Shiga Toxin (Stx) Phage-Encoded Lytic Genes Are Not Required for the Mouse Virulence of O157:H7 Escherichia coli Stx2-Producing Clinical Isolates

ABSTRACT Shiga toxin (Stx)-producing Escherichia coli (STEC) is a major cause of foodborne diarrheal illness in the United States and globally, and serotype O157:H7 is frequently associated with STEC outbreaks and sporadic cases in the United States. Severe systemic diseases associated with STEC are mediated by Stx types, particularly subtype Stx2a, encoded on inducible bacteriophages. We previously identified two STEC O157:H7 clinical isolates, JH2010 and JH2012, that exhibit a large difference in virulence in a streptomycin (Str)-treated mouse model. In this study, we aimed to identify a genetic basis for the difference in virulence between those strains. Comparison of the stx2a phage sequences showed that JH2012 lacks the lytic genes S and R on the phage genome. We also demonstrated that compared to JH2012 cultures, cultures of JH2010 released more Stx2 into the supernatant and were more sensitive to bacterial lysis during growth with ciprofloxacin (Cip), an inducer of stx phages. We therefore generated an stx2a phage SR deletion mutant strain of JH2010 to determine if those genes were responsible for the high virulence of that strain. We found that deletion of the SR genes from the stx2a phage in JH2010, and another O157:H7 strain, JH2016, resulted in increased cellular retention of Stx2, but there was no difference in virulence compared to the wild-type strains. Our results indicate that the stx2a phage SR genes are involved in Stx2 localization and phage-mediated cell lysis in vitro but that they are not required in wild-type STEC strains for virulence in a mouse model. IMPORTANCE The release of Stx from STEC has been thought to be tied to phage-mediated lysis of the host bacterial cell. In this study, we found that the stx2a phage lytic genes are not required for the virulence of pathogenic O157:H7 clinical isolates in a murine model of STEC infection or for release of Stx2a into the supernatant of bacterial cultures. These results point to an alternate mechanism for Stx2a release from STEC strains.

supportive, and even with successful recovery, the development of lifelong sequelae occurs in approximately 30% of patients (5). Due to the severity and mode of transmission of STEC infections, the economic burden of disease is high. The direct national costs associated with patient care and treatment are about $400 million per year, without including the indirect costs needed to investigate outbreaks and dispose of contaminated food (6).
The STEC virulence factor responsible for severe symptoms and HUS pathology is Stx. Stx is produced by the bacteria at the site of infection and has a systemic impact after it enters the bloodstream. The main site of damage is the kidney, where Stx enters susceptible cells expressing the toxin receptor globotriaosylceramide (Gb3) and inhibits protein synthesis (7). The Stx2 group is further divided into several subtypes. The prototype Stx2a and subtypes 2c and 2d are most commonly associated with the development of more severe symptoms and HUS (8). The stx genes are encoded as part of the late genes of lambdoid, temperate bacteriophages (stx phages). Like other lambdoid phages, the stx phage integrates into the host (E. coli) chromosome and can be induced to enter the phage lytic cycle by RecA, an SOS response protein that detects damaged DNA. Phage induction can be further stimulated by DNA-damaging stressors; for example, STEC cultures grown in the presence of the DNA-damaging antibiotic ciprofloxacin (Cip) exhibit increased stx phage induction and increased Stx production (9). Induction of the stx phage lytic cycle leads to downstream lytic cycle gene expression, including expression from stx and other genes involved in phage particle production, packaging, and release (10). Phage particles are released through a sequential process involving holin (S), endolysin (R), and spanin (Rz/Rz1) proteins that destabilize the bacterial cell membranes to facilitate host cell lysis (11). The release of phage particles through host cell lysis has been thought to lead to the concurrent release of Stx. Due to the link between stx phage induction with toxin production and release, antibiotics such as Cip are associated with more severe patient outcomes and are thus contraindicated for the treatment of STEC-infected patients (12).
The pathogenicity of STEC strains in patients is complex and involves both host and pathogen factors. Our previous study with a panel of O157:H7 STEC clinical isolates demonstrated that while the stx 2a phage plays a key role in streptomycin (Str)-treated mouse virulence, not all Stx2a-producing strains are lethal, even if they produce relatively high levels of Stx2a (13). We also found that O157:H7 isolate JH2010 produces low levels of Stx2c in addition to Stx2a, but Stx2c has no impact on the toxicity or virulence of the strain (13). In this study, we asked why Stx2a 1 O157:H7 STEC isolate JH2012 is avirulent in Str-treated mice, whereas JH2010 is lethal in that same model. Based on differences in Stx2 localization and stx 2a phage sequences between JH2010 and JH2012, we initially hypothesized that JH2012 was avirulent due to the absence of the stx 2a phage lytic genes S and R. However, our findings indicate that stx 2a phage lytic genes do not influence the in vivo mouse virulence of Stx2-producing O157:H7 clinical isolates.

RESULTS
Cellular localization of Stx2 from JH2010 and JH2012 with and without induction. To assess the cytotoxicity and stx phage induction capacity of cultures of JH2010 (mouse virulent) and JH2012 (mouse avirulent) in vitro, bacteria were grown overnight in Luria broth without yeast extract (LB-YE) with or without various concentrations of Cip to induce the stx phage(s). The cytotoxicity of culture samples was measured on Vero cells to calculate the 50% cytotoxic dose (CD 50 ). JH2010 and JH2012 exhibited similar cytotoxicity in whole-culture samples (2 to 4 Â 10 5 CD 50 /mL) (Fig. 1A). Because the whole culture sample included toxin from within the cells (cell-associated) as well as toxin released from the cells (supernatant), we next separated culture samples into cell-associated and supernatant fractions and determined the CD 50 values independently. We unexpectedly found that JH2010 cultures had significantly lower cytotoxicity in the cell-associated fraction and significantly higher cytotoxicity in the supernatant fraction than JH2012 cultures. The differences in cell-associated and supernatant cytotoxicity between JH2010 and JH2012 cultures were present when the strains were grown with or without Cip ( Fig. 1B and C). Both strains showed a significant increase in cytotoxicity of the whole culture when grown with 2.5 ng/mL Cip compared to no Cip (P , 0.0001 for JH2010 and JH2012). Taken together, these results indicate that both JH2010 and JH2012 produce high levels of Stx2 and respond to 2.5 ng/mL Cip induction similarly. However, the Stx2 from JH2012 did not localize toxin to the supernatant fraction to the same level as the Stx2 from JH2010.
Genomics of the stx 2a phages from JH2010, JH2012, and similar STEC strains. We hypothesized that the differential localization of toxin between JH2010 and JH2012 was due to genetic differences between the stx 2a phages. Annotated genomic sequence data for the parental, streptomycin-sensitive (Str s ) strains that gave rise to JH2010 and JH2012 are available at NCBI (see Table 1 for accession numbers). During our analysis of these DNA sequences, we found that the assembled genome for the strain that gave rise  to JH2010 was missing the entire stx 2a phage sequence. We reassembled the original Illumina sequencing reads using various stx 2a phage sequences available at NCBI as references and compiled the stx 2a phage sequence for JH2010 (see Materials and Methods for details). We then aligned the stx 2a phage DNA sequences from JH2010 and JH2012 to determine the synteny and homology of the phage genomes. We identified regions of 50% to 100% identity between the JH2010 and JH2012 stx 2a phage genomes (see Fig. S1), but it was initially unclear from the BLAST alignment if the identified gaps represented sequences that were missing, nonhomologous, or located in a different region of the phage genome. To determine synteny, we used the Easyfig viewer, which can identify rearrangement of homologous genes. We found that the JH2010 and JH2012 stx 2a phage sequence comparison contained some inversions and rearrangements, but the recombination, regulation, and replication regions were mostly nonhomologous ( Fig. 2). Investigation of the protein sequences encoded by these nonhomologous genetic regions resulted in similar NCBI annotations between JH2010 and JH2012 stx 2a phage proteins, despite the low percent nucleotide and amino acid identities (Table S1). However, unlike the recombination, regulation, and replication regions that were present but nonhomologous in both JH2010 and JH2012, two genes within the stx 2a phage lytic region (phage holin S and endolysin R) were present in JH2010 but missing from the JH2012 sequence. We were able to confirm by PCR that JH2012 genomic DNA was missing the stx 2a phage SR genes (Fig. S2).
We next searched the NCBI nucleotide database for stx phage sequences highly similar to those from JH2010 and JH2012. From that search, we identified the stx 2a phage genomes from strains PA28 and PA8. The predicted recombination, regulation, and replication proteins of PA28 and PA8 stx 2a phages were identical, respectively, to those from JH2010 and JH2012. Pairwise alignment of the stx 2a phage genomes showed that the sequences from PA28 and JH2010 were .99% identical (Fig. S3A). The JH2012 stx 2a phage sequence has two large sequence deviations from the PA8 stx 2a phage sequence: a 2,455-bp IS3 insertion element downstream of the stx genes and a 1,257-bp deletion of the SR genes. Aside from these deviations, the remainder of the stx 2a phage sequence comparison between PA8 and JH2012 was .99% identical (Fig. S3B). We isolated spontaneous Str-resistant (Str r ) derivatives of PA28 and PA8 (PA28-Str r and PA8-Str r , respectively). The cytotoxicity from overnight cultures of PA28-Str r and PA8-Str r was then compared to that of cultures from JH2010 and JH2012. All four strains exhibited similar cytotoxicity in the whole-culture fraction (Fig. 3A). However, the localization of toxin produced by JH2012 was significantly higher in the cell-associated fraction and lower in the supernatant fraction than that of the other three strains ( Fig. 3B and C). Since the stx 2a phages from PA8-Str r and JH2012 are highly similar (with the exception of the missing SR genes from JH2012), but the toxin localization is different, we concluded that the stx 2a phage recombination, regulation, and replication regions are not responsible for the decreased Stx release from JH2012. Additionally, because PA8-Str r encodes stx 2a phage lytic genes and exhibits fraction-specific cytotoxicity similar to that of JH2010, we hypothesize that the lack of stx 2a phage lytic genes in JH2012 prevents efficient Stx2a release from the bacterial host cell.
Role of stx 2a phage lytic genes in Stx2 release and phage-mediated lysis. To directly test whether the JH2010 stx 2a phage SR genes are responsible for the differential localization of cytotoxicity in JH2010 and JH2012, we replaced the SR genes of the JH2010 stx 2a phage with an antibiotic resistance gene (cat for chloramphenicol resistance). The resulting mutant strain, JH2010DSR 2af , showed similar levels of whole-culture cytotoxicity as JH2010 and JH2012 (Fig. 4A) and was further assayed for toxin localization. JH2010DSR 2af exhibited significantly increased cell-associated cytotoxicity but no significant difference in supernatant cytotoxicity compared to wild-type JH2010 ( Fig. 4B and C).
We next decided to complement the deletion in JH2010DSR 2af but were concerned that expression of SR in trans would cause unregulated bacterial cell lysis. Therefore, FIG 3 In vitro cytotoxicity from overnight LB-YE cultures of strains with similar stx 2a phage sequences. Samples were taken from the whole culture (A) or divided into cell-associated (B) and supernatant (C) fractions. Results are graphed as the mean log CD 50 /mL 6 SD (n = 3). Significance was calculated by one-way ANOVA with Šídák's test for multiple comparisons for each fraction. *, P , 0.05; **, P , 0.01; ***, P , 0.001; ns, not significant. Results are graphed as mean log (CD 50 /mL) 6 SD (n = 3). Significance was calculated by one-way ANOVA with Šídák's test for multiple comparisons for each fraction.**, P , 0.01, ***, P , 0.001, ****, P , 0.0001.
we created a derivative of JH2010DSR 2af , named JH2010DSR 2af ::SR-kan, in which the DSR 2af region was replaced with intact SR genes followed by a different antibiotic resistance cassette (kan). We found that JH2010DSR 2af ::SR-kan showed similar cytotoxicity across all fractions compared to JH2010 (Fig. 4A to C). Therefore, we concluded that while the stx 2a phage lytic genes increase the retention of Stx2 in the cell-associated fraction, they do not have a significant impact on the Stx release of JH2010.
Differential expression of genes associated with phage induction. We next analyzed the transcription of genes in overnight cultures of JH2010, JH2012, JH2010DSR 2af , and JH2010DSR 2af ::SR-kan. We targeted genes associated with the stx 2a phage lytic cycle to determine if we could identify differences in gene expression between the four strains (Fig. 5A). As expected based on the whole-culture cytotoxicity of the strains, all four strains had similar relative expression of genes associated with stx 2a phage induction (recA, lateearly phage gene q, and late phage gene Nu1 encoding the small terminase subunit) (Fig. 5B). Because JH2010 and JH2012 encode multiple phage lytic gene copies on other lambdoid phages within the bacterial genome (Table S2), we designed two sets of quantitative PCR (qPCR) primers to detect phage lysozyme transcripts (R): primers specific to the stx 2a phage R (R 2af ) and primers to detect transcripts with .85% nucleotide identity to the stx 2a phage R (R univ ). As expected based on the absence of R 2af from the strains, JH2012 and JH2010DSR 2af had significantly reduced relative expression of R 2af compared to JH2010 (Fig. 5B). However, there was no significant difference in the relative expression of R univ among the four strains (Fig. 5B), suggesting that a significant reduction in R 2af transcription does not impact the relative global expression of similar phage-encoded R genes. Among the four strains, we found that only the relative expression of R 2af reflected the differences in toxin localization exhibited by JH2010, JH2012, and JH2010-derived strains.
OD 600 differences among JH2010, JH2012, and JH2010-derivative strains after growth with Cip. We further examined the potential influence of phage lytic genes using a functional in vitro assay. The addition of sublethal concentrations of Cip to an STEC culture decreases the cell density of the culture due to phage-mediated bacterial host cell lysis (9), so we assessed cell lysis in JH2010, JH2012, and JH2010-derivative strains by measuring the optical density at 600 nm (OD 600 ) of overnight cultures grown in various concentrations of Cip. The results are displayed as the percent difference from the OD 600 of overnight cultures grown without Cip, because the baseline OD 600 of JH2010 was higher than that of JH2012, despite similar CFU per milliliter calculations (Fig. S4). The JH2010 and JH2010DSR 2af ::SR-kan cultures were more sensitive to cell lysis by sublethal Cip concentrations than the JH2012 and JH2010DSR 2af cultures, as shown by the drop in the relative OD 600 measurements (Fig. 6). These data suggest that under phage-inducing conditions, only the stx 2a phage lytic genes are required to mediate phage-induced bacterial cell lysis, even if their presence is not required for Stx release.
In vivo impact of the stx 2a phage lytic genes on STEC virulence. Since JH2012 is avirulent in Str-treated mice, while JH2010 is lethal even at doses as low as 10 2 CFU/ mouse (13), we hypothesized that deletion of the stx 2a phage lytic genes in JH2010 would lead to attenuated virulence and lower fecal cytotoxicity in infected mice compared to infection with the wild type strain. When Str-treated mice were orally infected with JH2010 or JH2010DSR 2af , we unexpectedly found that the deletion of the stx 2a phage lytic genes in JH2010 had no impact on the survival or fecal cytotoxicity of the infected mice ( Fig. 7A and B). We considered that the high level of Stx2a produced by JH2010 in vivo may prevent the detection of modest influences on mouse virulence. To address this issue, we evaluated strain JH2016, another stx 2a stx 2c O157:H7 STEC isolate (13). We used an intermediate infectious dose of 10 6-7 CFU/mouse because a preliminary experiment indicated that JH2016, unlike JH2010, is not fully virulent at that dose. JH2016 and JH2016DSR 2af exhibited similar in vitro cytotoxicity and toxin localization as was observed for JH2010 and JH2010DSR 2af (Fig. S5). We evaluated the virulence of JH2016 and JH2016DSR 2af in the mouse model at the intermediate dose. Mice infected with either JH2016 or JH2016DSR 2af exhibited a similar, intermediate phenotype of virulence, in contrast to the complete lethality of JH2016 at 10 10-11 CFU (Fig. 7C) (13). Unexpectedly, the JH2016-infected mice had lower fecal cytotoxicity on day 1 postinfection compared to that of the JH2016DSR 2af -infected mice (Fig. 7D), a result we did not observe in the JH2010-and JH2010DSR 2af -infected mice at the higher infectious dose (Fig. 7B).
We attempted to add the stx 2a phage lytic genes SR to JH2012 to test if that would alter the virulence of JH2012; however, we were unable to genetically manipulate the strain by lambda Red recombination or suicide vector, and we chose not to add the genes on a plasmid because we predicted that the vector-based expression of lytic proteins would be dysregulated and detrimental to bacterial growth. Instead, we tested the mouse virulence of strain PA8-Str r , which has a nearly identical stx 2a phage as JH2012 but has intact SR genes (Fig. S3B). We found that while JH2012 was avirulent, as previously observed, PA8-Str r killed 2/10 infected mice (Fig. 7D). However, mice infected with both strains exhibited similar fecal cytotoxicity on days 1 and 3 postinfection (Fig. 7F). The observed differences in mouse virulence and fecal cytotoxicity were not due to significant differences in fecal colonization of the mice infected with the wild-type or derivative strains (Fig. S6).

DISCUSSION
In this study, we discovered that the stx 2a phage lytic genes (S and R) were not required for the virulence of JH2010 or a second O157:H7 isolate, JH2016, in the Strtreated mouse model. These results from our STEC O157:H7 clinical isolates support those of a previous study in which the SR genes were not required for the virulence of a Citrobacter rodentium stx phage lysogen in a germfree mouse model (15). We did, FIG 7 Mouse virulence of wild-type and DSR 2af STEC isolates. (A, C, E) Survival of Str-treated mice infected with 10 10 to 10 11 CFU of JH2010 or JH2010DSR 2af (n = 4 to 5) (A), 10 6 to 10 7 CFU of JH2016 or JH2016DSR 2af (n = 10) (C), or 10 10 to 10 11 CFU of JH2012 or PA8-Str r (n = 10) (E). (B, D, F) Fecal cytotoxicity of mice infected as in panels A, C, and E, respectively (n = 4 to 5). Statistical significance was calculated using the log-rank test (for survival) and the unpaired, two-tailed t test on each day postinfection (for fecal cytotoxicity). **, P , 0.01. however, identify a possible small contribution of the stx 2a phage lytic genes to the virulence of strain PA8-Str r , which was modestly virulent in the Str-treated mouse model. We were unable to delete the stx 2a phage lytic genes in PA8-Str r to confirm that finding.
We found that that the JH2010 stx 2a phage lytic gene mutant strain JH2010DSR 2af exhibited a relatively high level of in vitro cytotoxicity in the supernatant (;10 5 CD 50 /mL), which is in contrast to the undetectable levels of supernatant Stx2 from a DSR 2af mutant of STEC strain EDL933 (16). We do not know the reason for the difference in Stx2 release from JH2010DSR 2af and EDL933DSR 2af , but bacterial factors or differences in assay sensitivity may be involved. As the stx 2c phage is not very active in JH2010 (13), we do not think that the stx 2c phage lytic genes are involved in significant Stx release, and the similarity in Cip-induced lysis of cultures from single-and double-stx phage-cured derivatives of JH2010 supports this hypothesis. Our data suggest that Stx2 can be released into the supernatant at high levels through alternate cellular mechanisms that do not involve the stx 2a phage lytic genes, possibly through a secretion mechanism previously reported to involve specific Stx amino acid residues (17).
We measured similar levels of cell-associated cytotoxicity for JH2010DSR 2af and JH2012, but JH2010DSR 2af had significantly higher levels of Stx2 in the culture supernatant. This higher supernatant cytotoxicity suggests that JH2010 generally trends toward slightly higher Stx2 production than JH2012. We hypothesize that the stx 2a phage present in JH2010 (and possibly PA28-Str r ) induces slightly more easily than the stx 2a phage in JH2012 (and possibly PA8-Str r ). This hypothetical induction phenotype is especially prominent when contrasting the high mouse virulence of JH2010 and PA28-Str r with the low mouse virulence of PA8-Str r and the avirulence of JH2012. Taken together, these findings suggest that the Str-treated mouse model is a more sensitive model for detecting differences in virulence among these STEC clinical isolates than in vitro assays. Future in vitro studies evaluating the toxin production of STEC clinical isolates may need to be optimized for growth and phage-inducing conditions to better replicate and predict the phenotypic differences observed in vivo.
Our findings clearly show that the stx 2a phage lytic genes are not required for the virulence of STEC O157:H7 strains in Str-treated mice. Since JH2012 was derived from a clinical isolate (13), we hypothesize that the stx 2a phage lytic genes are not absolutely required for human disease, similar to our results with JH2010DSR 2af and JH2016DSR 2af in Str-treated mice. Alternatively, it is possible that the patient infected with the JH2012 parental strain was more susceptible to infection or was treated with a phage-inducing antibiotic such as Cip. We did find that daily administration of intraperitoneal Cip to JH2012-infected Strtreated mice starting on day 6 post-infection led to death within 2 to 3 days (R. R. Atitkar, unpublished data), a finding which shows that JH2012 does have the capacity for increased toxin induction, which leads to mouse virulence. We made a similar finding previously for other avirulent O157:H7 STEC isolates (13). Finally, our results show that the stx phage lytic genes play a key role in Stx release in vitro, and future studies may further clarify their role in vivo and during the course of human infection.

MATERIALS AND METHODS
Bacterial strains and plasmids. All strains used in this study are described in Table 1. Bacteria were grown in LB without yeast extract (LB-YE; 1 L diH 2 O, 10 g tryptone, 5 g NaCl). Bacterial growth in liquid cultures was measured by absorbance at 600 nm (OD 600 ) and/or quantification of CFU per milliliter of culture (CFU/mL). To induce the stx phage, ciprofloxacin (Cip) was added to cultures at a range of concentrations from 0.625 to 10 ng/mL. Stx localization studies were carried out with overnight cultures that were either aliquoted directly (whole culture) or centrifuged and separated (cell-associated and supernatant). Before use in the Vero cell cytotoxicity assay, whole culture and cell-associated samples were sonicated until there was little to no cellular debris to increase the release of intracellular and/or membrane-associated Stx (sonication conditions, 1 min process time at amplitude 15, with 10 s on and 20 s off).
Mouse infection protocol. All mouse studies were conducted in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals (18) and were approved by the Institutional Animal Care and Use Committee of the Uniformed Services University. Mouse infections were conducted with 14 to 16 g male BALB/c mice (Jackson Laboratories). The mice were given streptomycin (Str) water (5 g/L) 2 days before infection until the experiment endpoint. On the day of infection, food was removed for 3 h prior to oral gavage with 10 10-11 CFU of STEC per mouse (unless otherwise noted). The mice were observed for clinical signs for 14 days.
The mice were weighed daily, starting 48 h before infection, except for the day of infection (day 0). Fecal pellets were collected from the infected mice at days 1 and 3 post-infection, and the samples were homogenized in phosphate-buffered saline (PBS). Supernatant from the homogenized samples was (i) plated on LB1Str or chloramphenicol to determine fecal colonization (CFU/g feces) and (ii) used in a Vero cell cytotoxicity assay to determine the fecal cytotoxicity (CD 50 /g feces).
Bacterial mutagenesis. The lambda Red recombinase system with plasmid pTP1215 (has a temperature-sensitive origin of replication and ampicillin, or Amp, resistance) (19) was used to create specific deletions in STEC strains through homologous recombination, as published previously (13). The mutagenesis primers listed in Table 2 were used to generate a PCR product with the cat gene for chloramphenicol (Cm) resistance flanked by 50 bp of homology to the target gene. The template for the amplification of cat was plasmid pHSG396 (TaKaRa catalog number 3396). Potential mutant strains were selected for Cm resistance, confirmed for the targeted deletion by PCR with the screening primers in Table 2, and confirmed for the loss of the mutagenesis plasmid by Amp sensitivity.
Vero cell cytotoxicity assay. The Vero cell cytotoxicity assay was conducted as described previously (13). Briefly, Vero cells (10 5 cells/mL) were seeded onto a 96-well flat-bottom plate, incubated for 24 h, overlaid with serial dilutions of toxin-containing samples, and incubated for an additional 48 h. The cells were then fixed with 10% buffered formalin and stained with crystal violet. The plates were rinsed with water and dried, and the absorbance at 590 nm was determined. The CD 50 was calculated as the inverse dilution of sample needed to kill 50% of untreated Vero cells.
RT-qPCR. RNA was isolated from overnight cultures using the Quick-RNA miniprep kit (Zymo Research). The RNA (100 ng) was used to produce cDNA using the QuantiTect reverse transcription kit (Qiagen). The converted cDNA mixture was stored at 220 degrees until used for qPCR. Using the SYBR Green qPCR kit (Qiagen), qPCRs were set up using the designated primer pairs (10 mM for each primer) and 1.5 mg cDNA with the appropriate controls. The primers and PCR conditions used are listed in Table 2. gyrB was used as the reference gene to calculate the change in threshold cycle values (DCt), and the results are presented as the log-transformed relative expression (log 2 22DCt ). Sequence comparisons. For strains JH2010 and JH2012, annotated genomes and Illumina reads were accessible through the NCBI nucleotide and SRA databases, respectively (Table 1). Based on discrepancies between in silico and laboratory PCR screening for Stx subtypes, we found that the NCBI genome for JH2010 did not contain the sequence of the stx 2a phage. Paired Illumina reads from JH2010 were assembled into contigs using CLC Genomics Workbench. Using Geneious Prime software, five of these contigs were aligned to the sequence of Escherichia phage PA28 (GenBank accession number NC_041935.1) and the stx 2a phage sequence from CDC PulseNet library STEC strain 08-3914 (GenBank accession number NZ_CP034808.1). The two remaining sequence gaps were filled using the 08-3914 stx 2a phage sequence and confirmed by PCR amplification and sequencing of the targeted gaps.
Individual genes and proteins were aligned using Clustal Omega to determine the percent identity. The alignment of assembled phage genomes was conducted using progressiveMauve to identify both homology and rearrangements. Nucleotide BLAST alignments of the phage genomes were visualized using either the BLAST Ring Image Generator (21) or Easyfig (22). Statistical analyses. All statistical analyses were conducted using GraphPad Prism version 9.0.2 (Windows).
Data availability. The JH2010 stx 2a phage sequence was submitted to NCBI (GenBank accession number OP797663).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 2.6 MB.