and zinc supplementation are commonly used in the swine industry (or in the case of therapeutical use of ZnO also to control postweaning diarrhea), heavy metals could accumulate in the environment, leading to a selection of ST34 isolates resistant to heavy metals and to IncHI2 plasmids frequently associated with various metal tolerance genes (Biswas et al., 2019).
IncHI2 plasmids have been reported to be diverse in terms of the overall genetic structure, while mcr-9 has been reported to be consistently located in the sil-cop region . As previously described , in this region, the core structure of all reported mcr-9 cassettes was rcnR-rcnA-pcoE-pcoS-IS903-mcr-9-wbuC, with the genetic content immediately upstream of mcr-9 mostly conserved. Differently, the gene content located downstream of mcr-9 was reported to be genetically diverse (silver resistance determinants and qseB-C regulators absent in most plasmids), and transposon elements were not identified.
In our plasmids, mcr-9 was located in a plasmid region (∼30 kb) lacking different genetic elements of the core structure, compared to the other plasmid sequences subjected to BLAST analysis (Figure 4). In particular, mcr-9 was flanked by the wpuC gene and the IS903 element with and additional IS element (IS26) located downstream wpuC, suggesting the potential ability to mobilize this gene. Heavy metal resistance genes were also identified (as arsenic and nickel) in the same region, but accordingly to previous findings, silver resistance determinants and qseB-C regulators were absent.
In conclusion, the spread of MDR S. Typhimurium, including monophasic ST34, has widely challenged the treatment options to control foodborne infections (Biswas et al., 2019). This can be of even more concern, especially when there is concomitant evidence of MDR genes against HPCIAs, such as colistin, extended-spectrum cephalosporins, and (fluoro)quinolones. Therefore, genomics and WGS-based surveillance are increasingly indispensable to achieve better insights also into the genetic environment and features of plasmid-mediated AMR and the relationships with bacterial pathogenic hosts, as in the case of such IncHI2 plasmids harboring mcr-9, that can be transferred horizontally also to major Salmonella serovars spreading along the food chain.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.