Bacterial strains used in this study are listed in Table 1. Cronobacter sakazakii ATCC 29544 (FARMER et al., 1980) was used as a host strain for bacteriophage isolation and propagation. All the bacterial strains were cultured in tryptic soy broth (TSB, BD Bacto, Franklin Lakes, NJ) or tryptic soy agar (TSA, BD Difco, Franklin Lakes, NJ) at 37°C for 12 h with an agitation speed of 220 rpm.

Feces and soil samples were collected from University animal farm of Seoul National University (Suwon, South Korea) for isolation of C. sakazakii-targeting bacteriophages. The 25 g samples were homogenized in 225 mL sterile Butterfield’s phosphate-buffered dilution water (0.25 M KH₂PO₄ adjusted to pH 7.2 with NaOH) using a blender (BacMixer 400, Interscience Laboratory Inc., France). The 25 mL of each homogenized sample was mixed with 25 mL of 2X TSB and incubated at 37°C for 8 h with 220 rpm agitation. After incubation, the culture was centrifuged (9,000 × g, 10 min, 4°C) and the supernatant was filtered using 0.22-μm-pore-size filters (Millipore, Billerica, MA). The 10 mL filtrate was mixed with 50 mL of TSB containing 5 mM CaCl₂ and 500 μL overnight culture of C. sakazakii ATCC 29544 and incubated at 37°C for 8 h with 220 rpm agitation, followed by centrifugation and filtration as described above. This supernatant was used for plaque assay. Tenfold serial dilutions of the last filtrate were dotted on the bacterial lawn (5 mL of 0.4% soft TSA containing 100 μL of overnight culture of C. sakazakii ATCC 29544 and being poured into TSA plate) and the plate was incubated overnight at 37°C to confirm the presence of bacteriophages via formation of plaques. Each single plaque on dotted or overlaid plate was picked with a sterile tip and dissolved in sterilized sodium chloride-magnesium sulfate (SM) buffer (50 mM Tris–HCl, pH 7.5, 100 mM NaCl, 10 mM MgSO₄·7H₂O). This process was repeated five times to isolate a single bacteriophage.

To propagate and concentrate the isolated phages, a culture of C. sakazakii ATCC 29544 with an optical density at 600 nm = 0.5 was incubated with SM buffer containing a single phage plaque and the incubation was carried out at 37°C until the culture became clear. After centrifugation of the cleared culture (15,000 × g, 10 min, 4°C), the supernatant was filtered with 0.22-μm-pore-size filter to remove cell debris and subsequently precipitated by the addition of polyethylene glycol (PEG) 6,000 (Junsei chemical, Japan) dissolved in 1 M NaCl solution at 4°C for 12 h. Centrifuged and precipitated phages were dissolved in SM buffer and the phage particles were separated by cesium chloride (CsCl) density gradient ultracentrifugation (25,000 rpm, 2 h, 4°C), followed by dialysis using SM buffer and storage at 4°C.

To determine the stability of CR8 and S13 under various temperature and pH conditions, each phage (~1 × 10⁸ PFU/mL) was incubated overnight at varying temperatures (4–75°C) and pH values (2 to 12). Thermal stability was estimated by incubating 1 mL of phage lysate (~10⁸ PFU/mL) across a temperature range from 4 to 80°C for 1 h. The stability of phages across different pH levels was assessed by exposing the phage lysate to SM buffers with varied pH values for 1 h. The viability of both phages was determined by measuring the concentration of the remaining activated phages using a spotting assay.

Phage CR8 and S13 stocks were diluted (about 1×10¹⁰ pfu/ml) with SM buffer and placed on carbon-coated copper grids. Following the removal of the excess phage suspension, negative staining was performed by adding 2% aqueous uranyl acetate (pH 4.0). The grids were examined by TEM (LEO 912 AB; Carl Zeiss, Germany) at 80 kV at the National Academy of Agricultural Science (Suwon, South Korea). Based on the examined morphologies, each phage was classified according to the guidelines of the International Committee on Taxonomy of Viruses (Fauquet et al., 2005).

Bacteriophage genomic DNA was extracted from the phage lysate according to Lambda DNA extraction protocol using the Phase lock gel (PLG; 5 PRIME, Hamburg, Germany). Whole-genome sequencing of the extracted phage DNA was carried out with a Genome Sequencer FLX (GS-FLX) titanium sequencer (Roche, Mannheim, Germany) and assembled with Newbler v2.3 (Roche) at Macrogen Inc., South Korea. Open reading frames (ORFs) were predicted using GeneMarkS (Besemer et al., 2001), Glimmer v3.02 (Delcher et al., 2007), and FgenesB softwares (Softberry, Inc. Mount Kisco, NY, USA), and ribosomal binding sites were identified using RBSfinder (J. Craig Venter Institute, Rockville, MD, USA). BLASTP (Altschul et al., 1990) and InterProScan (Zdobnov and Apweiler, 2001) programs are employed to annotate predicted ORFs, and the sequencing and annotation data were processed by Artemis14 (Carver et al., 2008). The CGView server program was used to generate a circular genome map of the phages (Stothard and Wishart, 2005). Sequence manipulations and genomic analysis were performed using CLC Genomics work-bench version 23 on a workstation at the Biopolymer Research Center for Advanced Material, Sejong University.

Phylogenetic analysis of amino acid sequences of the large subunit of the phage terminase was conducted. Sequence alignment was performed using MUSCLE, and the evolutionary distances among phages were calculated (Edgar, 2004). The phylogenetic tree was constructed using the neighbor-joining method in MEGA 11.0.13, with 1,000 bootstrap replicates (Saitou and Nei, 1987; Tamura et al., 2021).

For genomic comparison between phage CR8 and S13, their genomes were aligned using BLASTn and the results were visualized by Easyfig (Sullivan et al., 2011).

Cronobacter sakazakii ATCC 29544 derivatives with deletion of flgK, rfaC, fhuA, btuB, lamb, or ompC genes were constructed using the one-step gene inactivation method as described previously (Datsenko and Wanner, 2000). Briefly, the kanamycin resistance (Km^r) cassette from plasmid pKD13 was amplified by PCR using primers specific for each gene. Sequences of all primers for these constructs are listed in Table 2. The resulting PCR products were used to transform into C. sakazakii ATCC 29544 strain containing pKD46, and and integrated into the flgK, rfaC, fhuA, btuB, lamb, or ompC genes. Subsequently, the Km^r cassette was removed by introducing plasmid pCP20. For complementation of the deletion mutations, each gene of C. sakazakii ATCC 29544 was amplified by PCR using the primers containing restriction site (HindIII and NheI). The PCR products were digested with HindIII and NheI and ligated into HindIII/NheI-digested pBAD18 vector (Guzman et al., 1995). Plasmid constructs were confirmed by sequencing, and transformed into each deletion mutant strains. The protein synthesis was induced by the addition of 0.2% arabinose (final concentration) and phage infectivity was examined by spotting assay.

Cronobacter sakazakii ATCC 29544 wild-type or deletion mutant strains were grown exponentially until OD₆₀₀ reached to 1.0 to 1.5 and 900 μL of the culture was aliquoted into 6 microtubes (900 μL culture per 1.75-ml tube). Each culture was infected by phage (MOI = 0.01) with 5 min intervals and incubated until the first infected tube was incubated for 20 min. After incubation, all tubes were centrifuged (16,000 rpm, 1 min, 4°C) and the supernatant was filtrated (0.22-μm-pore-size filter) to remove bacterial cells. Overlay assay was carried out as previously described (Kim and Ryu, 2011), to determine the number of unadsorbed phage particles. For control, phage was added to fresh TSB without host bacteria.

An overnight culture of each tested bacteria (100 μL) was inoculated into 5 mL of 0.4% soft TSA and mixed by gentle vortexing. The mixture was poured into the 1.5% TSA plate (with 50 μg/mL ampicillin or kanamycin, if necessary) and placed at room temperature (RT) for 20 min to allow the agar to solidify. Then, 10 μL of tenfold serial dilutions of phage stock was dotted on the bacterial lawn. The plates were dried at RT for 20 min, followed by overnight incubation at 37°C. The susceptibility of the tested bacteria to phages was examined on the following day.

An overnight culture of C. sakazakii ATCC 29544 was 1% subinoculated into 50 mL of fresh TSB medium and incubated at 37°C with agitation for 1.5 h (OD at 600 nm = 0.4 ~ 0.6). Phages were added to the culture at an MOI of 1. Bacterial growth was monitored by measuring the OD₆₀₀ every hour after phage infection. An uninfected culture was used as a negative control.

A phage cocktail consisting of phage CR8 and S13 was evaluate for their ability to inhibit the growth of host bacteria. The phages CR8 and S13 were added to the bacterial culture prepared as above at an MOI of 1 each. To count the number of viable cells, samples taken from the culture were centrifuged (15,000 rpm, 1 min) to remove free phage and the cell pellet was resuspended in same volume of phosphate-buffered saline (PBS, GeneDEPOT, Barker, TX, USA) (pH 7.4). Then, it was 10-fold diluted with PBS, plated onto TSA plate, and followed incubation at 37°C for 12 h.

To isolate phage cocktail-resistant mutants, the high-titer phage cocktail overlay assay was conducted as described previously (Capparelli et al., 2010). Briefly, 100 μL of each phage (about 10¹⁰ PFU/mL, for each) and the same volume of overnight bacterial culture were added to molten 5-ml soft TSA (0.4% agar). The mixture was poured onto TSA plate and incubated at 37°C until resistant colonies appeared. Three colonies were picked and streak-purified 20 times on fresh TSA plate in the absence of phages. One of them was selected for further study and the maintenance of phage resistance was certified by spotting assay.

One microliter of the overnight culture was injected onto soft-agar motility plates (TSA medium containing 0.3% agar) to allow bacteria to swim in the plates. The plates were dried for 30 min at room temperature and incubated for 8 h at 37°C. As a positive and negative control, C. sakazakii ATCC 29544 wild type strain and flagella (ΔflgK) defective mutant strain were used, respectively.

LPS was extracted from bacteria grown overnight using hot phenol-water extraction as previously described (Kim and Ryu, 2012; Kim et al., 2018). The extracted LPS was subjected to deoxycholate polyacrylamide gel electrophoresis (DOC-PAGE) followed by fluorescent staining using the Pro-Q® Emerald 300 Lipopolysaccharide Gel Stain Kit (Molecular Probes, Cat. No. P20495; Eugene, Oregon, USA) according to manufacturer’s instructions. The gel was visualized under 300 nm UV by the Gel doc™ EZ System (Bio-rad). As a positive and negative control, C. sakazakii ATCC 29544 wild type strain and LPS (ΔrfaC) defective mutant strain were used, respectively.

Autoaggregation assay was conducted as previously described (Kim et al., 2018). Briefly, bacteria were cultured overnight, subcultured in fresh TSB medium at 1%, and then incubated at 37°C with constant shaking. Mid-exponentially grown bacteria were diluted with fresh TSB broth to achieve an OD₆₀₀ of 1.5 in glass tubes and left at room temperature.

Bacterial invasion ability into mammalian Caco-2 cells was assessed using a modified gentamicin protection assay (Kim et al., 2010). Bacteria were cultured overnight, then transferred to fresh LB medium at a 1% concentration and incubated at 37°C for 3 h with 220 rpm agitation. The OD₆₀₀-adjusted bacterial cells were pelleted, resuspended in 400 μL of EMEM for infection and used to infect monolayered cells. The infection was carried out for 1.5 h in the presence of 5% CO₂ and washed 3X with PBS. Gentamicin was treated to eliminate any bacteria that were adhering to cells for 1.5 h and washed 3X with PBS. Cells were then lysed with 1% Triton X-100 (500 μL/ well) for 15 min, cell lysates were serially diluted and plated onto LB plates to enumerate CFUs. Relative invasion capability was calculated as described previously (Kim et al., 2010) using following formula:

Out of five C. sakazakii-infecting phages isolated from environmental samples, the phages CR8 and S13 were selected for this study due to their superior lytic activity against C. sakazakii and their ability to target different host receptors, ensuring a more effective and robust phage cocktail. The two phages formed clear plaques without halos on the lawn of C. sakazakii ATCC29544. The stability of CR8 and S13 under various temperature and pH conditions were evaluated. Both phages maintained high thermal stability from 4°C to 45°C, but a significant decrease was evident at 55°C, with a notable drop in titer for both CR8 and S13. In a wide pH range (2 to 12), the phages remained highly stable, indicating that both CR8 and S13 phages have robust pH tolerance, which may be advantageous for various application where pH fluctuations occur (Supplementary Figure S1). Their host range studies determined by spotting assay on strains of Cronobacter, Escherichia coli, Salmonella Typhimurium, Listeria monocytogenes, and Bacillus cereus showed that these phages formed plaques only on Cronobacter genus (Table 1).

Morphological analysis revealed that both phages have a characteristic of the family Caudoviricetes with a contractile tail (Figure 1). Phage CR8 has an isometric head with the mean diameter of 78.6 ± 1.7 nm, and phage S13 has longish head with the major axis of 103.7 ± 4.1 nm and the minor axis of 74.7 ± 6.0 nm. Their tail length is similar to each other – non-contracted tail and contracted tail length of phage CR8 are 122.1 ± 2.3 nm and 58.1 nm, respectively, and those of phage S13 are 112.0 ± 2.1 nm and 55.0 ± 2.1 nm (Figure 1). According to the International Committee on Taxonomy of Viruses (ICTV, https://ictv.global/, accessed on 20 August 2024), CR8 is predicted to belong to the Certrevirus genus and S13 shares genetic homologies with phages in the Straboviridae family, genus Slopekvirus.

Whole genome analysis revealed that the genome sizes of CR8 and S13 are 149,162-bp and 182,145-bp, respectively and the GC contents are 50.8 and 40.2%, respectively. Phages CR8 and S13 contain 269 ORFs with 17 tRNAs and 270 ORFs with 25 tRNAs, respectively (Table 3; Figure 2). No genes related to virulence and antibiotic resistance were identified in the genome of both phages, demonstrating their safety in application. The majority of predicted genes encoded hypothetical proteins (86.6% for CR8 and 64.1% for S13). Additional details on each ORF are provided in the Supplementary Table S1. Meanwhile, phage CR8 genome shares 97% nucleotide identity with that of phage CR3, a C. sakazakii infecting phage (Shin et al., 2012), but they did not have similarity with other phages genomes publicized in GenBank database. Phage S13 genome showed little sequence homology with well-known T4 phage. Additionally, no sequence similarity between CR8 and S13 was observed at the nucleotide level (Supplementary Figure S2).

The ORFs were categorized into eight functional groups; structure/packaging, DNA replication/modification, host lysis, nucleotide metabolism, RNA modification, transcription, cell signal and other groups. Genes related to lysogenic functions such as integrase, transposase, excisionase, repressor, and the genome attachment site (attP) were absent in the genome of CR8 and S13, suggesting that the two phages are likely lytic phages. Phage CR8 has two genes encoding tail fiber proteins, CR8_028 and CR8_033, which are likely receptor binding protein (RBP) targeting the flagella of C. sakazakii. Meanwhile, phage S13 possesses genes encoding for multiple tail fiber subunits and putative RBP (S13_008), which likely targets the LPS of C. sakazakii. Interestingly, S13 contains a gene encoding putative RNA polymerase sigma factor which is utilized in gene expression at the middle and late stages (Reznikoff et al., 1985), suggesting their potential role in preferential transcription of phages genes over host genes. Genes related to host lysis, such as holin and endolysin, were not detected in the phage CR8 genome, but it contains a cell wall hydrolase (CR8_090), which shows 99% amino acid sequence identity with an endolysin of phage CR3. A gene encoding a putative lysozyme (S13_190) was identified in the phage S13 genome, but its amino acid sequence showed no homology with any known lysozyme from Cronobacter phages. It exhibited 71 and 69% homology with the lysozyme from Citrobacter phage Ci1 and Shigella phage Esh27, respectively.

The phylogenetic analysis of the two phages was conducted based on their large subunit of terminase (Supplementary Figure S3). The resulting trees include Cronobacter phages, alongside other phages from different bacterial hosts. CR8 closely aligns with Pectobacterium phage vB_PcaM_P7_Pc, a member of the genus Cetrevirus. Phage vB_PcaM_P7_Pc exhibited myovirus-like morphology with lytic life cycle. Phage S13 is positioned near the Cronobacter phage LPCS28, which is also lytic phage. The findings highlight potential of phages CR8 and S13 for use as biocontrol or as therapeutic agents.

To identify bacterial components that are required for phage infection, we decided to test six receptors known in Gram-negative bacteria (Rakhuba et al., 2010). To this end, we constructed deletion mutants of C. sakazakii ATCC 29544, each lacking one of the followings: flagella (ΔflgK), LPS (ΔrfaC), outer membrane transporter for ferrichrome (ΔfhuA), vitamin B12 (ΔbtuB), maltose (ΔlamB), and outer membrane porin OmpC (ΔompC). These mutants were challenged to test susceptibility to specific phages. Phage CR8 was unable to form any plaques on the lawns of ΔflgK deletion mutant and phage S13 failed to form plaques on the lawns of ΔrfaC deletion mutant (Figures 3A,B). However, the infectivity of both phages to C. sakazakii ATCC 29544 was fully restored when flgK and rfaC genes were reintroduced into their respective mutants (Figures 3A,B). To determine whether these genes are required for the initial step of phage infection, phage adsorption assay was conducted with the WT and deletion mutants. The ability of phage CR8 to bind to its host receptor decreased in the absence of flgK (Figure 3C). Most of the phage S13 efficiently bound to the WT strain within 20 min, while its binding to the ΔrfaC deletion mutant was completely abolished (Figure 3D). These results indicate that flagella and LPS serve as receptors for the phage CR8 and S13, respectively.

The ability of a phage cocktail to inhibit the growth of host bacteria was evaluated using bacterial challenge assays. Phage CR8 and phage S13 showed similar ability to lyse host bacteria with rapid early cell lysis, and the phage-resistance appeared 5 h after infection. On the other hand, a phage cocktail consisting of phage CR8 and phage S13 showed the prolonged period of growth inhibition up to 11 h, resulting in the delayed emergence of the resistant mutant (Figure 4A) and there was a significant reduction in the number of viable cells compared to single-phage treatments (Figure 4B).

Three phage cocktail-resistant C. sakazakii colonies were picked from high-titer overlay plates and streak-purified 20 times onto fresh TSA plates in the absence of phages to isolate a single mutant strain. At each streaking-purifying step, the newly formed colonies were subjected to spotting assay to confirm phage resistance and a single colony that exhibited consistent resistance across generations was selected for further study and named CSR strain. Given that CR8 and S13 target flagella and LPS as receptors, we hypothesized that the resistant mutant might lack flagella and LPS or exhibit loss of motility and altered LPS structure. To test these hypotheses, motility and flagella structure were examined. The CSR strain was defective in motility like ΔflgK deletion mutant, yet still displayed flagella on its surface (Figures 5A,B). LPS structure of CSR strain differed from that of WT and ΔrfaC deletion mutant; some parts of inner core and O-antigen regions of LPS were absent in CSR strain (Figure 5C). It should be noted that nothing was extracted in ΔrfaC deletion mutant, which produces a lipophilic moiety (lipid A-Kdo2; rough-type of LPS) (Figure 5C; Sirisena et al., 1992), because the hot phenol-water method is selective for smooth types of LPS over rough types (Henderson et al., 2013). Previous studies have demonstrated that the alteration in LPS leads to autoaggregation in C. sakazakii (Wang et al., 2012; Kim et al., 2018). Consistent with these findings, CSR cells precipitated more quickly than WT cells when left in broth culture at room temperature, an indicative of enhanced autoaggregation (Figure 5D).

Considering that the development of phage resistance often incurs fitness costs, including reduced virulence and heightened vulnerability to antimicrobials (Chart et al., 1989; Chan et al., 2016; Gordillo Altamirano et al., 2021; Salazar et al., 2021; Wang et al., 2021) or immune clearance (Roach et al., 2017), we aimed to evaluate invasion capability of the CSR strain to mammalian intestinal epithelial Caco-2cells. Consistent with previous studies, invasiveness of the CSR strain was significantly reduced by 90% compared to WT cells (Figure 6).