Introduction

Bacillary necrosis of pangasius (BNP) is one of the most important bacterial diseases in striped fish characterized by white spots on liver, kidney, spleen, and severe lesions of the internal organs and is caused by the pathogenic bacteria Edwardsiella tarda or Edwardsiella ictaluri1,2,3,4,5. In the Mekong Delta, Vietnam, although E. ictaluri was reported as a main causative agent in the striped catfish Pangasianodon hypophthalmus; which has caused big economic losses in several fish farms6,7, we currently identify the pathogen E. tarda in the BNP infection. Edwardsiella tarda, a Gram-negative, facultatively anaerobic, rod–shape bacterium, has a broad host range including reptiles, amphibians, mammals and humans1. The widespread overuse of antibiotics in fish farms has contributed to antibiotic resistance, ecological imbalance, and the over-accumulation of antibiotic residues in frozen catfish products8,9,10. Antimicrobial resistance (AMR) is increasingly prevalent in the treatment of E. tarda infections, with a high index of antibiotic resistance, ranging from mono-resistance to multidrug-resistant (MDR) phenotypic patterns, reported particularly in E. tarda isolates from catfish, tilapia, and clown knifefish in China, Malaysia, Thailand, and Vietnam11,12,13,14. Therefore, there is an urgent need to find an alternatively and natural safe agent to prevent or treat the E. tarda infections in the aquaculture industry.

Bacteriophages infect only bacterial cells and considered as the most abundant biological entity on the earth15. Bacteriophages offer several advantages over traditional antimicrobials: they exhibit high specificity toward their bacterial hosts, thereby reducing off-target effects on beneficial microbiota; and they are capable of self-amplification at the site of infection, which are increasingly prevalent in aquaculture systems15. In the case of Edwardsiella tarda, several studies have reported that phage administration can significantly reduce bacterial loads, enhance survival rates, and alleviate disease signs in fish species such as zebrafish (Danio rerio) and turbot (Scophthalmus maximus)16,17,18. In this study, we aimed to isolate, characterize, and evaluate the potential virulence and the stability of lytic bacteriophages against MDR E. tarda isolates, originating from the Mekong Delta, Vietnam. To the best of our knowledge, this is the first report of novel lytic phages infecting E. tarda isolates in striped catfish Pangasianodon hypophthalmus.

Results

Edwardsiella tarda causing the bacillary necrosis of pangasius (BNP) in the Mekong Delta

Several striped catfish farms in the Mekong Delta, Vietnam reported the clinical signs of bacillary necrosis of pangasius disease. We collected the kidney and spleen samples of the diseased striped catfish from these farms to isolate five Edwardsiella isolates. They have rod-shaped, Gram-negative, weak motility, the positivity of catalase, glucose fermentation, and glucose oxidase, but the negativity of cytochrome oxidase. The molecular identification using the 16S rRNA gene sequencing indicated that they are distant and placed in a distinct genetic group of E. tarda bacteria originated geographically from Mekong Delta, Vietnam (Fig. 1). In contrary, E. ictaluri bacteria; for example, E. ictaluri E1 strain, frequently indicated as the primary pathogen causing the enteric septicemia of striped catfish in Mekong Delta in previous years7; were placed in a clade closer to the species reported from United States and China (Fig. 1). In addition, multi-locus sequence typing (MLST) analyses performed using PubMLST and the Edwardsiella database, which consists of alleles from ten loci19: adk, atpD, dnaJ, gapA, glnA, hsp60, phoR, pyrG, rpoA, and tuf, also confirmed that the five Edwardsiella isolates are Edwardsiella tarda (unpublished data).

Fig. 1
figure 1

Phylogenetic tree of the E. tarda isolates using 16S rRNA gene sequences of Edwardsiella bacteria strains. The maximum likelihood tree was built by using MEGA software with 1000 bootstrap replicates. Branch numbers show bootstrap percentage following 1000 replications. Scale bar corresponds to phylogenetic distance of 0.01 nucleotide change or substitutions per site. The red asterisk and red box indicate E. ictaluri E1 lab strain and E. tarda isolates in this study, respectively.

Full size image

Antimicrobial resistance profiles revealed the multiple-drug resistant (MDR) E. tarda isolates

Antimicrobial susceptibility testing for E. tarda isolates was performed with six antibiotics, frequently used in aquaculture in Vietnam, such as kanamycin, gentamicin, doxycycline, amoxicillin, florfenicol, fosfomycin and classified in five antibiotic classes. All E. tarda strains were resistant to at least one antibiotic; importantly, kanamycin hardly be used to inhibit their bacterial growth (Table 1). Amoxicillin had the highly similar risk with the resistance rate up to 60%, while the others as gentamicin, florfenicol, and fosfomycin, their resistance rate were also 40%. Multidrug resistance (MDR) refers to the ability of bacteria to resist to three or more antimicrobial classes20. Remarkably, we identified the E. tarda isolate E24.19 as multidrug-resistant (MDR) to three tested antimicrobial categories.

Table 1 Antimicrobial resistance profile of five isolates E. tarda.
Full size table

The potential lytic Autographivirales bacteriophages can repress the multidrug-resistant (MDR) E. tarda isolates

To control and treat E. tarda pathogen, especially the MDR isolates, causing bacillary necrosis of pangasius disease in striped catfish in Mekong Delta, Vietnam, we conducted the alternative strategies of phage application; whereby we isolated three virulent bacteriophages from the water samples of fish farms in An Giang, Can Tho, Dong Thap, respectively, using E. tarda strain E24.1 as host and named them21 vB_EtA_WAG25P1 (WAG25P1 for short), vB_EtA_WCT72P1 (WCT72P1), and vB_EtA_DT115P1 (DT115P1). The plaque morphology of the phages was a clear and circular plaque that ranged in size from approximately 1 mm to 3 mm in diameter for phage WCT72P1, WAG25P1, respectively, while about 3 mm to 4 mm for phage DT115P1 (Fig. 2a). Transmission electron microscopy (TEM) data revealed that the E. tarda phages had icosahedral head with the diameter of 41.33 ± 4.99 nm and short, noncontractile tails (Fig. 2b), characteristics of the podovirus morphotype.

Fig. 2
figure 2

Characterization of Edwardsiella tarda bacteriophage vB_EtA_WAG25P1, vB_EtA_WCT72P1, vB_EtA_DT115P1. (a) Top agar overlay showing plaque morphology of bacteriophages. The scale bar represents 1 cm. The experiment was repeated twice. (b) Transmission electron micrograph showing the tailed Autographivirales morphotype of bacteriophages. The arrow indicates the head structure of phages. The scale bar represents 100 nm. The experiment was repeated twice. (c) The one-step growth curves of the bacteriophages reveal progression of phage per infected cell over time with error bars showing standard deviation (SD). The values are means of three biological repeats. Biological stability analysis of bacteriophages at (d) different temperatures and at (e) different pH conditions. The values are means of three biological repeats. Error bars indicate the standard deviation. Significance was defined as ** p ≤ 0.001 using Mann–Whitney U-test.

Full size image

We characterized further biological features of the phages to gain insight into their plaque morphologies. To determine the latent period and the burst size in the life cycle of new phage isolates on E. tarda, one-step growth assays were conducted. The latent period was the shortest time required between phage adsorption and host cell lysis to release of extracellular phage particles, while the number of virions released per infected cell was defined as the burst size. For phages WCT72P1 and WAG25P1, the latent period was ~ 35 min, with a burst size of ~ 45 ± 13 virions and ~ 61 ± 12 virions per infected cell, respectively. While for phage DT115P1 ~ 51 ± 10 virions were released in the shorter latent period of ~ 25 min (Fig. 2c). Remarkably, our E. tarda bacteriophages exhibited a shorter latent period than phage ETP-1 (60 min), which share the same Podovirus morphology17.

In order to determine the stability of E. tarda phages for further phage applications, we carried out assays of phage survival under various conditions of pH and temperature. The E. tarda phages were stable in pH range 6 to 10, while their activities significantly declined at acidic pH 2 to 5 and at very alkalic pH 11 and 12 (Fig. 2e). The phages were tolerant to a broad temperature range from 4 °C to 37 °C and with titers remaining above 80% of the starting titer, even at 50 °C (Fig. 2d). The biological features of the phages were summarized in Table S5.

The host range assays for E. tarda phages was also performed, and disclosed their highly specific ability of host lysis on our E. tarda collection, especially on MDR isolates as E24.17 and E24.19 Fig. 3a, Table 2, \(\text{ no}\) lysis on E. coli strains and E. ictaluri strain E1 which was reported in the previous research22 for phage PVN06 and PVN09. The E. tarda phages exhibited their appropriate biological features of thermo-stability, pH stability and narrow lytic spectrum for phage applications. Therefore, we further evaluated their in vitro lytic activities against the E. tarda isolates at MOI 1. The growths of E. tarda isolates as E24.1, E24.2, E24.3 and E24.17 were significantly inhibited by the phages within 6 h Fig. 3b, except the weak inhibition of phage WCT72P1 against E. tarda E24.17. In contrast, the regrowth of the bacterial isolate E24.19 started at approximately 3 h of post-infection. It seemed the appearance of the new phage-resistant E. tarda isolate, which was similarly observed at tetracycline-resistance E. coli strain caused by adaptive mutations on the bacterial chromosome23.

Fig. 3
figure 3

Host analysis of Edwardsiella tarda bacteriophages. (a) Host range testing of Edwardsiella tarda bacteriophages vB_EtA_WAG25P1, vB_EtA_WCT72P1, vB_EtA_DT115P1 on the E. tarda isolates. SM on the petri dish indicates the no phage control; 1, vB_EtA_WAG25P1; 2, vB_EtA_WCT72P1; 3, vB_EtA_DT115P1. The experiment was repeated twice. (b) The in vitro assay of bacterial growth reduction in liquid cultures of the E. tarda isolates by the bacteriophage agents. The blue line represents the OD595nm value of the no phage control; cyan, vB_EtA_WCT72P1; orange, vB_EtA_WAG25P1; and black, vB_EtA_DT115P1. The values are means of three biological repeats. Error bars indicate the standard deviation.

Full size image
Table 2 Host range testing of E. tarda bacteriophages.
Full size table

The complete DNA genome analysis of E. tarda bacteriophages

Three E. tarda phages exhibited similar biological properties, so we sequenced their whole genome to get insight into their genomic diversity. The E. tarda phages consisted of linear double-stranded DNA genome with their overall G + C content of 47% (Fig. 4a–c), consistent with other members of their related genera (Supplementary Table S1, S2). The genomic assemblies of the E. tarda phages (Fig. 4a–c), Supplementary Table S3) revealed that they have the similar sizes of 43.508, 43.999 and 44.166 Kbp; encoding 60, 64 and 62 Open Reading Frames (ORFs) which located on both the sense and anti-sense strands; for the phages WCT72P1, WAG25P1 and DT115P1, respectively. Through protein database analysis, the genome maps of the E. tarda phages (Fig. 4a–c) assigned to the putative ORFs were categorized into five core modules; based on their predicted functions; including DNA/RNA metabolism, phage tail, phage capsid and packaging, phage lysis and hypothetical protein. Interestingly, the E. tarda phages contain an RZ-like spanin protein classified to o-spanin type, an outer membrane lipoprotein in conjunction with the inner membrane i-spanin to facilitate the fusion of the inner and outer bacterial membranes, leading to host cell lysis and the release of progeny virions24,25. Phage morphology data combined to their ability of self-transcription relying on a phage-encoded RNA polymerase indicated that the E. tarda phages should be classified to the order Autographivirales according to the ICTV (International Committee on Taxonomy of Viruses) recommendation, due to the presence of RNA polymerase. Remarkably for further in vivo phage applications, their genomes did not contain any toxin genes, lysogeny associated genes, or genes encoding transfer RNA, integrase, virulence factors, or genes related to antimicrobial resistance.

Fig. 4
figure 4figure 4

Linear genome map of the phage vB_EtA_DT115P1 (a), vB_EtA_WAG25P1 (b) and vB_EtA_WCT72P1 (c). Annotations for predicted open reading frames (ORFs) are presented. The ORFs marks with different colors indicate categorized predicted functions of proteins. Two first plots represent a GC content (in black) and a GC skew (in violet) of the phage genome. (d) Comparative genome analysis of E. tarda bacteriophages using VIRIDIC software. The heat map indicates the relationship of three E. tarda phages with their close homologs in BLASTn and estimates the genomic similarity between phages. (e) Comparative genome alignment of Edwardsiella tarda bacteriophages and the most closely related Klebsiella phage RCIP0053 (accession no. OR532847) using the Easyfig genome visualization tool. The percentage of the nucleotide similarity at different regions between genomes is represented by the intensity from the gray to black color.

Full size image

The heat map of average nucleotide similarity between close relatives of the E. tarda phages using VIRIDIC software (Fig. 4d) indicated that phage WAG25P1 and DT115P1 exhibited the highest nucleotide identity to Klebsiella phage RCIP0053 (accession no. WPJ52799.1) with the sequence identity of 67.4% and 68.1%, respectively, while the phage WCT72P1 was closely related to Klebsiella phage RCIP0089 (accession no. WPJ55266.1) with the similarity of 66.6%. All the E. tarda phages had similarity to phages in the genus Gajwadongvirus, but with the similarity under 70%. The nucleotide identity scores between the E. tarda phages > 95% suggested that they are in the same species. Thus, the phages are the first representatives of both a new species and new genus, based on current ICTV guidelines26.

The genomic comparison also confirmed the conserved synteny between Klebsiella phage RCIP0053 and the E. tarda phages (Fig. 4e) with conservation in the arrangement of functional modules among the genomic sequences. In spite of the extremely low nucleotide dissimilarity between the E. tarda phages, we uncover the absence of three ORFs encoding hypothetical proteins with molecular size 3.5 – 15.5 kDa in WCT72P1, possibly the structural proteins, and the entire DNA polymerase instead of the separation of two subunits of DNA polymerase in WAG25P1. The ORFs lack could reduce the growth fitness of phage WCT72P1 in E. tarda isolates, with obvious phenotypes such as the small plaque (Fig. 2a), the low burst size (Fig. 2c), and the weak in vitro repression of E. tarda E24.17 host (Fig. 3b).

To further classify the E. tarda phages in terms of their taxonomy, the viral proteomic tree with their close homologs was generated using ViPTreeGen program to perform global amino acid sequence comparison using tBLASTx. The phages with genome similarity score between 0.027 – 1 were included in phylogenetic analysis (Supplementary Table S2). The phage WAG25P1, WCT72P1, and DT115P1 are in a cluster of new species belonging to the new unassigned genus closing the Gajwadongvirus genus in the Autosignataviridae family (Fig. 5a). In comparison to other reported phages, the E. tarda phages were also classified using a maximum likelihood tree for a hallmark protein of the terminal large subunit (Supplementary Tables S1). The phylogenetic analysis (Fig. 5b) confirmed their designation in a separate clade closing the clade containing Pseudomonas phage MR4 (accession no. QJD54743), Escherichia phage ECBP5 (accession no. YP009146419), Pectobacterium phage PP99 (accession no. YP009788797) of the Gajwadongvirus genus in the Autographivirales order.

Fig. 5
figure 5

Phylogenetic position of E. tarda bacteriophages. (a) A proteome tree generated by the ViPTreeGen using tBLASTx (mid-point rooted), and annotated using iTOL to determine the phage relationship with close homologs. It indicates that E. tarda bacteriophages as new species in one unassigned new genus of Autosignataiviridae family. The classified families, subfamilies and genera are labeled by colored stripes. (b) Molecular phylogenetic analysis by the maximum likelihood method of BLASTp hits of large termirase sequences (terL) of close homologs and E. tarda phages.

Full size image

Discussion

The climate or natural environment changes could facilitate the predominance of Edwarsiella pathogens27,28,29. Whole genomic analysis of Edwarsiella pathogens in the Mekong Delta, including E. tarda and E. ictaluri isolates, and their genomic comparison could imply how their genomes evolved to adapt the habitat changes, and also understand their pathogenesis and toxicity in the striped catfish Pangasianodon hypophthalmus. In addition, we also concern that the uncontrolled overuse of antibiotics is the main reason for the appearance of emerging multidrug-resistant E. tarda isolates. Indeed, several reports also showed the E. tarda isolate from farmed seawater in Korea exhibited the resistance to antibiotics as streptomycin, erythromycin, penicillin, cefaclor, and vancomycin17 or E. tarda isolate from fish farms in Pakistan with the resistance to amoxicillin, erythromycin, and flumequine30.

To control MDR E. tarda isolates, we identified three novel lytic bacteriophages to apply for phage therapy as an alternative antimicrobial strategy. The finding of the distinct and highly genetic similar E. tarda phages and their representatives in a new unassigned genus of the Autographivirales family implied the high prevalence of the members in Mekong Delta, Vietnam, which is extremely specific to the group of the E. tarda isolates only in this geographic location, not in others in Asia (Supplementary Table S4). Thus, it highlights that the isolation and characterization of novel lytic phages specific for the pathogenic bacteria varied by geographic region is extremely important for the success of phage biocontrol strategies. Furthermore, our phages exhibit superior characteristics compared to ETP-1, which has demonstrated high biocontrol efficiency in the zebrafish. These findings highlight the promising potential for our phages in controlling E. tarda infection in the striped catfish17.

Despite the encouraging in vitro results, several limitations must be considered before field application. Environmental factors such as water salinity, organic matter, and UV exposure can significantly impact phage stability and infectivity, thereby reducing their effectiveness under natural pond conditions31. Moreover, the maintenance of phage viability during feed processing and throughout passage in the fish gastrointestinal tract remains a major challenge. Importantly, the innate immune responses of fish, including phagocytosis and the secretion of antimicrobial peptides, could inactivate phages until they reach their bacterial targets, particularly in cases of systemic infection32.

To fully assess their therapeutic potential, the in vivo testing in striped catfish is a critical next step. These studies should evaluate phage efficacy in reducing bacterial load, enhancing host survival, and controlling disease symptoms under both laboratory and farm conditions. In addition, monitoring phage – host interaction during and after treatment is essential to ensure the long-term sustainability of this antimicrobial strategy33. The observed bacterial regrowth in our in vitro assays, e.g. E. tarda E24.19, underscores the need for a phage cocktail to improve treatment efficacy and to prevent the emergence of phage-resistant E. tarda strains.

Conclusion

Bacteriophages infecting E. tarda have been less investigated while this host bacterium is one of the most important fish pathogens causing bacillary necrosis of pangasious disease and leading to heavy economical losses. The detailed study of vB_EtA_WAG25P1, vB_EtA_WCT72P1, and vB_EtA_DT115P1 will improve our understanding of their genera. This research will likely lead to their use in treatments of the BNP disease and biocontrol methods after we conduct in vivo evaluations.

Methods

Collection and characterization of E. tarda strains

The kidney and spleen samples of diseased striped catfish Pangasianodon hypopthalmus were collected from fish farms (4–8 live fish/farm, one farm/province) where bacillary necrosis outbreak was occurring in the Dong Thap province (December, 2023), Long An province (December, 2023; and July, 2024), and Can Tho province (June, 2024), Mekong Delta, Vietnam to isolate the bacterial strains. Samples were cultured on the Edwardsiella ictaluri agar (EIA)34 at 28 \(^\circ{\rm C}\) for 24 to 48 h and stored on the tryptone soy broth (TSB, Merck) supplemented with 25% glycerol at -80 \(^\circ{\rm C}\). Five Edwardsiella isolates were collected and characterized biochemically in this study. They were subsequently identified as Edwardsiella tarda by ONT sequencing (Oxford Nanopore Sequencing) of 16S rRNA gene (KTEST, Ho Chi Minh City, Vietnam)35,36. Their 16S rRNA gene sequences were submitted to Genbank database under the following accession numbers PV254726, PV254727, PV254728, PV254729, and PV254730. The 16S rRNA gene sequences from E. tarda isolates and from the most similar Edwardsiella bacteria were aligned with MEGA v11 software (version 11)37 and phylogenetic trees were constructed by the same software using Maximum Likelihood method with 1000 bootstrap replicates. All experimental protocols for the care and manipulation of fish in this study approved by the Ethics Advisory Council on Animal Research of Can Tho University (CTU-AEC-3881/QD-DHCT). All methods were carried out in accordance with relevant guidelines and regulations, and are reported in compliance with the ARRIVE guidelines.

The antibiotic sensitivity of all isolates was tested by an agar disk diffusion method38, against six antibiotics commonly employed for treatment of Edwardsiella enteric septicemia disease: Kanamycin (30 µg), Doxycycline (30 µg), Amoxicillin (10 µg), Florfenicol (30 µg), Fosfomycin (50 µg), Gentamicin (10 µg). These antibiotic disks were purchased from Oxoid (USA) and stored according to the manufacturer’s instruction. These antibiotics are not on the list of the banned drugs according to Circular No. 10/2016/TT-BNNPTNT issued on 01 June 2016 by Ministry of Agriculture and Rural Development (MARD) in Vietnam. The E. tarda isolates were cultured on the trypton soy agar (TSA, Oxoid) at 28 °C overnight in the aerobic condition to standardize the bacterial suspensions. A number of distinct colonies from the freshly grown plate culture were harvested to suspend in a tube containing solution of 0.9% NaCl until the turbidity reached that of a 1.0 McFarland standard, as visually observed. Using a micropipette, 100 μl of the standardized suspension was spotted onto the surface of a TSA plate. The suspension was then spread using a sterile glass triangle rod. The plates were allowed to dry for a maximum of 15 min. The antibiotic disks (6 mm of diameter) were placed on the surface of the TSA plates applied with the E. tarda samples, and incubated at 28 °C for 24 h. The Clinical and Laboratory Standards Institute (CLSI, 2024) guideline was used to read the antibiograms of the bacterial isolates38,39. Multidrug resistance (MDR) is defined as the ability of bacteria to withstand the effects of three or more antimicrobial classes20.

Bacteriophage isolation and purification

The bacterial strain Edwardsiella tarda E24.1 was used as a host for bacteriophage isolation. Water samples gathered from the fish farms in 2024 in Dong Thap, An Giang and Can Tho province, Mekong Delta, Vietnam were used to isolate three bacteriophages vB_EtA_DT115P1, vB_EtA_WAG25P1, vB_EtA_WCT72P1, respectively. Briefly, we used the plaque assay method to isolate the bacteriophages40. The water sample was initially centrifuged at 6000 × g for 10 min and the supernatant used for subsequent steps. To enrich the bacteriophages for isolation, 5 ml of processed water sample was incubated with the bacterial host E. tarda E24.1 in TSB medium overnight at 28 \(^\circ\) C. The enriched sample (100 μl) was mixed with the bacteria host cells (200 μl in 5 mL of 0.4% (w/v) TSB agar, and then poured onto a 1.5% (w/v) LB agar plate, and incubated overnight at 28 \(^\circ\) C. The individual plaques were picked and transferred to 10 μl of SM buffer containing 50 mM Tris – HCl pH 7.5, 100 mM NaCl, 8 mM MgSO4, 0.2% (w/v) gelatin. The sample was vortexed and centrifuged at 12,000 × g for 5 min at 4 \(^\circ\) C. The supernatant was serially diluted in tenfold steps in SM buffer, for further rounds of purification. For each individual plaque, the purification was repeated for at least three rounds to obtain a clonal and pure phage population. The phage stock was created using the plaque assay method. Briefly, the mixture of phage and bacteria were plated and incubated overnight. Subsequently, it was incubated with 4 ml of SM buffer for at least 8 h at 4 \(^\circ\) C. The suspension on the agar top layer was centrifuged at 12,000 × g for 5 min at 4 \(^\circ\) C and then filtered with a 0.22 μm membrane filter. Phage stock was stored at 4 \(^\circ\) C. Phage particles are also stored in SM buffer containing 20% v/v glycerol in a -80 \(^\circ\) C.

Transmission electron microscopy (TEM)

Bacteriophage particles were prepared by negative staining with uranyl acetate 2%41. The morphologies of bacteriophages were observed by the transmission electron microscope (JEOL JEM-1010, Japan) at a voltage of 80 kV and an instrumental magnification of 25,000 – 30,000 at the National Institute of Hygiene and Epidemiology, Vietnam.

Bacteriophage stability at acidic and alkaline pH and at different temperatures

The biological properties of the phage in various conditions were examined. To test the temperature stability of phage, 1 ml of phage suspension at 109 PFU/ml was incubated at 4, 20, 25, 30, 37, 50 °C for 1 h. To test the pH stability, the pH of TSB medium was adjusted from pH 2 to 12 using 1 M HCl or 1 M NaOH and then 500 μl of phage suspension was mixed with an equal volume of TSB medium, and incubated at 30 °C for 24 h. The phage titer in the supernatant after treatment was measured using a plaque assay17,42. The survival phage titers in tested conditions were represented by the relative percentage of retained phage progenies. The experiments were repeated three times.

Host range analysis

The host range of phages was determined using the spot assay method. Five E. tarda isolates, E. ictaluri E1 from the Mekong Delta, Vietnam and E. coli strains (K12 ATCC 10,798 and DH5α ATCC PTA-1798) were used. Overnight bacteria cultures were diluted to 1% (v/v) in TSB medium, and then grown for 2 h to reach the OD595nm of 0.1 (~ 2 × 108 CFU/ml). Briefly, 3 μl of the phage suspension at 109 PFU/ml was dropped on TSB soft agar containing 200 μl of bacterial culture, and then incubated at 28 \(^\circ\) C. The clearance of the spot was graded at completely clear zone (+ +) or turbid zone ( +), or none visible (-).

One-step growth curve

Briefly, an overnight culture of bacteria was diluted to 1% (v/v) in TSB medium and used to subculture to reach an OD595nm of 0.1 (~ 2 × 108 CFU/ml). Phage stock was mixed with the bacterial culture at multiplicity of infection (MOI) 0.01 and the mixture was shaken at 150 rpm in 10 min for the phage adsorption. The supernatant containing unabsorbed phages was obtained by centrifugation at 12,000 × g for 10 min at 4 \(^\circ\) C and titrated to measure the initial adsorbed phages. The bacterial pellet was re-suspended in the same volume of TSB medium in a clean tube to allow their growth by shaking at 150 rpm. The newly released phages were quantified every 5 min using plaque assay method. The burst size was defined as the number of phages produced by one infected bacterium in the end of the latent period. The experiments were repeated three times.

Growth reduction assay of E. tarda isolates

In brief, 200 μl of bacterial culture in log phase (~ 2.108 CFU/ml) were mixed with 50 μl of phage suspension at MOI 1 in each well of 96-well plate. TSB medium was used for sample dilution. 50 μl of TSB medium was used for the no phage control. The microplate was incubated at 28 \(^\circ{\rm C}\) with shaking (150 rpm). The bacterial growth was determined at OD595nm in every 20 min for 6 h by Microreader Imark Bio-rad. The experiments were repeated three times.

Phage DNA genome purification, sequencing and bioinformatic analyses

Prior to phage DNA genome extraction, phage suspensions at ~ 109 PFU/ml were treated with DNase I and RNase at 37 \(^\circ{\rm C}\) for 1 h to digest bacterial nucleic acid. After the heat inactivation of enzymes at 80 \(^\circ{\rm C}\) for 15 min, 1.5 μl protease K was added for capsid lysis in one hour. The solution of phenol–chloroform–isoamyl alcohol, 25:24:1 (v/v/v), was used at an equal volume of the phage sample to purify the phage DNA genome. The samples were centrifuged at 12,000 × g for 5 min at 4 \(^\circ\) C and then obtain the aqueous layer. An equal volume of chloroform–isoamyl alcohol, 24:1 (v/v) were added to remove any remnant phenol. The DNA aqueous layer was obtained after centrifugation with similar program. Two volumes of ice-cold ethanol and sodium acetate at the final concentration of 0.3 M were added to precipitate the phage DNA genome. The samples were centrifuged at 12,000 × g for 15 min at 4 \(^\circ\) C. The DNA pellets were washed three times by ice-cold 70% ethanol. The DNA samples were air-dried and then dissolved in 100 μl of nuclease-free water. The phage DNA genomes were stored at -20 \(^\circ\) C.

Whole phage genome sequencing libraries were prepared by NEBNext kit dsDNA Fragmentase and NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB, USA). The phage sequencing libraries were performed on the DNBSeq-G99 (MGI) sequencer with paired end, 150 bp per read at the KTEST company (Ho Chi Minh City, Vietnam). Whole-genome sequence data of phages WAG25P1, WCT72P1, DT115P1 are available in the NCBI GenBank database under the following accession numbers PV165919, PV165920, and PV165918, respectively.

FastQC (version 0.12.1)43 was employed to assess the quality control of the raw sequence data. To remove contaminant adapters and read quality trimming, Trim Galore (Version 0.6.10)44 was used. SPAdes genome assembler (version 4.0.0)45 was used to assemble genomes with only assembler option. Pharokka (version 1.7.3) with default options was used for gene annotation46. The function of the predicted gene was confirmed using the BLASTp on the non-redundant protein sequences (nr) database in NCBI. To determine the intergenomic similarities of bacteriophages, VIRIDIC software (http://rhea.icbm.uni-oldenburg.de/VIRIDIC/) was utilized with the default BLASTn parameter47. Easyfig software (version 2.2.5) was used to visualize the genome comparison of bacteriophages48. To assess the appropriateness of bacteriophages as the potential candidates for phage therapy, their genomes were analyzed for the presence of lysogenic genes and recombinase genes. In addition, they were screened for virulence factors, toxins, and antimicrobial resistance markers using PhageLeads software (https://phageleads.ku.dk/)49. A phyloproteomic tree based on the proteome analysis of phages was constructed using ViPTreeGen (version 1.1.2)50 with the mid-point rooted, and annotated using iTOL (version 7.1.1)51. The tree was relied on a dissimilarity matrix generated by pairwise tBLASTx scores between bacteriophage genomes. The proteomic tree was generated with forty bacteriophage genomes in the Autographivirales order and one genome in the Chaseviridae family as an out group. For each genus, at least one representative phages were selected, so at least six phages were selected from each family in the Autographivirales order. The phages with genome similarity score between 0.027 – 1.000 were included in phylogenetic analysis (Supplementary Table S2). The large terminase (terL) amino-acid sequences from E. tarda bacteriophages and from the most similar phage were aligned with MEGA software (version 11)37 and phylogenetic trees were constructed by the same software using Maximum Likelihood method with 1000 bootstrap replicates.