The E. coli O157:H7 strain JEONG-1266 was isolated from the recto-anal junction (RAJ) of a super-shedder cattle located in Florida (Jeon et al., 2013). Genomes of other three E. coli O157:H7 strains, EDL933, EC4115, and SS17, were used as reference genomes for comparative analyses. EDL933 (ATCC43895) [NC_CP008957] was linked to a hamburger-borne outbreak in 1982 in the U.S. The clinical isolate, EC4115 [NC_011353], was isolated from a human at the time of 2006 spinach outbreak in Maine, U.S (Eppinger et al., 2011). Another reference strain, SS17 [NZ_CP008805], was collected from the RAJ of super-shedder cattle (Cote et al., 2015).

Genomic DNA of JEONG-1266 was isolated using QIAGEN DNA mini Kit. Briefly, 1 mL overnight culture of JEONG-1266 was centrifuged at 5,000 × g for 10 min. Following the manufacturer's protocol, the bacterial pellet was resuspended in buffer ATL and treated with 20 μL of proteinase K at 56°C for 30 min. The released DNAs were eluted with 100 μL of ddH₂O.

Polymerase chain reaction (PCR) was performed to determine the stx2 subtypes (i.e., stx2a and stx2c gene) carried by JEONG-1266. Two E. coli O157:H7 strains, EDL933 carrying stx2a and stx1 and FRIK2455 carrying stx2c, were used as controls. Primers used for detecting stx2a (stx2a-F: 5′-CTTTTCGACCCAACAAAGTTATGT-3′ and stx2a-R: 5′-CACAGTCCCCAGTATCGCT-3′) and stx2c (stx2c-F: 5′-TACTGTGCCTGTTACTGGGC-3′ and stx2c-R: 5′-ACAGTGCCCAGTATCGCC-3′) were designed according to previous study (Jeon et al., 2013). The PCR reaction condition was set as 94°C for 5 min; 94°C for 30 sec, 54°C for 30 sec, 72°C for 60 s for 30 cycles and a final extension time of 10 min at 72°C. Amplified PCR product was analyzed in 1% agarose gel electrophoresis and visualized by ethidium bromide staining.

The method of phage induction was previously described (Park et al., 2013). Briefly, cell lysis caused by phage induction was measured by optical density (OD). Overnight culture of bacterial strains was inoculated into Luria-Bertani (LB) broth. Mitomycin C (MMC) treatment (0.5 μg/mL) was followed when the OD₆₀₀ of the cell culture reached 0.7. EDL933 and DH5α were used as a positive and a negative control, respectively, for phage induction. For phage particle preparation, MMC was added to 25 mL of cell culture when OD₆₀₀ reached 0.7 to make a final concentration of 1 μg/mL. After 18 h incubation, unlysed cells and debris were removed by centrifugation at 3,700 × g for 20 min at 4°C, and the resultant supernatant was filtered with 0.22 μm-pore-size membrane (Fisher Scientific, USA). Precipitation of phage particles was performed using 0.25 volume of polyethylene glycol/NaCl solution (20% polyethylene glycol 8,000 and 10% NaCl). The polyethylene glycol/NaCl solution containing phage particles was incubated at 4°C overnight, followed by centrifugation at 12,000 × g for 1 h. The pellet was resuspended and stored in 1 mL SM buffer (0.58 g NaCl, 0.2 g MgSO₄·7H₂O, 0.01 g Gelatin per 100 mL of 1M Tris-Cl pH7.5). SDS-PAGE was used to check the phage protein profile after an aliquot of the phage suspension was mixed with SDS-loading buffer and boiled for 5 min.

To detect the expression of Stx2, Western blot was conducted as previously described (Jeon et al., 2014). Briefly, the exponential phase culture (OD₆₀₀= 0.7) of an E. coli O157 strain was treated with MMC to the final concentration of 1 μg/mL to induce the lytic cycle of phages in the E. coli O157 strain and concomitant stx gene expression. Next, the O157 cell culture was incubated for 18 h for complete cell lysis. The cell debris was removed by centrifugation (3,700 × g for 20 min at 4°C) and the cell-free supernatant was collected by using 0.22 μm-pore-size membrane filter (Fisher Scientific, USA). The total proteins in the supernatant were precipitated by adding 0.25 volume of 100% trichloroacetic acid (Fisher Scientific, USA) to the cell-free culture. The mixture was incubated on ice for 30 min and centrifuged at 12,000 × g, 4°C for 30 min. The protein pellet was finally washed using cold acetone. The total proteins were separated by 12% sodium dodecyl sulfate-polyacrylamide gel and transferred to an Immobilon-P polyvinylidene difluoride membrane with a pore size of 0.45 μm (MilliporeSigma, USA) for Western blot. The membrane was blocked using 5% skim milk (BD, USA) in TBST (10 mM Tris-HCL [pH 7.4], 150 mM NaCl, and 0.05% Tween-20) at room temperature for 1 h. Later, the membrane was incubated overnight at 4°C with monoclonal Verotoxin II-a subunit antibody (Meridian Life Science Inc., USA), and washed using TBST at room temperature for 1 h. The membrane was then incubated with HRP conjugated secondary antibody (GE Healthcare, USA) diluted 1:10,000 in TBST, washed with TBST, and incubated with a chemiluminescent substrate (GE Healthcare, USA). Finally, the membrane was exposed to Kodak BioMax film (Sigma, USA).

The ability of JEONG-1266 to adhere to human epithelial type-2 (HEp-2) cells was evaluated. EDL933 was used as a positive control and DH5α was used as a negative control. HEp-2 cells were maintained in Dulbecco's Modified Eagle Medium (DMEM; Coring, USA) composed of 10% (vol/vol) heat-inactivated fetal bovine serum at 37°C and 5% CO₂. Approximately 1 × 10⁵ HEp-2 cells were seeded into a 24-well polystyrene plate as measured using a hemocytometer and allowed to grow until 90% confluence. Bacterial cultures were grown overnight at 37°C in LB broth and washed with sterile phosphate-buffered saline (PBS) three times. The final cell pellet was resuspended in PBS to a final concentration of OD₆₀₀ = 1 (5 × 10⁸ CFU/mL). An aliquot (20 μL) of the resuspended culture was added into 480 μL of DMEM. Then 500 μL of DMEM containing 1 × 10⁷ CFU of bacteria was added to each well to obtain a multiplicity of infection (MOI) of 100. After a 3-h incubation at 37°C with 5% CO₂, the bacterial suspension was removed and 500 μL of new DMEM was added to each well for another 3-h incubation. The infected cell monolayer was washed using sterile PBS three times to remove any unattached bacteria. To detach the HEp-2 cells from the plate, 1 mL of 0.1% Triton X-100 in PBS buffer was added to each well and allowed to incubate for 5 min. The media was collected from each well, serially diluted in LB broth, and plated on LB agar plates to enumerate the number of attached bacterial cells. The adherence assay of each strain was conducted in triplicate. Statistical significance between the level of attachments of isolates was accessed using a student's t-test (α = 0.05).

The complete genome of JEONG-1266 was acquired using PacBio sequencing and annotated by NCBI annotation pipeline (Teng et al., 2016). The completed genome was circularly presented by CGview (Stothard and Wishart, 2005; Overbeek et al., 2014; Brettin et al., 2015). All the coding sequences (CDSs) were assigned to different subsystems based on their functions using RAST (Overbeek et al., 2014; Brettin et al., 2015). Mauve 2.3.1 with default setting (Darling et al., 2004, 2010) was employed to compare the whole-genome architecture of JEONG-1266 and reference strains. Pangenome analysis was conducted to identify unique genes among strains using Prokka annotation and Roary (Seemann, 2014; Page et al., 2015). To identify the position of prophage related genes, the PHAge Search Tool (PHAST) (http://phast.wishartlab.com/) (Zhou et al., 2011) with default setting was used. The positions of predicted prophage regions of JEONG-1266, as well as reference genomes, were presented using BLAST Ring Image Generator (BRIG) (Alikhan et al., 2011). To investigate the similarity of Stx2a-encoding prophages in JEONG-1266 and other three reference strains, the nucleotide sequences of the prophages were compared using BLASTn embedded in EasyFig (Sullivan et al., 2011). Genes encoding known and putative virulence factors in the genome of JEONG-1266 were identified using the Virulence Factor Database (VFDB) (Chen et al., 2016) in Pathosystems Resource Integration Center (PATRIC) (https://www.patricbrc.org/portal/portal/patric/Home) (Wattam et al., 2014). The false-negative virulence genes were manually curated by doing BLASTn against whole-genome sequences of each isolate. The multilocus sequence typing (MLST) of E. coli can be identified using two different sets of housekeeping genes (Wirth et al., 2006; Jaureguy et al., 2008). Both typing methods were employed by MLST 2.0 in the Center for Genomic of Epidemiology (http://www.genomicepidemiology.org/).

In silico genotyping was used to predict the pathogenicity of JEONG-1266. Two methods were used to identify the clade of JEONG-1266. An in silico Clade 8 rhsA was used to determine whether the isolate belongs to the Clade 8 based on the presence of a SNP (3468C) in the rhsA gene (Liu et al., 2009). Besides, another in silico analysis based on 32 SNPs was also conducted to determine the clade of JEONG-1266 (Manning et al., 2008). To identify LSPA-6 profile of JEONG-1266, sequences of six pairs of primers were applied for in silico PCR (Yang et al., 2004). An isolate showing an LSPA-6 profile of 111111 was defined as lineage I. A strain with LSPA-6 profile of 222222 or 222223 was defined as lineage II. Other genotypes were defined as lineage I/II (Stanton et al., 2014). The subtype of Stx2-encoding prophage was identified by BLAST the sequence of PCR primers against the nucleotide sequence of the Stx2-encoding prophage (Ogura et al., 2015).

A phylogenetic tree of 26 E. coli O157: H7 strains was constructed using Parsnp (Treangen et al., 2014). The numbers of SNPs between strains were calculated using the VCF file generated by Parsnp. Figtree (v.1.4.2) (http://tree.bio.ed.ac.uk/software/figtree/) was used to edit the phylogenetic tree.

The complete genome sequence of E. coli O157 strain JEONG-1266 [NZ_CP014314], EDL933 [NC_CP008957], EC4115 [NC_011353], and SS17 [NZ_CP008805] were deposited in NCBI. The accession numbers of the strains used for comparative genomics were listed in the supplementary data (Table S1).

The STEC O157:H7 strain JEONG-1266 contained a chromosome of 5,478,683 and a 95,910 bp pO157 plasmid, with a GC content of 50.5% (Figure 1A, Table 1). Its chromosome was shown to encode for 5,363 CDSs, 106 tRNAs, and 22 rRNAs, while the pO157 contained 100 genes. Compared to the three reference strains of STEC O157:H7 including a super-shedder strain (SS17) and two outbreak-associated strains (EDL933 and EC4115), the chromosome of JEONG-1266 was slightly shorter in size, containing fewer CDSs (Table 1). The number of rRNA genes was highly conserved, while that of tRNA genes differed between the four strains, with the highest tRNA gene number in JEONG-1266. Functional categorization of genes by RAST assigned 57% (3,082 of 5,363 CDSs) of genes with the 27 subsystems categories, leaving the other 43% of the genes unclassified (Figure 1B). Among the classified genes, 114 genes belonged to the category of virulence, disease, and defense (Figure 1B), which was focused in this study.

SNP-based phylogenetic analysis of JEONG-1266 was performed using the core genome sequences of 26 previously characterized E. coli O157:H7 strains isolated from diverse sources (Table S1). The 26 isolates clustered into three major clades, Clade I, II, and III (Figure 2A). Clade I contained clinical strains that have been associated with outbreaks and severe diseases in humans, while Clade II consisted of strains of bovine origin (FRIK966 and FRIK2000) and ground beef (EC869) (Figure 2A and Table S1). Clade III was composed of both bovine and clinical strains (Figure 2A and Table S1). JEONG-1266 belonged to Clade III, which consisted of 2006 multistate spinach outbreak strains (e.g., EC4115) (Eppinger et al., 2011) and the super-shedder bovine strains (e.g., SS17) (Cote et al., 2015). In particular, the SNP matrix data revealed that JEONG-1266 is most closely related to EC4115 (48 SNPs) and EC4191 (56 SNPs) among the spinach outbreak strains (Figure S1). The MLST profile showed that all of the Clade II and Clade III strains analyzed, including JEONG-1266, were typed as ST628 (Figure 2A). Four strains in Clade I (EDL933, Xuzhou21, TW14588, and EC4501) belonged ST822, while the other 2 Clade I strains (Sakai and 1044) displayed a different MLST type, i.e., ST296 (Figure 2A). Therefore, SNP-based phylogeny and MLST together support that JEONG-1266 strain is closely related with STEC O157:H7 from two particular sources; super-shedder bovine and the spinach outbreak.

To further understand the genomic characteristics of JEONG-1266, the whole-genome architecture of JEONG-1266 and the other 3 reference strains (SS17, EC4115, and EDL933) were analyzed using Mauve 2.3.1. JEONG-1266 and EC4115 have 7 synteny blocks (SBs), while SS17 and EDL933 have 8 SBs. The EDL933 genome contained inverted SB4 and SB6 (Figure 2B). In addition, an extra non-homology block was found in SS17 and EDL933. Consistent with the phylogenetic analysis results, the genome architecture of JEONG-1266 was most similar to that of EC4115.

E. coli O157:H7 genomes are characterized by the presence of multiple prophages whose sequences are highly variable among strains. Many non-phage genes with virulent functions are present in prophages, including stx and tccP (an effector for LEE T3SS) (Garmendia et al., 2005). PHAST analysis identified a total of 19 prophages, including 11 intact prophages, in JEONG-1266 (Table 2, Figure S2). When the chromosome sequences of JEONG-1266 was compared with that of EC4115 and SS17, it carried 20 and 79 unique genes, respectively (Table S2). These unique genes located in prophage regions, mostly encoding hypothetical proteins or phage-related proteins. Furthermore, genomic variations between JEONG-1266 and the reference strains were mainly detected in prophage regions, P1, P4, P8, P9, and P10 (Table 2, Figure 2B). Except for P10, all the other four variable prophages were predicted to be intact. The genome of JEONG-1266 was shown to have two intact Stx-encoding prophages, P12 and P14, each of which carried stx subtypes, stx2c and stx2a, respectively.

Besides stx genes, the presence of other virulence genes was analyzed to address the virulence potential of the JEONG-1266 strain. In the four strains, a total of 132 virulence genes were identified using VFDB in PATRIC and categorized into 8 major functional groups: adherence (30 genes), chemotaxis (4 genes), invasion (2 genes), iron binding and uptake (21 genes), toxin (8 genes), protease (2 genes), T3SS (26 genes), and T3SS effectors (40 genes) (Figure 3A). Eight toxin genes were characterized; hlyABCD for hemolysin, stx1AB for Stx1, and stx2AB for Stx2. JEONG-1266, SS17, and EC4115 strains shared a highly conserved virulence profile overall. Most of the virulence genes displayed high homology (>90%) to the reference genes in the VFDB (Figure 3A). The most divergent gene categories (<90%) to the database include adherence and chemotaxis. The adherence-associated genes included intimin (eae), cytotoxin (toxB), outer membrane adhesin Paa (paa), E. coli common pilus (ecpABCDER), curli (csgBEFG), type 1 fimbriae (fimABCDEFGHI), and flagellar (flgCGH and fliGMP). JEONG-1266 encoded conserved T3SS structural genes as well as a total of 41 putative T3SS effector genes, indicating the presence of a functional T3SS. The essential genes of the locus of enterocyte effacement (LEE) (Nataro and Kaper, 1998), including tir, eae, espA, and espB, were shown to be highly conserved. Taken together, these results indicate that JEONG-1266 possesses an array of virulence genes required for the full virulence potential of STEC O157, a characteristic that often lacks in strains of bovine origin.

Combinational characteristics of the lineage type (I, II, I/II), stx genotype, the insertion site of the Stx-encoding prophage have been shown to be conserved in STEC O157 and useful to determine virulence potential and evolutionary relationship of the pathogen (Stanton et al., 2014). To determine the lineage of JEONG-1266, in silico LSPA-6 typing was performed using nucleotide sequences of the 6 polymorphism markers (Yang et al., 2004). JEONG-1266 showed a lineage I/II profile (i.e., 211111) (Figure 3B), which is identical to those of SS17 and EC4115 (Stanton et al., 2014; Cote et al., 2015).

Clade determination was conducted based on 32 SNP loci on the chromosome of E. coli O157:H7 (Manning et al., 2008). The resultant SNP profile indicated that EDL933 belonged to the Clade 3 while JEONG-1266, EC4115, and SS17 belonged to the Clade 8 (Figure 3B; Table S3). The Clade 8 has been shown to be the most virulent clade of E. coli O157:H7 due to high levels of Stx expression and cytotoxicity (Neupane et al., 2011).

Two types of Stx, Stx1, and Stx2, show distinct cytotoxicity and Stx2 is more frequently associated with severe diseases in humans than Stx1 (Melton-Celsa, 2014). In addition, variable Stx2 subtypes have been identified, including Stx2a and Stx2c, and Stx2a is known to be more toxic to humans than Stx2c (Schmitt et al., 1991). Stx2a and Stx2c are typically encoded in different Stx-encoding prophages. In silico analysis showed that both stx2a and stx2c are present in JEONG-1266, EC4115, and SS17 but only JEONG-1266 and EC4115 shared the same insertion site for Stx2a-encoding prophage, the argW locus (Figure 3B). SS17 and EDL933 used yehW and wrbA loci as the insertion site, respectively. These results suggest that the Stx2a-encoding prophage of JEONG-1266 is more related to that of EC4115.

Taken together, combinational genetic characteristics of JEONG-1266 (i.e., the lineage/clade types, stx genotype, and the insertion site of the Stx-encoding phage) further support its high virulence potential, especially being a member of the hypervirulent Clade 8 with the stx2a/stx2c genotype.

Of the two types of Stx2, Stx2a and Stx2c, Stx2a is known to as a major risk factor for severe STEC infections. Ogura et al. identified four major subtypes of Stx2a-encoding prophage (α, β, γ, and δ) and found that strains carrying φstx2a-γ or φstx2a-α (particularly the former) can produce higher Stx2 levels than strains carrying other stx2a phage subtypes (Ogura et al., 2015).

The subtype of Stx2a-encoding prophage of JEONG-1266 was determined by in silico PCR analysis, using the nested primer sets designed for Stx2a-encoding prophage subtyping (Ogura et al., 2015). The results showed that Stx2a-encoding prophage carried by JEONG-1266 is φstx2a-γ. Since both EC4115 and EDL933 have been shown to carry φstx2a-γ (Ogura et al., 2015), the genome structures of the Stx2a-encoding prophage were compared among JEONG-1266, SS17, EC4115, and EDL933. The length of the Stx2a-encoding prophage in JEONG-1266 (71,736 bp), SS17 (71,730 bp), and, EC4115 (71,736 bp) were highly similar, while it was shorter in EDL933 (63,575 bp) (Figure 4). Next, the nucleotide sequences of the entire prophage genomes were compared by BLASTn. The similarity of the Stx2a-encoding prophages was as follows; 99.98% for JEONG-1266 vs. SS17 (100% coverage, 17 nucleotides difference, and 6 gaps), 100% for JEONG-1266 vs. EC4115 (100% coverage, 1 nucleotide difference, and 0 gaps), and 97.60% for JEONG-1266 vs. EDL933 (74% coverage) (Table S4). Stx2a-encoding prophage in JEONG-1266, EC4115, and SS17 have 102 CDSs, while the one in EDL933 has 83 CDSs. Therefore, in silico PCR typing and prophage genome analysis together showed that JEONG-1266, SS17, and EC4115 are highly likely lysogenized by the same φstx2a-γ, indicating their strong potential to produce high levels of Stx2 toxins. Our data showed that the Stx2a-encoding prophage of EDL933 is structurally different from that of JEONG-1266, SS17, and EC4115, although it was determined to be φstx2a-γ subtype.

Stx production of JEONG-1266 was analyzed by MMC-mediated induction of the Stx-encoding prophages. The EDL933 strain was used as a positive control and DH5α, a stx gene-negative strain, was used as a negative control. The OD₆₀₀ value of both JEONG-1266 and EDL933 decreased after MMC treatment, while the growth of DH5α was not affected by MMC treatment (Figure 5A). This result indicates that bacterial cells of JEONG-1266 and EDL933 were lysed by phage induction.

To investigate the production of Stx2 toxins by JEONG-1266, we conducted Western blot using Stx2-specific monoclonal antibodies in the presence and absence of MMC. Both JEONG-1266 and EDL933 produced Stx2 toxins with MMC dependency (Figure 5B); whether the toxins were Stx2a and/or Stx2c was not determined in this experiment since the Stx2-specific monoclonal antibody does not have differential specificities for Stx2a and Stx2c.

Adherence of the STEC O157 to the epithelial cells of the human intestinal tract is the initial step for its colonization and pathogenesis. STEC O157 EDL933 strain has been extensively studied in its ability to adhere to and colonize the human intestinal tract (Lewis et al., 2015; Cordonnier et al., 2017). To evaluate the adherence capability of JEONG-1266 to human cells, adherence assay was conducted using human HEp-2 cells with EDL933 and DH5α as a positive and negative control, respectively. JEONG-1266 showed a significant increase in adherence to the HEp-2 cells (P < 0.05) compared with DH5α (Figure 6). The adherence capability of JEONG-1266 was similar with that of EDL933 (P > 0.05), suggesting that JEONG-1266 is comparable to EDL933 in its ability to colonize the human intestinal tract.