Genomic Properties and Temporal Analysis of the Interaction of an Invasive Escherichia albertii With Epithelial Cells

Diarrhea is one of the main causes of infant mortality worldwide, mainly in the developing world. Among the various etiologic agents, Escherichia albertii is emerging as an important human enteropathogen. E. albertii promote attaching and effacing (AE) lesions due to the presence of the locus of enterocyte effacement (LEE) that encodes a type three secretion system (T3SS), the afimbrial adhesin intimin and its translocated receptor, Tir, and several effector proteins. We previously showed that E. albertii strain 1551-2 invades several epithelial cell lineages by a process that is dependent on the intimin-Tir interaction. To understand the contribution of T3SS-dependent effectors present in E. albertii 1551-2 during the invasion process, we performed a genetic analysis of the LEE and non-LEE genes and evaluated the expression of the LEE operons in various stages of bacterial interaction with differentiated intestinal Caco-2 cells. The kinetics of the ability of the 1551-2 strain to colonize and form AE lesions was also investigated in epithelial HeLa cells. We showed that the LEE expression was constant during the early stages of infection but increased at least 4-fold during bacterial persistence in the intracellular compartment. An in silico analysis indicated the presence of a new tccP/espFU subtype, named tccP3. We found that the encoded protein colocalizes with Tir and polymerized F-actin during the infection process in vitro. Moreover, assays performed with Nck null cells demonstrated that the 1551-2 strain can trigger F-actin polymerization in an Nck-independent pathway, despite the fact that TccP3 is not required for this phenotype. Our study highlights the importance of the T3SS during the invasion process and for the maintenance of E. albertii 1551-2 inside the cells. In addition, this work may help to elucidate the versatility of the T3SS for AE pathogens, which are usually considered extracellular and rarely reach the intracellular environment.

Diarrhea is one of the main causes of infant mortality worldwide, mainly in the developing world. Among the various etiologic agents, Escherichia albertii is emerging as an important human enteropathogen. E. albertii promote attaching and effacing (AE) lesions due to the presence of the locus of enterocyte effacement (LEE) that encodes a type three secretion system (T3SS), the afimbrial adhesin intimin and its translocated receptor, Tir, and several effector proteins. We previously showed that E. albertii strain 1551-2 invades several epithelial cell lineages by a process that is dependent on the intimin-Tir interaction. To understand the contribution of T3SS-dependent effectors present in E. albertii 1551-2 during the invasion process, we performed a genetic analysis of the LEE and non-LEE genes and evaluated the expression of the LEE operons in various stages of bacterial interaction with differentiated intestinal Caco-2 cells. The kinetics of the ability of the 1551-2 strain to colonize and form AE lesions was also investigated in epithelial HeLa cells. We showed that the LEE expression was constant during the early stages of infection but increased at least 4-fold during bacterial persistence in the intracellular compartment. An in silico analysis indicated the presence of a new tccP/espF U subtype, named tccP3. We found that the encoded protein colocalizes with Tir and polymerized F-actin during the infection process in vitro. Moreover, assays performed with Nck null cells demonstrated that the 1551-2 strain can trigger F-actin polymerization in an Nck-independent pathway, despite the fact that TccP3 is not required for this phenotype. Our study highlights the importance of the T3SS during the invasion process and for the maintenance of E. albertii 1551-2 inside the cells. In addition, this work may help to elucidate the versatility of the INTRODUCTION Diarrhea is one of the leading causes of infant mortality worldwide, mainly in the developing world. Many enteropathogens are associated with this disease (Kotloff et al., 2013), including Escherichia albertii, which is emerging as an important enteropathogen of humans and birds (Huys et al., 2003;Oaks et al., 2010;Gomes et al., 2020).
The ler gene, located in the LEE, encodes a transcriptional regulator, which positively regulates many EPEC virulence factor-encoding genes in the LEE region (Mellies et al., 1999), except for genes within the LEE1 operon (Berdichevsky et al., 2005). The Ler protein counteracts silencing by the H-NS global repressor, thus promoting the expression of the LEE genes (Mellies et al., 1999;Bustamante et al., 2001). The LEE1, LEE2, and LEE3 operons encode most of the structural components of the T3SS (Elliott et al., 1998), while LEE4 contains genes encoding the needle and the translocon proteins (EspA, EspB, and EspD) (Knutton et al., 1998;Ide et al., 2001). LEE5 contains the eae and tir genes, which encode the adhesin intimin and its translocated receptor, Tir, respectively (Jerse et al., 1990;Kenny et al., 1997). The interaction between Tir and intimin leads to reorganization of the host cell cytoskeleton, with effacement of the enterocyte microvilli and F-actin accumulation underneath the adhering bacteria, forming a pedestal-like structure. These alterations are referred to as attaching and effacing (AE) lesions (Moon et al., 1983;Knutton et al., 1989). Besides the LEE effectors, various T3SS-dependent non-LEE (Nle)-encoded effector genes have been described (Deng et al., 2004;Tobe et al., 2006;Wong et al., 2011;Serapio-Palacios and Finlay, 2020). Nle proteins have been shown to disturb the host cell cytoskeleton and tight junctions as well as to modulate the host inflammatory response (Dean and Kenny, 2009;Wong et al., 2011;Pearson et al., 2016). E. albertii strains also contain multiple non-LEE effectors (Ooka et al., 2015).
AE pathogens can use two distinct pathways to trigger F-actin for pedestal formation: Tir-Nck dependent and/or Tir-Nck independent. In the Tir-Nck dependent pathway, tyrosine residue 474 (Y 474 ) of the Tir receptor is phosphorylated by host cell kinase(s), which recruits Nck to initiate localized actin assembly (Kenny, 1999). On the other hand, the Nckindependent pathway uses a non-LEE effector termed EspF U / TccP (Tir cytoskeleton-coupling protein) that links to the C terminus of Tir (EPEC NPY 454 /O157:H7 EHEC NPY 458 ) via the host adaptor IRTKS/IRSp53. Both pathways activate N-WASP, which recruits and activates the Arp2/3 complex, culminating in actin polymerization (Campellone et al., 2004;Garmendia et al., 2004;Vingadassalom et al., 2009).
Most AE bacteria are extracellular pathogens, but several studies have shown that some of them can be invasive (Hernandes et al., 2008;Sampaio et al., 2009;Yamamoto et al., 2009;Pacheco et al., 2014). We have previously shown that E. albertii strain 1551-2, formerly classified as atypical EPEC (Vieira et al., 2001;Hernandes et al., 2006), is able to invade several epithelial cell lineages such as HeLa cells and differentiated intestinal T84 and Caco-2 cells (Hernandes et al., 2008;Yamamoto et al., 2009;Pacheco et al., 2014;Yamamoto et al., 2017). Moreover, we showed that the invasion process of this strain is dependent on the intimin-Tir interaction, since a coisogenic eae mutant was no longer invasive, even though its ability to adhere to epithelial cells was unaltered (Hernandes et al., 2008;Yamamoto et al., 2017). Furthermore, the 1551-2 strain persisted within Caco-2 cells for 24 h (Pacheco et al., 2014). However, further research is necessary to understand the steps for this colonization process, including analysis of effector protein functions and the expression of the LEE genes during this process.
To unravel the pathogenesis of the E. albertii 1551-2 strain, here we performed a genetic analysis of the LEE and non-LEE genes and evaluated the expression of the LEE operons in various stages of the interaction of this strain with differentiated intestinal Caco-2 cells. The ability of 1551-2 strain to colonize and form pedestals was also investigated in epithelial HeLa cells in detail.

Bacterial Strain
The E. albertii 1551-2 strain, which was formerly classified as atypical EPEC, was isolated from a child (23 months old) with diarrhea, in the absence of other recognized pathogens, during an epidemiological study on diarrhea, which was carried out in 1989 at the Universidade Federal de São Paulo (UNIFESP), Brazil (Vieira et al., 2001). The strain was routinely grown aerobically in Lysogeny Broth (LB) medium or Dulbecco's modified Eagle's medium (DMEM, Gibco, USA) at 37°C. As this strain is nalidixic acid resistant, it was cultured in the presence of this antibiotic (20 µg/ml), when necessary, and stocked in LB supplemented with 15% glycerol at -80°C.

Indirect Enzyme-Linked Immunosorbent Assay
Bacterial strains were grown in LB broth at 37°C overnight with shaking (200 rpm). After, bacterial cultures were inoculated (1:100) in 3 ml of DMEM and grown under the conditions described above. Culture supernatants were obtained by centrifugation at 15,700 × g for 10 min. For ELISA, the supernatants were used for coating MaxiSorp microplates (Nunc, USA). After coating (4°C for 18 h), wells were blocked with 5% non-fat milk in PBS, and incubated with anti-TccP rabbit serum (1:100) (Martins et al., 2017), followed by peroxidase-conjugated goat anti-rabbit IgG antibody (1:5,000) incubation. The assay was developed with o-phenylenediamine (OPD) and H 2 O 2 , and the absorbance was measured at 492 nm with a Multiskan EX ELISA reader (Labsystems, USA), using DMEM medium as a blank. Two EPEC strains, BA589 (O5:H2), which is a strong TccP2 producer, and BA1768 (O51:H40), which is a TccP and TccP2 non-producer, were used as positive and negative controls, respectively (Martins et al., 2017).

Immunoblotting
Secreted proteins from cultures grown for 6 h in low glucose DMEM at 37°C and 5% CO 2 were prepared as previously described . Briefly, bacterial cultures were spun down, supernatants were collected, filter sterilized using 0.22-mm filter units, and total secreted proteins were concentrated using Amicon Ultra Centrifugal filters (Millipore). Whole cell lysates were prepared by resuspending the bacterial cell pellets in lysis buffer (50 mM tris, 150 mM NaCl, 0.1%Triton-X-100, protease inhibitor cocktail, Sigma-Aldrich, pH7.4).

Immunofluorescence and Confocal Microscopy Assays
HeLa cells were cultured in 24-well plates with round glass coverslips or 8-well chamber slides (Lab-Tek) for approximately 48 h and then infected with bacterial strains (E. albertii 1551-2, O157:H7 EHEC 86-14 and their coisogenic mutants) for 6 h at 37°C in an atmosphere supplemented with 5% CO 2 . After the infections, cells were washed with PBS, fixed with 3.7% formaldehyde, permeabilized with 0.2% Triton X-100 in PBS, and blocked with PBS plus 2% BSA. For detection of TccP-Myc or TccP3-Myc, samples were then probed with mouse monoclonal anti-Myc (1:100) primary antibodies and Alexa Fluor 568-goat anti-mouse (Molecular Probes) secondary antibody at a dilution of 1:1,000. In addition, for Tir detection, rabbit polyclonal anti-Tir (1:2,000) primary antibodies and Alexa Fluor 568-goat anti-rabbit (Molecular Probes) secondary antibody at a dilution of 1:1,000 were used. Actin and DNA were stained with FITC-phalloidin (1:50) and DAPI (1:2,000), respectively. Slides were mounted with ProLong antifade and visualized with a Zeiss LSM880 confocal laser scanning microscope.
To perform the Confocal Immunofluorescence Microscopy Assay, after the adherence assay was done with HeLa cells, the preparations were fixed with 3.5% paraformaldehyde for 15 min at room temperature. Anti-TccP (1:500), and goat anti-rabbit secondary (1:250) antibodies were diluted in PGNS solution [PBS pH 7.2 with 0.1% gelatin (Sigma-Aldrich, Co., USA), 0.1% sodium azide (Sigma-Aldrich), and 0.2% saponin (Sigma-Aldrich)], and sequentially incubated at room temperature in a humidified chamber for 1 h. Slides were mounted on pH 9.0 buffered glycerol solution with 9 mM p-phenylenediamine and evaluated in a TCS SP5 II Tandem Scanner (Leica) confocal microscope with a 63× 1.40 N.A. immersion oil objective. Images were analyzed and processed with ImageJ (Schneider et al., 2012).

Fluorescence Actin Staining Assay and F-Actin Pedestals Quantification
An overnight bacterial culture grown in LB at 37°C was inoculated (~10 7 CFU/ml) onto HeLa or MEF Nck-/-cells that had been grown in 12-well microplates in DMEM supplemented with 2% FBS. To evaluate the kinetics of pedestal formation in cells infected with the E. albertii 1551-2, preparations were incubated for 2 h, 3 h, and 6 h at 37°C in an atmosphere of 5% CO 2, washed with PBS, and fixed with 3.7% formaldehyde for 20 min at room temperature. Coverslips were then washed with PBS and incubated with 8 µM of FITC-phalloidin (Invitrogen) solution for 30 min. After that, they were washed twice with PBS and Saline-Sodium Citrate buffer (SSC) (2x), treated with 100 µg/ ml of RNAse A (Sigma-Aldrich) for 10 min, washed again with SSC (2x), incubated with 1.7 µM of propidium iodide for 10 min, and, finally, washed with SSC (2x). Images were obtained in a Zeiss Axio Vert or in a Zeiss LSM880 confocal laser scanning microscope with a 63× 1.40 N.A. immersion oil objective. Pedestal formation was quantified by randomly imaging different fields of view while recording the number of cells showing F-actin accumulation foci. Results were presented as means of percentage (%) of infected cells with F-actin or the number of pedestals per cell ± standard deviation. All samples were tested in replicates and at least two independent experiments were performed. The statistical analyses were performed with GraphPad Prism Version 8.4.2. To compare more than two means, One-way ANOVA was used, followed by the post hoc Tukey HSD test, in which P values ≤ 0.05 were considered statistically significant.

Live Cell Imaging
Transformation of E. albertii 1551-2 was performed using pDP151, a recombinant plasmid that expresses mCherry (Invitrogen) and ampicillin resistance. The plasmid was introduced by electroporation in E. albertii competent cells, and transformants (1551-2 mCherry) were selected on LB agar with 100 mg/ml of ampicillin.
HeLa cells stably expressing Actin-GFP (Riedl et al., 2008;Gruber and Sperandio, 2014) were seeded on a coverslip-bottom dish and cultured for 24 h. E. albertii 1551-2 mCherry was incubated without shaking for 18 h in LB at 37°C. The cells were then infected with a 1:50 dilution of the bacterial overnight culture (10 8 CFU/ml) in DMEM supplemented with 2% FBS and 100 mg/ml ampicillin at 37°C. After 1.5 h of incubation, the cells were washed with PBS, and then, DMEM (supplemented with 2% FBS and 100 mg/ml of ampicillin, but without Phenol Red) was added to the preparations. The cells were visualized by livecell imaging with a Zeiss confocal microscope with a 63× 1.40 N.A. immersion oil objective. Images were taken every 2 min for 2 h.

Kinetics of the Interaction Between E. albertii and Caco-2 Cells
Quantification of the 1551-2 bacteria associated with Caco-2 cells was performed as described earlier (Pacheco et al., 2014;Romão et al., 2014). Briefly, Caco-2 cells were washed with PBS, 1.0 ml of fresh medium (DMEM supplemented with 2% FBS) was added to the cell monolayers, and the cells were then inoculated with suspensions (~10 8 CFU/ml) of bacterial cells grown overnight in LB and diluted 1:50 in DMEM, and incubated at 37°C for 1.5, 3, and 6 h. After each incubation period, cells were washed, and half of the wells were treated with gentamicin (100 µg/ml) that kills extracellular (cell surface-associated) bacteria but not intracellular bacteria (invaded bacteria, IB), while the other half of the wells remained untreated to quantify total associated bacteria (TB). After 1 h, both groups were lysed with 1% Triton X-100 and serial dilutions were plated onto MacConkey agar plates and incubated at 37°C for approximately 18 h. The resulting colonies were quantified, and invasion indexes were calculated: IB × 100/TB. All assays were performed in biological and technical triplicates, and results are presented as mean ± standard deviation.

Quantitative PCR Assay
To study the expression of the ler, escJ, escV, eae, and espA genes, the 1551-2 strain was grown in LB and incubated statically at 37°C for 18 h. Bacteria (~10 8 CFU/ml) were inoculated in Caco-2 cells in 24 well-plates for 1.5, 3, 6, and 24 h (6 h of interaction and 18 h of gentamicin treatment-100 µg/ml) at 37°C. After each incubation period, the preparation was treated with Trizol (Invitrogen, USA) and the RiboPure Bacteria Kit (Ambion) was used for total RNA extraction. cDNA synthesis was performed by the SuperScript First-Strand Synthesis kit for RT-PCR (Invitrogen, USA). The primers used to detect gene expression are listed in Table 1. The expression levels of each gene at different time points were compared by using the relative quantification method (Romão et al., 2014). Real-time quantification data were expressed as fold change in the expression levels of each gene at different time points. Data obtained at 3 and 6 h were compared with the 1.5 h data as a calibrator. For internal control, we used the RNA polymerase subunit alpha gene (rpoA). Total RNAs of all time-points were obtained from three independent assays performed in triplicate. Statistical differences were determined by the Student's t-test, and P values ≤ 0.05 were considered statistically significant.

RESULTS
Genetic Virulence Determinants of E. albertii 1551-2 The LEE PAI of strain 1551-2 is integrated into the tRNA-pheU gene, and the core region is highly conserved when compared to those of EPEC and EHEC (data not shown), which is consistent with a previous report for multiple E. albertii strains (Ooka et al., 2015;Gomes et al., 2020). The Tir sequence from 1551-2 presented regions equivalent to the O157 EHEC Tir Y 458 / EPEC Tir Y 454 , as well as an equivalent to the EPEC Tir Y 474 ( Figure 1A), which are involved in pedestal formation via the Nck-independent and Nck-dependent pathways, respectively (Kenny, 1999;Campellone and Leong, 2005;Brady et al., 2007). Twenty-one genes for non-LEE related T3SS effectors (13 families) were identified in strain 1551-2 ( Table 2). The T3SS effector-encoding genes identified in strain 1551-2 included a tccP/espF U homolog (protein id: AUS68249.1). This TccP-like protein is smaller than the previously reported TccP/EspF U and TccP2/EspF M proteins (Table S1) and contained only one complete proline-rich repeat sequence ( Figure 1B). The amino acid sequence of the whole TccP-like protein of strain 1551-2 showed 54%-55% and 50%-55% sequence identity to the TccP/ EspF U and TccP2/EspF M proteins of EHEC and EPEC, respectively ( Table S1). The N-terminal region and the proline-rich repeat showed 45-57% and 60-64% sequence identity to those of TccP/EspF U and TccP2/EspF M proteins, respectively, but the central region is totally different (Table S1 and Figure 1B). These features suggest that the TccP-like protein of E. albertii strain 1551-2 is a novel subtype of the TccP/EspF U family and was herein named as TccP3.
Although ETT2 has been recognized as a cryptic second T3SS in the E. coli/Shigella lineage, Ooka et al. (2015) have shown that many E. albertii strains harbor an apparently intact ETT2 region and suggested that the ETT2-encoded T3SS might be involved in the virulence of this pathogen. However, in strain 1551-2, the ETT2 region has been completely deleted. This deletion was probably induced by an IS1-mediated transposition ( Figure S1), but it is unknown whether this deletion occurred in vivo or in vitro.

Kinetics of Pedestal Formation in HeLa Cells Infected With E. albertii 1551-2
The 1551-2 strain was incubated with HeLa cells, and the pedestal formed was quantified in fixed cells after 2, 3, or 6 h of interaction. Our data indicated that the bacteria rapidly induced pedestal formation at 2 h ( Figure 2A). The number of actin pedestals increased exponentially between 3 and 6 h ( Figure 2B). Moreover, we performed live-cell imaging of E. albertii 1551-2 infected HeLa cells to better probe the kinetics of actin pedestal formation. This analysis revealed that actin pedestals moved on the cell surface during all time-lapses (Supplementary Video S1).

Contribution of TccP3 in the Recruitment of F-Actin in Infected HeLa Cells
A novel subtype of tccP was identified from the whole genome sequence analysis of 1551-2 and was designated tccP3. We attempted to investigate if this strain would be able to produce TccP3, as well as define the role of TccP3 in triggering F-actin for pedestal formation during the infection processes. By using an anti-TccP antisera, we demonstrated that 1551-2 strain can produce TccP3 ( Figure 3A). Moreover, we showed that the TccP3 protein co-localized with polymerized F-actin underneath adhered bacteria ( Figure 3B). As expected, deletion of the tccP3 gene in 1551-2 strain did not affect its ability to promote F-actin aggregation underneath adhered bacteria (data not shown), because, as previously reported, the 1551-2 strain has a functional Tir-Nck dependent pathway (Hernandes et al., 2008).
To investigate the contribution of TccP3 in the formation of Factin pedestals, the tccP3 gene was cloned in frame with a Myc tag, generating the recombinant plasmid pTccP3. We demonstrated that 1551-2 harboring this plasmid produced the TccP3-Myc In panel (B), the N-terminal region (56 amino acids) and central region of TccP/EspF U family proteins are indicated by gray lines. Proline-rich repeats (PRRs) and a partial repeat are indicated by arrows and a dashed arrow, respectively. See Table S1 for sequence identities of the whole proteins, N-terminal regions, central regions, and PRR1s between the proteins shown in this figure. fused protein ( Figure 4A), as well as that TccP3-Myc ( Figure 4B) and Tir ( Figure 4C) proteins co-localized with polymerized F-actin in infected cells. Immunoblotting with anti-EspB and anti-RpoA antibodies, used as controls for the secreted protein preparation and bacterial cell lysate, respectively, are shown in Figure S2. Distinct to what was observed in HeLa cells infected with EHEC 86-24 (pKC471), where a perfect colocalization between F-actin and TccP-Myc can be seen, in cells infected with the 1551-2 (pTccP3) strain, one can observe the staining of F-actin colocalizing or not with TccP3-Myc, which may be explained by the fact that strain 1551-2 has a functional Tir-Nck-dependent pathway (Hernandes et al., 2008). Further, we decided to evaluate the ability of TccP3 to trigger Factin in a Tir-Nck independent pathway by using an O157:H7 EHEC 86-24 strain. First, we demonstrated that the TccP3-Myc and Tir proteins co-localized with polymerized F-actin in infected epithelial cells ( Figure S3A), similar to that observed with the positive control TccP-Myc ( Figure S3B). By comparing in quantitative assays, the EHEC mutant 86-24DtccP with its complemented counterpart, 86-24DtccP (pTccP3), we observed only a very slight increase in the number of infected cells with visible F-actin staining ( Figure 5A), as well as in the number of sites corresponding to F-actin staining per cell ( Figure 5B), with these differences being not statistically significant. Representative images used to perform the quantitative FAS assays are given in Figure 5C.
Since E. albertii and EHEC are AE pathogens of different species, we then hypothesized that TccP3 could possess a distinct mechanism from TccP/TccP2 to trigger F-actin during the infection processes in the host cell. To test this hypothesis, we used Nck-/-MEF cells infected with the strains EHEC 86-24 and   E. albertii 1551-2, and their respective mutants in the tccP and tccP3 genes, to assess their ability to trigger F-actin underneath adherent bacteria in the absence of Nck. Surprisingly, the 1551-2DtccP3 did not lose the ability to trigger F-actin, suggesting that the 1551-2 strain use an Nck-independent pathway to form F-actin pedestals, in which TccP3 does not appear to perform any significant contribution ( Figure S4). As expected, the EHEC 86-24DtccP mutant strain did not form F-actin pedestals in the absence of Nck ( Figure S4). Together, these data suggest that TccP3 may only weakly trigger F-actin to promote pedestal formation, or that it may not be associated with establishment of this phenotype in vitro.

Expression Levels of the LEE Genes Are Constant During the Initial Stages of Bacterial Interaction With Differentiated Intestinal Caco-2 Cells
To evaluate the expression levels of the LEE genes during the initial stages of the interaction of 1551-2 strain with differentiated Caco-2 cells, we performed a kinetic study of the expression of five LEE genes, each representing one of the five polycistronic LEE operons, using the following time-points: 1.5, 3, and 6 h. No statistically significant change in the expression of the LEE genes was detected during this time-course (Figure 7), suggesting that the T3SS and its constant expression may be necessary for efficient colonization, and subsequent events throughout the time bacteria were in contact with intestinal cells.
The LEE Genes Are Overexpressed During Persistence Inside Caco-2 Cells Pacheco and co-workers (2014) reported that strain 1551-2 was able to persist and multiply in the intracellular environment up to 24 h when the population was twice as high when compared to the 6 h time point, and the Caco-2 monolayers were preserved during that period. In addition, transmission electron microscopy images showed that intracellular bacteria were imaged on top of actin pedestal-like structures inside vacuoles (Pacheco et al., 2014). Therefore, we decided to evaluate the expression of the LEE genes after 24 h of infection by qRT-PCR. Except for espA, the expression of the eight LEE genes investigated significantly increased at 24 h in comparison with 6 h (Figure 8). This suggests that the LEE-encoded T3SS may be necessary for 1551-2 strain to persist for more extended periods inside the host cell.

DISCUSSION
In this study, we reported genomic features of the E. albertii strain 1551-2, which was formerly classified as atypical EPEC (Vieira et al., 2001;Hernandes et al., 2006), and evaluated the expression of the LEE along the process of colonization of differentiated intestinal Caco-2 cells.
By studying the dynamics of the interaction of strain 1551-2 with epithelial cells, in this study, we found actin pedestals A B FIGURE 3 | Evaluation of TccP3 production and colocalization with F-actin accumulation underneath adherent E. albertii 1551-2 strain. (A) Detection of TccP3 production by the 1551-2 strain by ELISA. The recombinant protein TccP-His and atypical EPEC BA589 (TccP2 + ) were used as positive controls, while the atypical EPEC strain BA1768 (TccP -/TccP2 -) was used as a negative control. ***p < 0.001. (B) HeLa cells were incubated with E. albertii 1551-2 for 6 h, fixed and stained with anti-TccP sera, phalloidin-FITC, and DAPI, then evaluated by confocal microscopy. Actin filaments (green) clearly accumulate underneath adherent bacteria and colocalize (arrowheads) with TccP3 (red). Confocal microscopy image of a single focal plane; DAPI (blue), scale bar = 5 mm. moving on the HeLa cell surface during all time-lapses studied. Similar behavior was reported recently for an EHEC strain and a typical EPEC strain (Velle and Campellone, 2017). In addition, it was shown that EHEC extracellular motility and cell-to-cell transmission are driven by TccP/EspF U -mediated actin assembly (Velle and Campellone, 2017). These observations suggest that actin pedestal mobility may be necessary for bacterial internalization, and perhaps for bacterial persistence within cells.
Extending the genomic analyses of the 1551-2 strain, we have identified a gene encoding a new variant of the TccP/EspF U family, which was termed tccP3. Although we have accumulated evidence that the TccP3 protein is produced by 1551-2 strain,  and that this non-LEE effector is translocated to infected host cells and co-localizes with both polymerized F-actin and Tir, apparently TccP3 was not sufficient to efficiently trigger F-actin to form pedestals via a Tir-Nck independent pathway in the EHECDtccP background. A hypothesis that can explain the previously mentioned data is the fact that the proline-rich segment 27 IPPAPNWPAPTPP 39 , which harbors the motifs responsible for linking the TccP/EspF U bacterial protein to the SH3 domain of the host protein IRTKS (Aitio et al., 2012), is only partially conserved in the 1551-2 strain. Additionally, in the same study, the authors have identified that the amino acid tryptophan (W), present in the proline-rich segment 27 IPPAPNWPAPTPP 39 of the TccP/EspF U protein, is critical for its high affinity to the IRTKS/ IRSp53 host linker (Aitio et al., 2012). The 1551-2 strain lacks this tryptophan residue, harboring an arginine (R) in the corresponding position. All these observations allowed us to hypothesize that TccP3 probably has a low affinity to IRTKS/IRSp53, and thus is not efficient enough to trigger F-actin polymerization underneath adhered bacteria to form pedestals in the infected cells. It is also important to consider as a hypothesis the fact that an in trans complementation assays of the EDL933DtccP mutant strain with a plasmid encoding the TccP protein harboring only the N-terminus translocation domain and one proline-rich repeat, similar to TccP3, was unable to complement the loss of actin polymerization activity, distinct to that observed with the TccP protein with two proline-rich repeats (Garmendia et al., 2006). The assays carried out with Nck-null MEF cells, infected with wildtype E. albertii 1551-2 and its tccP3 derivative mutant, reinforced the data obtained with the in trans complementation of EHEC 86-24DtccP with tccP3, thus excluding a potential contribution of the TccP3 effector to polymerize F-actin underneath adherent bacteria. Of importance and surprisingly, 1551-2DtccP3 triggered F-actin polymerization in an Nckindependent pathway, indicating that 1551-2 strain can use both Nck-dependent, as previously observed (Hernandes et al., 2008), and Nck-independent pathways to form F-actin pedestals in infected host cells. However, the bacterial effector(s) involved in the Nck-independent F-actin recruitment remains unclear and will certainly be addressed in future studies. Very recently, it has been demonstrated that TccP/EspF U can reverse the anti-inflammatory response induced by EPEC in epithelial cells (Martins et al., 2020). However, if TccP3 could act similarly to modulate the inflammatory response of the host cell, it is a question that remains to be explored in further studies. Regarding TccP3, it may be noteworthy that our preliminary analysis of the prevalence of tccP3 in 243 E. albertii genomes (Ooka et al., 2019) identified only five tccP3-positive strains (including strain 1551-2). However, as there is considerable difficulty in searching for tccP family members in draft genome sequences due to the presence of internal repeats, more detailed analyses are required to determine the precise prevalence of tccp3 genes and their intactness.
Even though we have previously demonstrated the ability of 1551-2 strain to invade and persist in differentiated intestinal Caco-2 cells (Pacheco et al., 2014;Yamamoto et al., 2017), and the critical contribution of the interaction between the LEEencoded intimin-Tir proteins for the establishment of this invasive phenotype (Hernandes et al., 2008;Hernandes et al., 2013;Yamamoto et al., 2017), a kinetic study of these events and the LEE expression levels during the different stages of the infection processes were hitherto unknown. Quantitative assays  of the interaction of 1551-2 strain with Caco-2 cells showed important key points to assist us in understanding these processes. A good example of this was the definition of two essential time points to study the processes of adhesion and invasion of host cells by this pathogen (3 and 6 h, respectively). Corroborating our previous observation that interaction between the LEE encoded intimin and Tir proteins is an important step in the adhesion process (Hernandes et al., 2013) and essential for invasion (Hernandes et al., 2008;Yamamoto et al., 2017) of host cells infected with the 1551-2 strain, we observed in this study that all five operons of the LEE region were consistently expressed during these stages of the infectious process.
Importantly, LEE genes are up-regulated during the 1551-2 strain intracellular stage, pointing to the importance of their expression in maintenance of the strain inside the host epithelial cells. These data should be considered together with findings from our previous study with the 1551-2 strain. First, we reported that this strain was located within individual vacuoles surrounded by polymerized F-actin inside infected HeLa cells (Hernandes et al., 2008). Further, transmission electron microscopy of infected Caco-2 cells also showed F-actin rich pedestals underneath adherent bacteria associated with the cytoplasmic vacuole membrane. Salmonella-containing vacuoles are also surrounded by polymerized F-actin, with this morphological aspect of vacuoles being induced by proteins encoded by genes located in the SPI-2 (Salmonella pathogenicity island 2) (Guiney and Lesnick, 2005). Vacuole-containing bacteria surrounded by polymerized F-actin were also observed in the intracellular compartment of Caco-2 cells infected by atypical EPEC E110019 (Bulgin et al., 2009), obtained from one of the most severe diarrheal outbreaks due to diarrheagenic E. coli infection in 650 individuals in a school and 137 associated household members with diarrheal disease in Finland (Viljanen et al., 1990). The biological role of this event may be related to allowing the maintenance and replication of this pathogen within the cellular environment, thus protecting these bacteria from the host's defenses.
In summary, we showed the constant expression of the LEE genes through the distinct time-points of infection, thus reinforcing the importance of the LEE-encoded T3SS proteins during adhesion, invasion and bacterial persistence of E. albertii 1551-2 strain in the host cells. A better understanding of the roles of E. albertii T3SS may help to elucidate the versatility of this system for AE pathogens since AE bacteria are usually extracellular and rarely reach the intracellular environment.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

AUTHOR CONTRIBUTIONS
TG, RH, TH, and VS conceptualized the study. FR, FM, TO, FS, DY, AB-M, NJ, and WE contributed to the formal analysis. TG, TH, and VS were responsible for the funding acquisition. FR, FM, RH, TO, FS, DY, and AB-M carried out the investigation. TG, TH, RH, and VS helped with the project administration. TG, TH, WE, RH, and VS supervised the study. TG, TH, TO, WE, RH, and VS validated the study. FR, RH, and TG wrote the original draft. FR, FM, RH, TO, FS, DY, AB-M, TH, WE, NJ, VS, and TG reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.
SUPPLEMENTARY FIGURE 1 | Structural comparison of the ETT2 region of E. albertii CB9786 strain and the corresponding region of E. albertii 1551-2 strain. ETT2 was not detected in the 1551-2 strain.
SUPPLEMENTARY VIDEO SHEET 1 | Live cell imaging of the interaction between E. albertii 1551-2 and HeLa cells. After 1.5 h of interaction between E. albertii and HeLa cells, the images were acquired every 2 min for 2 h. The actin pedestals increased in size and moved around on the surfaces of the cells during the observed time. Romão et al. Invasive E. albertii Characterization