The Y. enterocolitica strains used in this study are listed in Supplementary Table S1. The reference strains ATCC 23715 (serotype O8) and CICC 21565 (serotype O3) were purchased from American Type Culture Collection (Manassas, VA, United States) and China Center of Industrial Culture Collection (Beijing, China), respectively. The 49 clinical isolates of Y. enterocolitica were kindly provided by the Yunnan Institute of Endemic Disease Control. After sequencing the 16S rRNA gene with the universal primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GGTTACCTTGTTACGACTT-3′) (Hao et al., 2016), identification of O3, O5, O8, and O9 serotypes was carried out with slide agglutination test by using diagnosis monoclonal antibodies against Y. enterocolitica (Wantai Biotech Co., Ltd.; Zhengzhou, China). Strains were cultured in LB (lysogeny broth, containing 5.0 g yeast extract, 10.0 g peptone, and 10.0 g NaCl per 1000 mL) medium (Bertani, 2004) or on plates with selective diagnostic CIN agar (Cefsulodin-Irgasan-Novobiocin; Oxoid, United Kingdom) at 37°C for 18–24 h (Orquera et al., 2012). For long-term storage, bacterial cultures were stored in glycerol [3:1 (v/v)] at −80°C.

The bacteriophage was isolated, purified and concentrated according to a previously described method (Gu et al., 2011). ATCC 23715 was used as a host to isolate the phage. Sewage samples were collected from a sewage treatment plant of Changchun City, Jilin Province, China. The phage was purified by the double-layer agar plate method for more than three times. To determine the host range of the phage, spot test and double-layer agar plate methods were used as described previously (Leon-Velarde et al., 2016).

The phage was concentrated as previously described (Synnott et al., 2009) and was spotted on a hydrophilic. Formvar-carbon-coated copper mesh for 2 min and negatively stained with 2% phosphotungstic acid for 30 s, after which the liquid was blotted with filter paper. Phage morphologies were observed under a TEM (H-7650; Hitachi, Japan) at 80 kV.

The multiplicity of infection (MOI) is defined as the ratio of the number of phages to the number of host bacteria (Czajkowski et al., 2015). The host strain was cultured in LB to a logarithmic growth phase (OD₆₀₀ nm = 0.6). The phage and host bacteria (10⁹ CFU/mL) were added to LB medium at different MOIs (10^–7, 10^–6, 10^–5, 10^–4, 10^–3, 10^–2, 10^–1, or 1) and cultured for 6 h at 37°C with shaking at 180 rpm. Phage titration was measured by the double-layer plate method after the incubation.

The one-step growth curve experiment was conducted with some modifications as explained elsewhere (Xi et al., 2019). The phage was mixed with an ATCC 23715 suspension at an optimal MOI of 1. After centrifugation at 4°C for 5 min (10,000 × g), the mixture was suspended in 10 mL of fresh LB medium and incubated at 37°C with shaking (180 rpm). The samples at different time were subjected to phage titer determination.

The phage genome was extracted from the concentrated phage preparation using a viral genome extraction kit (Omega Bio-tek Inc., Norcross, GA, United States). The genome was submitted to BGI Biotechnology Co. Ltd., for genome-wide sequencing via the Illumina HiSeq platform. Potential open reading frames (ORFs) were predicted and analyzed by BLAST (NCBI) and GeneMark (Jeon et al., 2019), and gene function block diagrams were mapped using CLC Main Workbench version 7.7.3 software (CLC Bio-Qiagen, Aarhus, Denmark) (Ji et al., 2019).

Phage stability experiments in vitro were performed for thermal stability and pH according to the method described previously (Zhang et al., 2013). Phage suspensions were inoculated in SM buffer (pH range 1.0–13.0) for 1 h at 37°C. In addition, the thermal stability was tested by using a phage suspension (3.75 × 10⁶ PFU/mL) that was incubated at 4, 25, 37, 50, 60, 70, and 80°C for 1 h. Phage titer was determined by double-layer plate assay. After 12 h of fasting, BALB/c mice were euthanized by intravenous injection of Fatal Plus (pentobarbital sodium) (100 mg/kg). The intestinal contents from the stomach, duodenum, jejunum, ileum, cecum, and colon of the mice were harvested. The contents were homogenized in 1 mL of phosphate buffered saline (PBS, pH 7.4) and centrifuged at 4°C for 1 min (14,000 × g) to obtain the supernatants (Bosak et al., 2018). The stability of the phage was detected with supernatant of PBS aspirate of corresponding parts of intestinal tract, and the phage titer was measured after incubation at 37°C for 10, 30, 60, and 90 min.

Referring to the previous study (Cheng et al., 2017), early exponential cultures of ZTYSG 21 were infected with the phage at five different MOIs (PFU/CFU ratios of 10^–6, 10^–4, 10^–2, and 1), and the mixture was incubated in tubes with shaking at 180 rpm at 37°C. In addition, the uninfected phage group was set as a negative control group. The colony count of the cultures was measured every hour.

Female BALB/C mice aged 6–8 weeks (18–20 g) were provided by the Experimental Animal Center of Jilin University, Changchun, China. Experimental animals were placed under specific pathogen-free conditions and provided sterile materials (cages, bedding, water, and food). All animal experiments strictly abided by the National Guidelines for Experimental Animal Welfare (Ministry of Science and Technology of China, 2006) and conducted according to the experimental practices and standards approved by the Animal Welfare and Research Ethics Committee at Jilin University (permit number: 20181227087). All efforts were made to minimize suffering.

To test whether phages could eliminate pathogenic bacteria from the intestinal tract of infected asymptomatic animals, a murine chronic enteritis model was established by previously described methods (Damasko et al., 2005). After being starved for 12 h, the mice were orally administered 0.1 mL ZTYSG 21 by using a feeding syringe. Prior to use, exponential-phase bacteria ZTYSG 21 were washed with PBS and resuspended in PBS to 2.05 × 10⁹ CFU/mL. Phage treatment was orally administered at 6 h post infection. BALB/c mice were randomized into three experimental groups: (i) Bacteria-PBS group of 15 infected mice treated with 0.1 mL of PBS at 6 h post infection; (ii) Bacteria-Phage group of 15 infected mice treated with phage (0.1 mL, 1.95 × 10⁹ PFU/mL) in a single dose at 6 h post infection; and (iii) PBS-Control group of 15 healthy mice treated with 0.1 mL of PBS as the controls. The experimental animals were monitored for clinical manifestation for 6 days. The experiments were repeated three times.

Colon and cecum tissues were collected under sterile conditions for microbiological analysis at the indicated times. Briefly, after 6 h of challenge, three mice were randomly selected from each group every 3 h for euthanasia by intravenous injection of Fatal Plus (pentobarbital sodium) (100 mg/kg) until 24 h post challenge. After 24 h, three mice were randomly selected from each group every 24 h for euthanasia, and the bacterial load was measured until 6 days after the challenge. The removed organs were homogenized in 1 mL of PBS buffer, then serially diluted with PBS, and plated on Y. enterocolitica selective medium (CIN) for counting the number of colonies forming units (CFU).

To determine the phage titer in the Bacteria-Phage group, the supernatants from each time point were centrifuged at 4°C for 15 min (12,000 × g) and filtered. The phage titer was determined by agar double-layer plate experiment (Bosak et al., 2018; Jeon et al., 2019).

To assess histological changes, the intestinal tissue samples harvested from euthanized mice were fixed in 10% zinc formalin and stained with hematoxylin and eosin (H&E). Representative tissue sections from each group were imaged. Tissues in cecum were scored blindly for inflammation and pathology using a 13-point system as described before (Barthel et al., 2003). The scoring included submucosal edema (0–3), polymorphonuclear granulocytes (PMN) infiltration into the lamina propria (0–4), goblet cells (0–3), and epithelial integrity (0–3).

Proinflammatory cytokines (TNF-α, IL-6, and IL-1β) in cecum, colon, and spleen tissues were measured. Briefly, three mice were randomly selected from each group at 12, 24, 48, and 72 h after infection. The removed organs were homogenized in 1 mL of PBS buffer, serially diluted with PBS and detected by an enzyme-linked immunosorbent assay kit (BioLegend, San Diego, CA, United States) (Cai et al., 2013).

SPSS version 13.0 software (SPSS, Inc., Chicago, IL, United States) was utilized for the statistical analysis. All experimental data were analyzed by one-way analysis of variance (ANOVA). Error bars represent standard deviation of the mean. P < 0.05 was considered statistically significant. ^∗P < 0.05; ^∗∗P < 0.01; and ^∗∗∗P < 0.001.

A total of 49 clinically isolated strains were identified as Y. enterocolitica by 16S rRNA sequencing. BLAST analysis showed that the 49 isolates were more than 99% identity to Y. enterocolitica strains identified and confirmed in the GenBank database (such as KC776767.1 and LR134492.1). The dominant bacterial serotype was O3 (40%, 18/49), followed by O8 (29%, 13/49), O5 (18%, 8/49), O9 (11%, 5/45), and NT (11%, 5/49) (Supplementary Table S1).

The phage isolated from the sewage was purified by a double-layer agar plate method and designated Yersinia phage X1 (Viruses; dsDNA viruses, no RNA stage; Caudovirales; and Myoviridae). Plaques of phage X1 were 5–7 mm in diameter surrounded by transparent haloes in the periphery that expanded with time (Figure 1A). As determined by TEM (Figure 1B), phage X1 had an icosahedral head of 70 ± 3 nm and a contracted tail of 90 ± 3 nm in length and belonged to Myoviridae. The host range analysis demonstrated that phage X1 possessed a broad host range and infected 27 (27/51) strains of Y. enterocolitica (2 standard strains and 25 clinical isolates), which included pathogens of the following various serotypes: O3 (14/18), O5 (3/8), O8 (6/13), and NT (4/5) (Supplementary Table S1). When the MOI was 10^–6, the phage titer reached the highest level, approximately 2.75 × 10⁸ PFU/mL (Figure 1C). The one-step growth curve results (MOI = 1) indicated that the latent period of phage X1 infection was 20 min. The burst size was approximately 290 particles/infected cell (Figure 1D).

Whole genome sequencing of phage X1 is available in GenBank (MN617773). The genome of phage X1 was 48,948 bp in length, with a G + C content of 47.9%. It was found to have 88 putative ORFs, which included 21 proteins of known putative function and 67 hypothetical proteins (Figure 2). ORFs 10–23 and 28–39 were observed to be transcribed on the positive strand and the remaining ORFs were transcribed on the negative stranded. The predicted functional proteins were divided into five modules according to their functions: hypothetical function, structural composition, DNA packaging, host lysis, and nucleic acid metabolism and replication. The host lysis module was the smallest, with only ORF 9 encoding endolysin. No genes related to drug resistance, lysogeny or bacterial virulence were found in the predicted functional genes of phage, at least based on the limited studies.

BLAST analysis showed that the whole genome sequence of phage X1 shows 99.50% sequence identity and 99% genomic coverage with PY100. However, there are differences between X1 and PY100. The size of PY100 genome is 50,291 bp which contains 93 putative ORFs. While, the genome of phage X1 is 48,948 bp and it contains 88 putative ORFs. Though a large part of the ORFs in phage X1 and PY100 show high identity with each other, 10 ORFs in phage X1 shows low query cover and identity with the corresponding putative ORFs that derived from PY100 (as seen from the Supplementary Table S2). It is worth noting that, comparing with PY 100, phage X1 lacks a head protein and an HNH endonuclease which was encoded by ORF58 and ORF91 of PY100, respectively. And four other hypothesis ORFs of PY100 were not found in X1. Additionally, the F0F1 ATP synthase subunit A which encoded by ORF28 of X1 was not found in PY100.

The lytic capability of the phage was stable at pH 4–11 for 1h, and almost no loss of activity was observed (Figure 3A). In addition, the sensitivity of phage X1 to various temperatures (4–80°C) is represented in Figure 3B. The activity of phage X1 was stable between 4 and 60°C, although phage activity could be maintained after incubation at 70°C for 1 h. Briefly, we found that phage X1 was highly stable over a wide pH and temperature ranges. Similarly, no significant loss of phage activity was observed when phage X1 was incubated with the intestinal contents at 37°C for 60 min, indicating that these phage particles remained stable under gastrointestinal conditions (Figure 3C).

The activity of phage X1 inhibiting ZTYSG 21 growth in vitro is shown in Figure 4. As we expected, colony forming unit number increased continuously in the negative control. In contrast, all phage-infected groups dramatically inhibited bacterial growth, although the antibacterial effect was slightly different under different MOIs. With higher MOI values, the bacterium was more sensitive to phage infection. After 4 h, the bacterial counts declined to approximately 10³–10⁴ CFU/mL and remained stable compared to 10⁸ PFU/mL in the negative control group.

The mice challenged with ZTYSG 21 did not die of infection. Therefore, the bacterial loads in the murine intestinal tract were measured. A single oral administration of phage X1 at 6 h post infection was sufficient to eliminate Y. enterocolitica in 33.3% of mice (15/45). The Bacterial-PBS group had a median bacterial load in the cecum and colon of 3.84 × 10⁷ CFU/g and 3.79 × 10⁷ CFU/g, respectively, at 18 h (Figure 5A and Supplementary Figure S1A). The treatment with phage X1 significantly decreased the bacterial load to 3.13 × 10³ CFU/g and 2.15 × 10³ CFU/g, respectively. After 24 h of infection, bacterial counts in the tissues decreased to approximately 10² CFU/g (the untreated mice were approximately 10⁴CFU/g). No colonies were found in the intestine of PBS group mice. Phage X1 had transient increase after oral administration with a slight uptick of the PFU value in the cecum (Figure 5B) and colon (Supplementary Figure S1B). The phage was completely cleared after 48 h of infection. It is worth noting that Y. enterocolitica also remained in mice, while phage X1 was cleared by the host immune system. Ten colony-forming units isolated from the germ-carrying mice in the Bacteria-Phage group were randomly selected and propagated. Interestingly, the isolated Y. enterocolitica strains were lysed by phage X1 suspension in vitro.

At necropsy, gross and histopathologic lesions in cecum from the different groups were examined. At 48 h post-infection, Y. enterocolitica infection induced cecum shortening and increased loss of cecal weight, accompanied by reduced colon size (Figures 6A,C). Microscopically, mice in the Bacteria-PBS group displayed greater cecal histopathology scores than mice in the Bacteria-Phage group (Figure 6D), represented by a loss of goblet cells, mucosal ulcerations, and odema in the submucosa (Figure 6B) and PMN were also observed in the lamina propria of the Bacteria-PBS group mice. In contrast, no significant lesions were observed in the phage-treated mice. The pathological changes were remarkably alleviated in response to X1 treatment.

We also measured the inflammation status in the animals. The levels of the proinflammatory cytokines TNF-α, IL-6, and IL-1β in the cecum (Figures 6E–G), spleen (Supplementary Figure S2A) and colon (Supplementary Figure S2B) of the Bacteria-Phage group were close to the level of healthy mice, whereas they were significantly lower than those of the Bacteria-PBS group (P < 0.05). Taken together, these results indicated that phage X1 could mitigate injury in the cecum against Y. enterocolitica infection in mice.