All methods were performed in accordance with the La Trobe University Ethics, Biosafety, and Integrity guidelines and regulations. Clinical isolates of A. hydrophila were obtained from specimen cultures as part of routine care. Informed consent was obtained from participants for their involvement and use of samples in this study. The study protocols were approved by the La Trobe University Ethics Committee, reference number: S17–111.

Aeromonas hydrophila bacteria was isolated from a deep wound infection (Strain AHB0117), a polymicrobial liver abscess on a background of cholangiocarcinoma (Strain AHB0148), a polymicrobial surgical site of infection (Strain AHB0116), diarrhea fecal samples (Strain AHB0139), and a scalp abscess due to trauma (Strain AHB0147), all de-identified. All strains were cultured in nutrient broth or agar (Oxoid, Australia) at 37°C aerobically. The bacterial strains were initially identified by Matrix Assisted Laser Desorption/Ionization – Time of Flight (MALDI-TOF; Bruker Daltonik, Germany). Conclusive identification was achieved by sequencing of the 16S rRNA region (see Table 1 for PCR conditions), as well as screening for intrinsic antibiotic resistance markers endogenous to A. hydrophila (see below). The 16s rRNA amplicons were purified using QIAquick^® PCR purification kits (Qiagen, Australia) and Sanger sequenced by the Australian Genome Research Facility (AGRF) in Queensland, Australia. The strains that were identified as A. hydrophila were used for subsequent bacteriophage screening.

Antibiotic sensitivity of the A. hydrophila clinical strains used in this study was assessed using the VITEK^® 2 analyzer (bioMérieux, Australia), according to the manufacturer’s instructions. Whole genome sequences of A. hydrophila strains published in the GenBank NCBI database were imported into CARD (Comprehensive Antibiotic Resistance Database) (Jia et al., 2017) and analyzed for genes coding antibiotic resistance. These genomes were also screened for CRISPR coded sequences using CRISPRFinder (Grissa et al., 2007). Identified sequences from (version 9.5.4) multiple strains were aligned in CLC genomic workbench and PCR primers designed from their conserved regions. The primer sequences and their PCR conditions are listed in Table 1 and amplicons were confirmed by sequencing (AGRF, Australia). Bacteriophage whole genome sequences were also screened for antimicrobial resistance genes (ARGs) and CRISPR as above.

Wastewater and fishpond samples from Victoria, Australia, were screened for bacteriophages by enriching the samples with A. hydrophila. In brief, 100 μL of log phase A. hydrophila was added to 10 mL of broth with 1 mL of filtered sample (0.2 μm cellulose acetate; Advantec, Australia). This enrichment was incubated for 4 days before filtration, and 10 μL of this filtrate was then spotted onto a bacterial lawn of A. hydrophila on agar to screen for the presence of plaques. Host range testing was performed on five clinical strains of A. hydrophila.

Aeromonas hydrophila strains in exponential growth phase, collected by centrifugation at 12,000 × g for 10 min and resuspended in fresh nutrient broth at a concentration of 0.6U (OD₆₀₀), were used for the one step growth experiments (Wang et al., 2016). The strain AHB0147 was used for one-step growth experiments involving LAh1–LAh5 bacteriophages while LAh6–LAh10 bacteriophage one-step growth experiments were performed using the strain AHB0116. One hundred microliters of bacteriophage were added to 900 μL of each A. hydrophila strain at a MOI of 0.01 and incubated at 4°C for 30 min to allow for adsorption. Adsorbed bacteriophage were collected by centrifugation at 12,000 × g for 10 min and pellet resuspended in 50 mL of fresh nutrient broth. The mixture was incubated aerobically at 37°C and bacteriophages assayed using aliquots collected every 5 min by centrifugation at 12,000 × g for 2 min at 4°C.

Bacteriophage particles were visualized by Transmission Electron Microscopy using a JEOL JEM-2100 transmission electron microscope (TEM) at 200 kV. Bacteriophage lysate was adsorbed onto 400-mesh formvar and carbon coated copper grids (ProSciTech, Australia) for 1 min. Grids were rinsed with milli-Q water and adsorbed phage particles were negatively stained twice using 2% (W/V) uranyl acetate (Sigma-Aldrich^®, Australia) for 20 s. Excess stain was removed using filter paper and grids air dried for 30 min. Images were captured on a Gatan Orius SC200D 1 wide-angle camera using the Gatan Microscopy Suite and Digital Micrograph Imaging software (version 2.3.2.888.0). The images obtained were further analyzed using ImageJ (version 1.8.0_112).

All chemicals were purchased from Sigma-Aldrich^® (Australia), unless stated otherwise. Concentrated bacteriophage stock (approximately 10¹¹ PFU mL^–1) was treated with 5 mmol L^–1 of MgCl₂ as well as RNase A and DNase I (Promega, Australia) to a final concentration of 10 μg mL^–1. The digest was incubated at room temperature for 30 min before polyethylene glycol precipitation at 4°C using PEG-8000 at 10% (w/v) and sodium chloride (1 gL^–1). Precipitated virions were recovered by centrifugation at 12,000 × g for 5 min to obtain a pellet which was then resuspended in 50 μL nuclease free water (Promega, Australia). Viral proteins were digested with 50 μg mL^–1 of proteinase K, 20 mmol L^–1 EDTA and 0.5% (v/v) of sodium dodecyl sulfate for 1 h at 55°C to release phage DNA. Bacteriophage DNA was separated from proteins by addition of an equal volume of phenol-chloroform-isoamyl alcohol (29:28:1) and carefully collecting the aqueous phase after centrifugation at 12,000 × g for 10 min. An equal volume of isopropanol and overnight incubation at −20°C was used to precipitate bacteriophage DNA. Bacteriophage DNA was then collected by centrifugation (12,000 × g for 5 min) before washing in 70% ethanol, air-drying, and re-suspending in 30 μL of nuclease free water (Promega, Australia).

Nextera^® XT DNA sample preparations kits were used to prepare phage DNA for sequencing according to the manufacturer’s instructions. Whole genome sequencing of the prepared libraries was performed on an Illumina MiSeq^® using a MiSeq^® V2 300 cycle reagent kit. Sequence reads were imported into CLC genomics workbench (version 9.5.4) and assembled de novo. Open reading frames (ORFs) were predicted and translated using CLC genomics workbench (version 9.5.4). Translated ORFs were analyzed using BLASTP (Mount, 2007) and tRNAs and tmRNAs predicted using ARAGORN (Laslett and Canback, 2004) and tRNAscan-SE 2.0 (Lowe and Chan, 2016). Bacteriophage genomes were also analyzed for CRISPR sequences using the CRISPR database (Grissa et al., 2007). Whole genome alignments of isolated bacteriophages and those against other Aeromonas spp. (sourced from GenBank) were conducted and a phylogenetic tree constructed by neighbor joining method with 1000 bootstrap replicates in CLC genomics workbench (version 9.5.4). The bacteriophage genomes were also assessed by MAUVE plugin (Darling et al., 2004) in Geneious (version 11.0.5)¹.

The capacity to disrupt A. hydrophila biofilms was determined by growing A. hydrophila mono-biofilms in a 96 well polystyrene plate (Greiner bio-one, Australia). The 96 well plates were inoculated with 100 μL of 10⁸ CFU mL^–1 log phase A. hydrophila in broth culture and a further 100 μL of sterile broth added. The cultures were then incubated aerobically at 37°C, shaking, for 4 days. Ten microliters of bacteriophage at a concentration of 10⁸ PFU mL^–1 was added to the established biofilms and for each experiment, heat inactivated (autoclaved) bacteriophage was used as a control to confirm that the effects on biofilm were the result of bacteriophage particles, rather than chemical residues or other matter in the preparation. Bacterial attachment was assayed according to Merritt et al. (2005). Briefly, plates were submerged in water to wash cells for 5 min before staining with 200 μL of 0.1% crystal violet for 10 min. The excess crystal violet was removed by submerging the plates in water for 5 min. The stained adherent cells were then solubilized in 70% ethanol and the absorbance in each well determined (at a wavelength of 550 nm) using a FlexStation 3 plate reader (Molecular devices, United States). Bacteriophages LAh7, LAh9, and LAh10 were tested on A. hydrophila strain AHB0116 biofilm whilst LAh1 was tested on biofilm formed by AHB0147.

Aeromonas hydrophila biofilms grown on glass slides were stained with 100 μL of SYBR^® gold (Eugene, OR, United States; 1 mg mL^–1) diluted in dimethyl sulfoxide (Sigma-Aldrich^®, Australia) and 3 μL of 1 mg mL^–1 propidium iodide (PI) in nuclease free water (Promega, Australia) for 30 min in the dark. The live/dead stained cells were mounted with 10 μL Vectorshield^® (Burlingame, CA, United States) on coverslips. The stained slides were visualized with an Olympus Fluoview Fv10i-confocal laser-scanning microscope (Olympus Life Science, Australia). PI stained DNA of membrane-compromised cells red while SYBR Gold^® stained DNA from both intact and membrane-compromised cells green.

Statistical tests were used to assess the capacity of bacteriophages to break down biofilms from clinical strains of A. hydrophila. Firstly, the Shapiro–Wilk test was used to assess whether, for each individual bacteriophage, biofilm absorbance at OD_550nm was normally distributed. As these data were found to be non-normally distributed, biofilm absorbance at OD_550nm for each phage was summarized in terms of the median rather than the mean, with the full five-number summaries presented in side-by-side boxplots. The interquartile range (IQR) was also calculated. Due to the non-normality of the data, the biofilm absorbance at OD_550nm for each bacteriophage was compared with that for every other phage using a non-parametric test – the Wilcoxon signed-rank test. p-values less than 0.05 were considered to be statistically significant. All statistical tests were performed in SPSS version 24 (SPSS Inc., United States).

Five clinical isolates identified as A. hydrophila were screened for ARGs by PCR amplification, revealing intrinsic multi-antibiotic resistance. These included genes coding for chloramphenicol resistance (5/5), major facilitator superfamily efflux transporter (5/5), CphA class B beta lactamase (5/5), FOX/MOX class C beta lactamase (4/5), and OXA-12 class D beta lactamases (5/5) (Figure 1). All strains used in this study were susceptible to ciprofloxacin, cotrimoxazole, and gentamicin.

Wastewater and pond samples collected from the cities of Bendigo and Melbourne, Victoria, Australia were screened for bacteriophages. Ten bacteriophages against five clinical strains of A. hydrophila were isolated. Of these, eight were Podoviridae (comprising five icosahedral and three elongated Podoviridae), one was a Siphoviridae and one was a Myoviridae virus. The novel bacteriophages were labeled LAh1–LAh10, with LAh1–LAh5 representing the icosahedral Podoviridae, LAh6, LAh8, and LAh9 representing the elongated version of the Podoviridae, while LAh7 and LAh10 were Siphoviridae and Myoviridae bacteriophages respectively (Figure 2). The elongated Podoviridae bacteriophages had capsid length ≈ 172 ± 10 nm, width ≈ 35 ± 2 nm and tail length ≈ 18 ± 1 nm while the icosahedral Podoviridae had capsid diameter ≈ 82 ± 4 nm and tail ≈ 8 ± 1 nm. The Siphoviridae bacteriophages had capsid length and tail length of ≈ 44 ± 3 nm and ≈232 ± 13 nm, respectively. Myoviridae bacteriophages had capsid diameter of ≈ 116 ± 13 nm and tail length ≈ 183 ± 5 nm (Figure 2). Specific features of the bacteriophages and their characteristics are summarized in Table 2 while their respective one-step growth curves are shown in Figure 3.

Illumina sequencing revealed novel and diverse genomes with several displaying a similar size of approximately 42,000 bp (LAh1–LAh5). While the genomes of LAh1–LAh5 were the most similar to each other, specific differences were seen, and these resulted in non-synonymous amino acid changes (differences in their genomes and amino acid sequences are highlighted in Table 3). Larger genomes were found in LAh7 (61,426 bp), LAh8 (97,408 bp), LAh9 (97,988 bp), and LAh6 (101,437 bp). LAh10 had the largest genome (260,310 bp) which is the also the largest reported to date against A. hydrophila. The genomes for LAh1–LAh10 were annotated and submitted to GenBank with accession numbers shown in Table 2. All bacteriophages had their genomes in a linear topology except LAh6 which had a circularly permuted genome. The number of ORFs varied with the genome size and ranged from 45 to 227 (Table 2). The GC% content ranged from 42.2% in LAh8 to 61.9% in LAh7. While no bacteriophage showed presence of tmRNA in their genomes, tRNAs were present in 4/10 of the bacteriophages (Table 2). The number or type of tRNAs were not associated with genome size. LAh6 and LAh8 had the highest number of tRNAs at 21 each, followed by LAh9 with 18 tRNAs and LAh10 with 4 tRNAs. Therefore, the bacteriophages with a lower GC% content had a higher number of tRNAs (Table 2). Bacteriophages LAh1–LAh5, and LAh7 did not have any tRNAs in their genomes. None of the bacteriophages isolated in this study had genes coding for putative CRISPR sequences, ARGs, toxins or chromosome integration genes in their genomes.

Prior to this study there were 30 reported bacteriophages against Aeromonas spp. including eight against A. hydrophila which have had their genome completely sequenced. The 10 bacteriophages reported here are the first lytic bacteriophages to be isolated against clinical strains of A. hydrophila. The complete genome sequences of the previously isolated bacteriophages against Aeromonas spp. sourced from GenBank and those from this study were compared. Figure 4 reveals the clustering of bacteriophages against clinical strains of A. hydrophila forming two clusters among other bacteriophages isolated using A. hydrophila strains from the environment but further away from those against other Aeromonas species.

These differences among the isolated bacteriophage genomes were further analyzed by Mauve whole genome alignment of LAh1–LAh10 (Figure 5). The differences in the bacteriophage cluster of LAh1–LAh5 is detailed in Table 3 to highlight functional differences between these bacteriophages. For the cluster of bacteriophages LAh6, LAh8, and LAh9, the Mauve alignment showed the presence of a genetic segment (Blue colored colinear block in Figure 5) in the region of the putative tail fiber genes of bacteriophages that possibly contributed to their differences in host range. While similar in their genomes, LAh6 and LAh8 are able to lyse A. hydrophila strain AHB0139 that LAh9 is not (Table 2). The genome alignments show close homology between LAh1 and LAh5, and a diversity of nucleotide sequence between genomes of LAh7 and LAh10.

The capacity to disrupt A. hydrophila biofilms was analyzed quantitatively by evaluating the biofilm mass remaining after bacteriophage treatment and estimating the viability of the remnant biofilm. A representative bacteriophage from each morphological group was analyzed for capacity to disrupt biofilm. LAh1 was used as an example of icosahedral Podoviridae, LAh9 for the elongated Podoviridae, LAh7, a Siphoviridae, and LAh10 a Myoviridae. All biofilm experiments were performed using A. hydrophila bacterial strain AHB0116 that was lysed by all bacteriophages isolated here except for LAh1, in which case the host AHB0147 was used. In all cases, the biofilm mass was significantly reduced in bacteriophage treated compared to non-treated groups (p < 0.001). The untreated biofilm had a median (IQR) absorbance at OD_550nm of 1.73 (0.86) whilst the largest value for the remnant biofilm after bacteriophage treatment was 1.20 (0.27), p < 0.001 (following treatment with LAh9). Treatment with LAh1 resulted in the lowest absorbance for the remnant biofilm [median (IQR) at OD_550nm of 0.35 (0.04)], significantly lower (p < 0.001) than those treated with all the other bacteriophages. LAh7 and LAh10 had statistically similar (p = 0.58) remaining biofilm with absorbance of median (IQR) at OD_550nm of 0.43 (0.29) and 0.48 (0.13), respectively (Figure 6). Of the bacteriophages tested on the A. hydrophila biofilms, those with higher GC% content and lower numbers of tRNAs [LAh1 (59.30%; 0), LAh7 (61.90%; 0), and LAh10 (47.50%; 4)] showed significantly greater capacity to degrade biofilms (p < 0.001) than that with lower GC% content and higher numbers of tRNAs [LAh9 (42.40%; 18)] (Figure 6).

The viability of the biofilm mass was investigated using SYBR Gold^® and PI live/dead staining. Figure 7 shows cells fluorescing green (whole population of cells making up the biofilm) and red (dead, membrane-compromised cells). The icosahedral Podoviridae Bacteriophage LAh1 on host strain AHB0147 was used in the biofilm viability experiments. Figure 7 shows untreated biofilm with a sparse population of membrane-compromised cells (Figure 7B1) compared to a dense total population (Figure 7B2). This indicated that cell population in the untreated biofilm was comprised of mostly membrane-intact cells. This was in contrast to the biofilms treated with bacteriophage (Figures 7A1,A2) in which the density was comparable between total population and membrane-compromised cells implying the cell population was mostly dead. In the bacteriophage treated biofilm, the total population and dead cells were both sparse. Treatment with heat inactivated (autoclaved) bacteriophage did not affect biofilm growth, indicating that Aeromonas biofilm disruption was the result of bacteriophage particles, and not other material in the preparation.