Neonatal and post-weaning diarrheal (PWD) disease affecting pigs during the first weeks after birth result in significant economic losses for the pig industry due to mortality, decreased weight gain, costs derived of treatment and handling, and vaccinations (Fairbrother et al., 2005; Luppi, 2017).

Among the etiologic agents, there are certain Escherichia coli pathotypes commonly implicated in enteric diseases in piglets, namely, Shiga toxin-producing E. coli (STEC) and enterotoxigenic E. coli (ETEC). STEC isolates carrying the shiga toxin type 2e (Stx2e) are the causative agents of edema disease in weaning piglets, while ETEC isolates encoding the heat stable (STa, STb) and/or heat labile (LT) enterotoxins cause secretory diarrhea in newborn and weaned piglets (Nagy et al., 1999; Fairbrother et al., 2005; Fairbrother and Gyles, 2012). Furthermore, some isolates harbor both the Stx2e and enterotoxin genes, being able to cause symptoms of edema disease and diarrhea in the same animal (STEC/ETEC) (Barth et al., 2011). A common feature of STEC and/or ETEC swine pathotypes is the expression of specific fimbrial adhesins which allow bacterial colonization of the mucosal surface. The most commonly reported are of types F4 (previously known as K88) and F18 (F107, 2134P, and 8813), both with different antigenic variants: three for F4 (ab, ac, and ad), with F4ac being the most prevalent, and two main variants for F18 (F18ab associated with edema disease and F18ac with PWD) (Westerman et al., 1988; Rippinger et al., 1995; Francis, 2002). Other associated fimbriae of lower prevalence include F5 (K99), F6 (987P), and F41, whose number of active receptors present on the intestinal epithelial cells decreases with age (Vu Khac et al., 2006; Luppi et al., 2016). Enteropathogenic E. coli (EPEC) is another pathotype found in pigs with enteric colibacillosis (Brand et al., 2017; Malik et al., 2017) which was first associated with diarrhea in infants (Shuman and Stock, 1956). EPEC isolates possess an outer membrane protein adhesin or intimin (Eae), responsible for intimate attachment of the bacteria to the host intestinal epithelium, which together with a complex secretion system leads to the development of the “attaching and effacing” (AE) lesion (Frankel and Phillips, 2008).

Colistin (polymyxin E) is one of the few cationic antimicrobial peptides commercialized in both human and veterinary medicine. In humans, colistin is used as a last resort against infections of multidrug-resistant (MDR) Gram-negative bacteria (Poirel et al., 2017). However, it has been extensively used since the 1960s in food animals, and particularly in swine with different purposes: therapeutically, prophylactically, and even for growth promotion (Rhouma et al., 2016). Since the description of the plasmid-mediated colistin resistance mcr-1 gene in late 2015 (Liu et al., 2016), several mcr genes have been described (Xavier et al., 2016; Borowiak et al., 2017; Carattoli et al., 2017; Liu et al., 2017; García et al., 2018; Hammerl et al., 2018; Rebelo et al., 2018). Livestock, and particularly pig farming, has been singled out as the principal reservoir for colistin resistance spread (Rhouma et al., 2016). Besides, co-occurrence on the same plasmid of mcr and extended-spectrum beta-lactamase (ESBL) or other beta-lactamase genes (such as carbapenemase) are being increasingly identified in isolates from different origins (Bai et al., 2016; Haenni et al., 2016). In addition, one of the major health concerns associated with E. coli is the role of certain clonal groups in the emergence and dissemination of antimicrobial resistance, namely those belonging to sequence type (ST) 10, ST69, ST131, ST405, ST410, or ST648 (Hansen et al., 2014; Falgenhauer et al., 2016; Johnson et al., 2017). Of particular importance is the ST131 due to its prevalence in community- and hospital-acquired urinary tract infections (UTIs) (Nicolas-Chanoine et al., 2014; Kallonen et al., 2017), and whose potential contribution to the global dissemination of antimicrobial resistances has been also highlighted in food-producing animals (Mora et al., 2010).

In the present study, we characterized a collection of E. coli isolates obtained in Spain during the period 2006–2016 from pigs suffering enteric colibacillosis with three main aims: (i) to define clonal groups of clinical importance in swine enteric colibacillosis; (ii) to analyze the rates of antibiotic resistance in pig farming in Spain, including mobile resistance to colistin (mcr gene); and (iii) to gain knowledge about the presence and zoonotic potential of ST131 isolates of livestock origin.

Pig diarrhea caused by E. coli is a worldwide economically important disease for the swine industry. While ETEC neonatal diarrhea can be effectively controlled by vaccination of pregnant sows, passive protection is quickly lost after weaning (Melkebeek et al., 2013; Matías et al., 2017). Currently, few vaccines are commercially available for PWD, and few provide protection against different E. coli pathotypes (Nadeau et al., 2017; Won and John Hwa, 2017). In the present study, we performed a wide epidemiological study on a collection of E. coli isolates recovered from 464 faecal samples of enteric colibacillosis in pig production farms of Spain over a 10-year period to know the E. coli traits (pathotypes) presently implicated in swine diarrheas, and how the use of antibiotics (including colistin) could have affected the selection and emergence of resistant strains. Most samples (73%) were of diarrheagenic cases occurred during the post-wean period which showed to be significantly associated with ETEC (67%) followed by aEPEC (21.7%). By contrast, aEPEC was more prevalent (60.3%) among diarrheas of suckling piglets, followed by ETEC (38.8%). STEC/ETEC or STEC were recovered in 11.3 and 0.9% of PW and neonatal diarrhea, respectively. It is noteworthy that the F4 colonization factor was identified among neonatal ETEC without significant differences compared to PWD ETEC (40.5 versus 27.7%) while, as expected, F18 was only present among PWD isolates.

Complete data on the prevalence, serotypes, and pathotypes are not frequently available, which makes comparisons difficult. However, previous analysis had already suggested differences between countries (Frydendahl, 2002; Blanco et al., 2006; Vu Khac et al., 2006). A recent study across Europe regarding ETEC and STEC PWD pathotypes describes a higher prevalence of F4 compared to F18 isolates in Belgium and The Netherlands, France, and Italy (Luppi et al., 2016). Although pairwise comparisons for each country gave no significant differences, the epidemiologic situation of PWD is clearly different from that found in our study for Spain (23.9% F4 isolates versus 51.5% F18 of ETEC, STEC, and STEC/ETEC isolates; P < 0.001). The reported association of F18 E. coli isolates with PWD in other countries varied widely, from 1.7% in Australia (Smith et al., 2010), 35% in Slovakia (Vu Khac et al., 2006), 39.3% in Denmark (Frydendahl, 2002), 53% in United States (Post et al., 2000), to 62% in Poland (Osek et al., 1999). High F18 prevalence (62.9% of 648 isolates) was as well reported in a study performed in Japan where, in addition, most isolates carried stx_2e gene (60.1% of 648 isolates) (Kusumoto et al., 2016) which describes a very different pathogenic profile in relation to other geographic areas. Thus, we found 10% of stx_2e-positive isolates among the Spanish collection of 481 E. coli, similar to the numbers reported in the European study of Luppi et al. (2016). Like the Japanese study, most of our stx_2e isolates (73.9%) were positive for F18.

The majority of ETEC associated with diarrhea in pigs appear to belong to a limited number of serogroups, being O8, O138, O139, O141, O147, O149, and O157 the most commonly reported worldwide. Similarly, STEC causing disease primarily belong to O138, O139, and O141 (Nagy and Fekete, 1999; Frydendahl, 2002; Fairbrother et al., 2005). The majority of ETEC and STEC isolates analyzed in this work belonged to the most common serogroups O8, O138, O139, O141, O147, O149, and O157. In addition, O15, O35, O45, O101, and O108 serogroups, less frequently reported, were also present within the ETEC isolates ranging between 15.5% for O108 and 1.8% for O35; likewise, STEC single isolates presented the infrequent O2, O22, O36, O142, and O158 serogroups. Apart from the 50 serogroups determined within 337 isolates, it is of note that other 144 (23.7%) isolates of different pathotypes did not react with any of the O1 to O185 antisera and remained as non-typeable (ONT).

We found in our study a high prevalence of aEPEC (156 out of 481; 32.4%) as presumptive agents of the clinical condition, and significantly associated with neonatal samples. For atypical EPEC, both animals and humans can be reservoirs and are known as pathogenic for children and young animals (Trabulsi et al., 2002; Watson et al., 2017). aEPEC have also been implicated in PWD in pigs but their pathogenicity still remains unclear (Malik et al., 2017). According to these and other authors, there exist differences of serotypes and intimin types among porcine eae-positive isolates in relation to the pathogenic potential, such as sero/intimin type O123:H11/eae-β1 or O45/eae-β1 (both determined in our collection) with AE activity on ileal villi and frequently occurring in diarrheagenic pigs (Zhu et al., 1994; Malik et al., 2017). In previous studies on swine diarrhea in Slovakia, we had found a much lower aEPEC involvement (3.2 and 0.9% of neonatal and PWD isolates, respectively) (Vu Khac et al., 2006, Vu Khac et al., 2007).

Since the discovery of the mcr-1 gene in late 2015, its detection has been reported globally, but its detection rate has been variable depending on the geographic region, source, and method of identification. Overuse of colistin in food animals is believed to have trigged the emergence and spread of mcr-1, consistently with the higher figures found in poultry and pig industry (Sun et al., 2018). Here, the characterization of 65 representative mcr-1 isolates, including 33 ETEC, 24 aEPEC, five STEC/ETEC, and three STEC showed that all were phenotypically resistant to colistin, and most (61 of 65) met the MDR definition of Magiorakos et al. (2012). MDR in pig industry has been previously reported in different studies, and associated with the widely use of aminoglycosides and beta-lactams in veterinary medicine (Rhouma et al., 2016). The 37 mcr-1 positive isolates recovered from PWD in Italy also showed genetically diverse pathotypes and all but one were resistant to ≥3 different antimicrobial families (Curcio et al., 2017). Li et al. (2017) found a higher prevalence of mcr-1 positie isolates among pathogenic E. coli from diseased pigs than from healthy pigs (45.1 versus 15.7%, P = 0.000); besides, resistance profiles of mcr-1 positive E. coli were more extensive than those of mcr-1 negative isolates. In our study, it is also of concern the finding of 11 out of the 65 mcr-1 isolates with MIC >32 mg/L for fosfomycin. Fosfomycin resistance is uncommon and primarily associated with specific chromosomal mutations, however plasmid-mediated resistance in livestock has been detected in Asian countries (Chan et al., 2014), and quite recently in France (Lupo et al., 2018). This antibiotic is considered an “old” broad-spectrum antibiotic such as colistin, reconsidered as a good candidate for treating infections caused by MDR microorganisms (Dijkmans et al., 2017).

Six ETEC and one STEC mcr-1 isolates were also carriers of ESBL genes. In addition, other seven mcr-1 isolates showed to harbor mcr-4 (three ETEC) and mcr-5 (two ETEC and two aEPEC) genes. We have not investigated if the double carriage of our isolates was in the same or different plasmids. Previous studies have reported the co-existence on the same plasmid of other resistance genes, such as bla_CTX-M, bla_CMY, bla_TEM, fosA, qnrS, floR, and oqxAB, in various combinations, alongside mcr-1 (Li R. et al., 2017). At the same time, new mcr genes have been successively discovered in plasmids of different origins, as well as co-transfer of variants in promiscuous plasmids (Sun et al., 2018). Since colistin is commonly thought as the last line antibiotic for the treatment of infections caused by MDR and extensively drug-resistant Gram-negative pathogens, such as carbapenem-resistant Enterobacteriaceae, it must be beared in mind that its use could lead to co-selection of colistin with other antibiotic-resistant isolates.

Of the 41 ETEC, STEC and STEC/ETEC mcr-1 isolates (Table 2), 18 were carriers of F18. Three F18 subtypes have been reported so far (F18ab, F18ac, and F18New) differentiated by five amino acids at specific positions (31, 57, 59, 83, and 122) with strong association between subtypes and pathotypes (Bosworth et al., 1998; DebRoy et al., 2009; Barth et al., 2011; Byun et al., 2013). Here, the F18 positive isolates subtyped as F18ac corresponded with the ETEC and STEC/ETEC pathotypes (15 and two isolates, respectively) while F18ab was detected in a single STEC isolate. Furthermore, the predicted amino acid sequences of the fedA genes identified in this study together with 13 sequences from Barth et al. (2011) were compared and their relatedness analyzed, demonstrating its polymorphic variability (Bosworth et al., 1998).Thus, 10 different amino acid sequences were found among the F18ac subtype of our collection, one of them (detected in four isolates) showed 100% similarity with the GQ325624 sequence identified by Barth et al. (2011) in Germany (Supplementary Figure S1). Interestingly, the deduced amino acid sequence of one isolate (FV18854) presented a novel glycine residue in position 59; however, it was classified as F18ac since the other four amino acids positions are equal to those described for this subtype (Supplementary Table S6) and clustered together with F18ac group in the phylogenetic tree (Supplementary Figure S1).

We found high diversity in the phylogenetic analysis of the 65 mcr-1 diarrheagenic isolates; however, more than 50% (38 out of 65) belonged to A-ST10 Cplx and from those, 29 isolates of different pathotypes showed CH11-24. The ST10 Cplx of E. coli, widely disseminated among animal and human intestinal samples both as a commensal or as a pathogen, is commonly encountered as antimicrobial susceptible but also linked with MDR and ESBL, and recognized as an emerging food-borne ExPEC lineage (Manges and Johnson, 2012). A recent study on genotyping of 68 mcr-1-like-positive E. coli from food animals at slaughter in Europe between 2002 and 2014, found a high genetic diversity among the isolates with 38 different STs, but being ST10 Cplx and, specifically the ST10, the most prevalent (El Garch et al., 2017, 2018). This finding is also consistent with the observation of Matamoros et al. (2017) who stated that within the overall diversity in the E. coli population, two lineages (ST10 and ST155) might function as reservoirs of the mcr-1 gene, the largest of which was linked to ST10 (Matamoros et al., 2017). In our particular case, it is also important to consider that ST10 Cplx is one of the main E. coli clonal complexes associated with porcine ETEC. In fact, Shepard et al. (2012) found that the majority of the porcine ETEC isolates belonged to three clonal complexes: 10, 23, and 165, and pointed out the association showed by ST10 Cplx and ST23 Cplx with certain resistance-associated elements, such as AmpC-type beta-lactamases, NDM-type carbapenemases, and other ESBLs. Other STs established in the present study such as ST1, ST29, ST42, ST100, or ST4247 have been previously associated to enteric pathogenic E. coli in pigs (Abraham et al., 2014; Kusumoto et al., 2016).

Interestingly, nine of the 24 aEPEC mcr-1 characterized in this work belonged to B1-ST29 and carried eae-β1 gene. They showed six serogroups, including O26, with two H-antigen combinations (H9 or H11). Additionally, two aEPEC belonged to the clonal groups O103:H2-B1-ST20 (eae-β1) and O145:H28-E-ST1034 of the ST32 Cplx (eae-γ1), respectively. Enterohemorrhagic E. coli (EHEC) is the causative agent of bloody diarrhea, the hemolytic-uremic syndrome (HUS), and thrombotic thrombocytopenic purpura (TTP). Besides the intimin (Eae) which confers the ability to cause AE lesions, EHEC harbors bacteriophage-encoded stx genes. Currently, the vast majority of EHEC infections are caused by isolates belonging to five O serogroups, namely O157, O26, O103, O111, and O145 (Mellmann et al., 2008). Eichhorn et al. (2015) analyzed by MLST the phylogenetic relationships in a collection of 250 isolates belonging to serogroups O26, O103, O111, and O145 defined as EHEC, obtained from different sources of isolation. As a result, the majority of the O26 and O111 EHEC isolates clustered into the ST29 Cplx. O103 isolates clustered mainly in ST20 Cplx, and most isolates of O145 were found within ST32 Cplx. In addition to EHEC, the ST29 Cplx cluster also included aEPEC isolates. According to these authors, the finding that aEPEC and EHEC isolates of non-O157 serogroups share the same phylogeny suggests an ongoing microevolutionary scenario in which the phage-encoded stx is transferred between aEPEC and EHEC. The concept of interconversion between STEC and aEPEC had been previously suggested by Bielaszewska et al. (2007) for O26 isolates by the loss, as well as by the gain, of the stx-encoding prophage through the lysogenic conversion. Applying the concept of bidirectional conversion, it could be hypothesized that the EPEC strains could function as pre-STEC strains that integrate the stx prophage into their genomes. However, this hypothesis together with the potential human pathogenicity of EPEC isolates from sources such as swine requires extensive research.

A highly important finding of the present study was the recovery of 18 ST131 isolates among the 464 fecal samples of diarrheanic pigs. As far as we know, there are only two reports of pig origin for this clonal group: the first obtained from pork meat in Denmark in 2003 (Trobos et al., 2009), and the second was a CTX-M-1 ST131 recovered from a gastrointestinal tract infection in a pig among 1,378 isolates analyzed in the frame of an ESBL monitoring program performed in Germany during the years 2006–2007 (Schink et al., 2013). None of the 18 ST131 detected in this work were carriers of ESBL genes but instead, most were MDR, harbored 12–15 virulence-gene traits and, surprisingly, seven isolates were mcr-1 carriers. Very few mcr-1 isolates have been reported belonging to the pandemic clone ST131, responsible for the high incidence of ExPEC infections as well as the worldwide dissemination of multidrug resistance (Nicolas-Chanoine et al., 2008, 2014). So far, ST131 carrying mcr-1 was first described in an isolate from chicken meat (Hasman et al., 2015), then in poultry (Ewers et al., 2016) and in a few human clinical isolates (Kuo et al., 2016; Sonnevend et al., 2016; Ortiz de la Tabla et al., 2017). To the best of our knowledge, this would be the first report of ST131 mcr-1 isolation in pigs. The antimicrobial susceptibility of the seven mcr-1 ST131 swine isolates of this work determined by MIC, confirmed that all but one isolate were colistin resistant. The presence of the mcr-1 gene in susceptible isolates has been described (Fernandes et al., 2016; El Garch et al., 2017). Here, we investigated the mcr-1 gene from the susceptible isolate with primers listed in Supplementary Table S3, and failed when trying to obtain the entire gene sequence, suggesting the gene is truncated or has been modified by agents such as insertion sequences like those previously reported (Terveer et al., 2017; Zhou et al., 2018); however, further work is necessary to determine the cause of loss of function. Besides, the seven mcr-1 ST131 swine isolates showed resistance against 6–12 antibiotics, including a ciprofloxacin-resistance determined in 1 isolate (Table 5). In a nationwide study performed in Spain during 2005–2012 in five hospitals of different regions, the ST131 accounted for 490 (16%) of the 2,995 isolates obtained from clinical human samples (Dahbi et al., 2014). The majority (78%) of the ST131 isolates belonged to the recently emerged fluoroquinolone-resistant ST131-H30R subclone and 61.6% to the H30-Rx additionally carrier of CTX-M15, while ST131-H22 subclone was the second most prevalent of the human collection. Dahbi et al. (2014) found in their work different patterns of associated antimicrobial resistances in relation to the fimH allele (fimH30 or fimH22). Ciprofloxacin, trimethoprim-sulfamethoxazole, gentamicin, and tobramycin resistances were significantly associated with fimH30 isolates in comparison with fimH22 isolates. In fact, they only detected one and eight isolates resistant to ciprofloxacin and trimethoprim-sulfamethoxazole, respectively, among 20 ST131-H22. Dahbi et al. (2014) also found association of the fimH alleles with the virulence profile, being all 20 fimH22 isolates of virotype D, subtypes D1, D2, D4, and D5. In the present work, we found high genetic variability within the eight fimH alleles of 18 ST131 isolates; however, fimH22 was the most prevalent (nine isolates) and the other seven alleles were fimH22 variants whose nucleotide and aminoacidic differences were compared with those reported by Dahbi et al. (2014) (Supplementary Tables S7, S8 and Supplementary Figures S2, S3). All our swine isolates conformed the D virotype and showed D5 (13 isolates) and D2 (two isolates) also described within the human collection. Worryingly, the seven swine ST131 carriers of mcr-1 also showed a higher number of resistances, including to trimethoprim-sulfamethoxazole (seven isolates), tobramycin (six isolates), gentamicin (six isolates), and ciprofloxacin (one isolate), more associated to the fimH30 subclone according to Dahbi et al. (2014). It is also remarkable that the 18 ST131 swine isolates exhibited virulence traits that satisfied the ExPEC and UPEC status according to Johnson et al. (2003) and Spurbeck et al. (2012) definitions. In the PFGE macrorestriction, comparison of our swine isolates with those of clinical human origin, two pig isolates clustered with two human ST131 D5 showing high similarities (≥85%).

In conclusion, the comprehensive characterization of this wide collection of E. coli isolates recovered from enteric swine colibacillosis in Spain provides valuable epidemiological information about the current pathotypes involved in this economically important pathology, as well as alerts about the worrisome presence of MDR clones such as ST131. Acquisition of mcr-1 by this specific clone means an increased risk due to its special feature of congregating virulence and resistance traits, together with its spread capability. Here, we give evidences of the potential human pathogenicity of ST131 D5 isolates of swine origin due to their genetic identity.

AM and JB conceived and designed the experiments. IG-M, VG, DD-J, SF-S, and MA performed the experiments. AM, JB, VG, IG-M, JB, and MB analyzed and interpreted the data. AM, VG, JB, and IG-M drafted the manuscript. AM, JB, VG, IG-M, JB, MB, MA, DD-J, and SF-S provided critical input and approved the final manuscript.

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.