Male BABL/C mice (23–25 g) aged 6–8 weeks were purchased and cultured in the experimental animal center of Yangzhou University. All animal experiments were performed in strict accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals approved by the State Council of the People’s Republic of China and the Animal Welfare and Research Ethics Committee at Yangzhou University.

Klebsiella oxytoca AD3 was isolated from clinical samples provided by Nanjing Stomatological Hospital in January 2021 and cultured in lysogeny broth (LB) on an orbital shaker at 37°C and 220 rpm. The 16S rRNA gene was amplified using primers 27F and 1492R for identification. K1, K2, K5, K20, and K54 primers were used for serotyping. RmpA, allS, ybtA, iucB, iroNB, fimH, ureA, uge, wabG, and wcaG were determined for virulence factors (Supplementary Table 1). Multi-locus sequence typing (MLST) was performed using the primers provided by the Institut Pasteur MLST.¹ Antibiotic resistance of bacteria was determined using the standard Kirby-Bauer disk diffusion method, and the interpretation of the data was performed using the standardized protocol of the National Committee for Clinical Laboratory Standards and designated as R (resistant), I (intermediate sensitive), and S (sensitive) (Supplementary Table 2).

Phage vB_Kox_ZX8 was isolated from fecal samples collected from the Nanjing Stomatological Hospital in January 2021. Briefly, 1 g of fecal sample was mixed with 1 mL of PBS, and the solution was mixed well, centrifuged at 8228 × g for 5 min, and passed through a 0.22-μm filter. The supernatant filtrate was added to 5 mL of K. oxytoca AD3 cultures at log phase (OD₆₀₀ = 0.6), and the solution was cultured at 37°C and 220 rpm for 2–4 h. The lysate was centrifuged at 18,514 × g for 2 min, and the phage in the supernatant was purified using the double plate method. Briefly, 300 μL of bacterial and 100 μL of 10-fold dilution series of phage were mixed and added to 5 mL of top LB soft agar, followed by pouring onto an LB agar plate and culturing at 37°C overnight. The next day, a single plaque was picked from the plate, inoculated into 5 mL of a log phase host, and cultured at 37°C and 220 rpm for 2–4 h. The cultures were centrifuged at 18,514 × g for 2 min, and the lysates were passed through a 0.22-μm filter. The above steps were repeated at least three times until a single phage was obtained.

The phage (MOI = 0.01) was added to 200 mL of log phase host and cultured at 37°C and 220 rpm for 2–4 h. The culture was centrifuged at 8228 × g for 10 min, and DNase I and RNase A (1 μg/mL) were added to the lysates and incubated at 37°C for 30 min. NaCl (1 M) was added to the supernatants and incubated in an ice bath for 1 h. After centrifugation at 8228 × g for 10 min, 10% (w/v) polyethylene glycol 8000 (PEG 8000) was added to the lysates and precipitated overnight at 4°C. After centrifugation at 18,514 × g for 20 min, the precipitate was resuspended in PBS and dissolved completely. An equal volume of chloroform was added to the above solution, and centrifuged at 4629 × g for 15 min, followed by transferring the upper water phase to a new centrifuge tube, and the above steps were repeated three times. The collected phages were concentrated in 100 KDa ultrafiltration tubes (Millipore, United States). The phage suspension was then stored at 4°C.

The morphology of the phage was observed using a transmission electron microscope (TEM) HT7800 (Hitachi, Japan). Before observation, the phage suspension was incubated on carbon grids (200 mesh) for 10 min, stained with 2% phosphotungstic acid for 3 min, and dried for 30 min.

The host spectrum of the isolated phages was determined using the spot-test assay on 25 bacterial strains isolated from patient fecal samples. The antibacterial activity was assayed against 10 K. oxytoca, 10 K. pneumoniae, 5 Escherichia coli, and 5 Proteus mirabilis strains (Supplementary Table 3). Briefly, 5 μL of phage (10⁸ PFU/mL) was spotted onto fresh bacterial lawns and incubated at 37°C for 6–8 h. The experiment was repeated three times.

The one-step growth curve of the phage was measured to calculate its incubation period, outbreak period, and platform period. The phage suspension was added to a fresh host culture (10⁸ CFU/mL) at a multiplicity of infection (MOI) of 0.01. The mixture was incubated at 37°C for 10 min and centrifuged at 18,500 × g for 1 min to remove the non-absorbed phages. The precipitate was re-suspended in LB broth (time zero) and cultured at 37°C and 220 rpm. Two parallel samples were taken every 10 min and centrifuged at 18,500 × g for 1 min. The phage titer in the supernatant was determined using the double agar plate method. Burst size was computed as the ratio of the final count of released phage particles to the initial count of infected bacterial cells during the latent period (Ciacci et al., 2018). The experiment was repeated three times.

Thermal stability test of phage: Phage suspension (300 μL, 10⁸ PFU/mL) was incubated at 4, 40, 50, 60, 70, and 80°C for 20–60 min. The phage titer was determined using the double-layer plate method.

pH stability test of phage: Phage suspension (300 μL, 10⁸ PFU/mL) was incubated at 37°C for 60 min under different pH values (2–11). The phage titer was determined using the double-layer plate method. The experiment was repeated three times.

The bactericidal effect of phages in vitro was evaluated by measuring the effect of phage on the number of bacteria in the culture tube. Different doses of phage vB_Kox_ZX8 (MOI = 10, 1, 0.1, 0.01, and 0.001) were added to log phase K. oxytoca AD3 (5 × 10⁸ CFU/mL) and cultured at 37°C and 220 rpm. The OD₆₀₀ of the culture was determined every 10 min using a Smart Microplate Reader (Tecan infinite M200 Pro, Switzerland). The experiment was repeated three times.

Phage DNA was extracted using a Virus Genomic DNA/RNA Extraction Kit (Tiangen Biotechnology Co., Ltd., China). Whole-genome sequencing of the phage was completed by Shanghai Bioengineering Co., Ltd. The phage DNA fragments with a length of approximately 500 bp were randomly interrupted by a Covaris ultrasonic crusher (Covaris, United States), and then purified using hieff NGS DNA selection beads (Yeasen Biotechnology Co., Ltd. China). The sequencing library was constructed using the NEB Next Ultra DNA Library Prep Kit for Illumina (NEB, United States), including terminal repair, adaptor ligation, DNA purification, and library amplification. The DNA library was sequenced on the Illumina hiseqpe150 sequencing platform after passing the quality test. The original sequencing data were filtered first and assembled using new blew 3.0 software.

tRNAs were predicted using tRNAscan-SE2.0² (Lowe and Chan, 2016). The virulence factors and drug resistance of phage genome were compared with Virulence Factors of Pathogenic Bacteria³ and The Comprehensive Antibiotic Resistance Database.⁴

The open reading frame was annotated using the RAST annotation server web (Aziz et al., 2008). Automatic annotation was manually reviewed using the BLASTp algorithm against RefSeq proteins deposited in the GenBank database. Trans-membrane helical domains and signal sequences were analyzed using Phobius (Käll et al., 2007). Visualization of the phage genome was performed using Easyfig (Sullivan et al., 2011). Phylogenetic analysis was performed using the phage terminase large subunit and major capsid protein of the subfamily Studiervirinae reported by the International Committee on Taxonomy of Viruses classification. The protein alignments were obtained using ClustalW, and the phylogenetic trees were generated using the maximum likelihood method with a bootstrap of 1000 in MERGA 6.0.

The Toxin Eraser Endotoxin Removal Kit (Genscript Biotechnology Co., Ltd., China) was used to remove endotoxins from the phage. The purified phage suspension was quantified using the ToxinSensor Single Test Kit with Standard (Genscript Biotechnology Co., Ltd, China). Endotoxin concentrations below 0.005–0.01 endotoxin units (EU)/PFU were deemed safe for injection in accordance with published data (Gangwar et al., 2021).

Each mouse in the experimental group was intraperitoneally (IP) injected with 100 μL of purified phage vB_Kox_ZX8 (5 × 10⁹ and 5 × 10⁷ PFU), and 100 μL of PBS was injected as a control, with five mice in each group. The weight and survival rates of the mice in each group were observed. The titers of phages in blood and tissues were determined using the double-layer plate method. The levels of tumor necrosis factor-α (TNF-α), interleukin 6 (IL-6), and interleukin 10 (IL-10) in organs were determined using ELISA.

Klebsiella oxytoca AD3 bacteria were cultured overnight and then centrifuged at 8228 × g for 2 min, followed by washing twice with PBS. Mice (six mice in each group) in the experimental group were IP injected with 100 μL of different doses of bacterial suspension (5 × 10⁸, 10⁸, 5 × 10⁷, 10⁷, 5 × 10⁶, and 10⁶ CFU) to determine the minimum lethal dose (MLD). MLD is the dose of bacteria that could cause all mice to die within 7 days. PBS (100 μL) was injected into the negative control group. The weight and survival rate of the mice in each group were recorded. Each mouse was IP injected with 100 μL of bacteria at the MLD to prepare the bacteremia mouse.

Each mouse was IP injected with 100 μL of K. oxytoca AD3 at the MLD. After 1 h, different doses of phage vB_Kox_ZX8 (5 × 10⁷, 5 × 10⁶, 5 × 10⁵ PFU) were IP injected into the experimental group, and 100 μL of PBS was injected as a control, with six mice in each group. The weight and survival rates of the mice in each group were recorded. The titers of K. oxytoca AD3 and phage vB_Kox_ZX8 in blood and organs were determined using bacterial monoclonal plate culture and double plate experiments, respectively.

After 48 h, the animals were euthanized, and tissues were collected. The organs were fixed with 4% paraformaldehyde for 24 h, dehydrated in alcohol, paraffin-embedded, sectioned, xylene dewaxed, hematoxylin and eosin stained, dehydrated in alcohol, and covered with glass. Pathological changes in the organs of the mice were observed.

Blood and various organs, including the eyeball, blood, heart, lungs, thymus, right upper liver, spleen, right kidney, and small intestine (2–4 cm downward from the pylorus) were collected immediately after the mice were euthanized. Blood was diluted with a PBS gradient immediately after collection. Each organ, including the small intestine, was homogenized with 1 mL of PBS and then diluted with a PBS gradient. The titers of bacteria and phage in blood and organ homogenates were determined using plate counting and the double plate method, respectively.

All data were analyzed using GraphPad Prism 6. Statistical analysis of significance was also undertaken using the GraphPad Prism program via unpaired t test with a p < 0.05 considered significant.

Klebsiella oxytoca AD3 was typed as ST367 and K1 serotype and encodes virulence factor genes rmpA, kfuBC, ybtA, fimH, uge, wabG, and wcaG. The strain is sensitive to most antibiotics but resistant to erythromycin, vancomycin, and tobramycin (Supplementary Table 2).

In this study, a phage named vB_Kox_ZX8 was isolated using K. oxytoca AD3. The phage formed small circular translucent plaques (diameter < 1 mm) on the lawns of the host (Figure 1A). Examination of phage morphology using TEM analysis showed that phage vB_Kox_ZX8 had an icosahedral head with a dimension of 53 ± 3.0 nm and a very short non-contractile tail (Figure 1B).

A total of 25 clinical isolates were used to evaluate the host range of vB_Kox_ZX8 (Supplementary Table 3). The results showed that phage vB_Kox_ZX8 only had lytic activity specific to K. oxytoca AD3. The narrow host spectrum of phages may be due to the limitations of the tested strains or the specificity of phage recognition sites.

A one-step growth curve was used to analyze the adsorption velocity, latency period, and burst size of the phage (Figure 2A). The results showed that phage vB_Kox_ZX8 had a latency period of 10 min, followed by a rise period of 60 min and a growth plateau of approximately 50 min. The burst size of vB_Kox_ZX8 was computed as 74 phage particles per infected bacterium.

The thermal stability test showed that the phage was relatively stable between 4 and 50°C. After incubation at 60°C for 20, 40, and 60 min, the phage titer (10⁸ PFU/mL) decreased to 2 × 10⁵, 5 × 10⁴, and 10³ PFU/mL, respectively. After incubation at 70°C, phage activity decreased to 2 × 10³ PFU/mL at 20 min and was completely lost at 40 min (Figure 2B). The pH stability test showed that phage activity was very stable at pH 3–11, and sharply lost at pH = 2 and pH = 12 (Figure 2C).

Different doses of phages (MOI = 10, 1, 0.1, 0.01, 0.001, respectively) were used to infect the host K. oxytoca AD3 in the log phase, and the OD₆₀₀ of the culture was determined using a Smart Microplate Reader to evaluate the killing effect of phages on the host. The results showed that the number of bacteria decreased gradually after the addition of phages, and the rate of decrease was proportional to the number of phages. Even if the MOI of phages was 0.001, the bacteria were almost cleared after 90 min (Figure 2D). This result showed that phage vB_Kox_ZX8 had a good in vivo anti-K. oxytoca AD3 effect.

The genome of phage vB_Kox_ZX8 (GenBank accession no. MZ424865) is a dsDNA that comprises 39,398 bp with a G+C content of 53.1%. tRNA was not predicted using tRNAscan-SE2.0. Virulence, toxin proteins, and drug-resistance genes were not detected. Whole-genome comparative analysis of phages showed that phage vB_Kox_ZX8 was most closely related to Klebsiella phage 2044-307w (95.05%), which is a member of the genus Przondovirus, subfamily Studiervirinae, family Autographiviridae.

The genome contains 46 coding sequences or open reading frames (ORFs) (Table 1). The ORFs of phage vB_Kox_ZX8 were divided into five modules: hypothetical protein, DNA metabolism, DNA packaging, phage structure, and phage lysis (Figure 3A). The main functional ORFs of phages are involved in DNA replication and degradation (ORF10, ORF11, ORF15, ORF17, ORF20, ORF21, ORF24, ORF26, ORF45, and ORF35); transcriptional factors (ORF13, ORF22, ORF30, and ORF31); phage capsid and scaffold (ORF3, ORF4, ORF5, and ORF6); phage tail (ORF1, ORF2, ORF7, ORF8, ORF41, ORF42, ORF43, and ORF44), phage assembly (ORF37 and ORF39), and phage lysis (ORF19, ORF38, and ORF40). ATG was proposed as a start codon for 41 ORFs, four ORFs used GTG, and one ORF used TTG. Signal peptides and transmembrane domains were found in seven ORFs: hypothetical protein ORF18, ORF29, ORF34, terminase large subunit ORF37, Rz-like lysis protein ORF38, holin class II ORF40, and internal virion protein B ORF44. The terminal large subunit and internal virion protein B are involved in the assembly of phage DNA. Holin forms pores on the inner membrane, and Rz-like lysis protein assists the lysozyme to enter the outer membrane from the inner membrane and cause complete cell lysis.

Phylogenetic analysis of the phage terminase large subunit and major capsid proteins showed that vB_Kox_ZX8 belonged to the genus Przondovirus, subfamily Studiervirinae, and family Autographiviridae (Figures 3B,C).

Compared with the control group, phage vB_Kox_ZX8 had no significant effect on the weight of the mice (Figure 4A). In the inflammatory response, phage caused an increase in pro-inflammatory factors (IL-6 and TNF-α) and a decrease in anti-inflammatory factors (IL-10) (Figures 4B–D). The levels of IL-6, TNF-α, and IL-10 in the serum, liver, and spleen began to change after 12 h or 24 h of phage injection, and the changes caused by the high dose of phage (5 × 10⁹ PFU) were more significant. The changes in inflammatory factors induced by phages showed a gradual upward trend within 48 h. At 48 h after administration of 5 × 10⁹ PFU phage to healthy mice, the IL-6 levels in the serum, liver, and spleen of mice were 1.5, 1.6, and 2.0 times that of the control group, respectively; the levels of TNF-α in the serum, liver, and spleen of mice were 1.8, 1.7, and 1.7 times that of the control group, respectively; the IL-10 content in the serum, liver, and spleen of mice was 0.8, 0.8, and 0.9 times that of the control group, respectively. The fluctuation of IL-6, TNF-α, and IL-10 levels in mice caused by phage were slight, which is not enough to cause obvious adverse reactions.

After intraperitoneal injection of 5 × 10⁷ PFU and 5 × 10⁹ PFU doses of phage vB_Kox_ZX8 to healthy mice, the phages disappeared in the blood after 3 and 5 h, respectively (Figure 5A). In mice injected IP with 5 × 10⁹ PFU, the phage titer in the blood reached the highest (6 × 10⁵ PFU/mL) at 30 min and lasted for 60 min, then dropped rapidly until it was completely undetectable. In mice injected IP with 5 × 10⁷ PFU, the phage titer in the blood reached the highest (2 × 10⁴ PFU/mL) at 15 min and lasted for 30 min, then dropped rapidly until it was completely undetectable. Phage vB_Kox_ZX8 had a higher titer and longer residence time in mouse organs because phages at 5 × 10⁷ PFU and 5 × 10⁹ PFU disappeared in organs within 24 and 48 h, respectively (Figures 5B,C). The titer of vB_Kox_ZX8 in the heart, thymus, lung, liver, spleen, kidney, and small intestine reached the highest levels within 1 h, and then gradually decreased. However, the titers of phages in the liver, spleen, kidney, and small intestine were significantly higher than those in the heart and lung after injection, which may be related to the injection method. At 12 h after 5 × 10⁷ PFU phage injection, the phage titer of the thymus was the highest, followed by the spleen. At 24 h after 5 × 10⁹ PFU phage injection, the phage titer of the spleen was the highest, followed by that of the thymus.

After the injection of K. oxytoca AD3, the mice were dispirited, and showed inverted hair, shivering, drowsiness, hunched back, weight loss, and death. Injection with a dose of 5 × 10⁶ CFU, which is regarded as the MLD, caused the death of all six mice (Figure 6A).

The weight of bacteremic mice rescued by phage vB_Kox_ZX8 increased gradually after 2 days (Figure 6B). The survival rate of mice rescued with 5 × 10⁷ PFU of phage reached 100%, and mice rescued with 5 × 10⁶ PFU and 5 × 10⁵ PFU of phage were 66% and 50% of survival rate, respectively (Figure 6C).

Therefore, 5 × 10⁷ PFU phages were selected to rescue bacteremia mice, and the therapeutic effect of the phage was evaluated by measuring the titer of bacteria and phage in mice. The results showed that the bacterial titers in the blood and organs of mice treated with phage decreased gradually and were cleared after 48 h, while the bacterial titers in the blood and organs of mice in the negative control group increased and killed the mice (Figures 7A,B). After the phage entered the body of the mouse, it was first reproduced through K. oxytoca AD3, then gradually decreased, and was completely metabolized after 48 h in the blood (Figure 7C). The changes in the number of phages in the organs seem to be related to the bacteria count, and they were cleared after 48 h (Figure 7D). In organs, the bacteria count was decreased by the phages in a short time (<6 h), followed a slight upward trend, and finally cleared after 48 h. Bacteria isolated from the organs and blood were still sensitive to phage vB_Kox_ZX8 at 12, 24, and 36 h post-infection, indicating that bacteria and phages co-exist between 12 and 36 h in mice. However, the phage-resistant strains were isolated after 12 h of coculture in vitro. The reason for this phenomenon may be that phages in mice have less selective pressure on bacteria, and a higher phage therapeutic dose should be considered.

After 48 h of infection with K. oxytoca AD3, the levels of IL-6 and TNF-α in the thymus, lung, spleen, kidney, and liver of mice increased significantly, while IL-10 decreased significantly, which was effectively alleviated via phage therapy (Figures 8A–C). K. oxytoca AD3 had the strongest pathogenicity in the small intestine, with obvious pathological changes, hyperemia, cell necrosis, and fuzzy structure (Figure 9A). The number of inflammatory cells in the thymus and spleen decreased significantly and inflammatory cell infiltration was observed in the liver (Figures 9B–D). No obvious pathological changes were observed in the kidneys and lungs (Figures 9E,F). Notably, the pathological changes in various organs in mice rescued by phage showed significant improvement.