Sixty-five pathogenic E. coli and 72 pathogenic S. enteritidis clinical isolates were used in this study (these strains were identified as different strains from different batches of diseased animals in Hebei, Shandong, and Jilin from 2010 to 2020, and drug sensitivity tests revealed different drug resistances). The bacterial strains were cultured in Luria–Bertani culture (LB) at 37°C and stored in 30% glycerol at −80°C.

Female BALB/c mice, 5 weeks of age (20–25 g in weight), were purchased from the Experimental Animal Center of Shandong, China.

All animal procedures were performed in strict accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals, approved through the State Council of the People’s Republic of China (1988.11.1), and with the approval of the Animal Welfare and Research Ethics Committee at Qingdao Agricultural University.

A pathogenic clinical strain of E. coli (serotype K88), hereafter named E. coli K88, was used in this study for phage isolation. Feces and seawater samples were collected from a pig farm in Shandong, China. Phage isolation was performed using the standard enrichment method as described before but with slight modifications (Lu et al., 2017). In brief, an overnight culture of 50 ml of E. coli K88 was cocultured with collected samples at 37°C for 24 h and then centrifuged at 12,000 rpm for 30 min. The supernatant was filtered with a sterile disposable membrane filter (0.22 μm). Then, 100 μl of the filtered supernatant was added to 100 μl of log-phase E. coli K88 and mixed with 3 ml of prewarmed NA top agar (0.7% agar), spread on an NA plate, and cultured for 4 h at 37°C. A single plaque was selected and picked from the plate. Then, the plaque was leached overnight with 0.9% normal saline. This procedure was repeated three times to obtain purified phages. Electron micrographs of purified phage particles were obtained according to a standard method (Kreienbaum et al., 2020). A 10-μl phage sample was dropped onto carbon-coated grids, negatively stained with 2% (w/v) aqueous uranyl acetate (pH 4.0) for 5 min, and air-dried. The samples were observed with a transmission electron microscope (HT7700, Hitachi, Japan) at 80-kV accelerating voltage.

Sixty-five E. coli and 72 S. enteritidis clinical isolates were used to test the host range of phage swi3 by the spot method. Briefly, 10 μl of diluted phage suspension (1 × 10⁵ PFU/ml) was spotted on each bacterial lawn on agar plates and incubated at 37°C for 4 h. The presence of plaques was examined. Additionally, the efficiency of plating was tested as described before (Ding et al., 2020).

To determine the optimal multiplicity of infection (MOI) of phage swi3, E. coli K88 was cultured in LB broth at 37°C with shaking at 250 rpm until the early exponential phase (5 × 10⁷ CFU/ml). Phage swi3 was cocultured with E. coli K88 in 10 ml of LB broth at different MOIs (0.0001, 0.001, 0.01, 0.1, 1, and 10) (Abedon, 2016). A culture of bacteria without phage swi3 (0 MOI) was used as a control. After 2.5 h of incubation, the culture was centrifuged at 10,000 rpm for 10 min, and then the supernatant was used to determine the phage titers of phage swi3 by the double-layered agar method. The MOI resulting in the highest phage titer was considered an optimal MOI and used in subsequent large-scale phage production.

The burst size and the latent period of phage swi3 were determined by one-step growth analysis as previously described (Sun et al., 2012). In brief, 200 μl of phage suspension (10⁶ PFU) was mixed with 200 ml of bacterial culture (5 × 10⁶ CFU/ml), incubated at 37°C for 5 min, and then centrifuged, and the cell pellet was resuspended in 500 μl of LB broth. Aliquots were taken, and the phage titers were immediately determined by the double-layered agar method. This assay was performed in triplicate.

The stability of phage swi3 was tested as previously described (Akhwale et al., 2019). Briefly, thermal stability was assessed by incubating the phage swi3 (4 × 10⁸ PFU/ml) at 40, 50, 60, 70, and 80°C, and aliquots (100 μl) were taken at 20, 40, and 60 min. To evaluate the stability of phage swi3 at various pH values (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, and 13), phage suspensions were incubated in LB broth with different pH values at 37°C for 1, 2, and 3 h and then assayed by the double-layered agar method. All assays were performed in triplicate.

Phage genomic DNA was extracted using a Virus Genome Extract DNA/RNA Kit (Tiangen, Inc., Beijing, China) according to the manufacturer’s instructions. Then, the extracted genomic DNA was sent to Biozeron Company (Shanghai, China) for high-throughput sequencing on an Illumina HiSeq platform and assembled with ABySS, v2.0.2¹. Contigs were assembled using the de novo assembly algorithm Newbler, version 3.0, with default parameters (Bao et al., 2014). Potential open reading frames were predicted and annotated using RAST² and GeneMark³ (Besemer and Borodovsky, 2005; Aziz et al., 2008). Additionally, the genome sequence, the terminase large subunit, and the tail-associated protein amino acid sequence of phage swi3 were compared separately in the GenBank database of NCBI, and all homologous phages were selected to construct the phylogenetic tree of phage swi3 using the neighbor-joining method with default parameters in MEGAX software. The genomes of phage swi3, Salmonella phage BP63 (KM366099.1), Salmonella phage LSE7621 (MK568062.1), Salmonella phage UPF_BP2 (NC_048649.1), and Salmonella phage vB_SenM_PA13076 (MF740800.1) were compared and performed using Mauve software with a default parameter (Darling et al., 2004).

To test the antibacterial activity of phage swi3, the 50% lethal dose (LD50) of E. coli K88 was first determined. In brief, 28 5-week-old female BALB/c mice were randomly divided into four groups. Each group was treated orally with different doses of E. coli K88 (10⁶, 10⁷, 10⁸, and 10⁹ CFU). The LD50 was determined based on the mortality of mice and used in subsequent challenge tests. To test the protective effect of phage swi3 on mice, 42 5-week-old female BALB/c mice were randomly divided into six groups (a–f). Groups a–d were inoculated orally with 10⁸ CFU of E. coli K88. Then, groups a–c were inoculated intraperitoneally with a single dose (10⁵, 10⁶, and 10⁷ PFU, respectively) of phage swi3 at 2 h after the bacterial challenge. Control group d was inoculated with an equivalent volume of LB instead of phage swi3, control group e was treated with 10⁷ PFU of phage swi3 without E. coli K88 challenge, and control group f only took an equal volume of LB orally. Then, the mice were observed daily for 5 days to record the clinical signs and cumulative mortality. At the same time, blood and feces were collected every 2 h on the first day and every 24 h after the first day to detect the metabolic dynamics of bacteria and phage swi3 in the mice.

The sequence data of swi3 were deposited at GenBank under accession number MT768059.

A novel virulent phage was isolated from E. coli K88 and observed by transmission electron microscopy. Phage swi3 has a regular icosahedral head (80 nm in diameter) and a contractile sheathed tail 120 nm in length (Figure 1). Thus, phage swi3 belonged to the family Myoviridae, order Caudovirales, following the current guidelines of the International Committee on Taxonomy of Viruses (ICTV). According to the novel universal system of bacteriophage naming, the suggested full name of phage swi3 will be vB_EcoM_swi3 (Lavigne et al., 2009).

The swi3 host range was determined using 65 E. coli strains and 72 S. enteritidis strains (Supplementary Table 1). Interestingly, phage swi3 showed characteristics of a wide range across species, not only having a certain lytic ability for six of the 65 E. coli strains but also having a lytic ability for 39 of the 72 S. enteritidis strains.

The biological characteristics of phage swi3 were measured by the double-layer plate method (Zhang et al., 2013). When the MOI was 1, phage swi3 had the highest titer of 6.4 × 10⁸ PFU/ml after proliferation. The one-step growth analysis showed that the incubation period of swi3 was approximately 25 min, after which there was a rapid increase in the number of released viral particles. It took about 75 min for swi3 to reach the growth plateau phase with a burst size of approximately 25 PFU/infected cells (Figure 2A). In addition, phage swi3 remained stable in the pH range of 6–8 (Figure 2B) and at a temperature less than 50°C for at least 1 h (Figure 2C).

The genome of phage swi3 was sequenced and analyzed. The general characteristics of the genome include a total of 52,782 bp with an overall GC content of 46.02%. Seventy-two orfs were predicted, 39 of which were positive-stranded, while the others were negative-stranded. Only 15 of 72 orfs were annotated as functional genes, including six structural-related genes (orf 19, orf 20, orf 21, orf 22, orf 27, and orf 33), seven genes associated with transcription and replication (orf 14, orf 18, orf 52, orf 53, orf 57, orf 59, and orf 72), and two lysis-related genes (orf 44 and orf 45). A detailed phage swi3 genome annotation showed that the orfs related to transcription and replication were mainly concentrated in the downstream part of the whole genome, while the structurally related orfs were mainly concentrated in the upstream part of the whole genome sequence (Figure 3).

The phylogenetic tree of phage swi3 was constructed based on the genomic sequence, the terminase large subunit (encoded by orf 19) sequence, the tail-associated protein (encoded by orf 27), and the tail tape measure protein (encoded by orf 33), using the neighbor-joining method with the default parameters in MEGAX software.

Based on the terminase large subunit, most comparable phages had less than 60% homology, a total of 10 phages in NCBI showed high similarities (>94%), and all of them belonged to Salmonella phages. Only Salmonella virus BP63 (KM366099.1) was included in ICTV, with 98% coverage rate and 97.73% identity. The phylogenetic tree showed that 11 phages belonged to two groups, phage swi3 and eight more phages belonged to the Rosemountvirus genus, and the other two phages belonged to the Loughboroughvirus genus (Supplementary Figure 1). Based on the whole-genome sequence, the phylogenetic tree was highly consistent with that of the terminase large subunit; all homologous phages were divided into two groups, and phage swi3 belonged to the Rosemountvirus genus. Additionally, all the homologous phages were Salmonella phages (Supplementary Figure 2). The genomic comparison results showed that no rearrangement or inversion occurred in the phage swi3 genome (Supplementary Figure 3).

Only two tail-related proteins were annotated among the known orfs: the tail-associated protein (encoded by orf 27) and the tail tape measure protein (encoded by orf 33). Their phylogenetic trees were constructed. Based on the tail-associated protein, only four comparable phages were found in the NCBI database with a high homology (>82.7%), and all of them were Salmonella phages. Other phages in the database showed less than 45% homology (Supplementary Figure 4). Based on the tail tape measurement protein, all homologous phages also belonged to Salmonella phages with more than 70% similarity (Supplementary Figure 5).

The LD50 of E. coli K88 on mice was determined as 10⁸ CFU, and the protective effects of phage swi3 were tested in a mouse model challenged with E. coli K88. There were obvious protective effects of phage swi3 in the bacteria-challenged mouse model (p < 0.05). After the challenge with E. coli K88, all mice in control group d died within 1 day; their lungs, liver, spleen, and kidneys showed varying degrees of bleeding and swelling, and E. coli K88 was isolated from these diseased organs. All mice in control groups f and e had good health until the end of the experiment, and no organ lesions were found (Supplementary Figure 6). In groups a–c, different doses of phage swi3 showed a good protective effect on E. coli K88-challenged mice. Compared with control group d, except for one mouse in group a that died on the second day, all mice survived, and no organ lesions were found (Figure 4A). The results indicated that a single intraperitoneal injection of swi3 (10⁶ PFU) at 2 h after oral challenge with E. coli K88 could sufficiently protect all mice without toxic side effects.

In addition, the dynamics of bacteria and phages in the blood and feces were measured. E. coli K88 entered the blood and reached its peak (10⁸ CFU/ml) 1.5 h after inoculation. Phage swi3 was intraperitoneally injected into the mice, detected in the blood in a short time, and then cleared E. coli K88 within 8.5–51 h in the blood; no bacteria or phage could be detected after 51 h (Figures 4B,C). Meanwhile, the dynamics of phage swi3 in feces were also measured. With the increase in phage inoculation dose, the phage detection time in feces was prolonged, ranging from 23 to 99 h (Figure 4D).