Swine ExPEC from China exhibit a highly diverse population structureThe ExPEC were isolated from diseased tissues of pigs from different piggeries in China, most of which were located in Central China. To ensure that each strain originated from distinct pigs, we exclusively isolated and preserved one representative ExPEC strain from each pig farm at a given time. In this study, we randomly selected 50-90 isolates for each year from 2011 to 2017 (Supplementary Fig. 1a), resulting in a total collection of 499 E. coli strains from 23 Chinese provinces, mainly from Hubei, Henan and Hunan provinces (Supplementary Data 1). These isolates were obtained from various organs of pigs, with 79.4% originating from the lung (Supplementary Fig. 1b). Each genome of the studied isolates was sequenced, producing approximately 1 Gbp of reads, which were subsequently assembled into contigs for further analysis.We first performed comparative genomic analysis to gain insight into the diversity and population structure of ExPEC from swine in China. Phylogenetic analysis revealed that over 50% of the isolates belonged to group B1 (54%), with 30% originating from group A, 6.2% from group F, and 5.6% from the recently defined group G (Fig. 1 and Supplementary Data 1). Very few E. coli isolates studied here resided in phylogroups B2, D and E, with proportion for each less than 3%. Based on in silico multilocus sequence typing (MLST), 490 of 499 E. coli could be assigned to 75 sequence types (STs), with 9 strains representing novel STs. The top 5 STs with the most isolates (>5%) including ST410 (14.8%), ST88 (9.2%), ST101 (8.2%), ST117 (5.6%) and ST156 (5.2%), accounting for above 40% of the entire collection (Supplementary Fig. 2a). Among these STs, ST410, ST88, ST101, and ST156 belonged to phylogroup B1, while ST117 was categorized in phylogroup G. The remaining E. coli exhibited a high degree of diversity, with 60 STs, each comprising fewer than 10 members. We also inspected the isolated organs and times of all the E. coli on the basis of phylogenetic analysis. Our findings revealed that strains from a special organ or time point were distributed throughout the whole phylogenetic tree, suggesting an absence of association between the source or time of isolation and the phylogroup or ST. We further assessed the diversity of this E. coli collection by serotype prediction. The high level of diversity was confirmed by the results about H- and O- serotype combinations. Two hundred serotypes were found (Supplementary Data 1 and Supplementary Fig. 2b). The most common serotype was O89:H9, which contained 31 strains accounting for 6.21% of the total collection. The remaining serotypes included less than 20 isolates, and 177 serotypes included less than 5 isolates.Fig. 1: Maximum likelihood phylogenetic tree of 499 swine-derived ExPEC isolates.From left to right, the four columns display the primary tissue sources, the isolation time, the phylogroups, and prevalent STs of the isolates. Source data are provided as a Source Data file.Additionally, to understand the epidemiology of these swine ExPEC strains, we compared them with all E. coli strains present in the EnteroBase database using core genome multi-locus sequence typing (cgMLST). As a result, we identified 39 cases of genetic relatedness between swine ExPEC and isolates collected from different sources, with a maximum separation of 10 alleles (Supplementary Data 2). Among these occurrences, 21 instances involved E. coli strains from the Enterobase database, with explicitly defined host types, including human, swine, and bovine. Human-host instances were the most frequently observed, with a total of 15 swine ExPEC isolates showing genetic relatedness to E. coli from various human body sites, such as blood and rectum. Among these, two strains belonging to ST648 (A71 and A91) and one human blood infection strain (EnteroBase ID, ESC_LA6470AA) differed by only 2 alleles, indicating very closely related genomes. Moreover, swine-derived ExPEC belonging to ST410, ST101, ST131, ST167, or ST744 were also observed to share a common clonal origin with human-derived E. coli. This result underscores evidence of the zoonotic potential of ExPEC strains originating from swine.Genes encoding virulence factors associated with extraintestinal infections are prevalent in swine ExPECVirulence factors (VF) play crucial roles in the interaction between E. coli and their vertebrate hosts. They have been used to define the pathogenicity features of E. coli, including those of different pathovars of intestinal pathogenic E. coli (IPEC) and ExPEC40. To infer the relationship between swine ExPEC and enteric E. coli, the characteristic toxins of enteric E. coli pathotypes were predicted. We found that toxin genes determining enteric diseases were rarely present in these ExPEC strains (Supplementary Fig. 3). Specifically, 49 strains harbored one or more of the following: the Shiga-toxin (STX), heat-labile (LT) enterotoxin, heat-stable (ST) enterotoxin, intimin, and fimbrial adhesins (F4, F5, F18, and F41). These strains mainly belonged to phylogroups F, A, and B1. It is noteworthy that these strains may represent hybrid pathotypes. Therefore, based on these virulence factors, we classified these 49 strains as four hybrid pathotypes, including ExPEC/ETEC, ExPEC/EPEC, ExPEC/EHEC, and ExPEC/EHEC/ETEC (Supplementary Data 3). The most prevalent hybrid pathotype is ExPEC/ETEC, comprising 39 strains in total. Roughly 70% of these strains contain the F5 fimbriae, all of which belonged to ST354/ST457.We subsequently surveyed the five categories of VFs commonly found in human ExPEC41 within the studied E. coli genome sequences. Regarding adhesins, genes associated with type 1 fimbriae (fim) and curli fibers (csg) were present in approximately 90% of the strains (Supplementary Data 3 and Supplementary Fig. 3). P fimbriae (papC) and antigen 43 (agn43) were identified in 10% and 18% of the strains, respectively, while other genes afa and foc were found in less than 5% of the strains. When considering iron uptake, the iron transport gene (sitA), aerobactin siderophore gene (iutA), and salmochelin siderophore gene (iroN) and were present in 69%, 66%, and 51% of the strains, respectively. While genes involved in yersiniabactin biosynthetic (irp2), heme uptake (chuA), and iron regulation (ireA) were detected in 36%, 16%, and 10% of the strains, respectively. When it comes to serum resistance, genes involved in protecting bacteria from complement-mediated lysis (traT), increased serum survival (iss) and colicin V synthesis (cvaC) were identified in 74%, 69%, and 32% of the genomes, respectively. Additionally, other genes within this category, namely ompA and kpsM, were detected in 11% of the strains. As for the invasins, gene ibeA was found in only 4% of the strains. In terms of toxin genes, those associated with vacuolating autotransporter toxin (vat) and serine protease (pic) were present in 7% and 6% of the genomes, respectively. Other toxin genes such as hly (alpha-hemolysin), cdt (cytolethal distending toxin), and cnf1 (cytotoxic necrotizing factor 1) were detected in less than 3% of the genomes.About 97% of swine ExPEC are classified as multidrug-resistant (MDR)In the past decades, antibiotics were heavily used in poultry and livestock industries in China42. It has been reported that E. coli from food animals developed extreme resistance to frequently used antibiotics29,43,44. To assess the antibiotic resistance profiles of this E. coli collection, we conducted susceptibility test of these isolates to 20 antibiotics from 7 different classes, and found that all the isolates showed resistance to at least one of the used antibiotics (Supplementary Data 4). Among the 485 isolates with dependable antibiotic susceptibility outcomes (each test was validated using control strains, and only those passing quality control were considered reliable), MDR (multidrug-resistant, resistance to antibiotics from at least three classes45) strains accounted for up to 97% of the isolates (Fig. 2a). In addition, 239 of the 485 (49.3%) MDR strains were resistant to agents spanning all 7 classes, with strains A376 and A430 exhibiting resistance to 18 antibiotics tested. Notably, among these MDR strains, 46 were hybrid pathotypes, of which 54.3% displayed resistance to agents across all 7 classes (Supplementary Data 4).Fig. 2: Prevalence of antibiotic resistance phenotypes in 485 isolates.a UpSet plot illustrating the diversity of antibiotic resistance profiles for ExPEC isolates. In the combination matrix at the base of the main bar chart, each column represents a distinct phenotypic profile. A colored dot within a column signifies resistance to at least one antimicrobial within a drug class, with the color corresponding to the respective drug class. The vertical bar chart displays the total number of isolates with specific phenotypic profiles, while the horizontal bar chart depicts the number of isolates resistant to each drug class. Sulfamethoxazole trimethoprim is a combination antibiotic, classified here as encompassing both sulfonamide and trimethoprim. b Histogram illustrating the prevalence of resistance to each test antimicrobial grouped by drug class in ExPEC strains. c The prevalence of antimicrobial resistance in four common STs of ExPEC. Source data are provided as a Source Data file.When the resistance to each antibiotic was analyzed, we found that most isolates were resistant to tetracycline (95.5%), ampicillin (95.5%), piperacillin (90.1%), sulfamethoxazole trimethoprim (89.9%) and chloramphenicol (85.2%) but only 0.8% of isolates were resistant to meropenem, imipenem, and tazobactam piperacillin (Fig. 2b). For aminoglycosides, about 70% of isolates were resistant to gentamicin and about 15.5% of isolates were resistant to amikacin. More than 65% E. coli showed resistance to each of the three fluoroquinolones, levofloxacin, moxifloxacin and ciprofloxacin. Additionally, resistance was observed in 10% to 60% of isolates for each of the other seven antibiotics within the beta-lactam class. Specifically, 51.8%, 13.4%, and 44.7% exhibited resistance to cefotaxime, ceftazidime, and cefepime, respectively, representing the third and fourth generation cephalosporins.To assess the commonalities and specificities of antibiotic resistance among strains from different lineages, we analyzed the resistance profiles of strains belonging to the four most prevalent STs (Fig. 2c). We observed similar patterns in the prevalence of resistance to various drugs within each ST. Nonetheless, notable differences were observed, particularly in the prevalence of resistance to three fluoroquinolones and cefepime, which were significantly higher in ST410 compared to other STs. Furthermore, among the four top prevalent STs, only strains belonging to ST410 exhibited resistance to three antibiotics, including tazobactam piperacillin, meropenem, and imipenem.Swine ExPEC genomes harbor diverse and abundant antibiotic resistance genesTo understand the genetic determinants of antibiotic resistance, we searched all the sequenced genomes for genes and point mutations contributing to antibiotic resistance. We predicted a total of 96 ARGs and point mutations of 3 chromosomal genes (gyrA, parC and parE), which could contribute to resistance to antibiotics belonging to 12 classes (Fig. 3a and Supplementary Data 5). We found that all genomes harbored at least one of these ARGs. Each of the 13 ARGs that determines the resistance to aminoglycosides was contained by more than 5% isolates, among which strA and strB were present in 68% strains. The most frequently occurring sulfonamide resistance gene was sul2, encoded by 85% isolates, which is the highest prevalence among ARGs. Among the 4 phenicol genes, floR had a prevalence of 73%, with others less than 35%. One of the three types of tetracycline resistance protein genes, tet(A), was present in more than 70% of the strains, while both tet(B) and tet(M) were found in less than 35% of the isolates. When genetic determinants contributing to the resistance against fluoroquinolones were considered, we observed that, in addition to the high prevalence of the chromosomal gene mutations, three ARGs were identified, with oqxA and oqxB as the most prevalent one (approximately 45%). Besides, we also found the genes conferring resistance to colistin (mcr-1) and fosfomycin (fosA3) in 23% and 17%, respectively, of the tested isolates. Among the genes encoding β-lactamase, the most prevalent one was blaTEM-1B which was contained by more than 67% isolates. Other beta-lactamase encoding genes had prevalence above 5% include blaCTX-M-14, blaCTX-M-55, blaOXA-1, blaCMY-2, blaCTX-M-65, blaTEM-1A, and blaTEM-141 most of which belong to extended-spectrum β-lactamase (ESBL) genes leading to the resistance to the three-/fourth-generation cephalosporins. When examining all ESBL genes present in our dataset, we found that as many as 262 strains harbored at least one ESBL gene. Moreover, among these, 25 strains were identified as hybrid pathotypes, specifically ExPEC/ETEC and ExPEC/EPEC (Supplementary Data 6). Additionally, 13 out of 15 strains with zoonotic potential also contained ESBL genes (Supplementary Data 2).Fig. 3: Diversity of genetic determinants of antibiotic resistance in 499 ExPEC strains.a Bar plot showing the percentage of genomes detecting each ARG or point mutation. Genes that contain point mutations leading to antibiotic resistance are marked with an asterisk. The color represents the class of antimicrobials to which genetic determinants confer resistance. The image displays only representative high-prevalence ARGs; you can find all the information in Supplementary Data 5. b Co-occurrence network of ARGs among ExPEC strains. The network was constructed based on the co-occurrence frequency of two-by-two ARGs in 499 genomes. The thickness and color depth of each link (edge) between ARGs (nodes) correspond to how frequently the two genes appeared together within the same genome. The image only visualizes genes that co-occurred together in at least 50 genomes. All connection information is available in Supplementary Data 8. Source data are provided as a Source Data file.We conducted an additional analysis comparing the prevalence of ARGs of isolates from different STs (Supplementary Fig. 4). We found that the most prevalent ARGs within each class were similar, which is consistent with those for all the isolates. However, there were some over-presented ARGs in each of the four top prevalent STs. In ST410, we detected higher prevalence of ARGs including aph(3’)-Ia, aac(6’)-Ib-cr, rmtB, blaCTX-M-55, blaOXA-1, blaTEM-141, catA1, catB3, oqxA, oqxB, qepA1, and tet(B). In ST88, the prevalence of resistance genes such as aac(3)-IIa, ant(3”)-Ia, blaCTX-M-65, qnsS1, and dfrA1 was higher compared to other STs. Additionally, mcr-1 and dfrA14 were more prevalent in ST101, and blaCTX-M-15 and sul2 were more common in ST117. Here, we further combined the comparative results of antibiotic resistance phenotypes among the four prevalent STs, especially noting that strains belonging to ST410 exhibited the highest resistance to three fluoroquinolones compared to the other STs. Therefore, we analyzed the distribution of genetic determinants responsible for fluoroquinolone resistance, including ARGs and chromosomal point mutations, within the ExPEC population (Supplementary Fig. 5). We observed mutations in three chromosomal genes (gyrA, parC, and parE) present in all ST410 strains, and the prevalence of fluoroquinolone resistance genes in ST410 was higher compared to the other prevalent STs. This could be one of the important genetic factors contributing to the prevalence of ST410. It’s worth noting that all ST410 strains were identified as MDR. Furthermore, a similar pattern was observed in ST354, which presented the hybrid pathotype ExPEC/ETEC, with most strains belonging to this ST type containing ESBL genes.Highly diverse antibiotic resistance gene co-occurrences contribute to the high prevalence of multidrug-resistant (MDR) ExPECTo investigate the genetic determinants behind the MDR and XDR (extensively drug-resistant, nonsusceptibility to at least one agent in all but two or fewer antimicrobial categories) E. coli, we examined the compositional profiles of ARGs in each genome. Among the 492 genomes containing at least 2 ARGs, we identified 406 combinations of distinct ARGs (Supplementary Data 7). More than 96% of the combinations were contained by only one or two genomes, with 340 genomes each harboring a unique set of ARGs. Each combination contained an average of 13 ARGs, with the largest one containing up to 28. More than two-thirds of the 406 combinations presented by more than 70% of genomes have over 10 ARGs, including 31 combinations found in 36 genomes containing more than 20 different genes. When drug classes targeted by these ARGs were predicted, we found that the putative MDR isolates accounted for up to 98% of the total, which is consistent with the test results described above. Our findings unveiled a striking diversity in the combinations of ARGs within the majority of the swine ExPEC genomes.To investigate the simultaneous transfer of multiple ARGs, we conducted co-occurrence network analysis on the entire set of ARGs (Supplementary Data 8). The most prevalent sub-network consisted of sul2, floR, strA, and strB, forming the central core of the entire network, with most of the other ARGs having connections to this sub-network (Fig. 3b). This sub-network was identified in more than 200 of the 499 genomes and had the potential to give rise to MDR E. coli strains resistant to antibiotics from aminoglycoside, chloramphenicol and sulfonamide drug classes. The other highly clustered ARGs included tet(A), blaTEM-1B, oqxA, and oqxB, conferring resistance to tetracyclines, beta-lactams, and fluoroquinolones, respectively. It is crucial to underscore the concurrent existence of ARGs targeting human drugs, such as blaCTX-M-55, blaCTX-M-14, fosA3, and mcr-1, in conjunction with ARGs targeting veterinary drugs. This issue carries notable significance, as the potential dispersal of veterinary drug-related ARGs, stemming from their misuse, indirectly amplifies the proliferation of co-existing ARGs against human drugs. Additionally, we observed that despite mcr-3 being detected in only five genomes (mcr-3.1 in four genomes and mcr-3.5 in one genome), it is noteworthy that two of these genomes concurrently carry both mcr-1 and mcr-3 (Supplementary Data 8).Both chromosomes and plasmids are vectors of antibiotic resistance genesTo decipher the vectors of ARGs, we chose 20 isolates from all the defined phylogenetic groups to obtain closed genomes. We obtained 62 closed replicons, 20 chromosomes and 42 plasmids, along with one incomplete plasmid sequence (Supplementary Data 9). The plasmids harbor one to four plasmid replicons, including IncFIB (n = 17), IncFIC (n = 14), IncI-gamma/K1 (n = 10), IncFIA (n = 8), and rep_cluster_2244 (n = 8). The average number of the ARGs contained by each chromosome was 2.5, adding up to 50 ARGs across all 20 chromosomes. The tetracycline-associated resistance gene tet(B) was exclusively identified in chromosomes, not in plasmids. Another ARG that exhibited over-representation on chromosomes is blaCMY-2 (Supplementary Data 10). Of the 43 plasmids, thirty-seven harbored at least one ARG, with an average count of 6.2 and the highest observed being 15. AGRs over-presented in plasmid genomes included aac(3)-IId, aadA2, mph(A), dfrA12, oqxA, oqxB, and aph(3’)-Ia. The top prevalent ARGs were all frequently found in both chromosome and plasmid genomes, such as sul1, sul2, floR, tet(A), blaTEM-1B, strA and strB, suggesting potential transfer between plasmids and chromosomes. In particular, we observed that certain ARGs have two copies located on distinct plasmids in the same isolate (Supplementary Data 9). For instance, in strain A430, each of the genes aac(3)-IId, aadA2, blaTEM-1B, dfrA12, and tet(A) was identified with two copies situated on separate plasmids. Additionally, we also identified integrons within the 20 complete genomes. A total of 16 integrons were detected, all of which were located within the plasmid genomes (Supplementary Data 11). Furthermore, we found that these integrons generally contain two ARG cassettes: aadA2/aadA5 and dfrA12/dfrA17. It is noteworthy that we observed these integrons closely associated with IS26 (Fig. 4a). This suggests that the accumulation and dissemination of specific ARGs in swine ExPEC were facilitated by the interplay between integrons and transposons.Fig. 4: Genetic contexts of vectors harboring ARGs with high co-occurrence frequency in 20 complete genomes.a Gene arrow maps illustrating ARGs and their neighboring ISs. The direction of the arrows corresponds to the strand where the genes are situated. Right-pointing arrows represent genes on the forward strand, while left-pointing arrows indicate genes on the reverse strand. Different colors of the arrows signify distinct gene types. The start and end positions of the genes within their respective molecules are visually depicted along the x-axis, measured in kilobase pairs (kbp). Highlighted blue lines emphasize potential co-transfer units of specific significance, while green boxes represent predicted integrons. The labels on the left side denote the names of each vector, and information about the plasmid’s replicon types and mobility (separated by ‘~~~‘) is provided above the dashed box. Gene names and IS elements, which conventionally appear in italics, are displayed in regular font for clarity. A412_chr exhibits only a subset of ARGs and ISs. b The stacked histogram illustrating the distribution ExPEC isolates for various phylogroups harboring genomic islands. c The prevalence of the plasmids in 499 ExPEC genomes. The heatmap illustrates the Jaccard similarity coefficient among pairs of plasmids. The stacked histogram on the left displays the number of genomes for different phylogroups harboring each plasmid. For clusters of highly similar plasmids (such as A333_p0, A430_p0, and A468_p0), we employ the prevalence of the plasmid with the shortest sequence among them as a representative measure for that type of plasmid. Source data are provided as a Source Data file.When the co-location of the ARGs were considered, we found that the five genes sul2, strA, strB, tet(A), and floR were clustered together on both chromosome and plasmid (Fig. 4a). The ARG cluster on the chromosome is delimited by IS3 and IS1H, forming Genomic Island 1 (GI1), while on the plasmid, it is framed by ISVsa3 and IS26, which are associated with IncI-gamma/K1 plasmid. Moreover, we observed that GI1 was detected in 30 isolates (Fig. 4b), while the Incl-gamma/K1 plasmid was present in 14 isolates (Fig. 4c). It’s worth noting that these isolates were distributed across phylogroups A, B1, D, F, and G. Another ARG cluster located on the chromosome, is consisting of sul2, strA, strB, and blaTME-1B, with IS5075 and IS26 serving as flanking elements, collectively forming Genomic Island 2 (GI2). It was identified in 30 isolates, primarily from phylogroup B1. Regarding plasmids, two similar gene clusters were observed, with both of these gene clusters being associated with IS26. One contains strA, strB, and blaTME-1B (A68_p1, A237_p0, A186_p1), while the other harbors strA, strB, and sul2 (A71_p0). These findings indicate that the transfer of this ARG cluster occurs in a diverse manner, facilitated by transposable elements (TEs) among chromosomes and also by conjugative mobilization through plasmids. Moreover, in addition to the ARG clusters, these conjugative plasmids carried an array of diverse ARGs, including aac(3)-IId, mph(A), cmlA1, sul3, and dfrA17, which confer resistance to antibiotics belonging to aminoglycosides, macrolides, phenicols, sulfonamides, and trimethoprim. These findings highlight the plasticity and heterogeneity of the gene clusters associated with strA and strB, enabling E. coli to combat a broad spectrum of antibiotic agents.Plasmids carried by swine ExPEC have the potential to transfer ARGs against important antibiotics used in human treatmentWe identified potential horizontal gene transfer facilitated by the plasmid by discerning similarities between the plasmids found in ExPEC carrying ARGs and those identified in bacteria from other hosts, especially human (Fig. 5). Among the 20 isolates selected for sequencing of closed genomes, A376 and A382 showed resistance to carbapenems. Two plasmids belonging to IncX3 (A382_p1) and IncFIA/FIC (A376_p2) harbored blaNDM-1 and blaNDM-5, respectively (Fig. 5a). Genomic analysis revealed a high similarity between plasmid A382_p1 and plasmid pNDM_MGR194 found in the human pathogenic bacterium Klebsiella pneumoniae46 (Fig. 5e). The main difference lies in the resistance gene carried by the pNDM_MGR194 and pNDM_MGR194-like plasmids, which is blaNDM-5. However, the insertion sequences on one side of these plasmids are the same. Three strains belonging to phylogroups A and B1 were found to possess plasmid A382_p1 in our swine ExPEC collection (Fig. 5e). Among these isolates, A386 and A499 were found to harbor the antibiotic resistance gene blaNDM-5, consistent with the observations made in pNDM_MGR194. For the blaNDM-5 gene, situated on plasmid A376_p2, it was found to be clustered with other ARGs conferring resistance to beta-lactams, aminoglycosides, sulfonamides, and trimethoprim. These ARGs on the plasmid are flanked by IS26, forming a potentially transposable unit. The plasmid A376_p2 exhibits significant similarity to the plasmid pLZ135-NDM in human E. coli, which contains two copies of this transposable unit. These indicate the potential transmission of blaNDM genes between swine E. coli and human pathogens through plasmids, highlighting the variability of the blaNDM gene on conjugative plasmids in the swine ExPEC.Fig. 5: Genetic contexts and comparisons of vectors containing ARGs against crucial antibiotics across various strains from different hosts.Gene arrow maps illustrating the presence of ARGs: blaNDM (Panel a), mcr-1 (Panel b), co-existence of fosA3 and ESBLs (Panel c), and co-existence of mcr-1, fosA3, and ESBLs (Panel d). The meaning represented by the depicted elements is similar to that of the elements in Fig. 4a. Black labels for vector names denote plasmids from our study, while each red label corresponds to plasmids from public databases that are most similar to the respective black-labeled plasmid. Information regarding the replicon types, mobility, species name of the strain, and the host of the strain (separated by ‘~~~‘) for public plasmids is provided above the dashed box. A333_chr only shows a partial representation of ARGs and ISs. Panel e: The similarity between plasmids from ExPEC and those from bacteria originating from other hosts, as well as the prevalence of plasmids within the ExPEC context. The meaning represented by the depicted elements is similar to that of the elements in Fig. 4c. Due to partial similarities and some distinctions among plasmids A251_p0, A412_p0, and A329_p0, and in accordance with the principle of plasmid incompatibility, it is not possible for a bacterium to simultaneously possess two such plasmids. Consequently, if the sequencing data of a bacterium mapped to the genomes of these three plasmids, each exhibiting a coverage exceeding 90%, we chose the result with the highest coverage. Source data are provided as a Source Data file.Seven copies of the colistin resistance gene mcr-1 were identified in the genomes of six isolates, with four copies located on four plasmids (A338_p0, A251_p0, A412_p0, A329_p0) and three copies found on two chromosomes (Fig. 5b and d). In all of these cases, the insertion sequence ISApl1 was found in close proximity to the mcr-1 gene, as reported previously47,48. Plasmid A338_p0 exhibits moderate genomic sequence similarity to plasmid pSCKLB684-mcr from K. pneumoniae. Plasmid A251_p0, A412_p0, and A329_p0 exhibit a high degree of similarity in terms of gene content; however, the latter two are capable of conjugative transfer, while the former is non-mobilizable. Plasmid A412_p0 and A329_p0 share a high level of genomic sequence identity with plasmids pSH15G1450 and pLS61394-MCR, both originating from Salmonella enterica isolated from children with intestinal infections in China49,50. And plasmid A251_p0 is homologous to plasmid pMCR-H8 from ESBL-producing E. coli H8, isolated from the mink farmer in Shandong Province of China51. Importantly, the A251 strain possesses zoonotic potential based on the cgMLST results (Supplementary Data 2). Besides mcr-1, these plasmids harbor a diverse array of additional ARGs conferring resistance to crucially important antibiotics for human treatment, including blaCTX-M-14 (cephalosporin), floR (florfenicol), fosA3 (fosfomycin), and oqxA and oqxB (quinolone). Analysis of plasmid prevalence showed that the conjugative plasmid A412_p0 and A329_p0 were present in 15 and 4 swine ExPEC strains, respectively, whereas the non-mobilizable plasmid A251_p0 could be found in 4 isolates (Fig. 5e).In addition to the fosA3 genes that co-locate with the mcr-1 genes, we identified an additional four fosA3 genes on the 20 genomes analyzed. Among these, one gene was located on the chromosome, while the remaining three were found on three different plasmids belonging to two distinct families, namely IncFIA and IncI-gamma/K1 (Fig. 5c). Furthermore, we observed that on these plasmids, two ESBL genes (blaCTX-M-15 and blaCTX-M-65) were found in close proximity to fosA3. Additionally, all fosA3 genes and ESBL genes were associated with IS26. Notably, among the three plasmids coexisting with mcr-1, the gene blaCTX-M-14 and fosA constitute a potential co-transfer unit with two IS26 elements serving as flanking sequences (Fig. 5d). Comparative genomic analysis showed that plasmids A163_p1 was highly similar to plasmids pHNHNC02 from a chicken-derived E. coli isolate52 (Fig. 5e). Significantly, plasmids A237_p1 and A68_p0 have a greater diversity of ARGs compared to their closest counterpart, plasmid pHNAH4-1 from a ceftazidime-resistant E. coli strain isolated from chicken feces in China which had no fosfomycin resistance gene53. This finding suggests that these two plasmids represent a novel type carrying both fosA3 and ESBL genes. Taken together, we observed that plasmids carrying the fosA3 gene have a wide host range, being found in different Enterobacteriaceae members of swine, poultry, and human origin.
