Virulence Markers and Phylogenetic Analysis of Escherichia coli Strains with Hybrid EAEC/UPEC Genotypes Recovered from Sporadic Cases of Extraintestinal Infections

Virulence genes from different E. coli pathotypes are blended in hybrid strains. E. coli strains with hybrid enteroaggregative/uropathogenic (EAEC/UPEC) genotypes have sporadically emerged causing outbreaks of extraintestinal infections, however their association with routine infections is yet underappreciated. We assessed 258 isolates of E. coli recovered from 86 consecutive cases of extraintestinal infections seeking EAEC and hybrid genotype (EAEC/UPEC) strains. Extensive virulence genotyping was carried out to detect 21 virulence genes, including molecular predictors of EAEC and UPEC strains. Phylogenetic groups and sequence types (STs) were identified, as well as it was performed phylogenetic analyses in order to evaluate whether hybrid EAEC/UPEC strains belonged to intestinal or extraintestinal lineages of E. coli. Adhesion assays were performed to evaluate the biofilm formation by hybrid strains in human urine and cell culture medium (DMEM). Molecular predictors of UPEC were detected in more than 70% of the strains (chuA in 85% and fyuA in 78%). Otherwise, molecular predictors of EAEC (aatA and aggR) were detected in only 3.4% (9/258) of the strains and always along with the UPEC predictor fyuA. Additionally, the pyelonephritis-associated pilus (pap) gene was also detected in all of the hybrid EAEC/UPEC strains. EAEC/UPEC strains were recovered from two cases of community-onset urinary tract infections (UTI) and from a case of bacteremia. Analyses revealed that hybrid EAEC/UPEC strains were phylogenetically positioned in two different clades. Two representative strains, each recovered from UTI and bacteremia, were positioned into a characteristic UPEC clade marked by strains belonging to phylogenetic group D and ST3 (Warwick ST 69). Another hybrid EAEC/UPEC strain was classified as phylogroup A-ST478 and positioned in a commensal clade. Hybrid EAEC/UPEC strains formed biofilms at modest, but perceptible levels either in DMEM or in urine samples. We showed that different lineages of E. coli, at least phylogenetic group A and D, can acquire and gather EAEC and UPEC virulence genes promoting the emergence of hybrid EAEC/UPEC strains.


INTRODUCTION
Escherichia coli colonizes the human intestine few hours after birth establishing a mutually beneficial relationship with its hosts. While they are restricted to the outer layer of intestinal mucus, these commensal E. coli strains rarely cause infections. However, some highly adapted E. coli lineages have evolved acquiring a broad range of virulence genes (VGs) that allows E. coli strains to adapt to, colonize and invade several anatomic sites (Kaper et al., 2004).
Pathogenic and commensal strains of E. coli are differently sorted into four major phylogenetic groups (phylogroups) named as A, B1, B2, and D (Doumith et al., 2012). Epidemiological studies have shown that extraintestinal pathogenic E. coli (ExPEC) strains are frequently classified as phylogroup B2 or D, while commensal strains are frequently sorted into phylogroup A or B1. Nevertheless, mechanisms of horizontal genetic transfer allows the exchange of VGs among phylogroups, which may promote the sporadic emergence of highly virulent strains belonging to commensal phylogroups A or B1. Additionally, each E. coli phylogroup may enclose heterogeneous groups of strains and different clonal populations, fact which imposes a more complex scenario in an attempt to establish epidemiological links between E. coli phylogroups and human infections (Dias et al., 2009). Thereafter, characterizing clonal structure within each phylogroup seems to be important toward recognizing subsets of clonal groups associated with distinct clinical features. The development of multilocus sequence typing (MLST) methods and the subsequent definition of sequence types (STs) pave the way to the recognition of highly virulent ExPEC clones with worldwide dispersion, such as the clones B2-ST131 and D-ST69 (Blanco et al., 2011;Nicolas-Chanoine et al., 2014;Petty et al., 2014).
On a molecular basis, an EAEC strain is classified as the isolate carrying the virulence plasmid pAA, which is marked by the presence of the predictor genes aatA (previously named as CVD432 probe) and aggR (the master virulence regulator in EAEC). Additionally, the pAA may harbor other EAECspecific VGs, including alleles of aggregative adherence fimbriae (aggA, aafA agg-3A, hdaA, agg-5A), dispersin gene (aap), and the plasmid-encoded toxin gene (pet). EAEC prototype strain 042 has an archetypal pAA plasmid displaying a complete set of virulence genes: aatA, aggR, aafA, aap, and pet. However, epidemiological studies have shown that wild type strains maintain pAA plasmids with a virulence load quite different from the one from archetypal EAEC plasmids. As an example, wildtype EAEC strains collected in epidemiological approaches may show a low frequency for the detection of aggregative fimbriae genes and pet (Pereira et al., 2007). Given their genetic plasticity, a minimum set of predictor genes (the occurrence of aatA along with aggR) was proposed to define typical EAEC strains (Kaper et al., 2004).
Molecular analyses of E. coli strains isolated from UTIs, bacteremia and neonatal meningitis lead to the recognition that these strains are phylogenetically distinct from commensal strains. Contrary to commensal strains, those pathogenic strains frequently belong to the phylogoup B2 or D (Russo and Johnson, 2000). Thereafter, the term extraintestinal pathogenic (ExPEC) was coined to classify strains recovered from extraintestinal infections but that appear to be incapable of causing diarrhea (Russo and Johnson, 2000). Differently from DEC infections, ExPEC infections rely on the bacterial translocation from intestinal lumen to an extraintestinal site, which frequently is the urinary tract.
E. coli strains recovered from UTI (uropathogenic E. coli-UPEC) are the most common cause of bacterial infections in humans, mainly among women, and account for around 100 million cases a year (Foxman, 2010). UPEC strains are a heterogeneous category of E. coli displaying a considerable number of well-recognized VGs ( Table 2) which can be combined into different genotypes. Despite that diversity, a set of four VGs has been reported as a predictor of UPEC strains. Any two of the genes chuA (heme receptor), yfcV (YfC fimbria) or vat (vacuolating autotransporter protein), when detected along with the gene fyuA (yersiniabactin siderophore receptor), can be used to differentiate UPEC strains from commensal and DEC strains (Spurbeck et al., 2012). Besides those VGs, the genes encoding fimbria P (pap, standing for pyelonephritis-associated pilus) and fimbria F1C (focA) play a pivotal role in ascending UTIs since they promote the colonization of proximal and distal tubular cells from human kidney (Korhonen et al., 1986;Marre et al., 1990).
Bacterial biofilms play an important role in medicine and is a serious issue mainly in urology. Bacteria that adhere to uroepithelial cells and form biofilms are more prone to cause pyelonephritis and even chronic or recurrent infections (Soto, 2014). Several studies reported that most of isolates collected from patients with recurrent infections were biofilm producers in vitro. Aside UPEC strains, the expression of biofilms has been considered a consensual virulence factor among EAEC isolates. In EAEC strains, biofilm formation is a complex event that may involve multiples adhesins and factors not devoted to adhesion (Pereira et al., 2010).
The remarkable genome plasticity displayed by E. coli strains has allowed the emergence of virulent strains displaying unusual arrangement of VGs, including arrays of VGs from different pathotypes detected in a single isolate (Gomes et al., 2016). Considering molecular markers from all diarrheagenic E. coli (DEC) pathotypes, EAEC virulence genes are among the most frequent markers of DEC strains reported in ExPEC strains isolated from sporadic cases of extraintestinal infections and outbreaks (Abe et al., 2008;Aurass et al., 2011;Olesen et al., 2012;Prager et al., 2014;Toval et al., 2014;Ang et al., 2016). The term "heteropathogenic E. coli" is now adopted to designate pathogenic E. coli strains that maintain phenotypic and genetic determinants from different E. coli pathotypes (Bielaszewska et al., 2014;Toval et al., 2014;Ang et al., 2016). This intriguing epidemiological scenario has shown how artificial and limited is the current genetic classification of strains into classical pathotypes of E. coli (Robins-Browne et al., 2016).
This study aimed to search for hybrid EAEC/UPEC strains recovered from sporadic extraintestinal infections in order to identify their genotypes and phylogenetic position among UPEC strains.

Ethics Statement
This study was approved by the FEPECS research ethics committee, which is linked to the Secretary of State for Health (Brasília-DF), under the registry number 782.067.

Samples and Strains
During a period of 5 months, microbiological samples were recovered from patients with E. coli-associated extraintestinal infections attended to in a tertiary hospital in Brasília-DF, Brazil. Two hundred fifty eight isolates of E. coli (3 isolates for each sample) were isolated from 86 consecutive cases of extraintestinal infections including urinary tract infection (79 cases); bacteremia (1 case); pneumonia (1 case); surgical site infection (2 cases); peritoneal cavity infection (1 case); and mucosa infections (2 cases) ( Table 1).

Phylogrouping, MLST, and Phylogenetic Tree
Definition of major E. coli phylogroups (A, B1, B2, and D) was performed by PCR as described by Doumith et al. (2012). Multilocus sequence typing (MLST) was performed in accordance with the Institute Pasteur scheme using eight housekeeping genes (dinB, icdA, pabB, polB, putP, trpA, defined as groups of two or more independent isolates that shared identical alleles at six loci. The correspondence between STs assignment by Pasteur's scheme and Achtman's scheme was based on results published in previous studies (Jaureguy et al., 2008;Clermont et al., 2015). In order to display the phylogenetic relationships among strains, MLST sequences were concatenated into a 2901-base-long super-gene (dinB-icdA-pabB-polB-putP-trpA) (Gadagkar et al., 2005). The dendrogram was constructed in MEGA6 applying the Maximum Likelihood method based on the Tamura-Nei model (Tamura et al., 2013). Initial trees for the heuristic search were obtained by applying the Neighbor-Joining method to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach. In order to test the accuracy of the phylogeny was applied the Bootstrap method with 1500 replications.

Samples of Human Urine
Healthy women with no history of consumption of antibiotics or anti-inflammatoreis within the last 15 days and with no clinical urinary symptoms were selected to donate urine samples. Donators were instructed, informed about the absence of health risk associated with the urine collection, and signed an informed consent term allowing the urine collection. Morning samples of urine (volume of 50 mL) were collected by spontaneous urination, centrifuged (3000 g/3 min), sterilized by ultrafiltration (0.22 µm) and preserved at −20 • . Samples of sterile urine were pooled (n = 3), supplemented with 0.5% of casamino acid and used as culture medium in biofilm assays.

Biofilm Assays
In order to test the biofilm formation, 96-well flat-bottom polystyrene plates were used as described by Wakimoto et al. (2004). Seventy seven strains (one strain per case including 3 hybrid EAEC/UPEC strains) were assayed for biofilm formation (Supplementary Table 1). Briefly, 200 µL of Dulbecco's Modified Eagle Medium (DMEM) or sterile samples of pooled human urine were set into each plate well and inoculated with 5 µL of overnight bacterial culture. The plates were incubated overnight at 37 • C without shaking. Afterwards, planktonic culture were discarded and formed biofilms were stained with crystal violet (CV) dye (15 min), washed once with 200 µL of phosphate-buffered saline and air-dried for 3 h. The absorbance (OD at 630 nm) reached by CV adsorbed on the well bottom was determined, and afterwards the bacterium-bound dye was released by the addition of ethanol (200 µL/well). The mean of the absorbances was used as a measure of the formed biofilms. Data were displayed as means obtained from three independent assays.

Statistical Analysis
Statistical analysis were performed on the software IBM R SPSS R Statistics (version 20). Results with p ≤ 0.05 were considered to be statistically significant.
The molecular predictors of UPEC chuA (85%) and fyuA (78%) were predominantly detected among ExPEC strains. Additionally, the genes for UPEC-specific fimbriae pap and yfcV were detected in 47 and 41% of the strains, respectively ( Table 3). UPEC VGs displayed an uneven distribution among phylogroups. The genes fyuA, pap e yfcV were more frequently detected in the extraintestinal phylogroups (B2 and D) than in the intestinal groups (A and B1) (p < 0.001-Fisher's exact test) ( Table 3). Moreover, the genes sfa, cnf and pic were exclusively detected in the phylogroup B2.
The molecular predictors of EAEC aatA and aggR were codetected in 3.4% (9/258) of the ExPEC strains recovered from 3 out of 86 studied cases (3.4%). Besides aatA and aggR, the UPEC genes fyuA and pap were always found forming the genotypes of these strains, which highlighted their hybrid nature: hybrid EAEC/UPEC strains ( Table 4). Hybrid EAEC/UPEC strains were recovered from two cases of UTIs (cases 63 and 85) and from one case of bacteremia (case 17) ( Table 4). Additionally, three strains were classified as phylogroup A (associated with the UTI case 63) and the six remaining strains were classified as phylogroup D (with each of the three strains associated with the UTI case 85 and the bacteremia case 17). Genes for the EAEC fimbria AAF-I (aggA) were also detected in the EAEC/UPEC strains isolated from the UTI case 63 (Table 4). Aside from these data, an interesting fact was that three strains recovered from an UTI case (case 55) tested positive for the EAEC dispersin gene (aapA), although they did not harbor any of the EAEC-predictor markers aatA or aggR (Table 4).

Allelic Profile and Sequence Type (ST) in Hybrid EAEC/UPEC Strains
Three hybrid EAEC/UPEC strains (each of them isolated from the cases 17, 63, and 85) had their allelic profiles and STs determined and compared with 19 contemporaneous UPEC strains, each from a different case chosen randomly (Table 5). Additionally, the equivalent Warwick ST was assessed in accordance with a previous report (Clermont et al., 2015) ( Table 5). Twelve allelic profiles were detected among the strains. The predominant allelic profile matched exactly with ST3 (Warwick ST69) (n = 8) and was followed by ST43 (n = 3) (Warwick ST 131). Five STs were of single occurrence (ST8, ST34, ST44, ST478, and ST479), and two STs were single-locus variants of ST43 and ST4 (Warwick ST73) ( Table 5). Concerning the hybrid EAEC/UPEC strains, two strains (17.1 and 85.1) were classified as ST3, (Warwick ST69) while the strain 63.1 was addressed to the ST478, which had a single occurrence. These findings uncover the heterogeneity of our collection and show that a considerable subset of strains (n = 11) belong to genetic lineages recognized by the epidemic potential (ST3 and ST43), including two of the hybrid EAEC/UPEC strains.

Phylogenetic Positioning of Hybrid EAEC/UPEC Strains
The phylogenetic relationship of hybrid EAEC/UPEC with typical UPEC strains, prototype EAEC 042, and genomic reference sequences were visualized in a dendrogram constructed   by alignments of concatenated housekeeping gene sequences (dinB-icdA-pabB-polB-putP-trpA) (Figure 1). Genotypes, phylogroups and STs were also displayed in order to highlight shared features detected among strains. Two of out three tested EAEC/UPEC strains shared with UPEC strains a well-defined clade (bootstrap value equals 94) featured by gathering strains (n = 8) with positive results for molecular predictors of UPEC (fyuA and chuA) and UPEC-specific fimbria Pap as well (Figure 1). Additionally, strains in this clade were addressed as phylogroup D and ST3 (Warwick ST 69). Outside the clade D-ST3, the strains showed a reduced positivity for pap (40% -6/15). The prototype EAEC strain 042 was positioned in a separate clade, although it also belongs to phylogroup D. A different phylogenetic position was displayed by the EAEC/UPEC strain 63.1, which was positioned in a different major branch sharing a clade with the reference sequence of commensal strain K12. Endorsing that phylogenetic positioning is the fact that both strains belong to phylogroup A (Figure 1).

DISCUSSION
EAEC is a heterogeneous category that has been recognized for gathering versatile pathogens since the late 1990's, when their epidemiological association with diarrhea involving children, adults, HIV-infected patients and travelers became more and more evident (Kaper et al., 2004). Besides that, EAEC strains has been recently reported as a sporadic etiological agent in extraintestinal infections, mainly in urinary tract infections (UTI). In Brazil, a previous epidemiological study on UTI showed that 3.5% (8/225) of the strains were classified as EAEC based on the presence of the virulence marker aatA (Abe et al., 2008). Along with others classical UPEC markers, 5 of out 8 aatA-positive EAEC strains also carried UPEC gene pap (pyelonephritis-associated pilus) along with EAEC gene aggA (aggregative adherence fimbriae I -AAF/I), which highlighted the hybrid aspect of these strains (Abe et al., 2008). This epidemiological scenario is similar to the results found now in our study. In our collection, 3.5% (9/258) of the strains were positive for aatA as well as for pap. Moreover, three of these hybrid EAEC/UPEC strains harbored aggA (EAEC fimbria AAF-I) along with pap (UPEC pilus). The gene pic (standing for protein-involved in colonization) was initially identified in the EAEC prototype strain 042 as a mucinase that displayed a pivotal role in the colonization of the intestinal mucus layer during diarrhea (Harrington et al., 2009). However, further experimental and epidemiological data broadened the role of pic endorsing its importance in extraintestinal infections such as UTIs, pyelonephritis and sepsis (Boll et al., 2013;Abreu et al., 2015). The gene pic was the most frequent EAEC marker detected among UPEC strains, reaching 13% of positivity in the Abe's study (Abe et al., 2008). Here, picpositive strains accounted for 4.3% of the strains. Despite the low overall prevalence, here pic was exclusively detected in strains belonging to the extraintestinal phylogroup B2 and, in this group, pic-positive strains accounted for 11.5% of the isolates.
Studies on the distribution of plasmid and chromosomal genes of EAEC proposed that some EAEC strains belonging to lineages diverse from typical EAEC strains can acquire and maintain the pAA plasmid, the major virulence element in EAEC strains (Jenkins et al., 2005). Therefore, the Group 2 of EAEC, as defined by Jenkins et al., comprises a more heterogeneous group of EAEC strains that emerges from acquisition of pAA by strains that does not retain chromosomal markers of typical EAEC strains (Jenkins et al., 2005). In order to explain the emergence of hybrid EAEC/UPEC strains, some authors have fomented similar discussions. Here, we showed that different alternatives for the emergence of hybrid EAEC/UPEC strains isolated from UTI could exist. Based on the phylogenetic analyses carried out with concatenated chromosomal genes, we showed that two hybrid EAEC/UPEC strains share a chromosomal backbone of UPEC strains. Therefore, this finding suggest that mobile EAEC genetic markers posed on pAA plasmid (aatA, aggR, and aapA) can be acquired by UPEC strains. Concerning the EAEC/UPEC strains 17.1 and 85.1 (both classified as phylogroup D-ST3[ST69]), our findings conflict with the paradigm stablished in the literature that poses that diarrheagenic virulence markers are rarely detected in UPEC strains from phylogenetic group D-ST3[ST69] (Blanco et al., 2011;Olesen et al., 2012;Skjøt-Rasmussen et al., 2013). However, a Wallace-Gadsden's report endorses our findings, showing that the ST69 complex has a propensity to acquire EAEC virulence genes and so it may have evolved as a progenitor lineage from which UPEC and EAEC strains belonging to ST69 emerged (Wallace-Gadsden et al., 2007).
On the other hand, E. coli lineages that frequently gather commensal strains (phylogroups A and B1) may also include highly virulent UPEC strains. The hybrid EAEC/UPEC strain 63.1, characterized for having the most complete set of tested virulence genes for EAEC (aatA, aggR, aapA, and aggA) and UPEC (fyuA, vat, focA, and pap), besides being positive for csgA and ag43, was classified as phylogroup A. E. coli strains with similar genetic arrangements have already been associated with UTI outbreaks. In 1991, it was detected in Copenhagen an UTI outbreak involving a hybrid EAEC/UPEC strain (Olesen et al., 2012). Further molecular characterization showed that 18 out of the 19 outbreak strains were positive for the EAEC specific genes aatA, aggR, aagA, and aap along with the UPEC genes sat and fyuA. Additionally, phylogenetic analyses showed that the hybrid EAEC/UPEC outbreak strains belonged to intestinal phylogroup A, which is known to gather commensal strains (Olesen et al., 2012).
The formation of biofilms is considered to be a pivotal step during infection processes developed in human mucosas. EAEC strains are renowned for their prolific ability to form biofilms, which are in general supported by the aggregative adherence fimbriae (AAFs) coded on the plasmid pAA. Cases and outbreaks of UTI involving heteropathogenic EAEC/UPEC strains have endorsed the suspicion that EAEC virulence genes add uropathogenic properties to E. coli strains. Concerning EAEC/UPEC strains involved in the Copenhagen UTI outbreak (genotype: aatA aggR aap aggA fyuA pic; phylogroup A) (Olesen et al., 2012), it was shown that the expression of AAF/I (gene aggA) allowed a pronounced increase of bacterial adherence to human bladder epithelial cells as well as it allowed biofilm formation at levels significantly higher than those of UPEC prototype strains (Boll et al., 2013). In our study, the AAF/Ipositive EAEC/UPEC strain 63.1 (genotype: aatA aggR aggA aapA fyuA vat focA pap csgA ag43; phylogroup A) showed an ability to produce invariable and perceptible biofilms in both DMEM and human urine. Although the biofilm levels had not been higher than those produced by the prototype EAEC strain 17-2 in DMEM, EAEC/UPEC strain 63.1 formed biofilms at higher levels than those formed by the strain 17-2 and at similar levels to those formed by prototype UPEC strains when tested with human urine.
Whether or not hybrid EAEC/UPEC strains have potential to cause diarrhea is an issue to be further evaluated. However, studies based on the extensive genetic characterization of typical EAEC strains have shown that phylogenetic patterns similar to those detected here are also found in diarrhea-associated EAEC strains. EAEC strains with phylogroup D and genotype aatA aggR aapA fyuA chuA were recovered from diarrhea and included the prototype strain 042 (Czeczulin et al., 1999;Okeke et al., 2010). Additionally, EAEC strains with phylogrup A and genotype aatA aggR aapA fyuA were also recovered from cases of diarrhea (Czeczulin et al., 1999;Okeke et al., 2010), including the strain C1192-92 recovered from a diarrhea case around the time of the UTI outbreak that occurred in F Compenhagen in 1991 (Olesen et al., 2012).
E. coli has a remarkable adaptation capacity that is mainly relied on its genomic plasticity (Touchon et al., 2009). Having a pan-genome with around 17,000 genes and 4700 genes per genome in average, no single E. coli strain can be assumed as a representative of the species. In this scenario, applying the genetic boundaries stablished for archetypal pathotypes in epidemiological studies involving E. coli has shown to be an elusive goal. Events involving hybrid or heteropathogenic E. coli strains have endorsed this perspective and blurred the classification of E. coli strains into classical pathotypes.

AUTHOR CONTRIBUTIONS
AP conceived the study and designed the experiments. AP and FL wrote the manuscript and were responsible for concepts, vision and direction of the study. FL, DN, Pd, and MA performed geno-and phylotyping experiments and analyzed the data. AP, Cd, and DN carried out the genetic sequencing. FC and Ld performed biofilm assays. LF and FL carried out the isolation and identification of E. coli strains. All authors read and approved the final manuscript.