Panel MP-9, a representative collection of clinical emerging non-O157 STEC strains, colloquially referred to as the “Big Six,” was obtained from the American Type Culture Collection (ATCC).¹ Strains are of serotypes O26:H11 (BAA-2196), O45:H2 (BAA-2193), O103:H11 (BAA-2215), O111:H8 (BAA-2440), O121:H19 (BAA-2219), and O145:NM (BAA-2192). Isolates were sequenced to closure, and the culture’s virulence was profiled in this study. Accessions for genomic reads, assembled annotated chromosomes and plasmids along with strain-associated metadata are provided in Table 1 and Supplementary Table S1.

Strains were cultured overnight at 37°C with shaking at 220 rpm in lysogeny broth (LB; Thermo Fisher Scientific, Asheville, NC, United States). To maximize total genomic DNA (gDNA) yields, bacterial overnight cultures were diluted to OD₆₀₀ of 0.03 in fresh LB medium and grown at 37°C with shaking at 220 rpm to mid-log phase (OD₆₀₀ ~ 0.5). Total gDNA was extracted using the Qiagen Genomic-tip 100/G Kit (Qiagen, Inc., Valencia, CA, United States) according to the manufacturer’s instructions. Genomic DNA was subjected to both long-read (Oxford Nanopore, Oxford, United Kingdom) and short-read (Illumina, Inc., San Diego, CA, United States) sequencing. For long-read Nanopore sequencing, gDNA was diluted to a concentration of 1.5 μg in 46 μL of nuclease-free water. The library was prepared using the Ligation Sequencing Kit (SQK-LSK109) with the Native Barcoding Expansion 1–12 (EXP-NBD104) according to the manufacturer’s instructions and sequenced on a MinION with the R10.3 SpotON Flow Cell (FLO-MIN111). Paired-end short-read libraries were prepared with the Illumina Nextera XT DNA Library Preparation Kit and sequenced on the MiSeq platform using the MiSeq reagent Kit (v3) with 600-cycles. Sequence reads in the fastq format were imported into Galaxy v.22.05 (Community, 2022). Default parameters were used for all software unless specified otherwise. Quality control of fastq files was assessed using FastQC (v.0.74 + Galaxy0).² Nanopore and Illumina reads were used for hybrid assembly using Unicycler assembler (v.0.5.0 + Galaxy1; Wick et al., 2017). The chromosomal dnaA and plasmid repA genes, if applicable, were designated as the zero point of the closed molecules prior to annotation using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP; Tatusova et al., 2016).

Chromosomes and plasmids were comprehensively analyzed and visualized in Blast Ring Image Generator BRIG (v.0.95; Alikhan et al., 2011) and MAUVE (v.2.4.1; Darling et al., 2008, 2010). Serotypes in the assembled genomes were confirmed in silico using the EcOH database (Ingle et al., 2016) in ABRicate (Galaxy v.1.0.1)³ with options—minid 80—mincov 80 (Community, 2022). Average nucleotide identities (ANI) using the E. coli strain BAA-2196 (O26:H11) chromosome as designated reference were calculated with FastANI (Galaxy v.1.3), based on MinHash mapping (Jain et al., 2018). Chromosomal repeats were identified with FindRepeats (v.1.8.2 + Galaxy1; Kurtz et al., 2004; Petkau et al., 2017). Virulence and antibiotic resistance genes (ARGs) were identified using VFDB (Liu et al., 2022) and ResFinder⁴ (Florensa et al., 2022), respectively. Boundaries and locations of intact, partial, or remnant prophages were identified using PHASTER (Zhou et al., 2011; Arndt et al., 2016) and MAUVE (v.2.4.1; Darling et al., 2008, 2010), followed by manual curation with BLASTn/p against the non-redundant NCBI databases (Camacho et al., 2009). Toxin subtypes of the carried Stx-bacteriophages were recorded in silico as described elsewhere by blastn of the carried toxins against an stx suballele database (Scheutz et al., 2012; Ashton et al., 2015; Carrillo et al., 2016). The EHEC phage replication unit (eru) subtype was assigned as described in Llarena et al. (2021) and Fagerlund et al. (2022) and Stx-prophages genomes were visualized in Easyfig (v.2.2.2; Sullivan et al., 2011). Mechanistics of phage insertion can create direct repeats (DR) and insertion sites were investigated for direct repeats (DR) and attachments sites (att) using NUCmer (v.4.0.0rc1 + Galaxy2) and BLASTn (Camacho et al., 2009). Lytic phage loci in ΦStx- and non-ΦStx-prophages were identified with Prophage Hunter (Song et al., 2019). Insertion sequence (IS) elements were identified and curated using ISEScan (v.1.7.2.3 + Galaxy0; Xie and Tang, 2017). Integrons were surveyed with Integron Finder (v.2.0.2 + Galaxy1; Néron et al., 2022). Genomic islands (GI) were detected with IslandViewer4 (Bertelli et al., 2017, 2018; Bertelli and Brinkman, 2018). Plasmid incompatibility groups were identified and analyzed with MOB-Typer (v.3.0.3 + Galaxy0; Robertson and Nash, 2018).

For toxin subtyping, the stx genes were aligned to a multifasta file comprised of all currently published stx-suballele nucleotide sequences (Scheutz et al., 2012; Carrillo et al., 2016; Bai et al., 2018; Yang et al., 2020) with BLASTn (Camacho et al., 2009). Heatmaps of cataloged genes were generated with iTol (v.6.8.1; Letunic and Bork, 2021). LEE islands were identified starting from the LEE1 operon gene espG to the espF gene in LEE4, and their comparative analysis was conducted and visualized using GeneSpy (Garcia et al., 2019). We determined the subtypes by aligning the intimin genes to the 27 currently published subtype sequences of eae in GenBank (α1-2, β1-3, γ, δ, ε1-4, ζ1 and 3, η1-2, θ1-2, ι1-2, κ, μ, ν, ξ, ο, π, ρ, σ; Supplementary Table S2) using BLASTn (Camacho et al., 2009).

The assembled ATCC MP-9 genomes along with E. coli strains EC4115 (O157:H7; Eppinger et al., 2011) and K-12 substrain MG1655 (Blattner et al., 1997) were imported into SeqSphere+ (v.8.3; Ridom GmbH, Münster, Germany) for gene-by-gene alignment, allele calling, and comparison (Jünemann et al., 2013). MLST typing was performed using targeted and whole genome schemas developed for E. coli (Foley et al., 2009; Zhou et al., 2020). We determined the Sequence Type (ST) by applying the 7-gene ST Achtman schema (Zhou et al., 2020). Allele sequences for the 7 genes (adk, fumC, gyrB, icd, mdh, purA, and recA) were accessed on the EnteroBase website⁵ and imported into Ridom SeqSphere+. A core genome (cg) MLST schema was developed using the closed chromosome of K-12 substrain MG1655 (GenBank accession U00096; Riley et al., 2006) as seed as previously described (Díaz et al., 2021). Core and accessory MLST targets were identified according to the inclusion/exclusion criteria of the SeqSphere+ Target Definer. The allele information from the targeted seven-gene schema and the defined core genome gene of the panel strains were used to establish phylogenetic hypotheses using the minimum-spanning method (Kruskal, 1956; Francisco et al., 2009) with default settings in Ridom SeqSphere+ (v.8.3).

Strains were cultured overnight (o/n) at 37°C with shaking at 220 rpm in LB. Overnight LB cultures were diluted to an OD₆₀₀ of 0.03 in fresh LB media, grown to early-log phase (OD₆₀₀~0.3) at 37°C, and then subdivided into two subcultures, LB and LB + MMC. Triggering the RecA-dependent SOS-response with MMC constitutes a major pathway of Stx₋phage mobilization (Kimmitt et al., 2000). Subculture LB + MMC was supplemented with MMC (Sigma-Aldrich, Saint Louis, MO, United States) at a final concentration of 0.5 μg/mL to mobilize the carried prophages, while subculture LB was used to evaluate spontaneous prophage mobilization. To confirm phage mobilization in MMC-treated cultures, growth curves were recorded in a 96-well plate (Corning 3,370, Corning Inc., Corning, NY, United States) on a BioTek Synergy H1 plate reader (BioTek Instruments, Inc., Winooski, VT, United States) recording OD₆₀₀ values for 6 h at 10 min intervals. All experiments were executed in two biological replicates.

In this study, we sequenced and comprehensively analyzed the pathogenomes and virulence traits of six non-O157 STEC strains. Strain panel MP-9 was obtained from ATCC, which is comprised of six strains representing each of the non-O157 STEC serogroups, colloquially referred to as the “Big Six” (Eklund et al., 2001; Johnson et al., 2006; Bettelheim, 2007; Hadler et al., 2011; Hegde et al., 2012; Gould et al., 2013; Vishram et al., 2021; Supplementary Table S1). STEC genomes house an extensive and partly repetitive phage complement that hampers assembly into closed genomes (Goldstein et al., 2019; Jaudou et al., 2022). In response, we applied a long- and short read sequencing hybrid strategy (Nyong et al., 2020; Allué-Guardia et al., 2022) that allowed us to provide the high-quality closed genomes, including carried plasmids (Figures 1, 2; Supplementary Figures S1, S2). The chromosomes have an average nucleotide identity of 98.8%, with a range from 97.6% to 99.8%, indicative of the substantial conserved chromosomal backbone of E. coli (Rasko et al., 2008; Jain et al., 2018). The chromosome size in this panel ranges from 5,288,508 to 5,840,137 bp with an average GC-content of 50.65%. When compared to non-pathogenic E. coli strain K-12 substrain MG1665, these STEC strains carry at least 648,833 bp of additional genetic information. Genome statistics along with strain-associated metadata are provided in Supplementary Table S1. In Figure 1, we compared the chromosomes using strain BAA-2196 (O26:H11) as the designated reference. In comparison to E. coli strain K-12, the Big Six strains acquired multiple mobile genome elements (MGE), including the hallmark ΦStx-prophages, which are major contributors of STEC genome evolution and diversification (Lawrence and Ochman, 1998; Rasko et al., 2008; Robins-Browne et al., 2016). Individual comparisons referenced to each of the strains can be found in Supplementary Figure S1. The mobilome on the chromosomes consisting of prophages, genomic islands, and IS elements contributes 22.4% to 28.7% of sequence information, in line with the assessment in other STEC (Perna et al., 2001; Delannoy et al., 2017; Supplementary Table S4). Neither chromosomal nor plasmid-borne integrons were detected. The prophages account for 13.9 to 21.2% of the chromosome, followed by genomic islands (5.4 to 6.9%), and IS elements (0.8 to 2.5%). If plasmid-carried IS elements are considered, the percentage of IS elements increases by 1.2 to 2.7%. The IS elements in this panel showed variations in both prevalence and numbers (Supplementary Table S4). ISEScan detected 726 IS elements and categorized them into 16 known families and 40 clusters, indicative of the plasticity present in these non-O157 STEC (Supplementary Figure S3). Eight of the 40 clusters were present in the six isolates, though their respective numbers between the strains vary considerably. We further note that BAA-2196 (O26:H11) and BAA-2215 (O103:H11) strains feature similar copy numbers in shared IS clusters distinct from the remainder of strains indicative of their close relationship (Iguchi et al., 2012; Ju et al., 2012; Supplementary Figure S3; Supplementary Table S4). Thirteen elements of the IS3-168 cluster were found in each of BAA-2196 (O26:H11) and BAA-2215 (O103:H11), compared to an average of 55 copies in other strains. Inversely, the IS66-46 cluster was found to have 48 and 36 copies in BAA-2196 and BAA-2215, respectively, while other strains carry an average of eight copies. Further, eight clusters are strain-specific, and 24 clusters are present in a subset of strains. This may suggest different dynamics in the propagation of these elements.

The mobilome is comprised of prophages, genomic islands, IS elements, and plasmids, which evolve at different rates and can be acquired and secondarily lost and thus are often not indicative of evolutionary relationships. To investigate the phylogenomic boundaries of the individual strains, we established a phylogenomic framework inferred from targeted MLST and core genome MLST (cgMLST; Figure 3). As expected for this heterogenous set of serotypes, the strains belong to distinct STs with a total of 14,340 allelic changes and 926 InDels (Figure 3; Supplementary Table S5). Their shared inventory was computed at 4,304 genes comprised of 3,148 core and 908 accessory loci, indicative of the extended conserved E. coli backbone (Abu-Ali et al., 2009; Lim et al., 2010; Eppinger et al., 2011; Yin et al., 2015). High-resolution core genome MLST typing revealed a close phylogenetic relationship of serogroup O26:H11 and O103:H11 strains, as previously suggested by MLST- and genome-wide single nucleotide polymorphisms (SNPs)-based analyses for these serogroups carrying flagellar antigens H2 and H11 (Iguchi et al., 2012; Ju et al., 2012). ST-21 (BAA-2196 O26:H11) and ST-723 (BAA2215 O103:H11) are only separated in their fumC allele and 357 allelic changes in the cgMLST analysis (Figure 3). This intimate relationship is reflected in the isolates’ shared chromosomal and mobilome inventories, such as virulence genes, prophages, and LEE island organization, as discussed below.

The hybrid-sequencing strategy further identified 14 functionally and phylogenetically diverse plasmids that range in size from 5,176 to 93,980 bp and belong to four incompatibility groups (Supplementary Figure S2; Supplementary Table S1). STEC often carry plasmids that contribute diverse virulence determinants (Kaper et al., 2004; Pilla and Tang, 2018). Virulence plasmids coding for hemolysin (hlyCABD), adhesin (toxB), and serine protease (espP) were found in all strains, except in strain BAA-2440 O111:H8 (Tatsuno et al., 2001; Kaper et al., 2004; Caprioli et al., 2005; Tozzoli et al., 2005; Johnson and Nolan, 2009; Figure 4). Colicins are synthesized to gain an advantage in the shared niche and are toxic to other bacterial strains (Cascales et al., 2007). Three strains, BAA-2440, BAA-2196, and BAA-2215, contained colicinogenic plasmids. Strain BAA-2440 O111:H8 codes for colicins E3 and D on plasmids pCol156-O111-1 and pCol-O111-2, respectively. The latter is phylogenetically related to plasmids pCol-O26-3 and pCol-O111-2 exhibiting a highly conserved plasmid backbone differentiated from each other by 30 SNPs and InDels (Figure 2). A Blastn query against the NCBI non-redundant database found related plasmids in Big Six serogroups O26, O103, and O111, and STEC serogroups O104, O157, and O165, among others, as shown in Supplementary Table S4 (Ogura et al., 2009; Yan et al., 2015; Sekizuka et al., 2019; Amadio et al., 2021). Strain BAA-2196 O26:H11 carries plasmid pO26-4, which is a multidrug-resistant plasmid encoding three ARGs (sul2, aph(6)-Ib, and aph(3″)-Ib) conferring resistance to sulfonamide and aminoglycosides (Hammerum et al., 2006; Bean et al., 2009; Messele et al., 2022; Supplementary Figure S2A; Supplementary Table S4). This broad host range plasmid shares high nucleotide similarity (>99%) and coverage (>99%) to plasmids found in E. coli, Shigella sp., Citrobacter freundii and Klebsiella pneumoniae (Iguchi et al., 2009; Ye et al., 2010; Kyle et al., 2012; Liu et al., 2015; AbuOun et al., 2021; Supplementary Table S4).

The prevalence of the identified chromosomal and phage- and plasmid-borne virulence genes revealed a considerable plasticity in the individual virulence complement. We surveyed chromosomes for virulence and resistance loci and analyzed their prevalence and distribution (Figure 5; Supplementary Table S6). In total, we identified 149 chromosomal virulence genes of which 113 are shared by all strains (Figure 5). The latter includes the phage-borne stx, along with genes that make up the LEE including its effectors (Moon et al., 1983; Mellies et al., 1999; Jores et al., 2004; Kaper et al., 2004; Sadiq et al., 2014; Franzin and Sircili, 2015). The strains feature four distinct siderophore types that facilitate iron acquisition in the iron limiting condition of mammalian hosts (Ratledge and Dover, 2000; Skaar, 2010; Caza and Kronstad, 2013; Sheldon et al., 2016). All strains possess enterobactin (ent), widely distributed in E. coli (Cox et al., 1970; Garcia et al., 2011; Mey et al., 2021). Yersiniabactin (ybt) and hydroxamate aerobactin (iuc) are present in the phylogenetically related strains BAA-2196 (O26:H11) and BAA-2215 (O103:H11; Iguchi et al., 2012; Ju et al., 2012; Figure 3). Hydroxamate aerobactin is also found in strains BAA-2440 (O111:H8) and BAA-2192 (O145:NM), and the heme utilization operon (chu) in BAA-2192 (O145:NM; Figure 5). We note here that siderophores such as ybt and chu have been proposed biomarkers for serotypes O26, O157, and O145 (Pasquali et al., 2018; Jarocki et al., 2019; Carbonari et al., 2022). Antimicrobial-resistant STEC, though uncommon, have been isolated from humans, animals, and food (Eppinger et al., 2011; Amézquita-López et al., 2016; Mukherjee et al., 2017; Greig et al., 2023; Lee et al., 2023). The ATCC MP-9 strains do not carry any chromosomal antimicrobial resistance loci other than the efflux pump gene mdf(A) (Edgar and Bibi, 1997), found in most E. coli (Ahmed et al., 2020; Moser et al., 2021; Zhou et al., 2022; Awosile et al., 2023; Lee et al., 2023).

Carriage of ΦStx-prophages is a virulence hallmark of STEC; and genomes can contain multiple ΦStx-prophages in diverse stx-suballele combinations (Huang et al., 1987; Eppinger et al., 2011; Krüger and Lucchesi, 2015; Rusconi et al., 2016; Allué-Guardia et al., 2022). Stx is a key virulence factor responsible for the severe symptoms associated with STEC infections such as HUS (Karmali et al., 1983). The panel strains carry either one or two ΦStx-prophages featuring suballeles stx_1a, stx_2a, and stx_2d (Figures 5, 6). Two suballeles have been associated with elevated cytotoxicity, stx_2a (Fuller et al., 2011; Hauser et al., 2020; Pinto et al., 2021) and stx_2d (McNichol et al., 2021). Suballele stx_2a was found alone (BAA-2219 O121:H19) or in combination with stx_1a (BAA-2196 O26:H11, BAA-2440 O111:H8). Two strains carry stx_1a only (BAA-2215 O103:H11, BAA-2193 O45:H2), or in combination with stx_2d (BAA-2192 O145:NM). As evident in the comparison of the individual subtypes (Figure 6), the prophages show a high degree of genomic plasticity, in particular upstream of the toxin locus, important for regulation and replication (Unkmeir and Schmidt, 2000; Yin et al., 2015). Variability in these regions has been linked to strain-level differences in Stx-production (Herold et al., 2004; Smith et al., 2014; Ogura et al., 2015; Yin et al., 2015; Llarena et al., 2021; Rodríguez-Rubio et al., 2021; Fagerlund et al., 2022; Zhang et al., 2022; Yano et al., 2023). In total, seven chromosomal sites are occupied (Supplementary Table S4), some of which are known ΦStx-phage targets (Serra-Moreno et al., 2007; Eppinger et al., 2011; Bonanno et al., 2015; Rusconi et al., 2016; Allué-Guardia et al., 2022). The ΦStx-phage integrases have undergone evolution that allows them to target distinct insertion sites. Stx-phages tend to primarily integrate at a specific site; however, the integrase demonstrates the capacity to detect alternate insertion sites for integration if the preferred site is already occupied or absent (Groth and Calos, 2004; Serra-Moreno et al., 2007; Casjens and Hendrix, 2015; Henderson et al., 2021). ΦStx_2a phages are inserted into either arginine tRNA argW or NAD(P) H dehydrogenase wrbA, and the ΦStx_1a phage, in analogy to some ΦStx_2a in wrbA, or alternatively in peptide chain release factor prfC, outer membrane protein ompW, the tRNA-dihydrouridine synthase dusA, or tmRNA ssrA, while the ΦStx_2d phage is disrupting the spermidine uptake gene potC (Figure 6). As evident in the occupation status of wrbA by either ΦStx₁ or ΦStx₂, there is no association between toxin suballele and insertion sites in line with previous observation (Groth and Calos, 2004; Serra-Moreno et al., 2007; Steyert et al., 2012; Henderson et al., 2021).

Carriage of the LEE pathogenicity island is responsible for the development of the characteristic attaching and effacing (A/E) lesions (Jerse et al., 1990; McDaniel et al., 1995; Sperandio et al., 1998; Schmidt, 2010; Stevens and Frankel, 2014; Franzin and Sircili, 2015). It is organized into polycistronic operons, LEE1 to 5, encoding T3SS components and regulators, chaperones, and effectors (Jerse et al., 1990; Mellies et al., 1999; Kirsch et al., 2004; Schmidt, 2010). Among the LEE-encoded proteins is intimin (Eae), an outer membrane adhesin that mediates the intimate bacterial attachment to the host’s intestinal cells (Supplementary Table S2). We detected eae subtypes β, ε, γ, and θ, and further located the respective boundaries of the islands (Figure 7). The LEE operon organization is conserved with minor rearrangements in BAA-2440 O111:H8 at espG/rorf1 as previously described in the O111:H- serotype (Ogura et al., 2009). As evident in Figure 7, the LEEs of β-eae + strains BAA-2196 O26:H11 and BAA-2215 O103:H11 exhibit syntenic organization and inventory, again suggesting a close relationship as established by our cgMLST analyses (Figure 3).

The actual disease outcome cannot be predicted from in silico virulence profiling, considering the complex interactions between infective agent, the host microbiota (Pruimboom-Brees et al., 2000; Gamage et al., 2006; Nguyen and Sperandio, 2012), and the infected patient (Wong et al., 2000; Dundas et al., 2001; Gould et al., 2009; Foster, 2013). Induction efficiency of the Stx-phages is positively correlated to Stx-production (Muniesa et al., 2004; Loś et al., 2009; Del Cogliano et al., 2018) and thus mobilization of Stx-phages is used as a means to assess the conferred pathogenic potential (Karch et al., 1999; Eppinger et al., 2022; Miyata et al., 2023). For the panel cultures, we recorded Stx-production traits under non-induced culture growth in LB and under phage mobilizing conditions by adding sublethal doses of MMC to the standard LB medium (Figure 8). In all cultures, toxin production was significantly elevated when grown in phage-inducing LB + MMC media. LB titers were undistinguishable between the cultures. In contrast, we observed culture-level differences in Stx-production capabilities upon MMC treatment. More specifically, we noted a correlation of Stx-levels to the respective stx status patterns of the strains. The class of ΦStx_2a phages carry a highly potent cytotoxin (Fuller et al., 2011; Hauser et al., 2020; Pinto et al., 2021) and are known to mobilize upon activation of the SOS-response (Bonanno et al., 2016; Zhang et al., 2019; Eppinger et al., 2022). In consequence, the Stx titers of the three stx_2a + isolates were all found exacerbated (Figure 8). Strain BAA-2219 (O121:H19), carrying only stx_2a, is the highest-level producer followed by stx_1a/stx_2a + strains BAA-2440 (O111:H8) and BAA-2196 (O26:H11). Significantly lower and comparable titers were found in the remainder of strains: stx₁ strains BAA-2215 (O103:H11) and BAA-2193 (O45:H2), as well as stx_1a/stx_2d strain BAA-2192 (O145:NM). One caveat using this methodology is that it cannot distinguish between the contribution of individual ΦStx-phages to the overall Stx titer (Skinner et al., 2014). For this reason, we further investigated the mobilization of individual ΦStx-phages and resulting stx expression in the three strains that co-harbor ΦStx_1a, ΦStx_2a, and ΦStx_2d phages (Supplementary Figure S4). Both phages carried by stx_1a/stx_2a + strains BAA-2196 (O26:H11) and BAA-2440 (O111:H8) respond to MMC treatment (Supplementary Figure S4). In the latter, ΦStx_2a copies and stx_2a transcripts exceed the respective ΦStx_1a numbers in both media, while in strain BAA-2196 the stx_1a and stx_2a transcript copies are comparable under non-induced growth in LB. In contrast, only the ΦStx_1a phage is significantly mobilized in stx_1a/stx_2d + strain BAA-2192 (O145:NM), and in consequence stx_1a transcripts surpass stx_2d copies upon MMC induction, while stx_2d copies are more abundant under non-induced growth in LB. Our observations suggest a considerable heterogeneity in ΦStx-phage mobilization, even within the same ΦStx-phage subtype (Muniesa et al., 2004; Yano et al., 2023). Overall, we observed a positive correlation between phage mobilization, toxin transcript levels, and produced titers (Figure 8; Supplementary Figure S4); in analogy to other studies (de Sablet et al., 2008; Bielaszewska et al., 2012).