Phages were isolated as described previously, with some modifications (Kim et al., 2022). Briefly, sewage samples were collected from a sewage treatment plant in Iksan and Gimje, Jeollabuk-do, South Korea, filtered, and mixed with an equal volume of 2 × Tryptic Soy Broth (TSB). E. coli NCCP 15961 and E. coli ER2738 were used as host strains for the enrichment of phages LEC2 and LEC10, respectively. Each host bacterium was inoculated into the mixture along with 2 mM MgCl₂ and CaCl₂, followed by incubation with shaking (180 rpm) at 37°C for 24 h. After incubation, the mixture was filtered using a 0.45-μm filter (Hyundai Micro, Seoul, South Korea) to remove residual bacterial cells, and serially diluted with sodium chloride and magnesium sulfate buffer (100 mM NaCl, 8 mM MgSO₄∙7H₂O, and 50 mM Tris–HCl, pH 7.5). The dilutions were spotted on each host bacteria E. coli NCCP 15961 and E. coli ER2738 overlaid plates and incubated at 37°C for 8 h to detect the presence or absence of phages in the final filtered liquid. The formed phage plaques were purified by successive pick-ups at least four times, followed by propagation through co-culture with the host strains E. coli NCCP 15961 and ER2738. After dialysis against sodium chloride and magnesium sulfate buffer, the phages were stored in glass vials at 4°C until further use.

The morphology of purified phages was analyzed using transmission electron microscopy (TEM). A 20-μL aliquot of the phage stock was placed on carbon-coated copper grids and left undisturbed for 10 min. The excess sample was removed and a drop of 2% uranyl acetate (pH 4.0) was added. Immediately after staining, the excess stain was removed, the grid was air-dried for 5–10 min before TEM imaging. The morphology of the phages was examined using transmission electron microscopy (TEM; Hitachi H-7650, Japan) at Jeonbuk National University, South Korea. Phage size was measured using ImageJ software v.1.52 (National Institutes of Health, Bethesda, MD, United States).

For the bacterial challenge assay, LEC2 (10⁹ PFU/mL) and LEC10 (10⁹ PFU/mL) lysates were added to exponentially growing E. coli NCCP 15961 and ER2738 at multiplicity of infection (MOI) of 0.01, 0.1, 1, and 10. Cultures were incubated with shaking at 37°C, and bacterial growth was monitored over a 10-h period by measuring the absorbance at 600 nm (A600) every hour using a microplate reader (Spark^®, Tecan, Switzerland).

The phage lifecycle was analyzed using a one-step growth curve, following a previously described method (Fulgione et al., 2019) with some modifications. Phage lysates were added at a MOI of 0.001 along with 10 mM CaCl₂ and MgCl₂ and incubated at 37°C for 5 min to allow phage adsorption. After 5 min, the cells were collected by centrifugation and resuspended in fresh TSB. After incubation at 37°C with shaking, the bacterial culture was sampled every 5 min, and phage titer was measured using a double-layer agar method.

The thermal stability of LEC2 (10⁸ PFU/mL) and LEC10 (10⁸ PFU/mL) lysates was evaluated by incubating them at various temperatures (4, 25, 37, 50, 60, 65, and 70°C) for 60 min using a heating block. Similarly, pH stability of the phage was assessed by incubating both phage lysates for 1 h at 37°C in TSB adjusted to pH 2, 4, 6, 8, 10, and 12. All experiments were conducted using 1 mL of phage lysate in 1.75 mL microcentrifuge tubes under static conditions. After temperature and pH treatments, phage titers were determined by enumerating phage plaques.

The host ranges of LEC2 and LEC10 were determined using a spotting assay with the bacterial strains listed in Table 1 and Supplementary Table S1. Each strain mixed with TSB soft agar (0.4%) was overlaid on TSA (1.5%) plates, and 10 μL of phage lysates (10⁹ PFU/mL) were spotted onto the solidified surface. The presence of a clear zone at the site of phage application (indicating no turbidity) signified strong lytic activity, whereas the absence of a clear zone (complete turbidity) indicated no lytic activity.

Phage cocktails were prepared by mixing equal volumes of LEC2 (10⁹ PFU/mL) and LEC10 (10⁹ PFU/mL). Five pathogenic E. coli strains (ATCC 43890, ATCC 43894, NCCP 14538, NCCP 15956, and NCCP 15961) were selected based on their clinical relevance and host range diversity. Each strain was grown individually to the exponential phase, and then combined in equal volumes to create a mixed culture. To evaluate the effects of individual phages and the phage cocktail in liquid culture, individual phage lysates and the cocktail were added to an exponentially growing E. coli mixture at an approximate multiplicity of infection (MOI) of 0.1. Bacterial growth was monitored over a 10-h incubation period by measuring absorbance at 600 nm (A₆₀₀) every hour using a microplate reader (Spark^®, Tecan, Switzerland).

Twenty fresh vegetables, including leafy salads, lettuce, and sprouted vegetables, were purchased from a local supermarket in Jeonju, South Korea. For each vegetable, a 30 g sample was taken and homogenized with 50 mL of TSB in a sample cup to ensure uniform distribution. From the resulting mixture, two 5 mL aliquots were separately transferred into individual tubes to create paired experimental samples. To one tube, 0.6 mL of a phage cocktail (prepared by combining LEC2 and LEC10 phage lysates at a 1:1 ratio, 10⁹ PFU/mL) was added, while the other tube received 0.6 mL of TSB as a negative control. This setup was designed to ensure that both treated and control samples originated from an identical initial mixture, minimizing variability prior to phage treatment. This process was repeated twice per vegetable. After phage or control treatment, the samples were incubated at 37°C for 2 h, a condition chosen to maximize phage lytic activity under unknown contamination status. Following incubation, the medium was serially diluted 10-fold, and 1 mL of each dilution was spotted onto 3 M Petrifilm™ E. coli/coliform Count Plates (EC). After 24 h incubation at 37°C, only colonies identified as E. coli based on morphology (i.e., black dots with a surrounding blue halo) were counted. This selective and differential medium allows the growth of only E. coli and coliform bacteria, enabling reliable enumeration of E. coli in the samples.

The E. coli MG1655 strain was used to analyze host receptors for phage LEC2 and LEC10. Based on previous research, key genes related to phage receptors in E. coli MG1655 were selected and analyzed (Hantke, 2020; Shin et al., 2024). To identify the receptors of the two phages, a spotting assay was performed using the wild-type (WT) E. coli MG1655 strain and mutant strains lacking specific receptor related genes, including E. coli MG1655 ΔbtuB, ΔfadL, ΔfhuA, ΔfliC, ΔlamB, ΔompA, ΔompC, ΔompF, ΔtolC, ΔwaaC, and ΔwaaG. The mutant strains used in this study were provided by the Molecular Food Microbiology Laboratory at Seoul National University.

Complementation of the target genes (waaC and ompC) in the deletion mutants E. coli MG1655 ΔwaaC and ΔompC was carried out by cloning the corresponding genes into the plasmid pUC19. The waaC and ompC genes of E. coli MG1655 were amplified using the primers waaC_F_HindIII, waaC_R_EcoRI, ompC_F_HindIII, and ompC_R_KpnI (Table 2). The PCR product for waaC was digested with HindIII and EcoRI, and ligated into HindIII/EcoRI-predigested pUC19. Similarly, for ompC, HindIII, and KpnI restriction enzymes were used for digestion and ligation. The ligated DNA was transformed into E. coli DH5α to construct the recombinant plasmid. The recombinant plasmids were then introduced into the deletion mutants by electroporation, and the transformants were selected on ampicillin plates (100 μg/mL). Finally, waaC and ompC gene expression was induced with 1 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG).

Phage DNA was extracted from a high-titer phage stock (10¹⁰ PFU/mL) using the Phage DNA Isolation Kit (Norgen Biotek, ON, Canada) (Naveen Kumar et al., 2024). The extracted phage DNA was sequenced using the Illumina MiSeq platform (Illumina, San Diego, CA, United States), and de novo assembly of the reads was performed using SPAdes v.3.13.0 (Prjibelski et al., 2020) at Sanigen Co. Ltd., South Korea. The phage nucleotide sequences were analyzed using BLASTN at the National Center for Biotechnology Information (NCBI). The open reading frames (ORFs) were predicted using RAST Server (Aziz et al., 2008; Salamov and Solovyevand, 2011) and annotated based on the functions of putative gene products using the NCBI online tool BLASTP¹ (Altschul et al., 1997). Antibiotic resistance and virulence factor genes were screened using the CARD (comprehensive antibiotic resistance database)² and UniProt,³ respectively. The CGview Server was used to generate a phage genome map (Li et al., 2021). Genomic comparisons were performed using Easyfig v2.2.2⁴ (Sullivan et al., 2011).

Phages were isolated from a sewage sample using E. coli NCCP 15961 and ER2738 as host strains. The isolated phages, vB_EcoS_LEC2 and vB_EcoS_LEC10, produced clear plaques on the lawns of their respective host bacteria. Morphological characterization using transmission electron microscopy (TEM) revealed that both LEC2 and LEC10 had icosahedral heads and long tails (Figure 1). Phage LEC2 possessed an icosahedral head measuring 96.6 ± 2.4 nm in length, accompanied by a tail of 102.1 ± 2.0 nm in length (n = 5). Similarly, phage LEC10 exhibited an icosahedral head with a length of 97.2 ± 1.4 nm and a tail measuring 106.4 ± 1.1 nm (n = 5). Further genomic analysis classified the phages within the Caudoviricetes order Straboviridae family, with LEC2 assigned to the genus Tequatrovirus and LEC10 to the genus Mosigvirus, demonstrating concordance between morphological and genomic classifications (Zhu et al., 2024).

The lytic activity of the isolated phages was assessed by evaluating bacterial growth suppression at different MOIs (0.01, 0.1, 1, and 10). LEC2 and LEC10 effectively lysed E. coli NCCP 15961 and ER2738 at all tested MOIs for up to 10 h (Figures 2A,B). Initially, bacterial growth increased steadily in the control groups and at lower MOIs; however, as phage activity intensified in an MOI-dependent manner, bacterial growth was gradually suppressed. At MOI of 10 and 1, both phages strongly inhibited bacterial growth from the beginning of incubation, maintaining OD₆₀₀ values close to zero for up to 10 h. At lower MOIs (0.1 and 0.01), bacterial control was less effective, likely due to an insufficient phage-to-host ratio, allowing bacterial escape or resistance mechanisms. However, even at these lower MOIs, OD₆₀₀ values gradually decreased with prolonged incubation, indicating that antimicrobial effects were still observed, albeit with a delayed onset of bacterial lysis (Figures 2A,B).

One-step growth curve were analyzed to determine the latent period (time before the first phage progeny are released) and burst size (number of phage particles produced per infected cell). LEC2 exhibited a latent period of 15 min, with a burst size of 22 PFU per infected E. coli cell (Figure 2C). In contrast, LEC10 demonstrated rapid intracellular replication and assembly, completing a shorter latent period of 10 min and a significantly larger burst size of approximately 189 PFU per infected cell (Figure 2D). Notably, the burst size of LEC10 was considerably larger than that of other Straboviridae E. coli phages (50–100 PFU/cell) reported in previous studies (Litt and Jaroni, 2017; Lu et al., 2022). Phages with short latent periods and high burst sizes are preferred for food biocontrol applications (Li et al., 2023). While LEC2 may not exhibit as much biocontrol potential as LEC10 based on burst size and latent period alone, its host specificity and potential synergistic effects within a phage cocktail warrant further investigation.

For the application of phages in food, their stability must be confirmed under various environmental conditions, including different temperatures and pH levels (Lewis and Hill, 2020). The stability of LEC2 and LEC10 was evaluated across a broad range of temperatures (−4 to 70°C) and pH values (2 to 12). Both LEC2 and LEC10 remained stable between 4°C and 50°C, with LEC2 retaining stability even at 60°C. However, at 65°C and above, phage titers gradually declined, and no plaques were detected at 70°C (Figures 3A,B). Similarly, both phages were stable under a wide range of pH (4–10) (Figures 3C,D). Under strongly acidic conditions (pH 2), phage activity decreased; however, many phages retained their ability to infect bacteria. Notably, even at pH 12, where most previously studied phages were completely inactivated, LEC2 maintained partial activity, demonstrating exceptional stability (Kitti et al., 2022; Zhu et al., 2022). These findings suggest that LEC2 and LEC10 have strong potential for the biocontrol of foodborne pathogens, as they remain stable under a range of environmental conditions, including varying pH levels, temperatures, and food processing conditions.

Generally, phages exhibit high specificity toward their target bacteria. Since phages LEC2 and LEC10 were isolated using E. coli as their host strains, it was expected that they would exhibit E. coli-specific activity. To experimentally confirm their host specificity, we evaluated their lytic activity against various major foodborne pathogenic bacteria, including Bacillus cereus, Cronobacter sakazakii, Listeria monocytogenes, Salmonella spp., Staphylococcus aureus, and Yersinia enterocolitica (Supplementary Table S1). As anticipated, neither LEC2 nor LEC10 displayed lytic activity against any of these tested non-E. coli strains, verifying their specificity toward E. coli. Broad lytic activity of phages against non-target bacteria can lead to unintended consequences, such as the emergence of phage-resistant strains or disruption of existing microbial communities. In this regard, the absence of such activity by our phage cocktail highlights its desirable specificity and supports its potential as a safe biocontrol agent. Next, we examined the host range of phages LEC2 and LEC10 more comprehensively using 34 different E. coli strains, including 17 pathogenic and 17 non-pathogenic strains (Table 1). The lytic activity was assessed by spotting phage lysates onto the lawns of the target E. coli strains and observing the formation of lysis zones and their clarity. The results showed that phages LEC2 and LEC10 exhibited a broad host range within E. coli, lysing 30 out of 34 (88%) and 28 out of 34 (82%) E. coli strains across different serotypes, respectively (Table 1). These included several NCCP strains that were originally isolated from clinical cases, indicating that the phage cocktail is effective not only against laboratory reference strains but also against strains derived from real infection sources. This observation of broad host range is consistent with previous research on the tendency of Straboviridae phages to have a broader host range compared to other phage families (Sørensen et al., 2021; Liao et al., 2024).

Importantly, LEC2 and LEC10 exhibited complementary lytic activity against the tested E. coli strains. Each phage was able to lyse strains that the other could not, ensuring broader host coverage. For instance, four strains (E. coli NCCP 14540, NCCP 13581, NCCP 13721, and NCCP 15656) that were resistant to LEC2 alone were successfully lysed by the LEC2-LEC10 phage cocktail, leading to plaque formation. Similarly, LEC10 alone was ineffective against certain strains that were susceptible to LEC2. As a result, the cocktail approach enabled complete lysis of all tested E. coli strains (Table 1). These findings highlight the effectiveness of combining phages with complementary host ranges to achieve broader bacterial control. The LEC2-LEC10 cocktail demonstrates strong potential as a biocontrol strategy, effectively targeting diverse E. coli strains, including naturally contaminated or unidentified E. coli in food (Oh et al., 2024).

To assess the bacterial suppression capability of the phage cocktail, individual phages and the phage cocktail were added to a mixture of five STEC strains that were arbitrarily selected from a subset of clinically relevant isolates representing diverse serogroups, and the differences in the inhibitory efficacy were compared. Among diverse E. coli strains, pathogenic E. coli is responsible for enteric diseases, including abdominal cramps, diarrhea, and vomiting (Hunt, 2010). In particular, Shiga toxin-producing E. coli (STEC) is associated with severe foodborne outbreaks and poses significant public health concerns (Mangieri et al., 2020). Given the importance of STEC, strains representing four major serogroups (O157:H7, O157, O103, and O26) were first selected for inclusion, and five representative isolates were then arbitrarily chosen for analysis: ATCC 43890, ATCC 43894 (both O157:H7), NCCP 14538 (O157), NCCP 15956 (O103), and NCCP 15961 (O26).

According to the host range data presented in Table 1, the five selected E. coli strains exhibited differential susceptibility to phages LEC2 and LEC10 (Table 1). Specifically, each individual phage failed to lyse at least one strain in the mixture, whereas all five strains were susceptible to at least one of the two phages. This experimental design enabled a direct assessment of whether combining the two phages could compensate for their respective host range limitations. As expected, the phage cocktail demonstrated superior bactericidal activity against the pathogenic E. coli mixture compared to individual phages. The limited inhibitory effect observed with individual phages is attributable to their inability to target all five strains, as evidenced by the progressive increase in OD₆₀₀ values over time. When each individual phage was added to the bacterial mixture, the growth of bacteria steadily increased from the early stages of cultivation reaching OD₆₀₀ values of 1.6 and 1.4, respectively, after 10 h of incubation, no difference compared to the negative control (without phage). In contrast, upon the combination of phages into a cocktail, significant growth inhibition was observed with OD₆₀₀ values remained below 0.2 for up to 7 h of incubation, indicating effective control over bacterial growth (Figure 4).

Despite the initial suppression of bacterial growth by the phage cocktail, an increase in OD₆₀₀ was observed over time, likely due to the emergence of resistant bacterial populations (Gurney et al., 2020). The growth of resistant strains can be addressed by employing a greater variety of phage combinations (Gvaladze et al., 2024).

Next, we evaluated the biocontrol potential of the LEC2 and LEC10 phage cocktail on fresh vegetables with unknown levels of naturally occurring E. coli contamination. Twenty fresh vegetables, including leafy salads, lettuce, and sprouted vegetables, were purchased from a local supermarket. Among these, E. coli contamination was found in two vegetables (#13, #19), while the remaining 18 showed no contamination (Figure 5). Treatment with the phage cocktail greatly reduced these unidentified environmental E. coli strains. In #13, although E. coli was detected at a relatively low count of 42 CFU/mL, phage cocktail application led to a substantial reduction of the bacterial population, leaving only a minimal amount after phage treatment. Notably, in #19, where E. coli contamination was too numerous to count, the phage cocktail demonstrated maximal inhibitory effects, leading to the substantial reduction of the majority of the bacterial population (Figure 5). Since the detected E. coli strains were unidentified, their pathogenic potential remained unknown. However, the ability of the phage cocktail to eliminate a broad spectrum of naturally occurring E. coli suggests its potential to also target pathogenic strains that may be present within the microbial population. This highlights the applicability of the phage cocktail beyond controlled laboratory settings, as it effectively reduced E. coli from fresh produce under real-world conditions. These results confirm the effectiveness of phages as biocontrol agents, demonstrating that the LEC2-LEC10 phage cocktail effectively targets diverse and unidentified E. coli strains in fresh produce, potentially including pathogenic variants. Despite the limited number of phages in the cocktail, each phage exhibited a sufficiently broad and complementary host range, allowing for robust biocontrol efficacy.

Although the phage cocktail significantly reduced the population of unidentified E. coli, trace amounts persisted. To eliminate these residual E. coli, higher MOI phage treatment or supplementation with additional phages exhibiting diverse host ranges may be required. This approach aligns with our previous result (Figure 2), which demonstrates that increasing phage concentration improves antimicrobial efficiency. Previous studies have also shown that enhanced bactericidal effects can be achieved by either increasing phage concentration (Lewis and Hill, 2020) or incorporating a more diverse phage cocktail (Korf et al., 2020; Erol and Keskin, 2024).

The host range of a phage is determined by the specific receptor that the phage recognizes on the surface of its host bacterium (Shin et al., 2012). These receptors may include outer membrane proteins, flagella proteins, or lipopolysaccharides (LPS) (Gambino and Sørensen, 2024). Optimizing a phage cocktail that targets multiple receptors increases the likelihood of effective bacterial control by broadening target recognition (Stone et al., 2019). In this study, we first observed that phages LEC2 and LEC10 exhibited distinct host ranges, suggesting differences in their receptor recognition. To confirm the molecular basis of their host range differences, we subsequently identified the specific receptors utilized by each phage.

To identify the host receptors used by phages LEC2 and LEC10, we evaluated their ability to infect and lyse wild-type E. coli MG1655 and various receptor-deficient mutants. The mutant strains contained deletions in genes related to key receptors, including btuB (vitamin B12 uptake protein), fadL (long-chain fatty acid transport protein), fhuA (outer membrane transport protein), fliC (flagellin), lamB (maltoporin), ompA (outer membrane protein A), ompC (outer membrane protein C), ompF (outer membrane protein F), tolC (outer membrane protein TolC), waaC (HepI transferase, involved in LPS inner core biosynthesis), and waaG (GlcI transferase, involved in LPS outer core biosynthesis) (Hantke, 2020; Shin et al., 2024). Phages LEC2 and LEC10 failed to form plaques on the lawns of E. coli MG1655 ΔwaaC and ΔompC, respectively (Figure 6). Gene complementation assays were performed by cloning the waaC and ompC genes into the expression vector pUC19 and transforming them into their respective deletion mutants. Phage infection was restored in the complemented strains (Figure 6), as evidenced by the appearance of lysis zones, confirming that LPS and OmpC serve as the receptors for LEC2 and LEC10, respectively.

The waaC gene encodes HepI transferase, which catalyzes the addition of L-glycero-D-manno-heptose (heptose) to the Kdo (3-deoxy-D-manno-oct-2-ulosonic acid) region of the LPS inner core (Piya et al., 2020; Han et al., 2024). This enzymatic step is essential for proper LPS structure, impacting bacterial outer membrane stability and phage susceptibility (Washizaki et al., 2016). Many phages, particularly those with myovirus morphotypes, recognize LPS as a primary receptor (Nobrega et al., 2018). Our findings (Figure 6) align with previous studies demonstrating that LPS modifications or deficiencies can drastically alter phage sensitivity. Given the essential role of LPS in E. coli, it is likely that phages such as LEC2 have evolved to target this ubiquitous structure to ensure efficient host recognition and infection.

Notably, phage LEC2 was unable to infect the ΔwaaC mutant but retained infectivity toward the ΔwaaG mutant. Both waaC and waaG are involved in LPS core biosynthesis; however, their roles differ. waaC is required for heptose addition to the Kdo region in the LPS inner core, while waaG encodes a glucosyltransferase that adds glucose to the heptose II residue (Yethon et al., 2000). A mutation in waaG results in LPS structure truncation after the inner core heptose residues, yet the inner core itself remains largely intact. In our study, phage LEC2 was unable to infect the ΔwaaC mutant but retained infectivity toward the ΔwaaG mutant. This suggests that LEC2 requires the complete inner core structure of LPS, as synthesized by WaaC, for successful infection. The ability of LEC2 to infect the ΔwaaG mutant indicates that the glucose I residue added by WaaG is not essential for its receptor recognition. Therefore, LEC2 likely recognizes the heptose residues within the inner core of LPS as its receptor.

Meanwhile, phage LEC10 was found to use OmpC as its receptor. OmpC, a major outer membrane porin in E. coli, facilitates the passive diffusion of small molecules and serves as a receptor for phages (Washizaki et al., 2016; Chen et al., 2020). The absence of OmpC in the mutant strain markedly reduced LEC10 infectivity, confirming its role as a critical receptor.

By identifying the specific receptors of each phage, we confirmed that the host range differences between LEC2 and LEC10, initially observed across multiple E. coli strains, were due to distinct receptor recognition. This validation further supports the rationale for optimizing a phage cocktail, demonstrating that combining LEC2 and LEC10 was an effective strategy for broadening bacterial targeting.