Animal experiments performed in this study strictly followed the national guidelines for experimental animal welfare announced by Ministry of Science and Technology of People’s Republic of China in 2006 (Guiding Opinions on Kindly Treating Laboratory Animals¹) and were approved by the Animal Welfare and Research Ethics Committee at Qingdao Agricultural University, Shandong, China.

The bacterial strains and plasmids used in this study are listed in Table 1. All P. aeruginosa strains were isolated from minks that died of hemorrhagic pneumonia in Shandong, China. Bacterial strains used for determining antibacterial spectrum were identified by 16S rRNA gene sequencing. In addition, antibiotic susceptibility of P. aeruginosa isolates was measured by the Kirby–Bauer disk diffusion method (Forozsh et al., 2012). Bacteria were cultured in Luria Bertani (LB) broth or LB agar (Hopebiol Biotech) at 37°C. The vector pCold TF (N-Trigger factor; Amp^r; His-Tag, Takara) was used for cloning and expression of the lysin gene in Escherichia coli BL21 (DE3). The plasmid pCANTAB 5E was used for display and expression of the holin gene in M13 phages.

Phages were isolated from sewage of a mink farm using a P. aeruginosa strain (L64) as the host by the double-layer agar method (Lu et al., 2017). The purification and concentration of phage ASP23 were conducted as described previously (Shi et al., 2020). The purified phages were deposited on copper grids and negatively stained with phosphotungstic acid (2% w/v) for 10 min. After drying, the morphology of single phage particles was observed using an HT7700 transmission electron microscope (TEM, Hitachi, Japan) at 80 kV (Zhang et al., 2013).

The one-step growth experiment of phage ASP23 was performed as described previously reported (Shi et al., 2020). The mixtures of phage ASP23 and P. aeruginosa strain L64 were collected at different time points, and phage titers were determined immediately using the double-layer agar method. In addition, the host range, thermal stability and pH sensitivity of phage ASP23 were determined as described previously with some modifications (Shi et al., 2020).

Minks (American mink, n = 30, 40–50 days old, average weight = 600 g) were divided into two groups (15 minks per group), including the control group and treatment group. To determine the therapeutic effect of phage ASP23, minks were challenged with P. aeruginosa strain L64 (4 × 10⁸ CFU/mink) by intraperitoneal injection. Two hours after challenge, minks in the treatment group received phages by oral gavage (10¹⁰ pfu/mink). At the same time, minks in the control group received an equal volume of phosphate buffered saline (PBS). Each group contained 15 minks. At different time points (2 h, 5 h, 9 h, 13 h, and 17 h), 3 minks in each group were humanely euthanized with carbon dioxide (Gu et al., 2016). Tissue samples (liver and lung) were collected, weighed, and homogenized in 2 mL sterilized PBS. Blood samples were collected by venipuncture and stored in tubes containing EDTA. Bacterial counts in blood and tissue homogenates were measured by plating serial dilutions on LB agar plates.

Bacteriophage DNA was prepared using a DNA Viral Genome Extraction Kit (Solarbio) after concentration. The genomic DNA was used to construct a 600-bp insert library using a NEBNext® Ultra™ II DNA Library Prep Kit for Illumina according to the manufacturer’s instruction. The genomic DNA of ASP23 was subjected to high-throughput sequencing using an Illumina HiSeq 2,500 sequencer (San Diego, United States). The complete genome sequence was assembled using CLC Bio (Aarhus, Denmark). Putative open reading frames (ORFs) were predicted using GeneMarkS² and RAST³ (Besemer et al., 2001; Aziz et al., 2008). The comparative genomic analysis was performed using BRIG (Alikhan et al., 2011). All predicted ORFs were annotated using BLASTp, and the protein domain analysis was conducted using Pfam.⁴ Putative tRNA genes were identified using tRNAscan-SE. The structural and functional features of putative proteins were analyzed using TransMembrane prediction using Hidden Markov Models⁵ and PredictProtein.⁶ The highly conserved amino acid sequences were used to construct neighbor-joining phylogenetic trees using MEGA 7.

The deduced amino acid sequence of LysASP was analyzed using BLASTp, and the sequence alignment was generated using ESPript 3.0.⁷ The sequences similar to LysASP were searched in the Protein Data Bank (PDB), and the structure of LysASP was predicted using Phyre2.⁸

The lysin gene was amplified from ASP23 phage genomic DNA using polymerase chain reaction (PCR) with a pair of specific primers containing NdeI and EcoRI restriction sites. The primers used are listed in Table 1. The PCR product was purified, digested with NdeI and EcoRI (Takara), and then cloned into the pCold TF expression vector with an N-terminal His6 tag to generate pCold TF-LysASP. The ligation product was transformed into competent cells (E. coli strain DH5α, Takara) using the heat shock method (Froger and Hall, 2007). Restriction enzyme digestions and sequencing were used to confirm the cloned fragment. Then the recombinant plasmid was transformed into E. coli BL21 (DE3).

The transformed E. coli BL21 cells were grown in LB medium containing ampicillin (50 μg/mL, Solarbio) until the OD₆₀₀ reached 0.6 ~ 0.8. The mixed culture was induced with 0.4 mmol/L isopropyl-β-D-thiogalactopyranoside (IPTG) at 16°C for 12 h. Cells were harvested by centrifugation (10,000 × g, 15 min) and resuspended in phosphate-buffered solution (PBS, Solarbio), followed by sonication on ice (22 kHz, 45 W, 3 s work, 2 s pause). After centrifugation and filtration, His-tagged proteins were purified using a His-tag Protein Purification kit (Beyotime Biotech, Shanghai, China). The purified proteins were dialyzed against PBS and separated by 12% SDS-PAGE with a molecular weight marker (15–130 kDa, Solarbio). The protein concentration was determined using a BCA Protein Assay Kit (Solarbio, Beijing, China).

Pseudomonas aeruginosa L64 was grown in LB broth to the exponential phase, and 1 mL of culture was pretreated with 2 mM EDTA for 5 min. After centrifugation (10,000 × g, 10 min), bacterial cells were washed twice and resuspended in PBS (pH = 7.4). Aliquots (100 μL) of bacterial suspension were added into sterile 96-well plates containing 100 μL of purified LysASP (0.6 mg/mL) per well, followed by incubation at 37°C. The OD₆₀₀ values were measured by a microplate reader at different time points. Simultaneously, the EDTA-pretreated bacterial suspension was added in PBS without LysASP and used as negative control. In the end, the colony-forming units (CFU) were determined. The experiment was performed in triplicate.

The lytic activity of LysASP against P. aeruginosa (PA1-PA11), E. coli (E1), Salmonella abortus equi (S1) and S. aureus (SA2) was determined based on OD₆₀₀ of bacterial culture measured by a microplate reader as mentioned above. P. aeruginosa strain L64 and PBS were used as positive and negative controls, respectively. All bacteria strains are listed in Table 1. The experiment was performed in triplicate.

The deduced amino acid sequence of holin was analyzed by BLASTp, and putative hydrophobic transmembrane domains were predicted using TMHMM 2.0.⁹ The holin gene was amplified from phage ASP23 genomic DNA using PCR with a pair of specific primers containing SfiI and NotI restriction sites. The primers used are listed in Table 1. The PCR product was purified, digested with SfiI/NotI (Takara), and cloned into the pCANTAB 5E phagemids to generate pCANTAB 5E-holin, in which a C-terminal E-tag was used. The ligation product was transformed into competent cells (E. coli strain TG1) using the heat shock method. Restriction enzyme digestions and sequencing were used to confirm the cloned fragment.

To produce recombinant fusion M13 phages, a single colony of pCANTAB 5E-holin transformed E. coli TG1 cells was grown in 2 × YT medium containing 2% glucose and 50 μg/mL ampicillin at 37°C with shaking (220 rpm) until the OD₆₀₀ reached 0.5–0.7. Then, helper phage M13KO7 was added in the medium at a multiplicity of infection of about 1:50, followed by incubation for 35 min. After centrifugation, cells were re-suspended in fresh 2 × YT medium containing 25 μg/mL kanamycin and cultivated for at 37°C with shaking (220 rpm) for 20 h. After centrifugation (13,000 × g for 20 min, 4°C), phages were concentrated and precipitated by adding 1/5 (v/v) of PEG/NaCl solution in the supernatant. After incubation on ice for 12 h, the mixture was centrifuged, and the pellet was resuspended in 800 μL PBS (0.01 mM; pH 7.4), followed by incubation at 4°C for 12 h. Finally, after centrifugation, the supernatant containing M13 phages was collected and stored at 4°C (Tanaka et al., 1995). The holin protein of recombinant phages was identified using mass spectrometry (MS)-based shotgun proteomics.

To determine the lytic range of HolASP (recombinant phages), which was made with M13KO7 displaying holin of ASP23, six strains of bacteria were tested (Table 1), including two Gram-positive bacteria (Staphylococcus aureus SA2 and Bacillus subtilis A01) and four Gram-negative bacteria (P. aeruginosa L64, E. coli E1, S. abortus equi S1, and Proteus mirabilis P1). Briefly, qualitative filter papers were placed onto freshly seeded lawns of bacteria, on which 10 μL HolASP (at a final concentration of 10⁸ virions/mL) was spotted (Song et al., 2021). After incubation at 37°C for 12 h, the production of transparent antibacterial rings was assessed. Simultaneously, the mixed solution of PBS and M13KO7 was used as a negative control to exclude the effect of M13KO7 on bacteria.

The six strains of bacteria mentioned above were used as indicator strains to determine the antimicrobial activity of recombinant phages (HolASP). Bacterial cells were washed and resuspended with sterile phosphate-buffered saline (PBS) to 5 × 10⁷ CFU/mL (Song et al., 2021). HolASP (at the final concentration of 10⁸ virions/mL) was added to the bacterial suspension, and the mixture was incubated at 37°C for 12 h. The antibacterial activity was expressed as CFU reduction following treatment. As a negative control, the bacterial strains were treated with elution buffer (PBS, 0.01 mM; pH 7.4) under the same conditions.

The complete genome sequence of phage ASP23 was deposited in GenBank under the accession number MN602045.

The experiments, including the one-step growth curve, temperature sensitivity and pH stability of ASP23, lytic activity of LysASP and antibacterial activity of HolASP, were performed in triplicate. The results were presented as means ± standard deviation (SD). Student’s t-test and analysis of variance were used to compare the differences between the numbers of bacteria in different organs using GraphPad Prism 6.0 (GraphPad, CA, United States). Differences were considered statistically significant when p < 0.05.

A P. aeruginosa phage was isolated from sewage and named as vB_PaeP_ASP23 (ASP23). The TEM image showed that phage ASP23 had an icosahedral head of 70 nm in diameter and a tail of 40 nm in length (Figure 1A), suggesting that the phage ASP23 belonged to the phiKMV-like phages genus. The one-step growth curve of ASP23 revealed that the latent period and lysis period were 10 and 45 min, respectively, and the average burst size was 140 pfu/infected cell (Figure 1B). The thermal stability test showed that ASP23 was stable below 60°C, but it was completely inactivated at 80°C after 40 min (Figure 1C). In addition, the titer of ASP23 could be maintained above 10⁶ PFU/mL within 20 min at 70°C. ASP23 could maintain high activity (10⁹ PFU/mL) over a broad range of pHs (5–12) for at least 3 h (Figure 1D), but it was inactivated at extreme pHs (below pH 4 or above pH 12). ASP23 exhibited lytic activity and lyzed 68% (15/22) of the P. aeruginosa strains (Table 2).

To determine the therapeutic effect of phage ASP23 against infection caused by P. aeruginosa L64, we treated minks with phage ASP23 after they were challenged with P. aeruginosa L64 for 2 h. As shown in Figure 2, phage ASP23 significantly reduced bacterial counts in the liver and blood early at 5 h after treatment, and bacterial count in the lung decreased significantly at 13 h after treatment. The result indicated that phage ASP23 had a therapeutic effect in minks against P. aeruginosa L64.

The ASP23 genome was 42,735 bp in size and had an overall G + C content of 62.15%, which was similar to lytic phages of the genus phikmvvirus, such as Lx18 (100%, GenBank Accession No. MN692672.2), φKMV (98.62%, GenBank Accession No. AJ505558.1) and phiNFS (GenBank Accession No. NC_047852.1). By calculating the average nucleotide identity (ANI) value and the in silico DNA–DNA hybridization (isDDH) identity, we found that the ANI and isDDH values between phage ASP23 and Lx18 were 100%, the ANI and isDDH values between ASP23 and phiKMV were 98.08 and 76%, respectively, and the ANI and isDDH values between ASP23 and phiNFS were 98.22 and 81.3%, respectively. A total of 54 putative open reading frames (ORFs) longer than 100 bp were predicted, and no ORFs associated with drug resistance or lysogenization were identified (Supplementary Table S1). Among them, 45 ORFs (83.33%) had an ATG start codon, five ORFs had a GTG start codon, and four ORFs carried a TTG start codon. All predicted ORFs were located on the sense strand of the genome. BLASTp analysis revealed that 25 ORFs were annotated to encode proteins of known function. Similar to phage φKMV, the genes of ASP23 could be classified into three clusters: early genes (class I), genes involved in DNA metabolism (class II), and genes encoding structural and lysis proteins (class III) (Figure 3). To investigate the evolutionary relationship between ASP23 and other phages, whole genome sequence, major capsid protein and DNA polymerase were chosen for the construction of phylogenetic trees. As shown in Figure 4, phage ASP23 was clustered with phage φKMV and located near the members of the genus Phikmvvirus, confirming their close evolutionary relationship. Thus, phage ASP23 could be classified as a member of the genus Phikmvvirus in the family Krylovirinae.

Similar to Phikmvvirus phages, the early region (ORF1-ORF21) ended after the DNA ligase gene, and these genes included DNA-binding protein (ORF15), DNA primase (ORF18), DNA helicase (ORF19), and DNA ligase (ORF21). The amino acid sequence of DNA ligase showed high identity with φKMV_gp17 and contained a conserved domain (PHA00454). Four ORFs involved in DNA metabolism were identified in the genome of phage ASP23. ORF23 appeared to encode DNA polymerase, and its C-terminal region contained a DNA_pol_A domain (pfam00476). ORF26 and ORF27 were predicted to encode exonuclease and DNA endonuclease VII, respectively. ORF31 was predicted to encode RNA polymerase, and it was not located in the early gene region but at the end of the DNA metabolism region. The structural module of ASP23 was comprised of twelve ORFs, which encoded structural and packaging proteins, capsid proteins, and tail-related proteins. Generally, the terminase is composed of two subunits (big subunit and small subunit) which are involved in DNA packaging (Catalano, 2000). In the ASP23 genome, ORF47 and ORF48 encoded small subunit and big subunit, respectively. For the lysis system of ASP23, ORF49 and ORF50 encoded holin and lysin, respectively. ORF49 shared sequence homology with a putative holin gene in phage phikF77, moreover, its small size and genome location showed that it might encode a holin. Furthermore, ORF49 was a membrane protein with a putative hydrophobic transmembrane domain (TMD) and had a hydrophilic C-terminus with several charged amino acids, which should be classified into class II holins.

The 483-bp lysin gene of phage ASP23, named as LysASP, contained 160 amino acid residues. Three sequences with high identity to LysASP were selected, and multiple sequence alignments were performed. The Pfam database revealed that it contained a lysozyme-like domain between residues 34 and 141 of LysASP (Supplementary Figure S1A). An alignment of LysASP with three phage lysins showed high similarity in the conserved domain, however, three non-conserved residues (amino acids 61, 64, and 76) were found in the conserved domain (Supplementary Figure S1B). Based on the sequence searches in PDB, LysASP shared the highest identity (30.6%, 59/196) with the Muramidase domain of SpmX (Supplementary Figure S2A; Randich et al., 2019). Furthermore, the structure of LysASP was predicted using Phyre2, and the confidence of the model was 100.0% (Supplementary Figure S2B). Although the predicted backbone structure of the LysASP shared high similarity with SpmX muramidase domain, their surface representations were different (Supplementary Figures S2C,D).

The LysASP gene was successfully amplified from ASP23 genomic DNA and was cloned into the NdeI/EcoRI sites of the pCold TF expression vector. The recombinant plasmid was confirmed by DNA sequencing. After induction with IPTG, the expression of the fusion protein was detected by SDS-PAGE (Supplementary Figure S3). As expected, the size of LysASP was ~69 kDa and most of the recombinant protein was found to be in the soluble fraction after sonication.

The lytic activity of LysASP was determined against bacterial cells. As shown in Figure 5, the OD₆₀₀ values of bacterial cells treated with LysASP decreased remarkably from 0.95 to 0.48, whereas the OD₆₀₀ values of bacterial cells in the control group remained constant. Besides, LysASP treatment significantly decreased bacterial counts of P. aeruginosa L64 at 2 and 4 h, compared to the control group. These results showed that, in combination with EDTA, LysASP had lytic activity against P. aeruginosa L16.

To test the lytic range of LysASP, we treated Gram-positive bacteria (S. aureus strain SA2) and Gram-negative bacteria (P. aeruginosa PA1-PA11, P. aeruginosa G3-G10, P. aeruginosa DL1-DL3, E. coli E1, and S. abortus equi S1) with LysASP. The results showed that LysASP exhibited antibacterial activity toward 15 strains of P. aeruginosa, but it could not lyse other three bacterial species, which was the same as the antibacterial spectrum of phage ASP23 (Table 2).

TMHMM analysis showed that holin of phage ASP23 is a membrane protein with typical holin traits (Supplementary Figure S4A), which consisted of 66 amino acids (aa). And disulfide bond, signal peptide and coiled helix were not predicted in holin. Holin of phage ASP23 had a hydrophobic TMD at the N-terminus and multiple positively charged amino acids at the hydrophilic C-terminus. Since holin of phage ASP23 shared structural characteristics with class II holins, it could be considered as a member of the class II family (Reddy and Saier, 2013).

The holin gene was successfully amplified from ASP23 genomic DNA and was cloned into the SfiI/NotI sites of the pCANTAB 5E phagemids. The recombinant phagemids were confirmed by DNA sequencing. The results of SHOTGUN proteomics indicated that holin protein was successfully displayed on the surface of recombinant phage. The titer of rescued recombinant phage was 6 × 10⁸ virions/mL (Supplementary Figure S4B).

The antibacterial abilities of HolASP are shown in Figure 6. HolASP produced clear rings on bacterial lawns of Staphylococcus aureus SA2 and Bacillus subtilis A01. However, HolASP had no lytic ability against three of the four Gram-negative bacteria, including E. coli E1, S. abortus equi S1 and P. mirabilis P1. Log-phase cultures of six tested strains were exposed to HolASP (final concentration, 6 × 10⁸ virions/mL) for 12 h. HolASP treatment resulted in log₁₀ CFU reductions of two Gram-positive bacteria (5.3-log10 CFU reduction of S. aureus SA2 and 4.48-log10 CFU reduction of B. subtilis A01), relative to the controls. Furthermore, HolASP showed slight lytic ability against its host bacterium P. aeruginosa L64 (1.6-log10 CFU reduction of L64). The results indicated that HolASP only had efficient lytic ability against Gram-positive bacteria.