Complete genome sequences of Aeromonas and Pseudomonas phages as a supportive tool for development of antibacterial treatment in aquaculture

Aeromonas and Pseudomonas are considered one of the most important fish pathogens among the etiological agents of bacterial fish diseases with capacity of hemolysis and biofilm formation [1‐3]. These aquatic bacteria are responsible for ulcer type diseases including ulcerative syndrome, bacteria haemorrhagic septicaemia, tail and fin rot, bacteria gill rot and dropsy [4‐10]. The increasing prevalence of bacterial infections leads to the indiscriminate use of antimicrobials that are the most common solution in combating pathogenic microorganisms. However, in the case of aquaculture, the range of authorized antibiotics is very narrow, e.g. the Food and Drug Administration (FDA) in 2014 approved only oxytetracycline, florfenicol, and sulfadimethoxine/ormetoprim [4]. In addition, the vast majority of bacteria pathogenic to aquaculture are resistant to multiple antibiotics [11, 12]. That forces fish farmers to look for alternative solutions that allow effective protection of breeding populations. Bacteriophages (phages) are one alternative solution that could be considered as a prospective anti-microbial strategy in aquaculture [1, 13‐24]. Bacteriophages are the most abundant biological entities on Earth. In almost all ecosystems that so far have been subjected to in-depth studies, it has been estimated that there are around ten phages for every microbial cell giving approximately10³⁰–10³¹ phages globally [25, 26]. However, their genomic sequences represent a small fraction in public databases. Only 1956 bacteriophage genomes are currently available in the NCBI genome database [National Center for Biotechnology Information (NCBI) Genome database https://​www.​ncbi.​nlm.​nih.​gov/​genome . Accessed 27 Nov 2018], of which 19 are genome sequences of phages infecting Aeromonas (phages belonging to Myoviridae, Podoviridae, and Siphoviridae families) and 144 infecting Pseudomonas (phages belonging to Myoviridae, Podoviridae, Siphoviridae, Cystoviridae, Leviviridae, Inoviridae families and 5 unclassified). Phages useful for therapeutic purposes must meet numerous criteria, from which the most important is their lytic nature. Only whole genome sequencing (WGS) enable an in-depth characterization of phages and the choice of the phages suitable for phage therapy. In this study, we present 7 new phages infecting Aeromonas and 4 new phages infecting Pseudomonas that significantly increase the genomic information of Aeromonas and Pseudomonas phages. Furthermore, we present results of bioinformatics analysis of genomic data assessing the likelihood of these phages to enter the lysogenic cycle and experimental data on their specificity towards a large number of bacterial field isolates representing different location. Taken together these data provide an essential basis for rational selection of bacteriophages for application in phage therapy of affected populations.

Thirty six Aeromonas spp. and nine Pseudomonas sp. bacterial strains were isolated from infected fish (Table 1.) and eleven bacteriophage strains were isolated from environment: seven active against Aeromonas spp. and four against Pseudomonas sp., for which 5 Aeromonas hydrophila and 2 Pseudomonas fluorescens strains were used routinely as the hosts (Table 2.).

For all of the phages we assessed the host range: in the case of the Aeromonas phages with a panel of 49 Aeromonas spp. isolates (A. hydrophila, A. salmonicida, A. sobria) and in the case of Pseudomonas phages with 9 P. fluorescens isolates (Table 3.). The resulting host range patterns were different for each tested phage. The broadest host range had 13AhydR10PP, 14AhydR10PP, 85AhydR10PP, 22PfluR64PP and 67PfluR64PP (42–51% of bacterial isolates were sensitive to these phages). Phage 25AhydR2PP had the narrowest host range with only 4/49 (8%) of bacterial isolates being sensitive to it. Overall, these phages showed different but complementary host ranges.

Visualization of bacteriophages by transmission electron microscopy showed that the tested bacteriophages 13AhydR10PP, 14AhydR10PP, 50AhydR13PP, 60AhydR15PP, 85AhydR10PP consisted of a polyhedral head and tail, which allowed classifying them in the order Caudovirales. In addition those phages were found to have a contracted tail characteristic for viruses belonging to the family Myoviridae. In contrary, phages 22PfluR64PP, 25AhydR2PP, 62AhydR11PP, 67PfluR64PP, 71PfluR64PP, 98PfluR60PP consisted of a polyhedral head and very short tail characteristic for viruses belonging to the family Podoviridae (Fig. 1).

Afterwards, isolation of DNA and restrictive analysis with enzymes: SspI and EcoRI were carried out. Obtained restriction profiles (Additional file 2: Figure S1.) allowed for the definition of initial genetic characteristics of the bacteriophages. Subsequently, after NGS sequencing (Additional file 1: Table S1.), a detailed genetic analysis of bacteriophages was performed (Table 4.). It was found that phages 13AhydR10PP, 14AhydR10PP and 85AhydR10PP possess genome sizes about 47–48 kbp and belong to double-stranded DNA viruses of Myoviridae family with circular genomes. They are homologues of bacterial viruses: Aeromonas phage 32, Aeromonas phage Asp37, Aeromonas phage 3, Aeromonas phage Ahp2 and Aeromonas phage 59.1. Moreover, their lifestyles were classified as lysogenic after conducting bioinformatic analysis, i.e. analysis of presence of genes coding for certain characteristic proteins (such as integrases or resolvases) among their ORFs and analysis of their amino acid sequences in PHACTS. They are highly similar to each other, with 96% query cover and 96% identity between 13AhydR10PP and 14AhydR10PP and 78% query cover and 89% identity between 13AhydR10PP or 14AhydR10PP and 85AhydR10PP. Phages 50AhydR13PP and 60AhydR15PP were also classified to Myoviridae family (Caudovirales order), containing linear double-stranded DNA (with circularly permuted genome) in size of approximately 165 kbp, but showing high similarity to the lytic Myoviridae bacteriophages specific against many bacteria from Aeromonas sp. They are highly similar to each other, with 94% query cover and 97% identity. PHACTS classified both of them as confidently lytic. Unclassified phage 62AhydR11PP with genome size of about 44 kbp showed low similarity with Aeromonas-infecting phages of Myoviridae family. It is also similar to the group of viruses with PLPE-like virion morphology. However TEM analysis allowed to classify this phage to Podoviridae family. Careful analysis of its ORFs allowed for the classification of this phage as probably lysogenic. Phages 25AhydR2PP, 22PfluR64PP, 67PfluR64PP and 71PfluR64PP belong to Caudovirales, Podoviridae, Autographivirinae with short, unshrinkable tails and icosaedral capsid containing linear double-stranded DNA of approximately 40 kbp in size. They showed a high similarity to lytic bacteriophages of T7 group specific to bacteria of the Aeromonas and Pseudomonas sp. Genome representations of these phages are linear with direct terminal repeats (DTRs). Among their ORFs no proteins responsible for lysogeny were found. Therefore, it was assumed that they exhibit lytic lifestyle. Phage 98PfluR60PP with genome size about 74 kbp has one reference genome in the NCBI database, i.e. Pseudomonas phage Littlefix active against Pseudomonas sp. It is classified into Podoviridae family, however with ORFs showing no or very little similarity to any known phage proteins therefore it was impossible to classify the genome of 98PfluR60PP as lytic or lysogenic on the basis of current knowledge. Labeled genetic maps (linear or circular depending on the genome) of all analyzed phages are presented in (Additional file 3: Figure S2, Additional file 4: Figure S3, Additional file 5: Figure S4, Additional file 6: Figure S5, Additional file 7: Figure S6, Additional file 8: Figure S7, Additional file 9: Figure S8, Additional file 10: Figure S9, Additional file 11: Figure S10, Additional file 12: Figure S11 and Additional file 13: Figure S12.). Afterwards, all genomes were subjected to phylogenetic analysis (Fig. 2.). The related phages can be divided into the following groups, along with the systematic classification given by GeneBank: Gr.1: 22PfluR63PP, 67PfluR64PP, 71PfluR64PP with the reference strain Pseudomonas phage PFP1, belong to the genus T7virus, Gr.2: 13AhydR10PP, 14AhydR10PP, 85AhydR10PP with the reference strain Aeromonas phage 32 are most likely to belong to the Myoviridae family, Gr.3: 25AhydR2PP together with the reference strain Aeromonas phage phiAS7 belong to the sub-family Autographivirinae, Gr.4: 98PfluR60PP with the reference strain Pseudomonas phage Littlefix belong to the Podoviridae family, Gr.5: 50AhydR13PP, 60AhydR15PP with the reference strain Aeromonas phage phiAS4 belong to the genus unclassified Secunda5virus.

The 62AhydR11PP phage appears to be unique. It has low similarity to phage Aeromonas 56 which is reflected on the phylogenetic tree, and they do not occur in the same clade. 62AhydR11PP bacteriophage due to the lack of similar sequences in the NCBI database and separation on the phylogenetic tree may belong to the group of phages that has not been recognized yet. Finally, all of the studied phages are deprived of any antibiotic resistance or virulence genes according to conducted bioinformatics analysis.

Among the entire population of phages only a few have the potential for use in phage therapy. It was previously estimated that only 50% of phages isolated from the environment are useful for therapeutic purposes [39]. However, the development of new analytical methods, including WGS, can change this proportion. Whole genome sequencing is an indispensable tool used in the study of phage biology [40‐43]. WGS facilitates a detailed characterisation of phages that allows them to be classified as useful for therapeutic purposes. Phages that are promising for phage therapy should be excluded of toxin genes, virulence factor genes, and genes responsible for lysogeny [44‐47]. In this study, we demonstrated 11 new bacteriophages among which 6 (25AhydR2PP, 50AhydR13PP, 60AhydR15PP, 22PfluR64PP, 67PfluR64PP, 71PfluR64PP) have a potential to be used in phage therapy due to confirmed lytic lifestyle and absence of virulence or resistance genes. At the same time, we observed (Table 3.) that most of the bacteriophages identified exhibited relatively narrow specificity to bacterial isolates. Selected Aeromonas phages revealed activity against only 8–35% of the whole examined Aeromonas collection but at the same time shows specificity towards A. hydrophila, A. salmonicida and A. sobria. This characteristic is contrary to previously described broad host range Aeromonas phages that act against only one Aeromonas species [13]. Narrow specificity of selected phages might create an obstacle in the development of effective phage treatment for Aeromonas sp. and Pseudomonas sp. infections. One potential way to overcome this challenge would be the creation of a multicomponent phage cocktail, consisting of completely characterized lytic bacteriophages. Using a mixture of bacteriophages to expand the coverage of treatment for heterogenous bacterial populations in bacteriophage therapy has shown high efficacy for many years. However, in most, if not all cases such bacteriophage mixtures do not consist of fully characterized bacteriophages [48, 49]. Selecting multiple components, that fulfil certain criteria, such as full genomic information, a lack of genes that create safety concerns, negligable risk of lysogenic lifestyle, and reproducible stability in the production environment, creates a significant challenge.

Bacteriophages isolated in this study have also significantly increased the knowledge about Aeromonas and Pseudomonas phages. In currently available genomes in the NCBI database among 19 described Aeromonas phages, 16 belong to Myoviridae (84% of whole population), only 2 to Podoviridae (10% of whole population) and 1 to Siphoviridae (6% of whole population). Myoviridae family members are most likely to be abundant in natural environments [13, 39, 50‐52] and when compared to public databases, it is represented on a similar level in this study (71%). A comparable situation can be observed for Podoviridae phage population which is represented by 14% of Aeromonas phages in this study. However among the described collection there was one unclassified phage proving to be unique comparing in the public database. In the case of Pseudomonas phages, 41 genomes from the database belong to Myoviridae (28% of whole population), 46 belong to Podoviridae (32% of whole population), 44 belong to Siphoviridae (31% of whole population), 4 belong to Cystoviridae (3% of whole population), 2 belong to Leviviridae (1% of whole population), 2 belong to Inoviridae (1% of whole population) and 5 are unclassified (3% of whole population). When compared to the publicly available genomes, in the collection described in this study, only the Podoviridae family is represented and no other phage families were observed in the analyzed set of Pseudomonas phages. The summary of these findings is presented in Table 5.

To conclude, 11 new bacteriophages were isolated and described on genomic level. Of these, only six have potential for phage therapy in aquaculture due to their proven lytic nature and their lack of antibiotic resistance and virulence genes. Four of them belong to the Podoviridae, while two to the Myoviridae family. The composition of these phages could be used as a therapeutic cocktail giving the cover of 41% of the Aeromonas and 44% of Pseudomonas pathogenic environmental isolates. Other phages described in this study should be excluded from any therapeutic cocktail composition due to identification of genes responsible for lysogenesis in their genomes.