In this study, 111 previously published aEPEC strains isolated during epidemiological studies conducted in Brazil were revisited (Vieira et al., 2001, 2010; Gomes et al., 2004). Seventy-eight strains (70.3%) were isolated from diarrhea cases and 33 (29.7%) from asymptomatic individuals. The EPEC pathotype was determined by the presence of the eae (intimin) gene and the absence of all the other genes related to any other IPEC pathotype, while the atypical subclassification was determined by the lack of the bfpA (BFP) gene (Vieira et al., 2001, 2010; Gomes et al., 2004).

The production of bacteriocins was assessed as described earlier (dos Santos et al., 2017). Briefly: strains were grown on three different solid culture media: Nutrient agar (NA), Tryptone yeast extract (TYE) (Difco Laboratories, Sparks-MD), and TYE plus 1% trypsin (Gibco^®) (TYE-T). The indicator strain E. coli C600, grown overnight (ON) in Tryptic Soy broth (TSB) (Difco), was diluted 1:100 in sterile saline solution (0.85% NaCl) and spread on the surface of the TYE, NA, and TYE-T media plates to obtain a bacterial lawn. Three microliters of ON TSB cultures of aEPEC strains were seeded over the indicator lawn in equidistant spots. E. coli strains GC138 and HB101 were used as positive and negative controls, respectively (dos Santos et al., 2017). After ON incubation at 36°C, the strains producing bacteriocin appeared surrounded by an inhibition zone of the indicator strain in TYE and NA. As a counterproof, the presence of trypsin in the TYE-T would abolish the bacteriocin effect, resulting in non-inhibited growth of the indicator strain by the bacteriocin-producer strains.

The strains that were positive in the assay were submitted to PCR to evaluate whether the bacteriocin that they produced belongs to bacteriocin types B, E1, E2, Ia, Ib, M, and microcin V using primers (Supplementary Table S1) and conditions described by Šmajs et al. (2010).

The hemolytic activity was assayed by inoculating 3 μL of an ON TSB culture in Columbia agar medium (Kasvi, Pinhais, PR) supplemented with 5% (v:v) of washed and defibrinated sheep blood and 10 mM CaCl₂ (Beutin et al., 1989). Hemolysis was detected after 3 h and 24 h of incubation at 37°C by a clear zone surrounding the inoculated strain. E. coli strains U4-41 and EDL 933 were the positive controls for alpha-hemolysin and enterohemolysin production, respectively. Klebsiella spp. was the negative control. The presence of the hlyA and ehxA genes was searched by PCR (Supplementary Table S2) for all hemolytic strains as previously described (Schmidt et al., 1995; Yamamoto et al., 1995; Johnson and Stell, 2000).

The serum resistance assay was done using a commercially acquired pool of human sera (Sigma-Aldrich, MO, United States). A logarithmic phase culture grown in TSB was washed twice using phosphate-buffered saline (PBS). The bacterial concentration was adjusted to 1 × 10⁷ CFU/mL in PBS. The 100 μL of bacterial suspension was mixed with normal serum (1:1 v) and incubated at 37°C for 24 h. Another bacteria-serum mixture was prepared using Complement-inactivated serum obtained by heating at 56°C for 30 min. At 0 and 24 h, 5 μL of the mixtures were plated onto MacConkey agar to detect strains that survived serum bactericidal activity during the evaluated period. The E. coli K-12 strain MG1655 was used as a serum-sensitive control, and the strain E. coli J96 (ExPEC prototype) was used as a serum-resistant control.

The genes iss and traT, considered involved in serum resistance, were searched by PCR in all aEPEC strains as previously described (Johnson and Stell, 2000; Dezfulian et al., 2003) using specific primers (Supplementary Table S2).

The phylogenetic origin of aEPEC strains was determined using the Clermont quadruplex PCR scheme (Clermont et al., 2013) followed by confirmation (for phylogroups A, C, D, and E), which enables the classification of E. coli into one of seven phylogroups (A, B1, B2, C, D, E, and F).

The aEPEC strains were molecularly classified as ExPEC by the presence of at least two genes among papC, afaBCIII, sfaDE, kpsMTII, and iutA (Johnson et al., 2003), and as UPEC when presenting at least three among vat, chuA, fyuA, and yfcV (Spurbeck et al., 2012). The PCR reactions were performed using primers and conditions described in Supplementary Table S3 (Le Bouguenec et al., 1992; Johnson et al., 1997; Johnson and Stell, 2000; Spurbeck et al., 2012).

Eight pathogenicity islands (PAIs) present in ExPEC prototype strains CFT073 (I_CFT073 and II_CFT073), J96 (I_J96 and II_J96), and 536 (I₅₃₆, II₅₃₆, III₅₃₆, and IV₅₃₆) were searched using primers and conditions (Supplementary Table S4) described by Sabaté et al. (2006) with minor modifications (da Silva et al., 2017).

The clonal relationship of aEPEC strains was assessed by RAPD using primers 1247 (AAGAGCCCGT) and 1283 (GCGATCCCCA) as previously described for E. coli (Nielsen et al., 2014). To determine a clonal relationship, the amplification pattern obtained for both primers was evaluated by the BioNumerics program version 7.6.3 (Applied Maths, Sint-Martens-Latem, Belgium). The similarity of 90% among the amplification profiles for both primers was established as a cutoff to determine the clonal relationship among the strains evaluated.

All PCR reactions were performed using GoTaq® Green Master Mix (Promega, WI, United States) and freshly boiled bacteria as templates. The PCR products were evaluated after electrophoresis in agarose gel.

The virulence of aEPEC strains carrying ExPEC characteristics was investigated “in vivo” using the Galleria mellonella model following procedures previously described (Santos et al., 2020b). An inoculum of 10 μL of the bacterial suspension containing 1 × 10⁷ CFU/mL was injected into the larvae prolegs using a hypodermic insulin syringe with a 31-gauge needle (Uniqmed, Chungcheongnam-do, KOR). The CFU injected inoculum was confirmed by plate count. The tests were performed in three independent assays, using five larvae for each strain tested. The E. coli strains J96 and MG1655 were used as positive and negative controls, respectively, and PBS injections were used as the procedure control. The E. coli JPN15, an EAF plasmid-cured EPEC strain derivative from O127: H6 strain E2348/69 (Levine et al., 1985) was used as an aEPEC control.

The Kaplan–Meier survival curve was used for survival analysis, and the differences between the groups were determined by the log-rank (Mantel-Cox) and Gehan-Breslow-Wilcoxon tests. The threshold for statistical significance was a p-value < 0.05. All analyses were performed using GraphPad Prism 9.2 (GraphPad Software, LLC).

In this work, we have searched for the production of bacteriocins, hemolysins, and serum resistance, which are recognized as contributors to the ExPEC virulence pathotype (Johnson and Russo, 2018) in aEPEC strains.

Production of bacteriocins in TYE and NA was detected in 24 (21.6%) aEPEC strains, with results of both media in accordance. All were confirmed producers by the test using the same media supplemented with trypsin. The bacteriocinogenic strains were submitted to molecular typing, and the bacteriocin of 16 (67%) were successfully genetically typed. The bacteriocins Ia and Ib were the most frequently identified and occurred together in six strains, while microcin V was identified in only one strain (Table 1). The bacteriocin types E1 and E2 were identified in three strains each, and the bacteriocin M was identified in another three strains, co-occurring with bacteriocin B in two of them (Table 1).

Twenty-one (18.9%) aEPEC strains were hemolytic when cultivated in Blood agar. For three aEPEC strains, hemolysis was detected after 3 h of incubation and in the others after 6 or 18 h of incubation. The hemolytic strains were evaluated by PCR for the presence of the genes hlyA (α-hemolysin) and ehxA (enterohemolysin), confirming the occurrence of one of these genes in 19 strains. Seventeen of the hemolytic aEPEC strains harbored hlyA and two ehxA. The remaining two strains, although hemolytic, were not detectable by the molecular method used.

The ability to survive the action of the Complement system active in normal human serum is a main virulence property that enables extraintestinal pathogens to succeed in passing through the bloodstream and gain access to extraintestinal sites.

The majority of aEPEC strains (n = 95, 85.6%) studied were viable after the 24-h challenge in the presence of human sera at a concentration of 50% in PBS. Among the resistant strains, only 13 survived during the challenge, reaching a low concentration in human sera, while the others kept high concentrations at the end of the assay. Sixteen strains were susceptible to human sera effects and did not survive the challenge. Thirty-eight (34.2%) strains were positive for traT, and only four of them did not display the serum-resistant phenotype. No strain was positive for the genes iss and kpsMTII with the primers used.

The phylogenetic origin of the 111 aEPEC strains as determined by PCR revealed that phylogroup B1 was the most frequent (54 strains—48.7%), followed by groups A (32 strains—28.8%), B2 (15 strains—13.5%), and E (9 strains—8.1%). Phylogroup D presented only one strain (0.9%). No strain from phylogroups C and F was identified. No difference was detected in phylogenetic distribution comparing strains isolated from cases and controls (Supplementary Table S6).

Searching the PAIs by PCR, it was found that about 55% of the aEPEC strains presented PAIs originally described in UPEC. Table 2 shows the distribution of these PAIs among the aEPEC strains according to their phylogroups. Five out of eight PAI markers searched were found in aEPEC, with only one PAI occurring in each strain, except for PAIs IV_536, and I_CFT073, which appeared combined in two strains, and PAIs II_536, and I_J96 in another two (Table 2). The PAIs most frequently detected were IV₅₃₆ (38 strains—34.2%), I_J96 (10 strains—9%), and I_CFT073 (9 strains—8.1%), comprising 55 out of 61 PAI-positive strains. No strain presented PAI markers for islands II_J96, I₅₃₆, and III₅₃₆.

Virulence factors related to ExPEC intrinsic virulence (afa/dra, sfa/foc, pap, kpsMTII, iuc/iut) and uropathogenic potential (vat, fyuA, chuA, yfcV) were searched. Many aEPEC strains presented at least one of these virulence genetic markers that are considered the hallmark of the capacity of the strains to cause systemic or urinary tract infections in murine models. Although many of these genes were spread in a variety of aEPEC strains (Table 3), only five could be classified as ExPEC (strains 0811–4, 2071–1, 1482–11, 2332–7, and 1551–3), and another one as UPEC (strain 3712–3) (Table 3), highlighting the existence of hybrid genotypes among aEPEC strains.

Regarding the presence of VFs associated with PAIs, except for fyuA and PAI IV, the other VFs did not hold this association in most strains. This was pointed out by the absence or rare co-occurrence of pap (related to PAI II_CFT073, PAI II₅₃₆, and PAI I_J96), sfa (related to PAI III₅₃₆), kps (related to PAI I_CFT073), and hlyA (related to PAI I_J96) among the aEPEC strains evaluated.

Table 4 presents a summary of the virulence properties that were detected in the aEPEC strains studied, comparing the frequency of each property by phylogroup.

Not surprisingly, the occurrence of the virulence properties followed the phylogenetic distribution, and most of the traits occurred in phylogroups A and B1, which are the most frequent among aEPEC strains. Serum-resistant strains were identified in all phylogroups, with phylogroup E concentrating the higher frequency of resistant strains than phylogroups A, B1, and B2 (Table 4). The five aEPEC/ExPEC hybrid strains (0811–4, 2071–1, 1482–11, 2332–7, and 1551–3) belonged to phylogroup B1, and the hybrid aEPEC/UPEC (3712–3) belonged to phylogroup B2.

The RAPD PCR methodology was used to evaluate the clonal relationship among the aEPEC strains, especially the six aEPEC strains identified as possessing hybrid genotypes, according to the amplification obtained using the primers 1247 and 1283 and considering a cutoff of 90% of similarity for both PCRs. RAPD highlighted the high heterogenicity of the aEPEC strains evaluated, with most not being grouped in a cluster (Supplementary Figure S1).

The hybrid aEPEC strains 0811–4, 1482–11, and 2332–7 are closely related, indicating a clonal relationship. The strain 2071–1 presents a similarity of >95% using primer 1247 but a similarity between 87 and 90% with primer 1283, which raised doubts about its relationship with the other three aEPEC strains. On the other hand, 1551–3 and 3712–3 were not closely related either to each other, or to the first group (Figure 1).

Considering the clonal relationship among 0811–4, 1482–11, and 2332–7, the aEPEC 0811–4 and the more distant strain 2071–1 were chosen to represent this group, being submitted together with 1551–3 and 3712–3 to genome sequencing.

The genome sequences revealed that the strains had predicted genome sizes ranging from 4,784,160 to 5,328,743 bp, with strains 0811–4 and 2071–1 displaying larger predicted genome sizes (Table 5). Three STs were identified among the sequenced strains: 0811–4 and 2071–1, which were clonally related, as suggested by RAPD, belong to ST381; strain 1551–5 belongs to ST29, and strain 3712–3 to ST1092 (Table 5).

A phylogenetic analysis conducted with 0811–4 and 2071–1 confirmed that they are clonally related (Supplementary Figure S2). Additionally, the pathogen detection tool showed that they belong to the same SNP cluster PDS000052783.4, differing by only 18 SNPs. Therefore, only 0811–4 was taken as a reference for the four aEPEC/ExPEC strains from ST381 for other evaluations.

A second phylogenetic analysis, including all hybrid strains, showed that these three strains belong to different EPEC/EHEC clonal groups, with 1551–3 belonging to the EHEC 2 clonal group, 0811–4 to EPEC 14, and 3712–3 to EPEC 4 (Figure 2).

The assembled genomes were submitted to VFDB, and the VF content was compared (Figure 3). It is noticeable that the VF content of the strains is slightly different, although 0811–4 and 1551–3 shared many VFs. In common, besides the LEE, all strains bear the genes related to the expression of CFA/I, ECP, curli, EhaB, UpaG, and Yersiniabactin (Figure 3). In addition, a diverse combination of auto transporters, including SPATES (Vat, Tsh, and Pic) and iron acquisition systems (Aerobactin, Salmonella iron transport system, and Heme receptor) was found (Figure 3). Additionally, the profile of VF genes used for molecular classification was identified as complete, including all the genes of operons, and, except for 0811–4, the strains did not share the intimin type or serotype. Strain 0811–3 harbors intimin type epsilon-1, while 1551–3 harbors epsilon-2, and 3712–3 intimin type zeta (Table 6).

The finding of hybrid aEPEC strains presenting UPEC or ExPEC molecular signatures drove us to search for similar genomic characteristics in a larger group of aEPEC strains that have their genomes accessible in the NCBI. The similar genomes, identified using MASH/MinHash, helped the construction of an additional phylogenetic tree that was added with the VFs used to molecular define the EPEC, EHEC, ExPEC, and UPEC pathotypes.

The tree’s analyses showed that all EPEC genomes closely related to 3712–3 were also hybrid aEPEC or EHEC bearing at least three of the required VFs to be classified as UPEC (Figure 4), suggesting that at least some of the strains that compose the EPEC4 clonal group (see Figure 2) are potentially extraintestinal pathogens. In contrast, the other two hybrid aEPEC in this study were unique in presenting ExPEC VFs within their clusters. Yet, it is worth mentioning that 0811–4 is clonal related with other three hybrid aEPEC strains (1482–11, 2071–1, and 2332–7) evaluated in the present study, pointing out that the spread of aEPEC/ExPEC hybrid genotypes is a reality among diarrheagenic strains isolated in Brazil.

In this study, the G. mellonella model was used to evaluate the pathogenic potential of three of those above-mentioned aEPEC strains that presented genotypic and phenotypic characteristics of ExPEC, meaning that they represented a hybrid class of pathogenic strains.

The results presented in Figure 5 show that all hybrid strains evaluated (aEPEC/ExPEC 0811–4 and 1551–3, and aEPEC/UPEC 3712–3) were capable of killing more larvae than the negative-control strains MG1655 and JPN15, a variant of prototype EPEC strain E2348/69 devoid of the virulence plasmid. However, the survival rate of larvae for the three hybrid strains was higher when compared to the prototype J96, being higher for 1551–3 (p = 0.0012) than for 0811–4 and 3712–3 (p = 0.01 and 0.02, respectively). There was no statistical difference in virulence among the three hybrid strains in this model (p < 0.05). These data suggest that the inoculum of 1 × 10⁵ CFU per larvae can differentiate diarrheagenic EPEC strains from those harboring a hybrid genotype. However, the inoculum used cannot distinguish the genomic diversity of the aEPEC hybrid strains.