Genomic Analysis of a mcr-9.1-Harbouring IncHI2-ST1 Plasmid from Enterobacter ludwigii Isolated in Fish Farming

This study analyzed the resistome, virulome and mobilome of an MCR-9-producing Enterobacter sp. identified in a muscle sample of seabream (Sparus aurata), collected in a land tank from multitrophic fish farming production. Average Nucleotide Identity analysis identified INSAq77 at the species level as an Enterobacter ludwigii INSAq77 strain that was resistant to chloramphenicol, florfenicol and fosfomycin and was susceptible to all other antibiotics tested. In silico antimicrobial resistance analyses revealed genes conferring in silico resistance to β-lactams (blaACT-88), chloramphenicol (catA4-type), fosfomycin (fosA2-type) and colistin (mcr-9.1), as well as several efflux pumps (e.g., oqxAB-type and mar operon). Further bioinformatics analysis revealed five plasmid replicon types, including the IncHI2/HI2A, which are linked to the worldwide dissemination of the mcr-9 gene in different antibiotic resistance reservoirs. The conserved nickel/copper operon rcnR-rcnA-pcoE-ISSgsp1-pcoS-IS903-mcr-9-wbuC was present, which may play a key role in copper tolerance under anaerobic growth and nickel homeostasis. These results highlight that antibiotic resistance in aquaculture are spreading through food, the environment and humans, which places this research in a One Health context. In fact, colistin is used as a last resort for the treatment of serious infections in clinical settings, thus mcr genes may represent a serious threat to human health.


Introduction
The emergence of colistin resistance in the last years is a serious threat to the treatment of infections caused by multidrug-resistant bacteria in human medicine. Consequently, colistin, a last resort antibiotic, is categorized by the World Health Organization (WHO) as one of the highest priority, critically important antibiotics for human medicine. The use of colistin in veterinary medicine has been prohibited in various countries. However, colistin is still an antibiotic extensively used in veterinary medicine for infections caused by Enterobacterales [1]. In aquatic animal species, colistin is also used to treat bacterial

Results and Discussion
The mcr-1 and mcr-9 variants are the most widespread mcr-family genes. The mcr-9 gene was identified in 40 countries through six continents, with 61.5% of the mcr-9-positive strains isolated in the United States [7]. In that study, S. enterica was the most common host species, especially in turkeys and chickens. Furthermore, other systematic reviews showed that isolates carrying mcr-9 were detected in 21 countries through six continents, mainly from Europe. mcr-9-positive isolates were disseminated by various genera and species of Enterobacteriaceae isolates among which Enterobacter spp. were predominant (37.0%) [21]. More than 50% of the isolates were from human origin, being 29.0%, 3.6% and 2.9% from animal, environmental and food, respectively. Here, an mcr-9-producing isolate (INSAq77) identified in a muscle sample of a commercial size S. aurata, collected during the winter season (March 2018), in a land tank from a fish multitrophic farming, in the south of Portugal is described. To our knowledge is the first description of mcr-9 gene in the aquaculture environment.
Average Nucleotide Identity (ANI) analysis performed by NCBI identified INSAq77 at the species level as E. ludwigii. The genome sequences of E. cloacae are 98.98% identical by ANI to the E. ludwigii, with 82.5% coverage of the genome. Indeed, INSAq77 isolate was identified as E. cloacae by the VITEK ® 2 automated identification system (BioMérieux, Marcyl'Étoile, France) and sequencing of the 16S rRNA gene. However, it is well known that precise species identification for the taxonomy of Enterobacter is complex [22] and that hsp60 gene sequencing showed a higher species diversity than MALDI-TOF [23]. Seven species have been grouped within the E. cloacae complex: Enterobacter cloacae, Enterobacter asburiae, Enterobacter hormaechei, Enterobacter kobei, E. ludwigii, Enterobacter mori and Enterobacter nimipressuralis, which share at least 60% similarity in their genome with E. cloacae [24].
E. ludwigii was first described as a novel Enterobacter species in 2005 [25]. All strains are naturally resistant to ampicillin, amoxicillin-clavulanic acid, and cefoxitin due to the production of a chromosomal AmpC β-lactamase. Antibiotic-resistant E. ludwigii has been found mainly in clinical samples [26], although a CTX-M-producing E. ludwigii was recently described in an environmental isolate collected from a wastewater treatment plant in India [27]. Recently, antibiotic-resistant E. ludwigii was also identified in India, in moribund goldfish collected from ornamental fish farms [28]. On the other hand, E. ludwigii has been suggested as a potential probiotic microorganism in agriculture and aquaculture [29,30].
INSAq77 strain was resistant to chloramphenicol (MIC 32 mg/L), florfenicol (32 mg/L) and fosfomycin (64 mg/L) but was susceptible to all other antibiotics tested; cefoxitin and amoxicillin/clavulanic acid are intrinsic resistances. The colistin MIC for MCR-9-producing E. ludwigii as 1 mg/L, within susceptible EUCAST breakpoint. Indeed, other studies have demonstrated that the presence of an MCR-9 enzyme not always is associated with colistin resistance [8,31,32]. Nevertheless, recent studies showed that mcr genes might enhance the survival ability of bacteria under clinical colistin pressure, thereby potentially leading to treatment failure [33,34].
This study also analyzed the resistome, virulome and mobilome of this MCR-9producing E. ludwigii isolated from farmed Sparus aurata. The analysis of WGS yielded 225 contigs, ranging from 237 to 244,787 bp. The draft genome contained a total assembly length of 5,276,953 bp, with estimated depth coverage of 30.7×; the GC content was 54.1%.
The MCR-9-producing INSAq77 E. ludwigii isolate belonged to the ST1342 lineage, first reported here. The whole-genome SNP-based phylogenetic tree using the 75 E. ludwigii genomes indicated that INSAq77 is not closely related to the other studied isolates ( Figure 1).
Indeed, INSAq77 has 12% of nucleotide sequence divergence with the closest strain (NZ_VLMJ00000000), an E. ludwigii isolated from the lung of a clinical patient, in 2016, in the USA (PRJNA553678) [35]. The two other MCR-9-producing isolates (NZ_JAGDFR00000000 and NZ_JAGDFs00000000) were grouped into another cluster.
In silico antimicrobial resistance analyses using ResFinder 4.1, with a threshold of 90% identity and a minimum length of 60%, revealed acquired genes conferring resistance to β-lactams (bla ACT-88 , here firstly identified), fosfomycin (fosA2-type) and colistin (mcr-9.1). Furthermore, a total of 21 genes were detected in silico by CARD RGI perfect, strict and loose algorithms, involved in efflux, transport and permeability, which might justify the florfenicol and chloramphenicol resistance identified by phenotypic methods (Tables 1 and S1). Indeed, INSAq77 has 12% of nucleotide sequence divergence with the closest strain (NZ_VLMJ00000000), an E. ludwigii isolated from the lung of a clinical patient, in 2016, in the USA (PRJNA553678) [35]. The two other MCR-9-producing isolates (NZ_JAGDFR00000000 and NZ_JAGDFs00000000) were grouped into another cluster.
In silico antimicrobial resistance analyses using ResFinder 4.1, with a threshold of  CARD loose algorithm (match bitscore less than the curated one blastp bitscore) [36] identified that INSAq77 also harbors a catA4-type gene (64.8% of identity), which might infer resistance to chloramphenicol. However, the low percentage of identity it is not enough to assure the phenicol resistance causality of catA4-type gene. Other resistance mechanisms might be involved; indeed, the multiple antibiotic resistance (mar) locus, a resistance-nodulation-cell division (RND) antibiotic efflux pump detected in silico by CARD RGI strict algorithm, has been reported to contribute to chloramphenicol resistance [37]. Furthermore, the multiple antibiotic resistance oqxAB-type locus, another RND multidrug efflux pump operon was detected, which has been reported to contribute to multidrug resistance [38]. Diminished susceptibility to different antibiotic classes (e.g., aminoglycosides, fluoroquinolones and tetracycline) were bioinformatically predicted (e.g., rsmA, adeF, ramA, acrABR and soxAS), although the isolate was phenotypically susceptible; this can be explained by the fact that efflux pumps are frequently associated with a low decrease in antibiotic susceptibility, which may not translate to a change in phenotype [39].
The acquired disinfectant resistance gene formA-type, a plasmid-mediated formaldehyde resistance mechanism [40], was also identified. The ability to survive aldehyde disinfection processes is clinically significant, with possible cross-resistance to antibiotics [41]. Furthermore, the INSAq77 isolate carried the terC virulence gene, commonly associated with IncHI2 plasmids and conferring resistance to tellurium, where soluble salts, especially potassium tellurite, were used clinically in humans as antimicrobial agents [42].
PathogenFinder predicted the strain as being "human pathogenic" with a probability of 77.7% due to the presence of 74 genes belonging to known pathogenic protein families (Table S2). Indeed, in addition to the known E. cloacae complex genes encoding pathogenic proteins and the homologous sequences of pathogenic proteins from Citrobacter koseri, Enterobacter spp., E. coli, S. enterica and Shigella spp. were found in the study. Mobile genetic elements (MGE), such as plasmids, prophages and transposons among others, are main drivers for the spread of antibiotic resistance [43]. In this study, nine insertion sequences (IS) were found using MobileElementFinder tool: IS26, ISKpn28, ISSen4-type, IS30-type, ISEcl1-type, ISKpn43-type, ISKpn24-type, IS100-type and ISPpu12-type.
A total of eleven prophage regions were also identified using PHASTER tool ( Figure S1), of which two regions were intact (PHAGE_Salmon_SEN4_NC_029015 and PHAGE_ En-tero_HK542_NC_019769), eight regions were incomplete and one region was questionable (PHAGE_Shigel_SfIV_NC_022749). Figure 2 shows the schematic representation of the phage-related proteins identified in the intact and questionable prophages. The size of the three prophages ranged from 16.7Kb to 32.6Kp with an average GC content of 52.7%. These prophages were firstly described in S. enterica subspecies salamae collected in the Czech Republic [44], E. coli isolated in Hong Kong and Shigella flexneri collected in Bangladesh [45], corroborating that MGE can be excised and integrated from the chromosome and MGE into each other. Indeed, several studies have already shown the presence of mcr-type genes in prophages [46,47], indicating the role of these MGE in the dissemination of antibiotic resistance. Furthermore, two cryptic prophages were detected by PathogenFinder (CP4-6 and CP4-57) which, although they do not form active phage particles or lyse their captors, can be considered the relatively permanent reservoirs of antibiotic resistance genes [48].
these MGE in the dissemination of antibiotic resistance. Furthermore, two c prophages were detected by PathogenFinder (CP4-6 and CP4-57) which, althoug do not form active phage particles or lyse their captors, can be considered the rela permanent reservoirs of antibiotic resistance genes [48]. Further bioinformatics analysis revealed the presence of five plasmid replicon ColE10, Col(pHAD28)-type, IncFIA(HI1)-type, IncR-type, IncHI2 and IncHI2A, th two linked to the worldwide dissemination of mcr-9 gene [7,21]. The mcr-9 gene found in different reservoirs (human, animal, food and environment), in various s of Enterobacteriaceae strains, mostly associated with IncHI2/IncHI2A plasmid rep [21]. Indeed, as observed in our study, several works demonstrated the prevalence o The mcr-9 gene was found in a 30,314 bp length contig, manually assembled after visual inspection and alignment of contigs (Aq77p_57, Aq77p_155, Aq77p_191, Aq77p_196, Aq77p_213) against themselves using CLC Genomics Finishing Module v.20.0.1 (QIAGEN, Aarhus); the GC content was 47.6%. The analysis of mcr-9-harbouring contig using the Microbial Nucleotide MegaBLAST analysis against the complete plasmids database revealed nine mcr-9-carrying IncHI2 plasmid sequences (>99.9% identity, >97% query coverage and e-value 0.0) from multiple species, collected in different antibiotic resistance reservoirs worldwide, including human clinical/colonization samples ( Table 2). Of notice, three plasmids of sequence type 1 (ST1), accordingly with the IncHI2 pDLST scheme, were collected from environmental samples during an extended bla IMP-4 -associated carbapenemase outbreak in an Australian hospital [50].  In INSAq77 IncHI2 plasmid and in all others studied here, except p565_1 from a C. freundii strain (NZ_CP038657, Figure 3), the mcr-9 gene was surrounded upstream by an IS903 element and downstream by a wbuC family gene, encoding a cupin fold metalloprotein, followed by IS26. , x FOR PEER REVIEW In INSAq77 IncHI2 plasmid and in all others studied here, except p565_ freundii strain (NZ_CP038657, Figure 3), the mcr-9 gene was surrounded upstr IS903 element and downstream by a wbuC family gene, encoding a metalloprotein, followed by IS26. The mcr-9 gene is frequently associated with the wbuC gene in differen species, suggesting an essential role of this wbuC gene for the activity of enzyme [51]. Indeed, this gene was proposed to have transferred together with whole fragment from Buttiauxella spp. [51]. On the other hand, qseB/qseC tworegulators were absent, which could explain the susceptibility to colistin o isolate. Indeed, based on previous studies, in the presence of subinhibitory con of colistin, the qseB/qseC regulatory system can induce the expression of the m which results in an increase in MIC values [32,51,52]. The mcr-9 gene is being d part of a mcr-9 cassette containing the rcnR-rcnA-pcoE-pcoS-IS903-mcr-9structure ( Figure 3) [21].
As shown in Figures 3 and 4, the genetic background immediately upstre 9 was consistent among IncHI2 mcr-9-bearing plasmids. The exception is the p an ISSgsp1 element from the IS66 family in the INSAq77 mcr-9-harbouring conti The mcr-9 gene is frequently associated with the wbuC gene in different bacterial species, suggesting an essential role of this wbuC gene for the activity of the MCR-9 enzyme [51]. Indeed, this gene was proposed to have transferred together with mcr-9 as a whole fragment from Buttiauxella spp. [51]. On the other hand, qseB/qseC two-component regulators were absent, which could explain the susceptibility to colistin of INSAq77 isolate. Indeed, based on previous studies, in the presence of subinhibitory concentrations of colistin, the qseB/qseC regulatory system can induce the expression of the mcr-9 gene, which results in an increase in MIC values [32,51,52]. The mcr-9 gene is being described as part of a mcr-9 cassette containing the rcnR-rcnA-pcoE-pcoS-IS903-mcr-9-wbuC core structure (Figure 3) [21].
As shown in Figures 3 and 4, the genetic background immediately upstream of mcr-9 was consistent among IncHI2 mcr-9-bearing plasmids. The exception is the presence of an ISSgsp1 element from the IS66 family in the INSAq77 mcr-9-harbouring contig, showing 100% of identity with an isolate of Klebsiella pneumoniae collected, in 2012, from a human patient in the USA (CP007734). Furthermore, in this study, the presence of the conserved nickel/copper operon (i.e., rcnA, rcnR, pcoE and pcoS genes), which plays a key role in cop-per tolerance under anaerobic growth and nickel homeostasis in bacteria, was also detected. Downstream of the mcr-9 gene, the nucleotide sequences, including that of INSAq77, were genetically diverse (Figure 3). Indeed, different regions were present among the IncHI2 plasmids also analyzed in this study, namely regions involved in resistance to copper (pcoABCDRSE), silver (silESRCBAP), arsenic (arsCBRH) and/or mercury (merEDACPTR).
Antibiotics 2022, 11, x FOR PEER REVIEW 10 of Figure 4. Schematic representation of the genetic environment of INSAq77 mcr-9.1-containing con with others IncHI2 mcr-9-harboring plasmids. Boxed arrows indicate direction of transcription all genes. Blue bars: normal tblastx matches; red: Inverted matches; depth of shading: percent blast match. Color-coding for the genes inside mcr-9 cassette: dark red, mcr-9 gene; cyan, mob DNA; purple, other genes; grey, other CDSs. The scale is represented in base pairs.
The abundance of mcr variants and alleles in bacteria isolated from aquatic reservo suggests that these enzymes may play another role, namely a defense system agai natural peptides and/or bacteriophages [53]. mcr-9 gene was firstly described in the US in a clinical S. Typhimurium isolate, which demonstrates the high transmission poten of this colistin resistance determinant and places this research in a One Health cont [54].

Study Design and Bacterial Identification
MCR-9-producing Enterobacter sp. INSAq77 was isolated from a seabream (S. aura The abundance of mcr variants and alleles in bacteria isolated from aquatic reservoirs suggests that these enzymes may play another role, namely a defense system against natural peptides and/or bacteriophages [53]. mcr-9 gene was firstly described in the USA, in a clinical S. Typhimurium isolate, which demonstrates the high transmission potential of this colistin resistance determinant and places this research in a One Health context [54].

Study Design and Bacterial Identification
MCR-9-producing Enterobacter sp. INSAq77 was isolated from a seabream (S. aurata) of commercial-size (500-1500 g), which was collected in March 2018 in a land tank from a fish multitrophic farming [55]. This station is in the Ria Formosa Natural Park (south of Portugal) with a semi-intensive production system. Animal welfare was safeguarded during production and transport accordingly with the European Commission SANTE/2016/G2/009 recommendations [56]. Species identification was performed by VITEK ® 2 Automated Identification System (BioMérieux, Marcy-l'Étoile, France), using GN ID card and by amplification of the 16S rRNA gene, as previously described [55].

Whole-Genome Sequencing
DNA was extracted from freshly grown overnight culture (MagnaPure 96 Instrument, Roche, Manheim, Germany) and was quantified using Qubit fluorometer (Thermo Fisher Scientific, Waltham, MA, USA). Dual-indexed Nextera XT kit was used to library preparation followed by paired-end sequencing (2 × 250 bp) on a MiSeq Illumina platform (Illumina Inc., San Diego, CA, USA), according to the manufacturer's instructions.

Plasmid Characterization
BRIG v.0.95 was used to perform a circular comparison between the complete sequence of INSAq77 mcr-9-harbouring contig and the highly similar plasmids detected by performing BLAST against the Microbial Nucleotide BLAST database for complete plasmids (https://blast.ncbi.nlm.nih.gov/Blast.cgi; accessed on 4 January 2022). The genetic environment of the mcr-9 gene was manually revisited using CLC Genomics Workbench v20.0.4 (QIAGEN, Aarhus, Denmark). EasyFig v2.2.5 was used for the visualization and comparison of mcr-9 genetic environment [60].

Conclusions
This work reinforces the knowledge that water environments play a crucial role in the spread of antibiotic resistance and that important antibiotic resistance mechanisms, such as mcr genes conferring low-or medium/high-level resistance to colistin, are also present in aquaculture. This fact, allows antibiotic-resistant bacteria to spread through food and through the environment, resulting in serious threats to human health [61].
The use of phenotypic methods to determine susceptibility to antibiotics may be a limitation, as they may not identify the low expression associated with the presence of a particular gene, as in this case. Thus, the implementation of high throughput methods in laboratories, such as the WGS, will make an important contribution to the detection of under-expressed genes, mostly when they are of clinical importance. Thus, the presence of antibiotic-susceptible isolates in different settings, such as the INSAq77 mcr-9-carrying strain isolated in aquaculture, highlights the risk of the silent dissemination of important resistance determinants, among which, in fact, the genes encoding such PMCR are an important example. Of concern is also the possible co-selection of antibiotic-resistant bacteria when exposed to heavy metals (copper and zinc), often used as growth promoters in aquaculture and terrestrial animal farms.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics11091232/s1, Figure S1: Presence of prophages in INSAq77 genome identified by PHASTER. A total of eleven prophage regions were identified, of which two regions were intact, eight regions were incomplete and one region was questionable. (a) Table with prophage characteristics; (b) Location of predicted prophages within INSAq77 contigs. Region types were marked with colors: intact (green), questionable (blue) and incomplete (red).; Table S1: Loose best-hit results (≥65% of identity), by predicted gene, obtained using the Resistance Gene Identifier (RGI).; Table S2: Results obtained from prediction of a bacteria's pathogenicity towards human hosts using PathogenFinder (https://cge.cbs.dtu.dk/services/PathogenFinder/, accessed on 4 January 2022). Results highlighted in green are those not matching protein pathogenic families.; Table S3: Demographic and genomic characteristics of the E. ludwigii isolates used for the phylogenomic analysis.

Data Availability Statement:
The authors confirm all supporting data, code and protocols have been provided within the article or through supplementary data files. The E. ludwigii whole genome shotgun (WGS) project has the project accession JABRPH000000000. The new bla ACT-type nucleotide sequence was submitted to the GenBank Database as bla ACT-88 with accession number MW887657, after request of the new allele number to NCBI (https://www.ncbi.nlm.nih.gov/pathogens/submitbeta-lactamase/; accessed on 4 January 2022).

Conflicts of Interest:
The authors declare no conflict of interest.