Whole Genome Sequencing and Characteristics of mcr-1–Harboring Plasmids of Porcine Escherichia coli Isolates Belonging to the High-Risk Clone O25b:H4-ST131 Clade B

Porcine Escherichia coli ST131 isolates are scarcely documented. Here, whole genome sequencing and core genome (CG) and plasmidome analysis of seven isolates collected from diarrheic piglets and four from pork meat were performed. All of the 11 ST131 isolates belonged to serotype O25b:H4 and clade B and showed fimH22 allele or mutational derivatives. The 11 porcine isolates possessed virulence traits that classified the isolates as avian pathogenic, uropathogenic, and extraintestinal pathogenic E. coli–like (APEC-, UPEC-, and ExPEC-like) and constituted virotype D. The CG was performed for all porcine isolates in addition to 73 ST131 reference isolates from different origins. Within clade B, the CG showed nine subclusters, allowing us to describe five new subclades (B6, B6-like, B7, B8, and B9). There was an association between subclade B6, PST43, virotype D2, and food origin, whereas subclade B7 included PST9 isolates with virotype D5 from diarrheic piglets (p = 0.007). The distance between human and porcine isolates from subclades B6 and B7 had an average of 20 and 15 SNP/Mb, respectively. [F2:A-:B1]-IncF, ColE1-like, and IncX plasmids were the most prevalent. Besides, IncF plasmids harbored a ColV region frequent among APEC isolates. Antimicrobial resistance genes conferring resistance to penicillin, tetracycline, quinolones, and colistin were the most common. The mcr-1.1 gene was detected in 5 of 11 porcine isolates, integrated into the chromosome of one isolate and into plasmids in the remainder isolates (two MOBH11/IncHI2-ST4, one MOBP3/IncX4, and one MOBF12/IncF [F2:A-:B1] supposedly cointegrated with an IncHI2). The surrounding environments of the mcr-1 cassette showed variability. However, there were conserved structures within the same plasmid family. In conclusion, CG analysis defined five new subclades. The ST131 porcine isolates belonged to new subclades B6 and B7. Moreover, porcine and clinical human isolates were strongly related. The 11 porcine ST131 isolates harbored a wide variety of plasmids, virulence, and resistance genes. Furthermore, epidemic plasmids IncX4 and IncHI2 are responsible for the acquisition of mcr-1.1 gene. We hypothesize that the APEC-IncF plasmid acquired the mcr-1.1 gene via cointegrating an IncHI2 plasmid, which is worrying due to combination of virulence and resistance attributes in a single mobile genetic element.

Porcine Escherichia coli ST131 isolates are scarcely documented. Here, whole genome sequencing and core genome (CG) and plasmidome analysis of seven isolates collected from diarrheic piglets and four from pork meat were performed. All of the 11 ST131 isolates belonged to serotype O25b:H4 and clade B and showed fimH22 allele or mutational derivatives. The 11 porcine isolates possessed virulence traits that classified the isolates as avian pathogenic, uropathogenic, and extraintestinal pathogenic E. colilike (APEC-, UPEC-, and ExPEC-like) and constituted virotype D. The CG was performed for all porcine isolates in addition to 73 ST131 reference isolates from different origins. Within clade B, the CG showed nine subclusters, allowing us to describe five new subclades (B6, B6-like, B7, B8, and B9). There was an association between subclade B6, PST43, virotype D2, and food origin, whereas subclade B7 included PST9 isolates with virotype D5 from diarrheic piglets (p = 0.007). The distance between human and porcine isolates from subclades B6 and B7 had an average of 20 and 15 SNP/Mb, respectively. [F2:A-:B1]-IncF, ColE1-like, and IncX plasmids were the most prevalent. Besides, IncF plasmids harbored a ColV region frequent among APEC isolates. Antimicrobial resistance genes conferring resistance to penicillin, tetracycline, quinolones, and colistin were the most common. The mcr-1.1 gene was detected in 5 of 11 porcine isolates, integrated into the chromosome of one isolate and into plasmids in the remainder isolates (two MOB H11 /IncHI2-ST4, one MOB P3 /IncX4, and one MOB F 12 /IncF [F2:A-:B1] supposedly cointegrated with an IncHI2). The surrounding environments of the mcr-1 cassette showed variability. However, there were conserved structures within the same plasmid family. In conclusion, CG analysis defined five new subclades. The ST131 porcine isolates belonged to new subclades B6 and B7. Moreover, porcine and clinical human isolates were strongly related. The 11 porcine ST131 isolates harbored a wide variety of plasmids, virulence, and resistance genes.

INTRODUCTION
Escherichia coli exists as part of the commensal microbiota in the mammalian digestive tract, as well as a zoonotic pathogen responsible for intestinal and extraintestinal infections in both humans and animals (Poirel et al., 2018). Treatment of multidrug-resistant (MDR) E. coli infections has become a serious clinical issue, especially with the emergence of high-risk clones, such as clone ST131 (Nicolas-Chanoine et al., 2008, 2014. ST131 clone is one of the most prevalent extraintestinal pathogenic E. coli clones (ExPEC). Although it remains unclear why this clone is prevailing over the others, antimicrobial resistance and virulence factors are suspected contributors (Dobrindt, 2005;Stoesser et al., 2016b). Even though ST131 clone is prevalent in human samples, it remains less frequent from animals and especially rare from porcine source (Reid et al., 2019). However, it has already been identified in many animal species, including in Antarctic pinnipeds . The majority of animal isolates belong to clade B (associated with fimH22), whereas in those that cause extraintestinal infections in humans predominate clade C (associated with fimH30) and especially those belonging to subclades C1-nM27, C1-M27, and C2 (H30Rx). This could be the reason of the underrepresentation of ST131 clade B isolates in the literature and sequence databases (Reid et al., 2019).
Besides, several subvariants for some of them have been reported, that is, 13 nucleotide and protein variants in the mcr-1 family (designated mcr-1.1 to mcr-1.13) (Partridge et al., 2018).
The mcr gene has been detected in many plasmid types, including IncI2, IncHI2, IncP, IncX4, IncY, IncF, and ColE10-like ones (Madec and Haenni, 2018) from different origins (Sun et al., 2018). Nevertheless, Matamoros et al. (2017) and García-Meniño et al. (2019) found that the majority of the mcr-carrier plasmids belonged mainly to four plasmid incompatibility groups: IncX4, IncI2, IncHI2, and ColE10-like. In contrast, it remains rare to find colistin resistance genes chromosomally encoded, in accordance to  studies, where the prevalence of chromosomal mcr carrier isolates was estimated in 4% of the analyzed isolates.
Genomic tools allow a compressive characterization of FBPs and the identification of clonal groups of bacteria that represent public health hazards (Kovac et al., 2017). However, epidemiological surveillance of epidemic plasmids related to antimicrobial resistances or/and virulence genes spreading in the bacterial population is still complicated because of the intrinsic plasticity of plasmids (Orlek et al., 2017).
In this study, we have performed the whole genome sequencing (WGS) analysis of 11 ST131 E. coli isolates from porcine samples. Our objectives were (I) perform a core genome (CG) analysis to establish the phylogenetic relationship of our isolates within clade B of ST131 E. coli isolates, (II) to determine and describe the genetic location of the resistance and virulence genes, (III) to investigate the role of mobile genetic elements (MGEs) in the dissemination of those genes, and (IV) to explore the genetic environment of the mcr-1 gene. As far as we know, this study would be the first one in which the whole genome of an extended collection of ST131 clade B isolates from porcine origin, collected from diseased animals and food, has been performed.

Epidemiological Background of the E. coli Collection
In the present study, we performed WGS analysis of a collection of 11 resistant E. coli O25b:H4-B2-ST131 clade B isolates, seven from piglets with diarrhea and four from pork meat. The isolates belong to extensive epidemiological studies accomplished in Spain. The seven isolates from diarrheic piglets were isolated during period 2006(García-Meniño et al., 2018 and the four from pork meat during the years 2011 and 2012 (Herrera Estévez, 2015).

WGS, Assembly, and Primary Analysis
Total DNA was extracted with the QIAmp DNA Mini Kit (Qiagen GmbH, Qiagen Strasse 1, Hilden, Germany). Libraries were prepared by using the TruSeq DNA PCR-Free protocol (Illumina, San Diego, California, United States) at the Genomics and Bioinformatics Core Facility (Centre for Biomedical Research of La Rioja). Paired-end 100-bp reads on fragments of 550-bp insert size were sequenced in an Illumina HiSeq 1500. Genomes were reconstructed by using PLACNETw (Vielva et al., 2017). Identification of Open Reading Frames (ORFs) and genome annotation of the assembled genetic elements was performed by using Prokka (Seemann, 2014). Genomes were in silico typed by the following databases: SerotypeFinder (Joensen et al., 2015), MLSTtyper (Larsen et al., 2012), and CHtyper (Camacho et al., 2009). Three different databases were used for the identification of antibiotic resistance genes: ResFinder (Zankari et al., 2012), CARD Resistance Gene Identifier (McArthur et al., 2013), and ARG-ANNOT (Antibiotic Resistance Gene-ANNOTation) (Gupta et al., 2014). Besides, the PointFinder database (Zankari et al., 2017) was used to determine point mutations. The VirulenceFinder (Ren et al., 2017) and the VFDB (Chen et al., 2005) databases were used to explore virulence factors.
The genetic environment of the mcr detected genes was manually revisited and compared to previously reported ones with CLC Sequence Viewer (version 8.9; Qiagen) and EasyFig tools (Sullivan et al., 2011).

CG and Phylogenetic Analysis
For phylogenetic analysis, we used the 11 genomes of the porcine E. coli sequenced in this study plus 73 full-genomes references from the ST131 clone retrieved from the Enterobase 1 and the NCBI Bioproject Database. 2 The CG was defined as the collection of genes present at least once in all the ST131 genomes analyzed, with more than 90% similarity and 90% coverage, as defined by Lanza et al. (2014). All 84 genomes used in this study and their accession number are available on Supplementary Table S2

Plasmid Analysis
Plasmid reconstruction from WGS data was performed by PLACNETw method (Vielva et al., 2017). In PLACNETw representation, most plasmids can be recognized by their replication initiation proteins (RIPs) and/or RELaxase proteins (REL), both of them considered as plasmid markers. Incompatibility groups (Inc) and pMLST subtypes were in silico determined with PlasmidFinder and pMLST (Carattoli et al., 2014), respectively, via the CGE online services. 3 Besides, reconstructed plasmids were subtyped according to the exact relaxase family subtype by phylogenetic comparison with previously defined relaxases subfamilies (Alvarado et al., 2012). Reconstructed plasmids and references belonging to the same Inc groups were compared by using BRIG (Alikhan et al., 2011) and EasyFig tools (Sullivan et al., 2011). All reference plasmids used in this study were recovered from NCBI database 4 (Supplementary Table S3).

Statistics
Normality test was performed (Shapiro-Wilk test). A nonparametric Mann-Whitney-Wilcoxon test on paired data was conducted. A significance level of 0.05 was used for all tests. All statistical analyses were carried out with XLSTAT software. 5

Pasteur Sequence Types and fimH Alleles
All of the 11 ST131 isolates belonged to serotype O25b:H4 and the phylogenetic group B2. The seven isolates from piglets with diarrhea showed the Pasteur sequence type (PST) 9, whereas the four isolates from pork meat showed PST43 (Table 1).

Resistance Phenotype
The resistance profile of the 11 ST131 porcine isolates was determined, showing that all but one exhibited an MDR phenotype. Amoxicillin/clavulanic acid resistance was present in 10 isolates, quinolone resistance in seven, colistin resistance in five isolates (with mcr-1 gene), cotrimoxazole resistance in four isolates, chloramphenicol resistance in two isolates, aminoglycoside resistance in two isolates, and tetracycline resistance in one isolate ( Table 1).

Whole Genome Sequencing
The draft genomes of the 11 ST131 porcine isolates yielded 32 to 78 contigs larger than 1 Kb, with assembly sizes ranging from 4.940 to 5.451 Mb (Supplementary Table S4). The WGS data were in complete agreement with our previous characterization performed with conventional methods.

CG and Phylogenetic Analysis
A total of 84 E. coli genomes were taken from different clades and subclades of ST131 in order to calculate the ST131 CG. The main objectives were in regard to insights of ST131 lineage and the relationship between isolates of different origins, with special attention to clade B of ST131 lineage. Because of that, the CG analysis was dominated by clade B (n = 49, 58%), and the remainder isolates belonged to clade C (n = 26, 31%), clade A (n = 8, 10%), and clade I (n = 1, 1%). The final collection represented sequences from human (n = 51, 61%), porcine (n = 19, 23%), avian (n = 8, 10%), and wild animals (Antarctic pinnipeds) (n = 6, 7%) sources.
The CG encompasses 2.98 Mb (length = 2,983,333 ± 7,338 pb), which comprised 3,100 total CDS. The single nucleotide polymorphisms (SNPs) analysis of the CG performed within ST131 lineage revealed a well-defined clade structure, similar to those previously described, including the three defined clades: A, B, and C. In total, 14,838 SNPs were present in the alignment. Furthermore, there was a distance of 7,742 to 8,565 SNPs between isolates from clade A and clade C; 6,508 to 8,675 between isolates from clade A and clade B; and 2,214 to 4,280 between isolates from clade B and clade C. The distance within clades varies greatly: being 36 to 1,782 SNPs in clade A, 8 to 2,051 SNPs in clade B, and 1 to 723 SNPs in clade C (Figure 1 and Supplementary Table S2).
All 13 MOB F12 /IncF plasmids were carriers of virulence genes. The genes iroBCDEN, iss, mchF, iutA, and iucABCD were constant ( Table 2). All the MOB F12 /IncF plasmids found in our study were structurally compared by BRIG software, using the pJIE186_2 (NC_020271) plasmid as reference (Supplementary Figure S2). The plasmids analyzed showed extensive sequence similarity over at least 100 kb, which includes not only backbone genes. The comparison emphasizes that all of them share an 80kb conserved region comprising the virulence genes described previously as well as cvaABC, sitABC, ompT, and hlyF.
The eight MOB P51 /ColE1-like plasmids identified in this study exhibited a wide size range (1.7-6.8 kb) and were  analyzed using the ColE1 plasmid (J01566.1) as reference (Supplementary Figure S3). The plasmids pLREC153_3, pLREC157_3, pLREC158_2, and pLREC161_3 (all of them >6.8 Kb) showed the highest homology to the reference plasmid. All of them carried the colicin E1 cea and immunity (imm) genes and the entry exclusion (exc1-exc2) system genes. On the other hand, plasmids from isolate LREC_160 were different among them and with the others, exhibiting a wide size range (1.8-5.6 Kb), and only pLREC160_4 carried the colicin E1 gene. Three MOB P3 /IncX plasmids were present among our isolates. We found pLREC157_2 that belong to the IncX1 subcluster and had a similar backbone than p2ESCUM (NC_011739.1) but codes for a completely different RIP (Supplementary Figure S4), whereas pLREC154_2 and pLREC160_2 belong to the IncX4 subcluster and were analyzed using pSH696_34 (JX258654.1) as a reference (Supplementary Figure S5). Because pLREC160_2 carried the mcr.1-1 gene, some previously reported IncX4 reference plasmids harboring the mcr gene were added. Compared to pSH696_34, plasmid pLREC160_2 and the other mcr carrier plasmids (pESTMCR and pICBEC7Pmcr) lacked two important backbone genes: the conjugative coupling protein gene (traG) and the replication protein encoding region (rep).
Structural comparison of the two MOB P12 /IncI1-ST171 plasmids and the single MOB P131 /IncL/M plasmid found within our isolates was also performed using pEK204 (EU935740) and pEC743-OXA48 (CP015071.1) as references, respectively. Both showed a high coverage of homology with the reference (Supplementary Figures S6, S7).
We also described five small (1.5 kb) multicopy plasmids in which any relaxase protein was found, categorized as cryptic. Those cryptic no-MOB plasmids were analyzed by using pEC10D (NC_017650.1) as reference (Supplementary Figure S8). BRIG comparison showed that four plasmids were highly similar to pEC10D and codes for the same replication protein (cryptic 1, from the HTH36 superfamily replication proteins), whereas pLREC159_6 codes for a different replication protein and was called cryptic 2 ( Table 2).
Finally, we analyzed the MOB V2 plasmids found in the isolates LREC_153 and LREC_157. A BLAST research was performed to identify a reference plasmid, being pEC0674 (MF684783.1) as the closest match. Two MOB V2 -like-related plasmids were also included for structural comparison (Supplementary Figure S9). Although no significant similarity was found between these latter two references and the pEC0674, the pLREC153_4 showed similarity with the replication protein, a hypothetical one and the Rec protein from the reference plasmid, whereas the pLREC157_4 only showed similarity to the latter two proteins.

Antimicrobial Resistance Genes
Many antimicrobial resistance genes (ARGs) were found in the 11 ST131 porcine isolates ( Table 2), including acquired resistance genes, point mutations, and efflux/transporter genes.
The phenotypic resistance profile in the studied genomes was due to genes carried by plasmids, except for quinolone resistance, which in our study is mediated by a Ser83Leu point mutation found in the chromosomal gyrA gene. In the case of LREC_158 isolate, it exhibited the bla TEM genes (−1c, −40, and −135) variants set into a chromosomal assembled contig. However, these genes were located into IncL/M, IncX4, IncI1, IncF, and IncHI2 in the other analyzed isolates ( Table 2).

Colistin Resistance Vehicles and Genetic Environment
Colistin resistance by mcr gene was found in five of the analyzed isolates. Therefore, the mcr-carrying plasmids were investigated in greater depth. The mcr-1.1 gene was located in two MOB H11 /IncHI2-ST4, a MOB P3 /IncX4 and a MOB F12 /IncF [F2:A-:B1], which was theoretically cointegrated with an IncHI2-ST-nt plasmid. Besides, the LREC_154 isolate showed the mcr-1.1 in a chromosomal location ( Table 2). In this study, any point mutation in the chromosomally encoded genes pmrA or pmrB was found.
The complete sequence of pLREC160_2 (MOB P3 /IncX4) was 40,221 bp in size and contains 69 predicted ORFs. A structural comparison against other reference IncX4 plasmids is shown in Figure 2, where pICBEC7Pmcr (34,992 pb, NZ_CP017246.1) was used as the internal reference. The structure of the mcr-1 carriers IncX4 plasmid remains stable if we exclude one ORF encoding for an IS91 family transposase present FIGURE 2 | Structural comparison between mcr-1 plasmids and IncX4 references. The alignment includes five references and the one IncX4 mcr-1-bearing plasmid found in our study. The plasmid pICBEC7mcr (IncX4) was used as a reference to match with the other plasmids with [pMCR_WCHEC1618, pECGD-8-33, pESTMCR, and pmcr1_IncX4] and without the mcr-1 gene [pIJE143]. The outer circle with black arrows denotes the annotation of reference sequence pICBEC7mcr. The image was generated using BRIG (default parameters with 90/70 as upper/lower threshold).
in pLREC160_2 and pICBEC7Pmcr. The BRIG comparison showed that the mcr-1-negative plasmid pJIE143 differs from pICBEC7Pmcr in a region of approximately 6 kb (encoding for parA, mcr, and pap2 genes). All the analyzed plasmids showed a typical IncX4 backbone, including the RIP gene (pir), conjugal transfer protein genes (tra) plus taxABC, and pilX operons (Supplementary Figures S10-S13).
Plasmid pLREC161_1 was 330,357 bp in size and contains 616 predicted ORFs. This MOB F12 plasmid belongs to the IncF [F2:A-:B1] incompatibility group and was theoretically cointegrated with an IncHI2 plasmid. A structural comparison with plasmid pMR0516mcr [F18:A-:B1] (KX276657.1) as an internal reference (225,069 bp) showed 100% identity in more than 70% of query coverage from BRIG (Figure 3). Genes involved in the conjugative transfer (tra, trb, and pili operons) in the toxin-antitoxin system (vagC) and the inhibition of SOS response (psiAB) were present. Most of the differences were observed in the resistance module. It is worthy of note that pLREC161_1 and pMR0516mcr plasmids contained APEC virulence genes and that the surrounding region of the mcr gene includes heavy metal resistance genes to tellurium (ter genes). Moreover, in Supplementary Figure S14, where pKP81_BE [F2:A-:B] (KU994859.1) was used as the reference, we observed 100% of identity in more than 78% of query coverage from BRIG with pLREC161_1. Interestingly, plasmid R478 (IncHI2 reference plasmid, U62007.2) also shared with pKP81_BE some backbone genes and the tellurium resistance module like pMR0516mcr (Supplementary Figure S14). A wide part of the backbone in pLREC161_1 (approximately 44 kb) was common with the widely studied IncF plasmid from E. coli-K12, used as nonmcr plasmid reference (from coord. 55 to 99 kb). The ancestral IncF plasmid of ST131 (pECSF1) was also included in BRIG comparisons (Supplementary Figure S15).
There were two MOB H11 /IncHI2-ST4 mcr-1-bearing plasmids: (I) pLREC159_1, which was 248,461 bp in size and contains 420 predicted ORFs, and (II) pLREC176_1, which was 278,079 bp in size and contains 523 predicted ORFs. Both showed 100% of identity over more than 75% of query coverage in BRIG comparison with pHNSHP45-2 (KU341381.1) (Figure 4). The backbone was almost identical; nevertheless, FIGURE 3 | Structural comparison between mcr-1 plasmids, IncF, and IncHI2 references. The alignment includes four references and the one IncF mcr-1-bearing plasmid found in our study. pMR0516mcr (IncF) was used as a reference to match with the other plasmids with [pKP81_BE (IncF)] and without mcr-1 gene [IncF plasmid from Escherichia coli K12, pECSF1 (IncF) and R478 (IncHI2)]. The outer circle with black arrows denotes the annotation of pMR0516mcr. The image was generated using BRIG (default parameters with 90/70 as upper/lower threshold). the resistance module was distinct between them. The result of this comparative analysis showed that a wide range of resistance genes could be found in the MDR module of IncHI2 plasmids. Moreover, all the IncHI2 plasmids analyzed presented a tellurium resistance module close to the mcr-1 cassette. The similarities between IncHI2 plasmids were high, even though their origin and parental isolate characteristics differ.
Afterward, a comparison between all our mcr-1-bearing plasmids was performed with pHNSHP45-2 (MOB H11 /IncHI2) as a reference, and as expected, almost none homologous region was found between plasmids from different families (Supplementary Figure S16). Nevertheless, pLREC161_1 (MOBF 12 /IncF [F2:A-:B1] plus IncHI2-ST-nt) plasmid shared a part of his backbone with the reference one, including the module encoding for tellurium resistance and the mcr gene, suggesting that the mcr-1 cassette was introduced into pLREC161_1 through the cointegrated IncHI2 plasmid.
We had also made a sequence alignment to analyze the surrounding region of our mcr-bearing contigs and compare their structure against plasmids references (Figure 5). Regarding plasmid families, all the IncX4 plasmids were identical in terms of the genetic arrangement in the vicinity of the mcr-1 gene, whereas the insertion site of mcr-1 gene in IncHI2 plasmid was variable. Interestingly, the mcr region in plasmids pMCR_WCHEC050613 (IncHI2) and plasmid pMR0516mcr (IncF) shared 100% of identity. It is also worth to note that the IncF mcr harboring plasmid pKP81_BE carried the tellurite resistance module surrounding the mcr gene like the IncHI2 plasmids.
Finally, the genetic arrangement of the mcr-1 cassette was investigated (Supplementary Figure S17). All sequences share a 2,542-bp DNA segment, containing mcr-1.1 and a putative ORF, encoding a Pap2 superfamily protein. However, some variability in the last 3 end of the pap2 gene was observed.
The genome reconstruction of LREC_154 showed that mcr gene was inserted into the chromosome. However, the mcr harboring contig included only mcr and pap2 genes. It was flanked at both ends by a partial ISApl1 sequence of 99 bp upstream and 405 pb downstream the mcr-1 cassette. In pLREC159_1, the mcr harboring contig was disrupted after a segment of 87 bp of DNA homologous to ISApl1 upstream the FIGURE 4 | Structural comparison between mcr-1 plasmids and IncHI2 references. The alignment includes four references and the mcr-1-bearing plasmids from our study pLREC176_1 and pLREC159_1, which are IncHI2 and pLREC161_1, which is an IncF plasmid with a presumptively IncHI2 cointegrated. The plasmid pHNSHP45-2 (IncHI2) was used as a reference to match with the other plasmids with [p2474-MCR1 (IncHI2), pECJS-59-244 (IncHI2), and pMCR_WCHEC050613 (IncHI2)] and without the mcr-1 gene [R478 (IncHI2)]. The outer circle with black arrows denotes the annotation of reference sequence pHNSHP45-2. The image was generated using BRIG (default parameters with 90/70 as upper/lower threshold). mcr gene. Probably the presence of the IS (repeated element) has interfered with the de novo reconstruction of the sequenced genome by the assembly of Illumina short reads. In addition, we identified the IS26 and an IS1294 surrounding the mcr-1 cassette in pLREC160_2.

Statistical Analysis
The variables (cdtB, cnf 1, hlyA, virotypes, PST, and subclades) did not follow a normal distribution (Shapiro-Wilk test: p < 0.0001). We performed the Mann-Whitney-Wilcoxon test to assess significative differences in the distribution of features (virotype D5| subclade B7| PST9| cnf1| hlyA vs. virotype D2| subclade B6| PST43| cdtB variables; p = 0.007) between ST131 isolates from piglets with diarrhea and pork meat.  1) containing mcr-1 gene with our sequences and other frequently used reference sequences analyzed in this article. Open arrows represent coding sequences (green for tellurium resistance, yellow for ISApl1, red for mcr-1, and blue for pap2) and indicate direction of transcription. The arrow size is proportional to the gene length. The shadow parallelograms denote genetic regions that exhibit sequence homology among different segments. Light shadow denotes regions with a lower level of sequence identity (99%) by BLAST. The image was generated using EasyFig (default parameters). In the present study, we performed WGS analysis of a collection of 11 representative ST131 isolates obtained from piglets with diarrhea and pork meat in Spain. All of the 11 ST131 isolates belonged to serotype O25b:H4 and clade B and presented a high number of virulence genes typical of E. coli that cause extraintestinal infections. In addition, the 11 porcine isolates harbored traits that genetically satisfied the criteria for designation as ExPEC, UPEC, and APEC status being designated as APEC-, UPEC-, and ExPEC-like isolates because of their site of isolation. Although seven of the 11 ST131 isolates were isolated from piglets with diarrhea, none of them were positive for virulence genes of intestinal pathogenic E. coli .
The CG SNP-based phylogenetic tree of 84 human and animal ST131 isolates analyzed in this study showed (Figure 1) that they are mostly clustered into three well-defined clades (A, B, and C), and similarly to Petty et al. (2014) study, we have observed that clade A is the most divergent from clades B and C. Our results also agree with those published by other authors (Ben Zakour et al., 2016;Matsumura et al., 2017;Liu C. M. et al., 2018;Decano and Downing, 2019) regarding that clade B isolates are displayed into multiple subclades. We identified the six subclades (B0, B1, B2, B3, B4, and B5) described by Ben Zakour et al. (2016) and five new ones (B6, B6-like, B7, B8, and B9). Notably, the 11 ST131 porcine isolates from this study were grouped in the new subclades B6 (four isolates from pork meat and virotype D2) and B7 (seven isolates from piglets with diarrhea and virotype D5). The new subclade B6-like also includes two isolates of porcine origin, whereas the new subclades B8 and B9 included only ST131 isolates from Antarctic pinnipeds. In contrast, avian isolates of virotype D4 were distributed in the subclades B2, B3, and B5.
The newly described B6 and B7 subclades were primarily associated with porcine isolates, but also include some isolates causing bacteremia and other extraintestinal infections in humans. Therefore, it is remarkable how those human isolates were closely related (46-60 SNPs average distance) with porcine isolates. Those results suggest the ecological overlap of human and animal isolates. Recently, Reid et al. (2019) compared the genomes of two ST131 clade B fimH22 isolates obtained in Australia, the first one isolated in 2007 from a healthy piglet and the second one obtained in 2009 from a human urinary tract infection. This study showed that the two isolates were highly related, separated by only 20 core SNPs (Reid et al., 2019). Thus, the analysis of the phylogeny confirmed E. coli ST131 as pathogen capable of frequent interspecies movements.
The plasmidome of E. coli ST131 clade B from porcine origin had shown variability and some interesting differences concerning the actual published data. IncF plasmids are a common feature in all ST131 isolates , having at least one per isolate in this study. Specifically, F1:A2:B20 and F2:A1:B-plasmids are more commonly associated with the C1-H30R clade and C2-H30Rx, respectively (Johnson et al., 2016). In contrast, F2:A-:B1 plasmids were dominant in the porcine clade B isolates used in this study. This type of plasmid was absent in 35 non-ST131 E. coli porcine isolates from colibacillosis (García-Meniño et al., 2019), suggesting that IncF plasmids vary greatly within the different lineages. Reid et al. (2019) reported 23 F-types among 282 isolates from a mixture of origins belonging to the clade B of ST131, and the F2:A-:B1 IncF subtype was one of the commonest (32%).
In accordance with previous studies, plasmids carrying virulence genes belong to the IncF family Nicolas-Chanoine et al., 2017). All the IncF plasmids found in this study conserved the ColV region, frequently detected among avian pathogenic E. coli (APEC) (Liu C. M. et al., 2018). This region comprises the iss (increased serum survival) gene, the iroBCDEN gene cluster (encoding the salmochelin siderophore system present in Salmonella enterica spp. and some ExPEC isolates), the iucABCD operon (encoding the enzymes of the biosynthetic pathway for aerobactin), iutA (outer membrane receptor of ferri-aerobactin complexes) gene, cvaB and cvaC (production of colicin) genes, sitC (Salmonella iron transporter C), and an ompT-hlyF-mig14 cassette (outer membrane related genes). Some APECs cause human infection (Liu C. M. et al., 2018); nevertheless, ColV plamids were absent in specific human clades, so they may not be a necessary feature to cause human infection (Reid et al., 2019). It is currently unknown which advantages could the IncF plasmids confer to the bacterial cell apart from virulence and resistance to antibiotics. However, Johnson et al. (2016) has hypothesized that they must play multiple roles in the host success because ST131 clone has adapted to harbor these plasmids at lower fitness cost and has stabilized them through mutational events. We assume that the presence of both, resistance and virulence genes, on the same MGE is worrying for public health, as shown in our study in IncF plasmids such as pLREC161_1 and pLREC 159_2. For instance, the evolutionary analysis performed by Ben Zakour et al. (2016) proved that ST131 emergence was adapted by isolates that had acquired first a subset of virulence genes followed by the gain of antibiotic resistances.
ColE1-like plasmids were the second more prevalent in our collection and the replication protein could not be determined; these results are in accordance with the findings of Lanza et al. (2014). Furthermore, ColE plasmids have been reported as mcr-4 and mcr-5 carriers Carattoli et al., 2017;García-Meniño et al., 2019). The ColE1-like plasmids from this study did not carry resistance genes. However, they might be an important adaptive weapon. Moreover, we described one MOB P131 and two MOB V2 plasmids, which are not frequently documented and therefore absent among the description given by Lanza et al. (2014) over ST131 plasmidome, although another MOB P131 plasmid has been recently found from a bla OXA−48 E. coli ST131 isolate collected from a patient with urinary infection (Stoesser et al., 2016c), and MOB V2 plasmids have been described in other STs, such as ST69 from a human urine sample and ST93 from chicken meat samples (de Been et al., 2014). Besides, MOB V2 plasmids pLREC153_4 and pLREC157_4 showed structural homology (around coordinates 3.7-7 kb in Supplementary Figure S9) with the pEC0674 plasmid (6 kb), an mcr-5 carrier plasmid belonging from an ST29 E. coli isolate recovered from porcine feces in Germany (Hammerl et al., 2018). Finally, and also barely documented, were the cryptic plasmids found in our study. Similar small no-MOB plasmids such as pLREC158_3, pLREC159_5, pLREC159_6, pLREC160_7, and pLREC161_4 have been reported among ST131 isolates. The role of these plasmids is still unknown, but researchers have suggested that they could be providing useful mobilization tools for ST131 adaptation .
Antimicrobial resistance genes conferring resistance to penicillin/ampicillin, tetracycline, quinolones, and colistin were the most common determinants. García-Meniño et al. (2019) described chromosomal mutation (PmrB V161G or PmrA S39I) in addition to the plasmid mechanism (mcr) in 16 (46%) of non-ST131 colistin-resistant porcine E. coli isolates. Here, any complementary chromosomal mechanism related with system PmrAB was found. Nonetheless, some drug efflux genes were found in plasmids. Members of the lnu (previously lin) gene family (mediate lincosamide resistance) (Achard et al., 2005;Roberts, 2008) and the mef (B) gene, which coexists with the aadA and sul3 resistance genes (pLREC161_2). This combination has previously been described in plasmids from porcine E. coli isolates and mediates macrolide resistance (Liu et al., 2009).
Currently, the mcr-1 cassette was found in a large diversity of plasmids and clonal lineages of E. coli. The broad adaptation of colistin resistance all over the world could be due to the ease with which the mcr-1 gene integrates into various regions (Hadjadj et al., 2017). There is an association with successful plasmid families (Matamoros et al., 2017;Zhong et al., 2018). On the other hand, any association with clonal lineages of E. coli is described . However, the description of mcr-1positive E. coli ST131 isolates is exceptionally worrying. Because ST131 has once been responsible for the worldwide appearance, ESBL resistances due to the acquisition of specific IncF epidemic plasmids harboring bla CTX−M−15 gene (Nicolas-Chanoine et al., 2008, 2014Mathers et al., 2015) and therefore major therapeutic failure could come due to the association of mcr-1 gene with other broad-spectrum resistance mechanisms, such as ESBLs and/or carbapenemases (Du et al., 2016;Haenni et al., 2016;Li et al., 2016;Olaitan et al., 2016;Skov and Monnet, 2016). Further, clade B of ST131 has been signaled out as a potential foodborne uropathogen and as a stronger early biofilm producer than clade C of ST131 (Nicolas-Chanoine et al., 2017;Liu C. M. et al., 2018;Flament-Simon et al., 2019).
In this study, the genetic context of mcr-1 gene was diverse. Recent studies propose that the transposon Tn6330, flanked at both ends by ISApl1, has led the mobilization of mcr-1 gene. Therefore, four structures of mobile elements carrying mcr-1 have been described: (I) the composite transposon Tn6330 (ISApl1-mcr-1-orf-ISApl1 structure); (II) a single ISApl1 upstream (ISApl1-mcr-1-orf structure); (III) a single ISApl1 downstream (mcr-1-orf-ISApl1 structure); and (IV) and sequences lacking ISApl1 altogether (mcr-1-orf structure), although a few other truncated or interrupted versions have been reported (Pham Thanh et al., 2016;Snesrud et al., 2016). In the study carried by Snesrud et al. (2018), prevalence rates of 11% for structure I, 22% for structure II, 1% for structure III, and 66% for structure IV (with no associated IS) have been described. Besides, the upstream ISApl1 element was observed in 78% of IncHI2 plasmids (Matamoros et al., 2017), whereas the mcr-1-orf structure has been described as dominant in small mcr-1-IncX4 plasmids (Li et al., 2017b;Sun et al., 2017). These results are in line with our findings. The five ST131 colistin-resistant isolates from this study presented the mcr-1.1 allele (KP347127.1) and showed the mcr-1-orf structure.
However, pLREC159_1 (IncHI2) presumptively harbored the ISApl1 upstream, whereas any copy of ISApl1 was present around the mcr-1 gene in pLREC160_2 (IncX4). Nevertheless, pap2 gene and ISs IS26 and IS1924 were present in pLREC160_2. Zhong et al. (2018) suggested that for IncX4 plasmids IS26/26 structures could be involved in mcr-1 gene mobilization. The BRIG comparison showed that the mcr-1-harboring IncX4 plasmids were genetically homologous between each other as previously described , suggesting that mobilization of the mcr-1 cassette was quite stable within this type of plasmids. Furthermore, pLREC159_1 and pLREC176_1 were almost identical in structure around the mcr-1 gene and were related to the presence of tellurium resistance genes, as well as for pLREC161_1. This heavy metal resistance lends HI2 plasmids to success in wastewater environments (Reid et al., 2019).
Moreover, a highly relevant finding of the present study was the appearance of chromosomally encoded colistin resistance. We found a presumptive ISApl1-mcr-1-orf-ISApl1-like structure. This genetic arrangement was described as the most prevalent by  in bacterial chromosomes. It has been proposed that the lack of one or both of the ISApl1-elements implies stabilization of the mcr-1 gene (non-transposable) (Snesrud et al., , 2018. Thus, chromosomally encoded colistin resistance might mean a more stable inheritance in ST131 porcine lineage. Nevertheless, it also implies less mobilization among isolates. Some studies have proved that stopping using colistin in positive mcr-1 carriers porcine farms has resulted in the disappearance of the resistance (Randall et al., 2018). However, because of the MDR profile of our ST131 isolates, this might not be that easy, due to coselection by other antimicrobials and host fitness adaptation. The fitness effects of carrying the mcr-1 gene were evaluated by Wu et al. (2018). They proved that IncI2 and IncX4 mcr-1-positive plasmids conferred fitness advantage for its host. In contrast, IncHI2 plasmids imposed a slight fitness cost and competitive disadvantage.
To sum up, it seems that epidemic plasmid types IncX4 and IncHI2, which have been reported as the more prevalent carriers of mcr-1 gene, are responsible for the acquisition of the colistin resistance in the high-risk clone ST131 of E. coli in our isolates from porcine origin. Worryingly, the mcr-1 gene has also integrated an IncF plasmid from ST131. The clinical significance of this category of plasmids is therefore highlighted because they are responsible for the epidemic dissemination of ESBL resistance worldwide (Johnson et al., 2016). We hypothesize that IncF plasmids such as pLREC161_1 had acquired the mcr gene via cointegrating an IncHI2 mcr-1-harboring plasmid. The mcr-carrying IncF plasmid pMR0516mcr that has been described by McGann et al. (2016) also shared 89 kb with pHNSHP45-2 (mcr-1-carrying IncHI2 plasmid), and possibly the same phenomenon had occurred.
The impact that humans have on the environment and other species is evident. A multidisciplinary point of view, taking into account the environment, animals, and human for the surveillance of health, is given by the concept of "One Health" (Destoumieux-Garzón et al., 2018). This philosophy fits perfectly the actual global situation. Measures to reinforce prevention and proper management for overall antimicrobials might be the right strategy to fight against antibiotic resistance.
Further studies to understand why some plasmids evolve to become genetically stable are necessary to be able to anticipate the evolutionary dynamic of high risk clones.
As a limitation, it is worth mentioning that short reads obtained by WGS do not always allow a perfect genome reconstruction due to the high amount of repeated sequences in the bacterial genomes. Long read sequencing technologies will probably bring new perspectives in future researches.

CONCLUSION
Clade B of ST131 showed a huge genetic diversity, and five new subclades were defined (B6, B6-like, B7, B8, and B9). The majority of ST131 porcine isolates belong to new subclades B6 and B7. An association between the phylogeny, the virotype, and the origin of porcine isolates was established. Some porcine and human clinical isolates were highly related. Most porcine ST131 isolates are MDR (91%) and carry many ARGs. Colistin resistance was introduced through MGEs and had been able to stabilize chromosomally. IncX4 and IncHI2 epidemic mcr-1-harboring plasmids are responsible for acquired colistin resistance encoded by mcr-1.1 gene. The surrounding environment of the mcr-1 cassette is variable but within the same family of plasmids, insertion, and stabilization had common structures. The plasmidome of ST131 clade B is distinct from other clades within ST131 clone and contains an impressive variety of different plasmids. Characteristically ST131 porcine isolates have APEC [F2:A-:B1]-IncF plasmids. ColE1like plasmids and IncX plasmids are also frequently observed. This behavior could be related to the clonal success due to its ability to easily rearrange DNA by horizontal gene transfer.

DATA AVAILABILITY STATEMENT
The datasets of the 11 LREC genomes for this study can be found in the NCBI sequence databases as part of BioProject SUB5714329. With accession codes SAMN11936814 to SAMN11937940. Accession in NCBI is PRJNA546088.

AUTHOR CONTRIBUTIONS
S-CF-S, MT, AM, VG, IG-M, DD-J, and AH undertook the laboratory work. MT and JB conceived the concept for the manuscript and designed the experiments. All authors provided the critical input and contributed to the writing of the manuscript and approved the final version.

FUNDING
This study was supported by projects: PI16/01477 from Plan Estatal de I+D+I 2013-2016, Instituto de Salud Carlos III (ISCIII), Subdirección General de Evaluación y Fomento de la Investigación and Fondo Europeo de Desarrollo Regional (FEDER); AGL2016-79343-R from the Spanish Agencia Estatal de Investigación (AEI) from the Spanish Ministerio de Economía y Competitividad (MINECO) and FEDER; and ED431C2017/57 from the Consellería de Cultura, Educación e Ordenación Universitaria of Xunta de Galicia and FEDER.

ACKNOWLEDGMENTS
S-CF-S acknowledges the FPU programme for her grant (FPU15/02644) from the Secretaría General de Universidades, Spanish Ministerio de Educación, Cultura y Deporte. IG-M and VG acknowledge the Consellería de Cultura, Educación e Ordenación Universitaria, Xunta de Galicia for his predoctoral grant (ED481A-2015/149) and her postdoctoral grant (ED481B2018/018), respectively. AM acknowledges the Ministerio de Educación, Cultura y Deporte (Spain) for the mobility grant PRX16/00023 for teachers and researchers from the Programa Estatal de Promoción del Talento y su Empleabilidad, Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016. She also expresses her gratitude to Dr. Mark Achtman and his group for their invaluable help with Enterobase during the stay at the Microbiology and Infection Unit, Warwick Medical School, University of Warwick, Coventry, United Kingdom.