Analysis of the Virulence Profile and Phenotypic Features of Typical and Atypical Enteroaggregative Escherichia coli (EAEC) Isolated From Diarrheal Patients in Brazil

Enteroaggregative Escherichia coli (EAEC) is an important agent of acute and persistent diarrhea in children and adults worldwide. Here we report a characterization of 220 EAEC isolates, 88.2% (194/220) of which were typical and 11.8% (26/220) were atypical, obtained from diarrheal patients during seven years (2010-2016) of epidemiological surveillance in Brazil. The majority of the isolates were assigned to phylogroups A (44.1%, 97/220) or B1 (21.4%, 47/220). The aggregative adherence (AA) pattern was detected in 92.7% (204/220) of the isolates, with six of them exhibiting AA concomitantly with a chain-like adherence pattern; and agg5A and agg4A were the most common adhesin-encoding genes, which were equally detected in 14.5% (32/220) of the isolates. Each of 12 virulence factor-encoding genes (agg4A, agg5A, pic, aap, aaiA, aaiC, aaiG, orf3, aar, air, capU, and shf) were statistically associated with typical EAEC (P < 0.05). The genes encoding the newly described aggregate-forming pili (AFP) searched (afpB, afpD, afpP, and afpA2), and/or its regulator (afpR), were exclusively detected in atypical EAEC (57.7%, 15/26), and showed a significant association with this subgroup of EAEC (P < 0.001). In conclusion, we presented an extensive characterization of the EAEC circulating in the Brazilian settings and identified the afp genes as putative markers for increasing the efficiency of atypical EAEC diagnosis.


INTRODUCTION
Enteroaggregative Escherichia coli (EAEC) is one of the six diarrheagenic E. coli (DEC) pathotypes, which is defined as E. coli isolates that produce the aggregative adherence pattern (AA) on cultured epithelial cells (HeLa and/or HEp-2) in vitro (Nataro and Kaper, 1998;Kaper et al., 2004;Croxen et al., 2013). The AA pattern is characterized by a bacterial arrangement that resembles stacked bricks on both epithelial cells and the coverslip (Nataro et al., 1987). EAEC is a cause of diarrheal disease in children and adults, living in both developing and developed countries, and are associated with cases of persistent diarrhea and diarrheal outbreaks worldwide (Huang et al., 2006;Estrada-Garcia and Navarro-Garcia, 2012;Hebbelstrup Jensen et al., 2014), besides the fact that EAEC can also be isolated from asymptomatic subjects (Gioppo et al., 2000;Araujo et al., 2007;Lima et al., 2013Lima et al., , 2019aDias et al., 2016;Durand et al., 2016).
The pathogenicity of EAEC is characterized by the colonization of the intestinal epithelium, followed by the production and secretion of enterotoxins and cytotoxins, which result in secretory diarrhea (Nataro and Kaper, 1998). The establishment of the AA pattern has been associated with the presence of five distinct aggregative adherence fimbriae (AAF/I-AAF/V) Czeczulin et al., 1997;Bernier et al., 2002;Boisen et al., 2008;Jønsson et al., 2015), which belong to the chaperone-usher subclass, as well as afimbrial adhesins (Monteiro-Neto et al., 2003;Bhargava et al., 2009). Very recently, a type IV pilus, termed aggregate-forming pili (AFP), encoded by genes located in the afp operon, was identified to be responsible for the establishment of the AA pattern by a hybrid EAEC/STEC (Shiga-toxin producing E. coli) isolate of serotype O23:H8, that lacks the AAFs described so far (Lang et al., 2018).
Once adhered to the intestinal epithelium, EAEC produces several enterotoxins and cytotoxins, which include: the enteroaggregative E. coli heat-stable enterotoxin (EAST-1), encoded by the astA gene, the plasmid-encoded toxin (Pet), the protein involved in colonization (Pic), and autotransporter proteases such as SigA and SepA, which were initially described in Shigella flexneri. Another important virulence factor associated with the pathogenicity of EAEC is the protein dispersin (Aap), which facilitates bacterial dispersion in the intestinal epithelium (Sheikh et al., 2002). Dispersin secretion occurs via an ATPbinding cassette (ABC) transporter system, encoded by a cluster of five genes (aatPABCD) located in the aggregative adherence plasmid (pAA). The aatA gene, which corresponds to the former EAEC probe CVD432, is one the most used target for EAEC identification nowadays (Baudry et al., 1990;Nishi et al., 2003;Lozer et al., 2013).
The AggR protein, encoded by the aggR gene located in the pAA, is a global virulence regulator, which activates the expression of several plasmidial and chromosomal virulence factor-encoding genes in the EAEC 042 prototype strain, such as the AAFs, as well as a chromosomal pathogenicity island (PAI), termed aaiA-aaiP, which encodes a Type VI Secretion System (T6SS) Morin et al., 2013). Some studies employed genes of the aaiA-aaiP operon, such as aaiA, aaiC, and aaiG, as an additional chromosomal marker to enhance the efficiency of EAEC diagnosis (Lima et al., 2013;Andrade et al., 2014;Havt et al., 2017;Hebbelstrup Jensen et al., 2018). Based on the presence of the aggR gene, EAEC isolates can be subgrouped into typical (aggR + ) and atypical (aggR − ) (Kaper et al., 2004).
Moreover, other potential virulence factor-encoding genes that may contribute to the pathogenicity of EAEC isolates, such as aar (AggR-activated regulator), shf (which encodes a protein involved in biofilm formation initially identified in Shigella flexneri), air (Enteroaggregative immunoglobulin repeat protein), and capU (which encodes a homologous hexosyltransferase) have been described (Czeczulin et al., 1999;Sheikh et al., 2006;Santiago et al., 2017).
The goal of this study was to better understand the diversity of typical and atypical EAEC isolates, obtained from diarrheal patients in Brazil, regarding the somatic (O) and flagellar (H) antigens determination (serotypes), phylogenetic classification, presence of 25 virulence factors-encoding genes (e.g., adhesins, toxins, secreted proteases, type VI secretion system), adherence pattern produced on HeLa cells, and the susceptibility to 11 distinct antimicrobial drugs.

Enteroaggregative Escherichia coli (EAEC) Isolates Used in This Study
A total of 220 EAEC isolates, obtained from stool samples of sporadic cases of diarrhea occurred in Brazil, were included in this study. Of note, 203 EAEC isolates were obtained over 6-years of active epidemiological surveillance (2011)(2012)(2013)(2014)(2015)(2016), performed at the Instituto Adolfo Lutz (IAL), a public health laboratory and the Brazilian reference center for diarrheagenic E. coli identification, as previously published (Ori et al., 2019), as well as 17 EAEC isolates, obtained during the year of 2010, at the same Institute.
In this study, EAEC was defined as E. coli isolates that possess the aatA gene (former CVD432 probe), and typical and atypical EAEC were differentiated based on the presence of the aggR gene in the former group (Kaper et al., 2004).

Serotyping
The somatic (O) and flagellar (H) antigens from 203 EAEC isolates, obtained from 2011 to 2016, were determined in our previous epidemiological surveillance study (Ori et al., 2019). In the 17 remaining EAEC isolates, obtained during the year of 2010, the O and H antigens were determined by standard agglutination tests, with O (O1-O181) and H (H1-H56) antisera produced at Instituto Adolfo Lutz, São Paulo, Brazil (Ewing, 1986).

Phylogroup classification
The classification of the 220 EAEC isolates studied into distinct phylogroups (A, B1, B2, C, D, E, and F) and Escherichia clades were performed using the quadruplex Polymerase Chain Reaction (PCR) method as previously described by Clermont et al. (2013). Subsequently, EAEC isolates classified in the phylogroup B2, with the following genotype: arpA − , chuA + , yjaA − , and TspE4 + , and all isolates assigned in the phylogroup F were tested in an additional Triplex PCR, using primers and PCR conditions as recently described (Clermont et al., 2019), in order to confirm these isolates as B2 and F or to reclassify them in the phylogroup G.

Detection of Virulence Factor-Encoding Genes
The presence or absence of the EAEC diagnosis markers aatA and aggR was confirmed in all 220 isolates, and subsequently, these isolates were tested for the presence of genes encoding for: adhesins (aggA, aafA, agg3A, agg4A, agg5A, afpB, afpD, afpP, afpA2, afpR, and eibG), toxins (pic, pet, astA, sigA, and sepA) and other virulence factors (aap, aaiA, aaiC, aaiG, orf3, aar, air, capU, and shf ), by PCR DNA amplification. PCR was performed using GoTaq Green Master Mix (Promega, Madison, WI, USA) with 0.34 µM of each of the primers. All primer sequences and PCR assay conditions used for the detection of virulence genes are described in the references cited in Table S1. PCR products were subjected to agarose gel electrophoresis, prepared with Trisborate-EDTA (TBE) buffer, and visualized with SYBR Safe DNA Gel Stain (Invitrogen, CA, USA).

Adherence Assays
The adherence patterns of the EAEC isolates was evaluated in HeLa cells as previously described (Cravioto et al., 1979), with some modifications. Cells were cultured in DMEM (Dulbecco's Modified Eagle Medium-high glucose; Sigma-Aldrich; St. Louis, MO, USA), supplemented with 10% fetal bovine serum (Gibco; Grand Island, New York, USA), with 1% antibiotic (Penicillin-Streptomycin, Sigma; St. Louis, MO, USA) in 5% CO 2 at 37 • C. For the adherence assay, ∼1 × 10 5 cells were cultivated in a 24-well microplate, using the same medium described above and containing glass coverslips in each well of the microplate, for 48 h until reaching a confluence of ∼80%. Then, the cells were washed three times with phosphate-buffered saline (PBS, pH 7.4), 1.0 mL of DMEM, supplemented with 2% fetal bovine serum and 1% D-mannose (Sigma, Saint Louis, MO, USA) was added, the cells were infected with 20 µL of an overnight bacterial suspension in LB broth (containing ∼10 8 CFU/mL), and the adherence assay was performed in 5% CO 2 at 37 • C. After 3 h of incubation, the infected cells were washed six times with PBS and then fixed with methanol (Dinâmica, SP, Brazil) and stained FIGURE 1 | Representative images of the distinct patterns of adherence exhibited by the EAEC isolates studied in adherence assays performed with HeLa cells. The adherence patterns are illustrated as follow: (A) aggregative adherence pattern, (B) aggregative adherence pattern (solid arrow) occurring concomitantly with the chain-like adherence (dashed arrow), (C) chain-like adherence pattern, and (D) diffuse adherence pattern. All the adherence patterns illustrated in this picture were observed in adherence assays performed with 3 h of infection. Scale bar = 10 µm. with May-Grünwald (Sigma; St. Louis, MO, USA) and Giemsa (Dinâmica, SP, Brazil) for identification of the adherence pattern by light microscopy analyzes. When the EAEC isolates did not adhere to the epithelial cells or the adherence pattern could not be defined, the adherence assay was repeated in a 6 h bacteria-cell interaction assay, with a washing step after 3 h of incubation, as previously described (Rodrigues et al., 1996), stained with May-Grünwald/Giemsa and then observed by light microscopy. To eliminate potential bias in the adherence pattern classification, each slide was analyzed by three distinct blinded examiners, both to define the adherence pattern exhibited by each EAEC isolate, as well as to select the representative images to be included in Figures 1, 2. In cases of divergence among the three examiners, the EAEC isolate was submitted to a new adherence assay and reevaluated in a blinded-manner until a consensus was obtained.

Statistical Analyses
Differences observed between typical and atypical EAEC were investigated using a Chi-square with Yates' correction or twotailed Fisher's exact test, as appropriate, and a significance level of 5% (P < 0.05) was adopted to consider the differences statistically significant.
Considering the novelty of the association of the genes of the afp operon and/or its regulator (afpR) with atypical EAEC, we compiled the main features of these 15 EAEC isolates in Table 5. Most of them were identified in the phylogroup A (80.0%, 12/15) and produced AA in 3 h assay (73.3%, 11/15) (Figure 2), even though they were heterogeneous regarding the serotype, and virulence profile exhibited ( Table 5).

DISCUSSION
The importance of EAEC as an etiological agent of diarrheal diseases has been reinforced in the last decade (Boisen et al., 2012;   UND, EAEC isolates exibithing undefined adherence, characterized by the occurrence of few bacteria randomly attached on the HeLa cells surface. b EAEC isolate (IAL 5749) that harbored the afpR gene, but lacked the genes from the afp operon investigated ( Table 5).  Lima et al., 2013;Durand et al., 2016;Chattaway et al., 2017;Ori et al., 2019). However, some questions remain to be addressed, such as the use of the adherence pattern as the gold standard for diagnosis, how to identify truly pathogenic isolates in this heterogeneous group of isolates, and which would be the best molecular markers to improve the diagnosis of EAEC, mainly atypical EAEC. Several of the virulence factor-encoding genes investigated here were statistically more frequent among typical EAEC, which is not surprising since the majority of these genes were identified, and functionally characterized, in the typical EAEC prototype strain 042 (Harrington et al., 2006;Chaudhuri et al., 2010;Hebbelstrup Jensen et al., 2014). Similar findings were also observed by other Brazilian authors that compared typical and atypical EAEC isolates, especially with the pic, aap, shf, and agg4A genes being more frequent among the typical EAEC isolates (Andrade et al., 2017;Taborda et al., 2018;Guerrieri et al., 2019).  ( Case-control studies, performed in distinct geographic regions, have been able to demonstrate the association of some genes encoding virulence factors with diarrheal disease. pic, which encodes a protein with hemagglutinin and mucinolytic activity in vitro , was statistically more frequent among diarrheal Peruvian children than in healthy subjects (Durand et al., 2016), as well as sepA, encoding an autotransporter protease originally described in Shigella spp., was associated with diarrhea in children from Bamako, Mali (Boisen et al., 2012). A case-control study performed in Northeastern Brazil demonstrated the association of pet (Plasmid-encoded toxin) and aafA (AAF/II fimbrial subunit) with diarrhea (Lima et al., 2013). Moreover, the occurrence of typical EAEC, harboring concomitantly aatA and aaiC, was more frequent among malnourished than in nourished children (Havt et al., 2017), thus reinforcing the need for greater attention to asymptomatic infections by this pathogen.
In typical EAEC the establishment of the AA pattern is mediated by five distinct AAFs (Jønsson et al., 2017a), and the occurrence of genes associated with the biogenesis of AAF/I-AAF/IV has already been reported among Brazilian EAEC isolates, with their frequency varying according to the geographic regions studied Zamboni et al., 2004;Lima et al., 2013;Guerrieri et al., 2019). However, the occurrence of Brazilian EAEC isolates harboring the agg5A, alone or concomitantly with agg3A, is described for the first time in this study. A study performed with Danish EAEC isolates identified six harboring the major pilin subunits from both AAF/III (agg3A) and AAF/V (agg5A), with these genes located in a novel pAA variant (Jønsson et al., 2017b).
Recently, Lang and coworkers described a new fimbria, termed aggregate-forming pili (AFP), in a hybrid EAEC/STEC isolate of serotype O23:H8 (Lang et al., 2018). The genes responsible for the biogenesis of AFP have ∼52% of nucleotide identity with the genes responsible for encoding the bundleforming pili (BFP), a type IV pilus that mediates the localized adherence (LA) phenotype of typical enteropathogenic E. coli on epithelial cells surface (Scaletsky et al., 1984;Girón et al., 1991;Donnenberg et al., 1992). The afp and bfp operons have a similar gene organization (Tobe et al., 1999;Lang et al., 2018), despite the presence of an additional afpA2 gene, that is not required for AFP biogenesis (Lang et al., 2018). The deletion of the complete afp operon, afpA and afpR genes (encoding an AraC-like regulator), but not afpA2, led to the loss of fimbriae production and drastically reduced the ability of the EAEC/STEC isolate in adhering to epithelial cells in vitro (Lang et al., 2018).
Using an NCBI nucleotide BLAST analysis with the afp operon genes as a query sequence, Lang et al. (2018) identified 17 afp-positive E. coli isolates, which were also positive for EAEC markers, such as aatA and aap, but lacking aggR, which is a genotype of atypical EAEC. This observation motived us to deeply investigate the occurrence of the afp genes in other EAEC isolates, including both typical and atypical. To the best of our knowledge, this is the first study to demonstrate the association of genes from the afp operon, as well as, a gene encoding its regulator (afpR) with the atypical EAEC subgroup, thus providing a new putative molecular marker to increase the diagnostic efficiency of this pathogen.
Moreover, some of the isolates classified as typical EAEC (aatA + /aggR + ) produced the CLA or DA patterns on HeLa cells (3 and 2 isolates, respectively), similar to the observed in previous studies that compared adherence pattern and molecular markers of the distinct DEC pathotypes (Gomes et al., 1998;Gioppo et al., 2000;Scaletsky et al., 2002;Spano et al., 2017). A Brazilian study demonstrated that E. coli isolates exhibiting CLA on epithelial cells harbored genes commonly used as a marker of the EAEC pathotype, i.e., aatA and aggR, as well as virulence factor-encoding genes, such as pet, astA, and aggC (Gioppo et al., 2000). These data demonstrate that the EAEC definition is still ambiguous since E. coli isolates producing the AA pattern on HeLa cells could be devoid of EAEC molecular markers of this pathotype, as well as E. coli positive for such genes could produce adherence patterns distinct from AA. We believe that further studies using whole-genome sequencing analysis, in combination with phenotypic features of the EAEC isolates and clinical information of the patients could provide greater accuracy to the molecular markers necessary for the EAEC diagnosis, as well as, in the identification of truly pathogenic isolates.
In conclusion, although we have characterized the EAEC isolates studied in terms of phylogroups, serotypes, adherence patterns, and genes that encode virulence factors, the main message of this study is the association of the aggregate-forming pili (AFP)-encoding genes with the atypical EAEC subgroup. Our study showed that the afp operon was found only in E. coli isolates that carry the aatA gene, but are devoid of the genes encoding the EAEC virulence regulator and the aggregative adherence fimbriae (aatA + /aggR − /AAFs − ), thus providing a novel putative marker for increasing the efficiency of atypical EAEC diagnosis.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
This study was analyzed and approved by the Botucatu Medical School Ethical Committee for human experimentation (CAAE 09994419.6.1001.5411). Written and informed consent from the participants was not required as approved by the aforementioned Ethics Committee.

AUTHOR CONTRIBUTIONS
RH, LS, and WE conceptualized the study. RD, LS, RH, and WE contributed to the formal analysis. RH and RD were responsible for the funding acquisition. RD, RT, MV, MC-N, LS, TG, WE, and RH carried out the investigation. RD, RT, MV, and MC-N worked on the methodology. RH, RD, and LS helped with the project administration. RH, LS, TG, and WE supervised the study. RH, LS, TG, and WE validated the study. RD and RH wrote the original draft. RD, RT, MV, MC-N, LS, TG, WE, and RH reviewed and edited the manuscript.

FUNDING
This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES) -Finance Code 001 and CAPES/PROEX 0708/2018.