Extended-spectrum β-lactamases (ESBLs) and AmpC β-lactamases are the most widespread mechanisms of cephalosporin resistance in E. coli in animals and humans worldwide, including the Asia Pacific region (Jean and Hsueh, 2017; Naas et al., 2017). Some of the ESBLs and AmpC β-lactamases described in E. coli have spread globally in the last two decades, including CMY-2, which is the most common AmpC enzyme (Naas et al., 2017), and CTX-M, which is the most widespread ESBL type (Jean and Hsueh, 2017); however, there are marked differences in the CTX-M variants occurring across sources and geographical areas. In Vietnam, CTX-M-1 group enzymes including CTX-M-15 and CTX-M-55, and CTX-M-9 group enzymes including CTX-M-14, CTX-M-27, and CTX-M-65 have been described in E. coli from diseased and healthy humans, chickens, pigs, meat, and vegetables (Bui et al., 2015; Zurfluh et al., 2015; Nguyen et al., 2016; Hoang et al., 2017; Jean and Hsueh, 2017).

In most cases, ESBL/AmpC-encoding genes are located on mobile genetic element, and many are plasmid-mediated and transferable between bacteria of different species. Studies from Europe have shown that farmers share highly related and even identical ESBL/AmpC E. coli and/or ESBL/AmpC plasmids with their livestock (Madec et al., 2017). It is likely that this risk is enhanced within countries with higher level of animal to human contact due to low hygienic standards in their livestock production, such as in pig production in Vietnam (Hoang et al., 2017), but this remains to be further investigated.

Knowledge and control of ESBL and AmpC in E. coli from animals is crucial to preserve the effect of medically important antimicrobials such as cephalosporins. Therefore, ESBL/AmpC monitoring in E. coli from animals and food has been mandatory in EU countries since 2014 (Commission implementing Decision 2013/652/EU) and such monitoring is a suggested focus area by the World Health Organization in low-and middle-income countries (The ESBL Ec Tricycle AMR surveillance project)¹.

In Vietnam, knowledge on the ESBL/AmpC variants in E. coli from pigs and pig farmers is limited, and thorough characterization of the genetic location of the ESBL/AmpC-encoding genes in E. coli is lacking. Moreover, most genomic studies that address transmission of resistance genes between animals and humans rely on tracking clonally evolving lineages excluding extrachromosomal plasmids despite plasmids being the primary carriers of antibiotic resistance genes across the pathogens (David et al., 2020).

The objective of this study was therefore to characterize ESBL/AmpC genes in E. coli in pigs and pig farmers in northern Vietnam and determine their location and context on mobile genetic elements to estimate the level of transfer between pigs and humans. Information from this study illustrates the role of different genetic elements in the spread of ESBL/AmpC genes between human and animal hosts.

It is widely suggested that the livestock sector may be an important source of antimicrobial resistance genes, including ESBL/AmpC encoding genes, found in humans. The transfer mainly happens through horizontal spread of mobile elements (Zaja̧c et al., 2019; Dantas Palmeira and Ferreira, 2020). To elucidate the level of transmission of ESBL/AmpC genes in E. coli from pig and pig farmers in Vietnam, we performed a detailed genetic analysis of E. coli strains, which harbored such genes, and we analyzed the location of genes in the genome. This included annotation of transposons and insertion sequences located around the resistance genes, the phylogenetic relatedness of the plasmid components reconstructed from strains, and genetic relatedness of the strains, all based on whole genome sequences. This approach provides a holistic answer to the transmission of resistance genes between hosts, either by transfer of strains or by spread of mobile genetic elements, and thus it extends the approaches used in most previous studies of spread of resistance genes and mobile genetic element between hosts (Kudirkiene et al., 2018; Zaja̧c et al., 2019; Baniga et al., 2020). Analysis of the plasmids and the genetic contexts along with the strains diversity allow to investigate the diversity, distribution, and transmission dynamics of resistance plasmids in the strains (David et al., 2020). The ESBL/AmpC β-lactamases genes detected in the analyzed strain collection of E. coli were present across pigs and pig farmers suggesting a distribution of the same genes across hosts, however, the most commonly observed gene differed between the two hosts; bla_CTX-M-55 was the most commonly observed among the pig isolates and bla_CTX-M-27 among human isolates. The genes detected have previously been described from diseased and healthy humans, chickens, pigs, and food products in Vietnam in studies where the human cases were suggested to emanate from animal origins (Nguyen et al., 2016; Hoang et al., 2017; Hinenoya et al., 2018).

The predominant ESBL gene, bla_CTX-M-55 was mainly plasmid mediated, and was found to be carried on different types of conjugative plasmids with highly variable genetic contexts around the ESBL genes, just as the insertion transposases downstream of the gene were also mostly different. Previous reports have found bla_CTX-M-55 on chromosomal fragments as well as on IncHI2 (Zhang et al., 2019), Incl2 (Lv et al., 2013), and IncF and Incl2 (Lupo et al., 2018) plasmids. The diversity observed in types of plasmids suggested limited transmission bla_CTX-M-55 between pigs and farm workers. The phylogenetic analyses of the plasmids carrying bla_CTX-M-55 supported that plasmids in strains from pigs and humans were unrelated, e.g., different replicon types were found in the two hosts. Another supporting observation was the fact that the plasmids did not share reference plasmids in NCBI to any large extend. For instance, no more than five contigs with bla_CTX-M-55 showed similarity to a common plasmid reference in the database [plasmid accessions MN823991 (5), AP023198 (5), MN158989 (5), CP034747 (4)]. Overall, these findings with regard to plasmids carrying bla_CTX-M-55, as well as plasmid carrying bla_CTX-M-14, bla_CTX-M-15, and bla_CMY-2 indicated that horizontal gene transfer in a real-life situation between pigs and pig farm workers happens with so low frequency that it will only be detected by very intensive sampling strategies. However, laboratory-based conjugation experiments shows that transfer of ESBL-encoding plasmids between E. coli strains is more frequent (Liu et al., 2019). Plasmids harboring bla_CTX-M-55 were confirmed to be conjugative in our investigation with similar conjugation transfer frequencies as plasmids encoding bla_CTX-M-27. Whether transferability of plasmids between strains under field conditions is as efficient as in the laboratory remains to be investigated. The presence of bla_CTX-M-55 on diverse plasmid replicon types with different transposons on various E. coli lineages show the heterogeneity of the plasmids carrying this ESBL gene, and suggest multiple introductions of these genes independently in each host (David et al., 2020). Together with the high variation in genetic context around the bla_CTX-M-55 gene, these results show that bla_CTX-M-55 in farmers is not necessarily horizontally acquired from close contact with animals and further that there is limited transmission of this type of resistant E. coli from pigs to farmers, and vice versa. The sample size is, however, not large enough to rule out that transmission occurs on a low level, given the fact that the plasmids are transferable.

The gene bla_CTX-M-27, was the second leading ESBL gene found in pigs and pig farmers. It was located in a relatively conserved genetic context in all isolates with the same flanking transposons in all the strains and all carried by the same conjugal IncF types of plasmid. The strains carrying the gene, on the other hand did not show resemblance. This is evidence of horizontal spread of the IncF plasmids types carrying bla_CTX-M-27 in the study area. The same types of conjugative IncF plasmids were found in E. coli strains from pigs and farm workers (Supplementary Table S1), however, the direction of transfer of plasmids cannot be established from the current study. The gene bla_CTX-M-27 seems to be generally associated with conjugative plasmids of the IncF replicon type including FIA, FIB, and FII (Fernandes et al., 2020; Matsuo et al., 2020). The phylogenetic analysis of the plasmids revealed that they shared a larger core-genome cluster compared to the plasmids harboring bla_CTX-M-55. In contrast to the observation with the plasmids carrying bla_CTX-M-55, many of the bla_CTX-M-27 plasmids shared reference plasmids in NCBI (plasmid accession CP049168 shared by 11 strains and KX008967 by nine strains), further supporting the low genetic variation for this type of plasmids. We identified a novel plasmid species harboring bla_CTX-M-27 in three strains. Its closest reference in NCBI, Salmonella Typhimurium plasmid pB71, is a non-ESBL, multi-drug resistance plasmid, and it is likely that it has experienced an insertion of bla_CTX-M-27, however, other ways including intermediate plasmid species, could also have led to the formation of this plasmid. The size of the plasmid was approximately 150 kb (Supplementary Table S1 and Supplementary File 1). It is an IncF replicon type plasmid, and encoded multi drug resistance with mcr-3, aadA1, aac(3)-Iid, aph(3″)-Ib, aph(3′)-Ia, aph(6)-Id, mef(B), cmlA1, floR, sul3, and dfrA12 genes in addition to the ESBL gene. Overall, this novel plasmid is similar to the other bla_CTX-M-27 plasmid reported in this study and elsewhere (Fernandes et al., 2020; Matsuo et al., 2020).

Moreover, our comparative analyses of contigs carrying the IncF types of plasmids also revealed another plasmid that showed complete identity with E. coli plasmid pESBL57, Accession MT230319 initially harboring bla_CTM–M-97 in NCBI. Our analysis indicates that bla_CTX-M-14 has replaced bla_CTM–M-97.

Five of the seven strains where bla_CMY-2 was detected carried the gene in the chromosome. The other AmpC genes detected, bla_CMY-42 and bla_DHA-1, were plasmid-mediated with consistent genetic contexts between isolates from pigs and humans, indicating that these genes are transmitted between hosts. In other studies, bla_CMY-2 was reported to be chromosomally located (Harada et al., 2010), which suggests that this gene is less likely mobilizable and can only be spread together with the host strains. In many cases, the plasmids containing ESBL/Ampc β-lactamases also carried mcr-1 or mcr-3 genes encoding colistin resistance, as well as the presence of metal/biocide tolerance genes. Co-occurrence of ESBL/AmpC, colistin resistance as well as other antimicrobial resistance genes such as qnrS1, mef(B), mph(A), cmlA1, sul1, sul3, tet(A), dfrA12, aadA1/2, aac(3)-IIa was common in most isolates, and resistance genes were often occurring on the same plasmid (Supplementary Table S1). This corroborate previous resports that the majority of the strains analyses had multidrug resistant status (Dang et al., 2018). The plasmids identified therefore represent an important source of multidrug resistance, and due to their location on conjugative plasmids, they may transfer resistances of significant public health importance, and such plasmids need particular attention to track their global emergence, evolution and spread (Lupo et al., 2018; Zaja̧c et al., 2019). Our findings further corroborate studies showing co-carriage of mcr-genes in many ESBL isolates (Lupo et al., 2018; Zaja̧c et al., 2019) although, in those studies, the mcr-encoding plasmids are not the same as the plasmid carrying the ESBL genes.

In our study, ESBL/AmpC and other antimicrobial resistance genes were frequently co-located with heavy metal resistance genes on the same plasmids. Heavy metal exposure due to usage in livestock feed, e.g., use of zinc oxide in pig feed, can therefore co-select for antimicrobial resistance in bacteria (Ludden et al., 2019; Peng et al., 2020). In support of this, recent studies have reported that antimicrobial resistance is associated with tolerance to heavy metals existing naturally or used in food animal production, including zinc oxide and copper (Rensing et al., 2018; Cheng et al., 2019; García et al., 2020). Studies have also reported occurrence of metal tolerance genes in E. coli from post-weaning diarrhea in piglets with no documented exposure to metals, probably reflecting that over the years, metal resistance genes have been widely spread among E. coli in pigs (García et al., 2020).

The phylogenetic analysis of the isolates revealed wide genetic variation with no evidence of transmission of strains between pigs and farmers, and vice versa. All the isolates were of different MLST types, and none of them was predicted to be of known pathogenic sub-types (Supplementary Table S1). Thus, ESBL and AmpC encoding E. coli in pigs and pig farmers in Vietnam belong to unrelated commensal groups with both related and unrelated plasmids. A recent study has shown that E. coli causing bloodstream infections in the United Kingdom were not acquired from livestock, and that sharing of mobile elements between animals and humans was infrequent (Ludden et al., 2019). Moreover, a large diversity was reported in E. coli clones and plasmid types in France from food animals corroborating our findings (Lupo et al., 2018). Thus, our findings provide further evidence that plasmids carrying ESBL/AmpC β-lactamase encoding genes occur in genetically unrelated commensal E. coli. Strain sharing between the two hosts was infrequent and so was the mobile genetic element except for bla_CTX-M-27, and this underlines that one cannot only focus on the resistant strains when analyzing spread patterns of resistance genes.

Despite that the bacterial strains from 2015 (Dang et al., 2018) and 2018 were epidemiologically related, i.e., matching fecal samples from workers and pigs collected from the same pig farms, they showed genetic variations and were predominantly commensals occurring independently in each host. Moreover, commensal E. coli are less studied than pathogenic E. coli with the former usually found genetically distinct (Ahmed et al., 2017). This may hide the occurrence of frequent transfer of mobile elements across strains as suggested by the conjugation experiments. Furthermore, as mobilizable genes, ESBL carriage is dynamic and could affect the consistency of their contexts in cross-sectional studies. However, the fact that sampling at two separate time frames did not show any significant fluctuations in the genetic context of ESBL genes in pigs and farm workers suggests that most of the ESBL genes are less frequently shared between the two hosts. Nevertheless, future longitudinal studies would provide further and improved insights.

A main limitation in this study is that only short-read technologies were used to generate the whole genome and plasmid sequences, Long and short-reads hybrid assemblies do not show significantly different outcomes (David et al., 2020), however, using combinations of the two techniques, closing of circular (plasmid) sequences is possible. To overcome the limitation, we used plasmid profiling to confirm the results of the predicted plasmids and their sizes, and conjugation experiment were further used to confirm the transferability of the plasmids between strains. We suggest that the relatedness of the plasmids encoding the resistances and the consistency of the genetic context around the ESBL/Ampc genes is considered along with the phylogenetic relationship between the strains.

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

AD and JO: conceptualization and supervision. YH and VB: methodology and software and data curation. YH, VB, AD, JO, SD, and DT: validation and writing—review and editing. YH: formal analysis, writing—original draft preparation and visualization. YH, SD, and DT: investigation. YH, VB, AD, and JO: resources. AD: project administration and funding acquisition. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest