Virulome and genome analyses identify associations between antimicrobial resistance genes and virulence factors in highly drug-resistant Escherichia coli isolated from veal calves

Food animals are known reservoirs of multidrug-resistant (MDR) Escherichia coli, but information regarding the factors influencing colonization by these organisms is lacking. Here we report the genomic analysis of 66 MDR E. coli isolates from non-redundant veal calf fecal samples. Genes conferring resistance to aminoglycosides, β-lactams, sulfonamides, and tetracyclines were the most frequent antimicrobial resistance genes (ARGs) detected and included those that confer resistance to clinically significant antibiotics (blaCMY-2, blaCTX-M, mph(A), erm(B), aac(6’)Ib-cr, and qnrS1). Co-occurrence analyses indicated that multiple ARGs significantly co-occurred with each other, and with metal and biocide resistance genes (MRGs and BRGs). Genomic analysis also indicated that the MDR E. coli isolated from veal calves were highly diverse. The most frequently detected genotype was phylogroup A-ST Cplx 10. A high percentage of isolates (50%) were identified as sequence types that are the causative agents of extra-intestinal infections (ExPECs), such as ST69, ST410, ST117, ST88, ST617, ST648, ST10, ST58, and ST167, and an appreciable number of these isolates encoded virulence factors involved in the colonization and infection of the human urinary tract. There was a significant difference in the presence of multiple accessory virulence factors (VFs) between MDR and susceptible strains. VFs associated with enterohemorrhagic infections, such as stx, tir, and eae, were more likely to be harbored by antimicrobial-susceptible strains, while factors associated with extraintestinal infections such as the sit system, aerobactin, and pap fimbriae genes were more likely to be encoded in resistant strains. A comparative analysis of SNPs between strains indicated that several closely related strains were recovered from animals on different farms indicating the potential for resistant strains to circulate among farms. These results indicate that veal calves are a reservoir for a diverse group of MDR E. coli that harbor various resistance genes and virulence factors associated with human infections. Evidence of co-occurrence of ARGs with MRGs, BRGs, and iron-scavenging genes (sit and aerobactin) may lead to management strategies for reducing colonization of resistant bacteria in the calf gut.

Introduction Escherichia coli are Gram-negative facultative anaerobes that are commensal members of the bovine gut as well as frequent members of environmental (non-animal) communities. Most E. coli are non-pathogenic, but a small subset has been linked to a range of mild to severe human diseases. These typically include self-limiting gastrointestinal (GI) infections as well as extra-intestinal infections such as bladder/urinary tract infections (UTIs), prostatitis, wound infections, pneumonia, sepsis, and meningitis in newborn babies. Infections are primarily caused by strains that carry various suites of virulence factors (VFs), but opportunistic infections can be caused by any strain, even those lacking major VFs. E. coli causes a significant number of GI infections annually in the United States and is responsible for 80% of UTIs [1][2][3]. Treatments for non-Shiga-toxigenic infections typically involve antimicrobial therapy, but pathogenic and non-pathogenic E. coli can be resistant to these drugs; some are multidrugresistant (MDR) and can cause difficult-to-treat infections in humans and animals. Further, the E. coli population can serve as a reservoir of resistance genes that can transfer from commensal to pathogenic strains, or to other pathogenic organisms, such as Salmonella enterica [4][5][6].
Antimicrobial-resistant infections are an on-going human and animal health threat on a global scale, causing an extremely high, but not well-quantified, number of medical complications and fatalities each year [7][8][9]. Like antimicrobial-susceptible organisms, infections caused by resistant organisms can be nosocomial, community-acquired, waterborne, or foodborne. Foodborne and waterborne antimicrobial-resistant E. coli infections typically occur from fecal contamination of produce, meat, milk, eggs, and surface and drinking waters; communityacquired resistant E. coli infections, although transmitted differently than foodborne and waterborne infections, can be caused by strains that have a natural food animal host reservoir, such as poultry and cattle.
Beef cattle, dairy cows, and dairy calves are well documented reservoirs of antimicrobialresistant bacteria and pathogens, but antimicrobial resistance carriage in veal calves remains understudied [10][11][12][13]. Calves raised for veal are usually the male calves from dairy herds. In the United States, milk is a major component of their diet until they are 16 to 18 weeks of age. About 15% of marketed veal calves are "bob veal" which are sold from birth up to three weeks of age. Recent studies have shown that dairy calf feces are a significant source of resistant bacteria and typically harbor a different suite of antimicrobial resistance genes (ARGs) and a greater concentration of ARGs than older lactating and dry cows [10][11][12][13][14]. However, the genetic mechanisms or management practices responsible for these age-related differences in resistance carriage remain unknown. Further, the characteristics of resistant bacteria shed by veal calves, which are raised under significantly different management practices than replacement dairy calves, have not been adequately studied. The aim of this study was to comprehensively evaluate the genomic characteristics, virulence profiles, and ARGs in MDR E. coli collected from veal calf feces, as well as the genomic features that co-occur with these genes and may influence resistance carriage in the veal calf gut. We further compared the genomic distance between isolates to investigate the relatedness of isolates collected from animals on different farms. and VFs in the isolates was visualized using ForceAtlas2 algorithm on an interactive network inference, Gephi version 0.9.1 (scaling 10, edge weight influence 1) [29,30]. ForceAtlas2 is a force-directed algorithm used for network spatialization where nodes repulse each other like charged particles while edges attract their nodes [30]. These parameters were chosen for clarity of the nodes in the network. The edges (curves) in the network links a gene to the host isolate.
Core genome single nucleotide polymorphisms (SNPs) were identified by aligning the 85 E. coli genomes used in this study with 118 publicly available Escherichia genomes representing the eight major phylogenetic groups (A, B1, B2, C, D, E, F, and G), E. coli Cryptic Clades, and the near neighbors E. fergussoni and E. albertii using the Harvest package [31]. ParSNP within the Harvest package was run with the following parameters, -c (to force inclusion of all genomes in the analysis) and -x (to enable recombination filtering), and the complete chromosome of E. coli K-12 substr. MG1655 (NCBI accession: NC_000913.3) as the reference genome (-r). These SNPs were then used to infer a maximum likelihood tree with 1000 bootstrap replicates under default settings using the Randomized Axelerated Maximum Likelihood program (RAxML) [32]. To interrogate the relatedness of isolates collected from different farms, high quality SNPs (hqSNPs) were identified by aligning the cleaned and curated reads of presumptive related genomes to the E. coli K-12 MG1655 genome and retaining those SNPs that have a minimum of 10X sequencing read coverage by using the program Lyve-SET [33].
Among the MDR isolates, 35 unique plasmid replicons were detected (Table 1). IncFII and IncFIB (both often carried on the same plasmid) were the two most frequently detected replicons, followed by ColRNAI, IncFIA, IncI1, IncQ1, Col(MG828), and IncA/C2. Since the sequencing chemistry used in this study could not result in the assembly of completely closed plasmids, a co-occurrence analysis between resistance genes, VFs, and plasmid replicons was conducted to identify which plasmids may be potential carriers of certain ARGs, MRGs, BRGs and VFs ( Table 2). The IncFIB replicon was positively associated with the IncFII, IncFIA, and IncB/O/K/Z replicons, as well as four ARGs, the sit system (sitABCD), and the aerobactin (iucABC-iutA) operon. The IncFII replicon was similarly positively associated with the sit system (sitABCD), and aerobactin (iucABC-iutA) operon, as well as seven ARGs. IncFIA replicons were associated with IncFIB and IncFII replicons as well as the sit system, aerobactin operon, 12 ARGs, and the copper resistance operon. IncA/C2 replicons were associated with 11 ARGs. BRG qacEΔ1 was associated with IncFIA, IncFII, and IncFIC replicons, while sugE1 was associated with IncI1, IncQ, and IncX1 replicons.
Although the focus of this study was on highly drug-resistant E. coli isolates, we randomly selected a smaller subset of susceptible isolates as comparators to better evaluate the characteristics of the MDR isolates. Similar to the MDR isolates, the most common ST among the susceptible isolates was ST10. However, the most common phylogroup among these isolates was B1 ( Table 1). The most frequently detected plasmid replicons were IncFIB, ColRNAI, and IncFII, which were detected in 11, 7, and 6 genomes, respectively. IncI1, IncQ1, and IncA/C2 replicons were not detected among the susceptible isolate genomes. EHEC and EPEC-associated VFs absent in MDR genomes were identified in susceptible isolates. These include efa-1/ lifA (EHEC factor for adherence/lymphostatin), eae, paa (porcine attaching-effacing associated protein), ler (LEE encoded regulator), hlyABCD (hemolysin), stx, cif (cycle-inhibiting factor), nleACD (non-LEE-encoded effectors), and plasmid-encoded regulator (per) (Fig 3).    Based on a nonmetric multidimensional scaling (NMDS) analysis, the virulomes of MDR isolates were somewhat different than those of susceptible isolates (ANOSIM R = 0.1698, P = 0.001) (Fig 4). Since some VFs are integral to within-host survival, we assessed if there was an association between VFs, which may confer a selective advantage in the host gut, and the MDR phenotype. We analyzed the relative abundances of these accessory genes and compared these abundances in the MDR genomes to the susceptible genomes. In total, there were 40 VFs that were differentially abundant between the two groups (Fisher's exact test, two-tailed, q < 0.05). Of these, 22 were more abundant in susceptible genomes and 18 were more abundant in MDR genomes (Fig 3). The VFs more abundant in susceptible isolates included stx1AB, stx2AB, ler, tir, eae, cif, paa, per, hlyABCE, ehxA and nleABCD (q < 0.05). The VFs more abundant in MDR genomes were iron acquisition genes sitABCD and iucABC-iutA (aerobactin), and pap P fimbriae (q < 0.05).  The presence of all resistance genes and VFs in all strains (MDR and susceptible) were visualized in a network interface (Fig 5). In this study, most of the susceptible isolates were grouped into a separate cluster from the MDR isolates in the network structure indicating variation in the resistance genes and VFs repertoires in susceptible versus MDR isolates, which is congruent with the results of the NMDS analysis (Fig 5).
Several clonal strains with high levels of genomic similarity, based on the core genome SNPs and ARGs, were isolated from different veal operations (Fig 1). Isolates from farms E and H (ARS-CC11278 and ARS-CC11291) collected 122 days apart differed by 20 SNPs and two isolates from farms B and G (ARS-CC11328 and ARS-CC11288) collected over 7 days differed by 17 SNPs.

Discussion
Antimicrobial resistance has been well-documented in dairy and beef cattle and recent studies have demonstrated that younger calves harbor a greater abundance of resistant bacteria than older animals [10][11][12][13][14]. However, resistance in veal calves, which are considered a separate production class from dairy and beef calves during the Food and Drug Administration drug approval process, remains under-studied. Dairy calves raised as replacements for lactating cows and veal calves are managed differently and fed different diets. Replacement dairy calves are initially fed a diet of milk or milk replacer, followed by gradual introduction of hay and a solid calf starter. Once weaned, typically at 8-9 weeks of age, they are fed an exclusively solid feed. This phased transition to solid food assists in the development of a functional rumen. The diets of veal calves, on the other hand, typically include milk or milk replacer (made from

PLOS ONE
whey and whey protein) until marketed at 16-18 weeks. Bob veal calves are fed either colostrum, waste milk, and/or milk replacer for approximately three weeks when they are sold.
Although several studies have evaluated the prevalence of antimicrobial resistance on veal farms [34], none to-date have evaluated the non-ARG genetic features that co-occur with ARGs and may be associated with persistence or selection of resistance in the calf gut. Further, the public health risk posed by these bacteria remains unknown. Here we analyzed 66 MDR isolates from non-redundant veal calf fecal samples and compared these to a smaller subset of susceptible isolates with the aim of further understanding the diversity of MDR strains shed by these animals, as well as the genomic features that may be responsible for their persistence in young calves.

High genotypic diversity with an observed predominant phylogroup and genotype
Results of this analysis demonstrate that there is a high level of diversity among MDR E. coli isolated from veal calf feces, but the group was dominated by phylogroup A-ST Cplx 10 strains (33% of all isolates). These data suggest that the MDR and susceptible strains from veal calf feces, in general, are associated with different lineages of E. coli. It appears that MDR E. coli shed in veal calf feces are more likely to be phylogroup A, while susceptible isolates are more likely to be B1. Currently it is unknown if certain phylogroups are more likely than others to acquire transferrable resistance, but previous studies have described this phenomenon, albeit some identified similar trends as observed in this study (group A strains having a high level of resistance), while others showed that different groups are more likely to be resistant than phylogroup A [35][36][37][38][39][40]. These studies characterized E. coli from a variety of non-bovine matrices and isolates from studies in which E. coli was recovered from feces were similar in phylogroup distribution to those presented here [41,42]. Studies focused on bovine feces indicate that randomly selected generic E. coli are predominantly group B1 [43][44][45][46], while extended spectrum β-lactamase (ESBL)-producing E. coli were more likely to be phylogroup A [47], the latter being consistent with our results.

ExPEC-associated sequence types repeatedly identified among MDR isolates
The predominant ST among the MDR and susceptible isolates was A-ST10, which is a "globally distributed" ST that is commonly isolated from a wide diversity of hosts, environments, and regions [48]. It is therefore not surprising that A-ST10 is common among E. coli from veal calf feces. More than half of the MDR isolates were identified as STs that are frequently isolated from human infections, including gastrointestinal and extra-intestinal infections. The pandemic ST131, which is the current leading cause of ExPEC infections globally, was not detected among any of the isolates, but ST69, ST410, ST117, ST88, ST617, ST648, ST10, ST58, and ST167, which are among the leading causes of non-ST131 ExPEC infections globally were all identified repeatedly, except for ST58 and ST167, which were identified once each [49][50][51][52]. A significant number of the isolates encoded VFs involved in ExPEC infections, such as fyuA (yersiniabactin), sit operon (Sit system), iucABC-iutA (aerobactin), chuA (heme binding protein), and pap operon (P fimbriae), which were particularly abundant in ST69, ST117, ST410, and ST648 genomes. Of particular interest, of the 33 isolates identified as ExPEC-associated STs, 27 were bla CTX-M -encoding strains, and 26 encoded the azithromycin resistance gene mph(A), indicating that these potential ExPEC strains encoded resistance to antibiotics of human clinical significance.
Based on these data, there is an appreciable prevalence of ExPEC-associated STs among the MDR fecal E. coli isolated from veal calves. Previous studies have identified poultry as a significant reservoir of ExPEC isolates causing human bladder infections [50,[53][54][55], and results of this study indicate that veal calves may harbor similar strains. However, this study only takes into account the STs and VFs of these isolates and does not definitively identify them as pathogens. Further, more research needs to be conducted to evaluate the abundance of potential ExPEC strains in relation to the total E. coli population in the veal calf gut.

ARGs, MRGs, BRGs, and their co-occurrence
The most frequently observed antimicrobial class to which ARGs were identified were aminoglycosides, β-lactams, sulfonamides, and tetracyclines, in decreasing order of frequency. Antimicrobial usage data on these operations were not available and national data on antimicrobial usage on veal operations is lacking, unlike dairy and beef calves for which these data have been periodically tabulated. Veal calves are considered a separate production class from dairy and beef steer calves during the FDA drug approval process, so antimicrobial usage in these animals cannot be accurately extrapolated to usage in veal calves. Currently, aminoglycosides (streptomycin), β-lactams (ampicillin and amoxicillin), sulfonamides (sulfabromomethazine, sulfamethazine, sulfaethoxypyridazine, and sulfamethazine), bacitracin, and tetracyclines are approved for oral administration in veal calves in the United States [56,57].
Oxytetracycline and chlortetracycline have been included in scour (diarrhea) medication and supplemented in milk replacers fed to calves and may, in part, select for bacteria encoding resistance to these antibiotics. β-lactams, specifically ampicillin and amoxicillin, can be administered orally and intramuscularly for the treatment of bacterial enteritis and bovine respiratory disease (BRD), a significant cause of morbidity and mortality in calves and leads to considerable economic losses. Tulathromycin (macrolide) (subcutaneous administration), and ceftiofur (β-lactam, veterinary cephalosporin) (subcutaneous or intramuscular administration) can also be used for the treatment of BRD. Intramuscular administration of ceftiofur and florfenicol in dairy calves has been associated with a transient increase in resistant fecal E. coli [58]. Neomycin, an aminoglycoside that has been used to prevent scours in dairy calves, is not approved for oral administration in veal calves. What is not known is how historical use of antimicrobials within the birth herd (where neomycin is approved for use in replacement calves) can influence the presence and types of resistance carried by in the veal calves after they are moved from the source herd to the veal farm. These neonatal exposures should be considered a potential source of resistance that may remain within the veal calf gut after transitioning to a veal farm.
Some of the most common resistance genes among the MDR isolates confer resistance to antimicrobials approved for use in veal calves [59]. All but one MDR isolate encoded tetracycline resistance genes. β-lactamases, sulfonamide resistance genes, and aminoglycoside resistance genes were detected in most MDR isolates, and the macrolide resistance gene mph(A) was detected in a considerable number of isolates.
In addition to direct treatment with antimicrobials, calves can be exposed to antimicrobial residues in colostrum from cows treated with intramammary antibiotics at the time of dry-off (initiation of break from lactation) for mastitis treatment and prevention. Dairy operations often treat mastitis with first and third generation cephalosporins (β-lactams) and calves are sometimes fed unsaleable waste milk containing antimicrobial residues from treated lactating cows [60]. Since resistance genes are often co-located on mobile elements, exposure to one antimicrobial may select for multiple resistance genes [61]. For example, based on the genetic co-occurrence data from isolates in this study, exposure to neomycin or oxytetracycline could potentially select for trimethoprim (dfrA12), phenicol (floR), and/or sulfonamide (sul2 and sul3) resistance genes.
The co-occurrence of metal and biocide resistance genes with antibiotic resistance genes is notable and has been identified previously [10,[62][63][64][65]. Our isolate genomics results confirm the metagenomic analysis of Liu et al. [10], which showed a similar relationship in the metagenomes of dairy calf feces. Our analysis confirmed that MRGs and BRGs are associated with some ARGs, but there is also evidence of a negative cooccurrence between all metal resistance genes and the most frequently identified antibiotic resistance genes (sul2, aph(3'')-Ib, and aph (6)-Id). Silver (sil genes) and copper (pco genes) resistance had the most frequent positive cooccurrence with ARGs, with some of these conferring resistance to antibiotics of human health significance such as bla CTX-M-15 (ESBL) and mphA (azithromycin resistance). Associations between silver and ARGs have been noted previously, although some of these are not consistent with our findings [66,67]. Congruent with our results, silver resistance was found to be positively associated with bla CTX-M in E. coli by Sütterlin et al. [68]. However, this was only observed with CTX-M-15 in our analysis. Copper is present in animal feeds, including colostrum, milk, milk replacer, and calf starter, and the positive co-occurrence between pco genes and bla CTX-M-15 and mphA indicates a potential selection for ARGs due to this dietary component. A positive co-occurrence between quaternary ammonium compound (QAC) resistance gene qacEΔ and bla CTX-M-15 and mphA was also observed, as well as between QACresistance gene sugE1 and the ESBL gene bla CMY-2 . QACs are among some of the antiseptics and have been used on farms for cleaning surfaces and equipment. It is unknown if QACs were used on these veal operations, but exposure of the dams to QACs prior to calving could potentially result in exposure of the calves to these compounds, or transmission of QAC-resistant bacteria from dam to calf.
Among these E. coli isolates there was a notable occurrence of genes conferring resistance to antibiotics of public health significance such as CTX-M, as well as quinolone resistance gene qnrS1, aminoglycoside and fluoroquinolone resistance gene aac(6')Ib-cr, and azithromycin resistance gene mph(A). Extended-spectrum β-lactamases (ESBL) are the most common proteins responsible for β-lactam resistance, and E. coli that harbor these genes are typically resistant to extended spectrum cephalosporins and monobactams. From a public health perspective, this is significant since β-lactams are among the most frequently prescribed antimicrobials globally, and ESBL-producing Enterobacteriaceae are considered a serious public health threat by the Centers for Disease Control (CDC) [https://www.cdc.gov/drugresistance/ pdf/threats-report/2019-ar-threats-report-508.pdf]. The ESBLs identified in these genomes (bla CTX-M-and and bla CMY ) are particularly notable since they are known to confer resistance to the 3 rd generation cephalosporin, ceftazidime, ceftriaxone, cefotaxime and the 4 th generation extended spectrum penicillin/β-lactamase inhibitor piperacillin/tazobactam [World Health Organization Essential Medicine Watch Group Antimicrobials, ttps://apps.who.int/ iris/rest/bitstreams/1237479/retrieve]. Resistance to β-lactams has been increasing worldwide and CTX-M lactamases are among the most prevalent ESBLs in human infections. bla CTX-M-1 and bla CTX-M-15 are globally distributed. bla CTX-M-15 is notable because it is associated with pandemic ST131, but in these veal isolates it is mainly associated with A-ST10 Cplx strains. CTX-M genes were identified in strains with ExPEC VFs that were also STs associated with ExPEC infections (ST69, ST410, ST617, ST648, and ST167). The presence of plasmid-mediated quinolone resistance (PMQR) genes aac(6')Ib-cr and qnrS1 is significant as fluoroquinolones comprise a group of broad spectrum antibiotics of critical importance in animal and human health. Both of these genes increase the quinolone minimum inhibitory concentration (MIC) which gives these strains a competitive advantage in the presence of a fluoroquinolone challenge [69]. These genes were identified in ST69, ST617, and ST167 isolates that also encoded ExPEC VFs. Azithromycin has been historically used to treat Gram-positive infections but has shown promise as an alternative to treat infections with Enterobacteriaceae that may be resistant to other commonly used therapeutics [70]. Therefore, resistance to this antibiotic is potentially an emerging public health threat' and should be closely monitored.

Virulome differences and VFs associated with MDR and susceptible genotypes
On average, the virulomes of MDR strains and susceptible strains were also somewhat different according to the NMDS ordination analysis and the analysis of similarities test. Although it is clear that the virulence profiles of some MDR isolates are more similar to those of some susceptible isolates, it should be noted that the presence of these VFs does not confirm that the isolates are human pathogenic strains, but only indicates the potential for these strains to cause disease in humans. The health statuses of the animals were not reported and for some of these VFs their role in the pathogenesis in calves is not well-defined or are not known to cause disease in these animals. Some of these VFs are also known, or presumed, to enhance colonization of the mammalian gut, and therefore act as fitness factors that may confer a competitive advantage in the calf gut, regardless of disease outcome for the animal. Interestingly, our analysis indicated that stx1AB and stx2AB were enriched in the susceptible isolates when compared with their presence in the MDR isolates. Similarly, eae and tir, both located on the locus of enterocyte effacement (LEE) and involved in adherence to the human small intestine wall in EPEC and EHEC, were enriched in the susceptible strains and absent in the MDR strains. Their presence in susceptible strains is not surprising, but their absence in the MDR isolates is noteworthy. MDR STEC are occasionally shed by cows and calves [71][72][73], but among the animals sampled in this study they represent an undetectable minority based on the number of isolates collected and/or sequenced. Future work should investigate any potential interplay between carriage of stx and LEE and the presence of ARGs, or if there are veal management practices that select against MDR STEC.
Two accessory plasmid-borne iron acquisitions systems, sitABCD (Sit system) and iucABC-iutA (aerobactin), were significantly more abundant in the MDR isolates. Iron is a common limiting factor of bacterial growth and replication and is vital for many bacterial processes [74]. These two systems are involved in scavenging extracellular iron within the host environment, most notably the human gastrointestinal system and urinary tract. In E. coli, these two systems are primarily found on IncFIB plasmids, which are known to also encode multiple ARGs. IncFIB plasmid replicons were detected in all but one MDR isolate encoding sitABCD and/or iucABC-iutA genes and has previously been shown to encode both ARGs and iron acquisition systems [75,76].
Milk has a low iron content and milk-fed calves are at a high risk of anemia [77,78]. We hypothesize that the low input of iron into the calf gut may, in part, select for bacterial strains that encode accessory iron scavenging systems thereby allowing these organisms to outcompete strains lacking these systems. Similar to the phenomenon of antibiotic administration selecting for bacteria encoding complementary resistance genes, low iron environments potentially select for strains with genes encoding iron acquisition systems. Since these systems are co-located on resistance gene-encoding plasmids, the low iron input to the calf gut may coincidentally select for MDR strains in the absence of antibiotic administration and may synergistically act with resistance genes as simultaneous selection pressures to select for these strains. Similarly, P fimbriae genes (pap), involved in binding to glycolipids of the human urinary tract epithelial cells, were more abundant in MDR than susceptible isolates, but their role in binding to young bovine intestinal cells has not been evaluated. The differential enrichment of VFs in susceptible versus resistant strains has been previously identified in human isolates, but the selection pressures driving these trends are currently unknown [79,80]. We suggest that such differential enrichment of accessory genes in MDR isolates confers an advantage upon these strains in the calf gut.
Based on genomic comparisons, closely related strains were isolated from different veal farms. Animals on these premises were primarily acquired from auction houses or buying stations, where animals from many different farms are typically commingled, and therefore exposed to a large suite of bacteria. This could include MDR E. coli, which they could then transmit to other animals in their cohort, either by direct contact during transport to the farm or at the farm, or through intermediary means such as farm workers or fomites. There is also the potential that different calves shedding highly similar strains were born at the same dairy farms on which they were exposed to the same microbial communities, and therefore could potentially be colonized by clonal copies of MDR E. coli that are endemic in their birth herd. Individual veal calves could not be traced back to their herd of origin to investigate this possibility. The repeated isolation of highly similar strains from different sources within a highly diverse E. coli population [43] suggests they haven an enhanced ability to persist within the veal farm environment. We have previously demonstrated that clonal MDR E. coli can be isolated from different animals (and animals of different ages) on the same farm, suggesting that transmission of E. coli occurs between animals and that some MDR strains may be selected for, or persist, in the bovine gut [15]. Results of this analysis suggest that MDR E. coli have the potential to spread between animals at auction houses and on dairy farms; transmission of these bacteria can spread between farms when co-colonized calves are sold to different veal farm operations.
This study demonstrates that MDR E. coli in veal operations are highly diverse but dominated by phylogroup A/ST Cplx 10 strains. Further, a significant proportion of these MDR strains are similar to ExPEC isolates known to cause infections and many encode VFs involved in colonization and virulence outside of the human intestine, particularly in the urinary tract. The encoded VFs include iron-scavenging systems, most likely co-located with resistance genes on plasmids, that may enhance the colonization of the low-iron, milk-fed calf gut environment. This analysis also demonstrated that ARGs of human health significance and MDR E. coli strains are circulating among veal calves in the same and different farms. Although this work focused on veal calves, it has relevance outside of this production system and future work focused on antimicrobial resistance in other systems or environments should evaluate the multiplicity of factors that may influence, or be associated with, the carriage of resistant bacteria. Research aimed towards mitigating the carriage of resistance in food animal production should consider the role of management practices, not just limited to antimicrobial administration, in the carriage and maintenance of resistant organisms.
Supporting information S1