Pseudomonas syringae pv. tomato DC3000 (Pst) were cultured at 28°C in Luria-Bertani (LB) medium (tryptone, 10 g/L; NaCl, 10 g/L; yeast extract, 5 g/L) containing 50 μg/ml rifampicin, in sealed tubes on a rotary shaker at 180 rpm or on Petri plates (in this case medium was supplemented with 1.4% agar). For growth rate evaluation, bacterial suspension was cultivated in 96-well plates. Growth was evaluated by continuously measuring the optical density of the culture at 600 nm wavelength (OD₆₀₀) on TECAN Infinite^® 200 PRO (Switzerland).

Bacteriophages were isolated from pepper fruits with bacteriosis symptoms, typical for bacterial speck induced by Pst (Pernezny et al., 1995). The initial amount of bacteriophages in the samples was amplified by the enrichment method (Zaika et al., 2013; Kharina et al., 2015). The macerated tissue from brown rot lesions was transferred to liquid LB medium and incubated for 48 h at 28°C without shaking. After incubation, the suspension was centrifuged (3,400 g, 25 min, Thermo Heraeus Multifuge 3S Centrifuge, Sorvall 75006445 Swing Bucket Rotor), and the supernatant was mixed with chloroform 1:1 to remove bacteria. The upper aqueous phase was then plated on a bacterial lawn by the agar overlay method (Kropinski et al., 2009). Separate phage plaques were picked and transferred to 0.9% NaCl sterile solution (1 ml). Infectivity of isolated bacteriophages was tested on Pst by spot test. Briefly, Pst was incubated overnight in LB broth at 27°C and 180 rpm (OD₆₀₀ = 3–4). Then, 200 μl of bacterial culture was added to 6 ml of 0.7% LB soft agar (preheated up to 45°C) and poured onto Petri dishes containing solid agar (LB medium supplemented with 1.4% agar) (both LB agar layers containing 50 μg/ml rifampicin). The plates were left to dry for 10 min. Subsequently, 5 μl of the phage lysate (10⁹ PFU/ml) was spotted on the bacterial lawn and incubated overnight at 28°C for formation of lysis zones.

Phages were propagated and purified as described previously (Bonilla et al., 2016). Plaques of different morphology were picked from a plate with numerous negative colonies using a sterile Pasteur pipet and individually plaque-purified further. Plaques were resuspended into a microcentrifuge tube containing 1 ml SM buffer (5.8 g NaCl, 2.0 g MgSO₄⋅7H₂O, 50 ml 1 M Tris-HCl pH 7.4, in 1 L dH₂O), vortexed for 5 min, and then centrifuged at 3,400 g for 5 min. Using a double-agar layer method, 10 confluent-lysis plates were obtained. Next, 5 ml of SM buffer were poured on top of the plates, and plates were gently shaken for 15 min at room temperature. Buffer from the top of the plates was collected and centrifuged at 4,000 g for 20 min. The resulting supernatant was filtered through a 0.22 μm syringe filter (Millex-GS Syringe Filter Unit, Millex, Germany), and 0.1 volumes of chloroform were added. Samples were vortexed and centrifuged again at 3,400 g for 5 min; the supernatant was then stored at 4°C. Phage titer was assessed by the agar overlay method. Phage samples were then concentrated by ultracentrifugation at 50,000 rpm for 6 h (The Optima™ XPN, Beckman Coulter, rotor 70-Ti). The pellet was resuspended in 1 ml of SM buffer.

The samples were prepared for electron microscopy by applying 5 μl of concentrated purified virions (5.0 × 10⁹ PFU/ml) onto formvar-coated copper transmission electron microscopy (TEM) grids (3 mm, 300 mesh, Agar Scientific Ltd., Essex, United Kingdom) and incubated for 10 min at room temperature. Grids were then negatively stained with 2% uranyl acetate. Specimens were examined using the transmission electron microscope FEI (Morgagni 268D).

The phage host range was assessed by spotting 5 μl of phage suspension on plates with bacterial lawn, prepared similarly as for the phage purification and titer evaluation (Kropinski et al., 2009). The following bacterial strains were used: Pseudomonas syringae pv. tomato DC3000; P. syringae pv. tomato CUCPB5112 hrcC-, derived from Pst DC3000; P. syringae pv. syringae 4073; P. syringae pv. morsprunorum OPPB; P. syringae pv. maculicola 4326; P. syringae 8511; P. syringae pv. syringae 1022; P. syringae pv. atrofaciens 4394; P. syringae pv. coronafaciens 9030; P. aeruginosa; P. putida P-14; P. putida B-115T; P. pseudoalcaligenes; P. fluorescens DBM3506; P. veronii DBM3208; P. fluorescens cmm 2115; P. chlororaphis 4341; Pantoea agglomerans pv. gypsophilae 2406; Agrobacterium rhizogenes 3113; A. tumefaciens LBA4404; Clavibacter michiganensis ssp. michiganensis 7035; Erwinia amylovora 8/95; E. amylovora 5032; E. carotovora pv. carotovora IVIA 241/1; Pectobacterium carotovorum; P. carotovorum ssp. carotovorum 1077; Ralstonia solanacearum 2505; Xanthomonas arboricola pv. juglandis CCM 1449; and X. vesicatoria 2101 (Table 1). After 24 h incubation at 26°C, phage efficiency was evaluated by the presence/absence and relative size of the lysis zones on the bacterial lawn.

Phage adsorption dynamics was assessed as previously described (Adriaenssens et al., 2012). Briefly, overnight culture of Pst DC3000 in liquid LB medium supplied with 50 μg/ml of rifampicin was diluted to an OD₆₀₀ = 0.4 and mixed with phage suspension at a multiplicity of infection (MOI) of 0.001. Immediately after mixing, 100 μl of the suspension was transferred into 2 ml tubes containing 850 μl LB medium with 50 μl chloroform. This stage was repeated every minute into the new tubes. Mixtures were then shaken for 10 min to remove bacterial debris. The supernatant was serially diluted (1:10), and the amount of non-adsorbed phages was assessed by the agar overlay method. Adsorption rate constant was calculated according to Rombouts et al. (2016) as [k = (2.3/(B × t)) × log(P0/P)], where B is the bacterial titer at time zero, P is the phage titer, and t is the time.

Phage thermal and pH stability was tested by incubating a phage suspension of 10⁹ PFU/ml in SM buffer at different temperatures (from -20 to 50°C) and in pH buffer ranging 1–13 (SM buffer adjusted to a given pH with NaOH or HCl) for 6 and 24 h and quantifying the phage titer by spot test.

Killing curves were obtained after infecting the Pst DC3000 (OD₆₀₀ = 0.1) with phages at MOI of 0.1 and 0.01. Mixtures of Pst with the studied phages, as well bacterial and phage controls, were prepared in transparent 96-well plates and incubated at 26°C for 5 h on a microplate reader (TECAN Infinite 200 PRO, Switzerland); adsorption at λ = 600 nm was measured every 5 min.

The frequency of bacterial resistance was determined using the previously described method (Rombouts et al., 2016). Phage and bacteria were plated at MOI 10 to obtain the full lysis of the bacterial lawn on plates. After 72 h of incubation, emerging colonies were counted, and four independent colonies (two for each phage) were cultured and retested for resistance against both phages. The frequency of phage-resistant mutant occurrence was calculated by dividing the number of the resistant bacteria by the total number of sensitive bacteria.

Phage genomic DNA was extracted using the protocol for phage DNA extraction with phenol:chloroform (Center for Phage Technology, Texas A&M University). Briefly, phage lysates were precipitated with PEG (10% PEG-8000, 1 M NaCl final concentration) and incubated at 4°C overnight. The PEG-precipitated sample was centrifuged at 10,000 g for 30 min, and the pellet was resuspended in 5 mM MgSO₄. The subsequent steps were the addition of 1.25 μl of DNase I and RNase (20 mg/ml) and incubation at 37°C for 1 h. Afterward, 1.25 μl of proteinase K of 20 mg/ml stock (20 μg total), 25 μl of 10% SDS stock (0.5% final concentration), and 20 μl of 0.5 M EDTA, pH 8.0 (20 mM final), were added, and the sample was incubated for 1 h at 60°C. Phenol:chloroform (1:1) was then added before centrifugation for several times, using decanted supernatants each successive time. Samples were incubated with 3 M NaOAc (pH 7.5) and 2.5 volumes of ice-cold ethanol (100%) overnight at -20°C. Next day, the samples were centrifuged at 10,000 g for 20 min in the bench centrifuge, and the pellet was washed with 70% ethanol twice. The pellet was then dissolved in TE buffer and stored in -20°C before proceeding with the next steps. The REPLI-g Single Cell Kit (Qiagen, Venlo, Netherlands) was used to increase the amount of DNA via whole-genome amplification (WGA) as per manufacturer’s instructions. To prepare the input for phage whole-genome sequencing, 200 ng of whole-genome-amplified phage genomic DNA was subjected to sonication using a Covaris S220 focused ultrasonicator (Covaris, Woburn, MA, United States) using a protocol for a target fragment length of 550 bp. Fragmented phage genomic DNA was then used to prepare DNA fragment libraries for 250 bp paired-end read sequencing using the TruSeq DNA Nano Low-Throughput Library Prep Kit (Illumina, San Diego, CA, United States) according to the manufacturer’s protocol. Adapters #2 and #4 from TruSeq DNA Single Indexes Set A (Illumina) were used for Eir4 and Eisa9 DNA fragment libraries, respectively. The resulting libraries were inspected for the quality and quantity of their contents using an Agilent 2100 bioanalyzer (Agilent, Santa Clara, CA, United States) with a High Sensitivity DNA Kit (Agilent), as well as a Qubit fluorometer (Invitrogen, Waltham, MA, United States) dsDNA high-sensitivity quantification assay (Invitrogen). Sequencing of the libraries was performed on the Illumina MiSeq system (Illumina) with a 500-cycle MiSeq Reagent Kit v2 nano (Illumina), with Eir4 and Eisa9 libraries being two of the 12 pooled libraries with different barcodes.

Demultiplexed raw read statistics of sequenced Eir4 and Eisa9 libraries were collected with the help of FastQC v0.11.9 (Andrews, 2010), revealing that the sequencing run has yielded 46,264 (Eir4) and 47,759 (Eisa9) read pairs. Read trailing base removal, as well as quality (bases below Phred quality of 20 removed) and length (50 bp minimum length of the read post-trimming) trimming, was performed using bbduk from the BBmap package (Bushnell, 2014). The trimmed read pairs were then subjected to de novo assemblies using Unicycler v0.4.8 in normal mode (Wick et al., 2017), yielding “circular” contigs of 40,177 and 40,730 bp length for Eir4 and Eisa9, respectively. Raw read mapping unto the assembled contigs representing complete genomes of the studied phages revealed that the mean depth of both is more than 550×.

These circular assemblies were then subjected to BLASTN analysis (Altschul et al., 1990) against all the complete viral genome sequences publicly available (nr/nt database restricted to virus taxid: 10239; other settings default). The total top-scoring hit to Pseudomonas phage JOR (MZ826346.1) with query coverage of 99% and 98.8% identity was documented for the Eir4 genome assembly as a query (e-value 0.0). The Eir4 genome assembly was then reverse-complemented and “linearized” to begin with base 30416 of the circular contig to ensure the assembled Eir4 genome collinearity with the complete genome of the already publicly available phage JOR. The circular assembly of Eisa9 was similarly reverse-complemented and “rearranged” to ensure the collinearity with the complete genome of the top-scoring Pseudomonas phage PollyC (NC_042104.1) hit, the alignment to which resulted in 87% query coverage of 77.34% identity when the Eisa9 complete genome was used as a query (e-value 0.0).

The rearranged linearized contigs representing complete phage genomes were then used as an input for an open reading frame (ORF) and tRNA gene prediction using tools available in the DNAmaster v5.23.6¹ [for ORF prediction, Glimmer (Delcher et al., 2007) and GeneMark (Besemer and Borodovsky, 2005); for tRNA, Aragorn (Laslett and Canback, 2004) and tRNAscan (Lowe and Eddy, 1997)]. During the ORF prediction, four possible start codons (ATG, GTG, CTG, and TTG) were considered, and only ORFs encoding for a product with a length greater than 30 amino acids (aa) were documented and subjected to further supervised product functional annotation as described in Zrelovs et al. (2021a), with a modification that 3’-AUUCCUCCACUAG-5’ was used as an anti-Shine-Dalgarno sequence of Pseudomonas syringae pv. tomato (host species of phages Eir4 and Eisa9) for free_align.pl (Starmer et al., 2006).

For packaging strategy/expected genome physical termini type prediction, a set of TerL amino acid sequences from phages with an experimentally verified packaging strategy used previously by Merrill et al. (2016) was used to build a packaging strategy prediction maximum-likelihood tree based on TerL homology of different packaging-type employing phages with the addition of respective protein sequences from Pseudomonas phages Eir4 (UGL61101.1) and Eisa9 (UGL61147.1). Multiple amino acid sequence alignment of TerL proteins was performed using MAFFT v7.453 in automatic mode (Katoh and Standley, 2013), a maximum-likelihood phylogeny was reconstructed in IQ-TREE v 2.0.6. (Nguyen et al., 2015) under VT + F + R4 (Müller and Vingron, 2001; Soubrier et al., 2012) as the best-fit substitution model as determined by ModelFinder (Kalyaanamoorthy et al., 2017), while allowing for polytomies. Ultrafast bootstrap (UFBoot) with 1,000 replicates was used to determine the branch support (Minh et al., 2013). The resulting tree was midpoint rooted and visualized using FigTree v.1.4.4,² and shapes of distal nodes of branches having less than 95% UFBoot support were manually removed using Inkscape v1.0.1.³

To infer the genome nucleotide sequence evolutionary relationship of phages Eir4 and Eisa9 with other sequenced Pseudomonas sp.-infecting phages from the family Autographiviridae (podophages encoding their own RNA polymerase), the respective phage complete genome sequence accession numbers were retrieved from ICTV Virus Metadata Repository (VMR number 18, October 19, 2021, MSL36) for Pseudomonas-infecting phages belonging to the family Autographiviridae that are currently recognized by ICTV (hereafter referred to as VMR subset; n = 49). Accession numbers of the top 10 highest-scoring BLASTN hits not seen in this VMR subset were also retrieved for both Eir4 and Eisa9 complete genome queries. To be noted, for Eisa9, nine out of these top 10 non-VMR subset hits were phages of hosts other than Pseudomonas that had very little intergenomic similarity to Eisa9 but were nevertheless retained for context. After downloading the aforementioned phage complete genome entries, the resultant dataset consisting of 71 phage complete genome nucleotide sequences (n = Eir4 + 10 + Eisa9 + 10 + 49 = 71) was used as an input for VIRIDIC v.1.0 (Moraru et al., 2020) to calculate pairwise intergenomic distances between the phages.

Phylogenies were reconstructed for three of the selected phage Eir4 and Eisa9 marker proteins known to be evolutionary conserved among related Autographiviridae phages: major capsid protein (MCP; Eir4: UGL61088.1; Eisa9: UGL61137.1), terminase large subunit protein (TerL; Eir4: UGL61101.1; Eisa9: UGL61147.1), and RNA polymerase (RNAP; Eir4: UGL61062.1; Eisa9: UGL61131.1) using 22 presumed respective protein amino acid sequences (from phages Eir4 and Eisa9, as well as the 10 highest-scoring BLASTp hits against the respective Eir4 and Eisa9 protein amino acid sequence queries).

Multiple sequence alignments (MSAs) were performed using ClustalO v.1.2.4 in the automatical alignment option selection mode (Merrill et al., 2016), and the respective MCP, TerL, and RNAP MSAs were used as an input for maximum-likelihood phylogeny reconstruction in IQ-TREE v.2.0.6 (Nguyen et al., 2015) under WAG + G4 (Yang, 1994; Whelan and Goldman, 2001) (for MCP) and LG + G4 (Yang, 1994; Le and Gascuel, 2008) (for TerL and RNAP) as the best-fit amino acid substitution models according to the Bayesian information criterion after ModelFinder inference (Kalyaanamoorthy et al., 2017) and allowing for polytomies. Ultrafast bootstrap (UFBoot) with 1,000 replicates was used as a measure of branch support (Minh et al., 2013). Resulting trees were midpoint rooted and visualized using FigTree v.1.4.4 (see text footnote 2), and shapes of distal nodes of branches having less than 95% UFBoot support were manually removed using Inkscape v1.0.1 (see text footnote 3).

For pairwise genome nucleotide sequence/genome organization comparison of Pseudomonas phages Eir4 (OL581611) and Eisa9 (OL581612) and their closest known relatives with the standing in the current phage taxonomy (phage 17A (NC_048201) for Eir4 and PollyC (NC_042104) for Eisa9), as well as phage T7 (NC_001604; as arguably the most known and studied Autographiviridae phage), the respective complete annotated genomes were retrieved in the form of annotated GenBank entries. These entries were then manually color-coded by the addition of a “/color = R G B” qualifier for every ORF based on the functional annotation of its product as seen in the downloaded entries (without reannotating anything). Complete genomes of these selected phages (as well as Eir4 and Eisa9) with coding sequences colored based on their product function were then used as an input for Easyfig v.2.2.3 using the TBLASTX and BLASTN comparison options and drawing only the hits with a minimal length of 50 and maximum e-value of 1e-5 (Sullivan et al., 2011).

Biological activity testing in planta was based on the approach of Álvarez et al. (2019). A. thaliana plants were cultivated on a solid medium and infected by flood inoculation (Ishiga et al., 2011). Stratified sterile seeds (n = 9–10 per plate) were cultivated on a solid MS/2 medium containing 2.2 g Murashige and Skoog medium (Duchefa, Netherlands). Seedlings were grown in Petri dishes for 12–14 days under a long-day photoperiod (16 h/8 h light/dark regime) at 100–130 μE m^–2 s^–1 and 22°C. On day 14, plates were flooded with a suspension of an overnight culture of Pst DC3000 in 10 mM MgCl₂ (OD₆₀₀ = 0.01) containing 6.25 μl of Silwet L-77 in 50 ml of suspension for 2–3 min and phage at MOI = 4. Control (mock) plants were flooded with phage suspension and buffer only. Samples were collected at 2 dpi (days post inoculation); each sample contained pooled tissues from rosettes of two to three plants (roots were cut off with scissors). Tissues were homogenized in 2 ml Eppendorf tubes with 1 g of 1.3 mm silica beads using a FastPrep-24 instrument (MP Biomedicals, United States). The homogenate was serially diluted in 10 mM MgCl₂ and pipetted onto LB plates containing rifampicin. The colonies were counted after 1–2 days of incubation at 28°C.

To investigate the dynamics of phage interaction with host bacteria, we performed adsorption tests and killing curve experiments. Both for Eir4 and Eisa9, 98% of phages were irreversibly adsorbed to the Pst host cell within 1 min, with adsorption rate k’ = 5.8 × 10^–10 for Eir4 and 6.34 × 10^–10 for Eisa9. This is quite similar to some other Pseudomonas phages (Sillankorva et al., 2008), but slower than phage φ6 (Vidaver et al., 1973). Killing curves were generated by monitoring the optical density of Pst suspension in liquid LB media at 600 nm (OD₆₀₀) in the presence of phages at MOI = 1 and 0.01. In the presence of Eir4, OD₆₀₀ rapidly decreased after 80 min (MOI = 1) and after 100 min (MOI = 0.01). For Eisa9, OD₆₀₀ decreased after 125 min for MOI = 1 and after 150 min at MOI = 0.01 (Figure 1B). Interestingly, the OD₆₀₀ decrease was only partial in the case of Eisa9, while the OD₆₀₀ for Eir4 dropped to the values of blank samples. We hypothesize that it might be a result of a slower phage life cycle, so the high concentration of bacteria that propagated in media by the time of burst was sufficient to fill the culture with cell debris. Another possible explanation could be the surfacing of the slow-growing phage-resistant bacteria.

After incubation of full lysis plates for 72 h, we detected the emergence of phage-resistant bacterial colonies. For Eir4, an average of 116 ± 10 colonies was observed per plate, which corresponds to the frequency of resistance of 3.5 × 10^–6, and for Eisa9, an average of 294 ± 5 resistant colonies appeared per plate, with the frequency of phage resistance emergence being 0.9 × 10^–5. Four independent randomly selected phage-resistant cultures acquired from such colonies appeared to be cross-resistant to both Eir4 and Eisa9 phages, although the mechanism of such resistance remains yet to be elucidated.

Assessment of thermal stability revealed that both Eir4 and Eisa9 were stable at a range of moderate temperatures from 4°C (storage temperature) to 26°C (Pst growth). A marked decline in phage viability was observed after 24 h of preheating at 37°C or freezing at -20°C (Figure 1C). This corresponds to the optimal temperatures for virulence of Pst in greenhouse conditions on host tomato plants (Gullino et al., 2009). As for the pH stability, both phages maintained infectivity in neutral or slightly alkaline conditions (pH = 7–9), while lower or higher pH was significantly reducing phage titers after 24 h of incubation (Figure 1D). Both phages regularly produced clear plaques as an indication of a non-lysogenic cycle.

Genomes of phages were sequenced, annotated, and deposited to GenBank as Pseudomonas phages Eir4 (accession OL581611) and Eisa9 (accession OL581612). Both genomes were similar in terms of genome length, GC% content that resembled that of their host (e.g., ∼58.4% GC content for P. syringae pv. tomato strain DC3000 chromosome complete sequence; NC_004578.1), coding capacities, and predicted ORF number (Table 2), as well as ORF functions (Supplementary Tables 1, 2). Although due to the WGA that was partaken before whole-genome sequencing to increase the amount of genomic phage DNA, the elucidation of the exact genome termini using raw read pile-up inspection onto the assemblies (e.g., using PhageTerm or manually) was impossible (bias of fragment non-randomness introduced by WGA); the completeness of the acquired phage genomes was implicated by the “circularity” of the genome de novo assemblies. Inside the virion, however, the genomes of studied phages Eir4 and Eisa9 should likely represent linear dsDNA molecules of 40,177 and 40,730 bp or longer (in case of short direct terminal repeat presence, which is the most likely packaging strategy employed by both Eir4 and Eisa9 based on the TerL aa sequence-based packaging strategy prediction; Supplementary Figure 2), so the beginning of both genomes was chosen according to the related phages already deposited in GenBank.

The presence of ORF coding for a DNA-dependent RNA polymerase in both of the genomes allowed us to classify both of these podophages as members of the phage family Autographiviridae (Adriaenssens et al., 2020). There is absence of almost any genomic nucleotide sequence similarity, as well as very little pairwise amino acid sequence identity, between the proteins with the same predicted function from Eir4 and Eisa9. This hinted at quite a distant evolutionary relationship between them, which is, given their enormous diversity, not that uncommon even for the phages infecting the same host and coming from the same source as in the case of Eir4 and Eisa9.

The genome organizations of Eir4 and Eisa9 differed too. Eir4 genome organization was reminiscent of that from arguably the most well-known Autographiviridae phage—T7—with the ORF being responsible for encoding RNAP preceding a presumable DNA replication, modification, and repair functional ORF module, which also contains a lysozyme-encoding ORF in between other non-functionally related protein-encoding ORFs and is followed by ORFs encoding proteins necessary for virion morphogenesis, with the remainder of lysis proteins being located within it at the 3’ end of the genome, with a terminase small subunit located between the holin- and spanin-encoding ORFs and the TerL-encoding ORF following the spanin-encoding ORFs as the last one in the morphogenesis functional module and the penultimate ORF in the genome of Eir4 (Figure 2 and Supplementary Table 1). The genome of phage Eisa9, however, has an organization that is quite different—the RNAP-encoding ORF is preceded by the presumable DNA replication, modification, and repair functional ORF module that does not contain any lysis-associated genes within it and is followed by a morphogenesis ORF module which contains an ORF encoding holin before two of the module’s last ORFs (which encode both terminase subunits), with endolysin and both i- and o-spanin-encoding ORFs following the morphogenesis module after two hypothetical protein-encoding ORFs (Figure 2 and Supplementary Table 2). Neither of the phages was found to encode any proteins necessary for temperate life cycle, complementing the microbiological observations that these phages are strictly lytic. No known AMR genes or putative virulence factors beneficial to the host were found to be encoded by either of the phages as well. The genome contents of Eir4 and Eisa9 suggest that both of these phages look like suitable candidates to be further evaluated for possibilities of their application in P. syringae pv. tomato phage-mediated biocontrol.

Investigation of possibilities for lesser taxon assignments for phages described herein revealed that Eisa9’s closest-known phage relative is the Pseudomonas phage PollyC, which is so far the only representative of Pollyceevirus virus genus. The intergenomic similarities between Eisa9 and PollyC (67–68%; Supplementary Figure 1; BLASTN e-value 0.0), however, fall behind the currently accepted genomic nucleotide sequence similarity genus-level assignment criterion (≥ 70%) for phages (Turner et al., 2021). In contrast, based on the phylogenies of the selected marker genes (Figure 3), as well as more-detailed genome organization and sequence comparison (Figure 2), Eisa9 might still be somewhat of an “edge case,” meaning that Eisa9 could potentially be proposed as an exemplar isolate of a novel species within the existing Pollyceevirus genus, rather than being viewed as a sole representative of a putative new phage genus. However, further analyses regarding the Eisa9 taxonomical standing are more fit for an ICTV taxonomic proposal and are out of the scope of this paper.

As for phage Eir4, based on the performed analyses, many closely related phages were found. Eir4 thus can convincingly be considered a member of the Studiervirinae subfamily and Ghunavirus genus, with its closest known relative having a standing in the current phage taxonomy being Pseudomonas phage 17A (∼91% intergenomic similarity with Eir4; BLASTN e-value 0.0). Yet, surprisingly, we found at least eight more phages without current standing in the official phage taxonomy available at public biological sequence repositories that show striking similarity to Eir4 (≥ 95% intergenomic similarity—isolates of the same species; Supplementary Figure 1; Turner et al., 2021).

Interestingly, the metadata associated with their annotated complete genome submissions indicate a very broad geographical distribution of such phage isolates that were previously recovered mostly from water sources in contrast to plant material (as in the case of phage AH05 and Eir4 within this study), which can definitely be proposed as different isolates belonging to the same novel phage species (Supplementary Figures 1, 3 and Table 3). Moreover, Roary pangenome analysis (Page et al., 2015), performed on the annotated genomes of the isolates seen in Table 3 under default settings (minimum percentage identity for BLASTp being 95%), has shown that the pangenome of representatives of this novel phage species encodes for 64 proteins with 44 of them being present in all nine of the isolates (including Eir4), while 54 of the products are present in two or more genomes. Speaking of Eir4 genome contents in this regard, only products of ORFs #1 (hypothetical protein), #45 (predicted tail fiber protein), #46 (hypothetical protein), and #50 (predicted o-spanin) from Eir4 were considered singletons under these rather conservative pangenome analysis settings. Of note, we did not manage to find any annotation for an o-spanin in the genomes of other phages presented in Table 3 (Supplementary Figure 3), which seems to be simply overlooked by the annotators of these phages, as it is an ORF enclosed within i-spanin-encoding ORF in the genome of Eir4 (as well as well-known phage T7), and predicting spanins usually requires particular attention from an annotator.

In order to investigate the efficiency of phages in planta, we took advantage of the well-studied interactions between Pst and Arabidopsis thaliana plants. Two-week-old A. thaliana Col-0 plants were flood-inoculated with the Pst or a mixture of Pst and phages to model the potential field application of phage preparations. In this setup, phages are premixed with bacteria immediately before the inoculation, so part of the phage is potentially being adsorbed on bacteria before contacting the plant. However, the relatively low MOI requires phage propagation in planta to successfully inactivate all bacteria. At 2 dpi, whole plant rosettes were harvested for the quantification of bacteria that propagated inside leaf tissues (Figure 4A). In samples inoculated with a mixture of Pst with phage Eir4, the load of Pst was significantly lower in comparison to plants treated with Pst only. The presence of phage Eisa9, however, did not influence Pst propagation in planta. Having no significant decrease in Pst growth in the presence of Eisa9 might be a result of a slower phage life cycle (burst occurring only after 120 min at MOI = 1, Figure 1B) or the contribution of newly selected resistant bacteria (as the frequency of resistance emergence appeared to be more than twofold higher for Eisa9). However, while Pst-inoculated plants exhibited growth reduction and leaf yellowing in comparison to mock-treated ones, both Pst + Eir4 and Pst + Eisa9 plants showed less yellowing and almost normal growth (Figure 4B). In the frame of potential field application, this suggests that both Eir4 and Eisa9 are good candidates for future field trials as they repress the development of Pst-caused disease symptoms, even though only Eir4 significantly reduced the bacterial propagation within the plant tissues. The mechanism of the Eisa9 effect on disease development, thus, remains to be investigated further.