Discovery and Characterization of a Novel Pseudomonas Phage of a New Genus

Phage discovery and cultivation

Freshwater was collected from Clear Creek, TX, USA, in a sterile 50 mL Falcon tube. This was centrifuged (4400 g, 20 min), and an aliquot was filtered (0.22 μm Durapore PVDF membrane, MilliporeSigma). About 0.5 mL of filtrate was mixed with 0.7% (w/v) top agar and 100 μL of overnight PAO1 culture and spread on an LB agar plate. This was incubated overnight, and plaques were inspected the next morning. The plaque of interest was isolated by stabbing with a pipette tip to remove a small plug of agar and mixed in phage buffer (10 mM Tris, pH 8, 100 mM NaCl, 1 mM MgSO₄). This was then serially streaked (three times) to ensure homogeneity. A plate lysate was prepared by mixing 30 μL of phage plaque isolate with 100 μL of PAO1 overnight in 3 mL of 0.7% (w/v) top agar and spread on LB agar plates. The next morning, cleared top agar was scraped and mixed with 3 mL of phage buffer, vortexed and then spun at 4400 g for 10 min. The supernatant was filtered (0.22 μm Durapore PVDF membrane, MilliporeSigma) and stored at 4°C.

Host range determination

We selected 25 random clinical isolates from the TAILΦR library. We deliberately selected isolates only from unique patient cases. These clinical isolates represented cases from across the United States, from different institutions and different sites of infection. We grew these isolates as lawns in 0.8% (w/v) top agar and spotted serial dilutions of ΦJB8. We incubated overnight and counted plaques in the morning. We performed this experiment in triplicate and averaged the plaque counts. We calculated the efficiency of plating (EOP) compared with PAO1 as (number of plaques on clinical isolate)/(number of plaques on PAO1).

Phage adsorption assay

We performed this assay as previously described in work by our laboratory. ³¹ Briefly, we used PAO1 cells (OD600 = 0.1, ≈3 × 10⁸ CFU/mL) and ΦJB8 (MOI = 0.1) and measured adsorption at 37°C. Quenching (dilution 1:100 into ice-cold LB) was performed at various time intervals, and fraction of free phage compared to input was determined after clearance of infected cells (centrifugation 10 mins, 4000 g). We included a condition with ΦJB8 but no cells to control for decay in warm LB media. We performed the experiment in biological triplicate. We analyzed the data with GraphPad Prism, fitted an exponential decay curve, calculated k_abs and report the mean and SEM.

Electron microscopy

Imaging was performed at Baylor College of Medicine’s Cryo-EM Advanced Technology Core Facility (Houston, TX, USA). Phages were deposited on grids (Quantifoil 2/2 200Cu +2 nm ThinC [Quantifoil Micro Tools GmbH, Jena, Germany]), negatively stained (2% w/v uranyl acetate) and imaged using a JEOL 1230 electron microscope (JEOL, Japan) at 80 kV with a 4k x 4k Gatan Ultrascan CCD Camera (Gatan, AMETEK). Images were taken at various magnifications and representatives were selected. Processing including contrast adjustment and scale bar addition was performed in NIH ImageJ.

Bioinformatic analysis of ΦJB8

Phage DNA was extracted using the Universal Pathogens Kit (Omega-Biotek) per the manufacturer’s instructions. ΦJB8 was sequenced using the Illumina platform by Novogene (Novogene, China). Briefly, the sequencing libraries were made using NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB, USA) per the manufacturer’s directions, and index codes were added. The PCR products were purified using the AMPure XP system. The size distribution was verified using the Agilent2100 Bioanalyzer. Final quantification was by qPCR. The final library was sequenced using an Illumina NovaSeq instrument (paired-end 150 bp reads). We filtered reads and discarded anything smaller than 50 base pairs using BBDuk (version 38.84), trimmed using a Q20 quality score, assembled using Geneious 2022.0.1 (native assembler with default settings for medium sensitivity with detection of structural variants of any size) and these assemblies were annotated using the RASTtk pipeline.^32,33 To generate a phylogeny based on nucleotide sequence, we used BLASTn (discontiguous BLAST, standard settings) and then visualized the phylogeny in NIH Genome workbench. Genetic comparisons between nearest neighbors were made using VIRIDIC. ³⁴ Proteomic comparisons were made using VIPTree. ³⁵

Isolation of lysogens

We followed the protocol of Altamirano and Barr. ³⁶ Briefly, we spotted 5 μL of ΦJB8 (3.6 × 10⁵ PFU/mL) on PAO1 top agar (0.8% w/v) and incubated overnight. We then picked presumably ΦJB8-resistant cells from the mesa and streaked them on an LB agar plate and incubated overnight. We picked 100 isolated colonies and propagated them on patch plates for 3 serial passages to remove any externally adherent phage particles. After 3 passages the colonies were picked by touching with a sterile pipette tip and very gently streaked on the surface of a top agar (0.8% w/v) coated plate that had been inoculated with 100 μL of PAO1 overnight growth, incubated overnight and examined for zones of clearance. ³⁶

Verification of lysogeny

We grew each potential lysogen overnight in LB and inoculated 2 μL of this overnight into 198 μL of LB in an untreated 96-well polystyrene plate (VWR, PA, USA). ³⁶ We grew this for 6 h at 37°C with shaking. We took the media from each well and clarified it by centrifugation (4000 g, 20 min) and then filtered the supernatant (0.22 μm Durapore PVDF membrane, MilliporeSigma). We titrated this supernatant using standard methods, pictures were taken and phage concentration (PFU/mL) was calculated. ³⁷ This experiment was performed in biological triplicate from separate colonies of each strain. To test resistance to ΦJB8, we picked colonies of each presumed lysogen and used them as the lawn in 0.8% (w/v) top agar and spotted dilutions of the original ΦJB8 phage, incubated overnight, imaged and counted plaques. We performed this experiment in biological triplicate from independent colonies of each lysogen. We report the mean and standard error.

Genetic analysis of potential lysogens

We selected 10 clones grown from a ΦJB8-generated mesa that did not exhibit zones of clearance upon streaking on PAO1 as ΦJB8-exposed controls. We also selected the 10 clones from a ΦJB8 mesa that did exhibit clearance (positives) as well as the PAO1 ancestor. These were grown overnight in LB, and then 1.5 mL of culture was harvested by centrifugation, and the pellet was rinsed thrice in PBS. We purified DNA from the pellets using a Power Soil Kit (Qiagen) following the manufacturer’s instructions. Illumina sequencing was done by Novogene. Reads were processed as described earlier. We used Bowtie2 to align reads from each pool (PAO1, controls and positives) to the ΦJB8 genome. ³⁸ We assembled contigs from each sequenced strain using SPAdes (v3.15.3, default settings). We then took each individual contig one by one and attempted to align the contig against the ΦJB8 genome and then the PAO1 host genome using BLASTn (discontiguous megablast, default algorithm parameters). We recorded whether the contig aligned to the phage, to the host or both.

Discovery and initial characterization

As freshwater sources are rich in Pseudomonads, we searched freshwater samples from Clear Creek, TX, USA (29.514347°N, 95.104101°W) for Pseudomonas phages. We plated filtered water samples in PAO1 top agar and identified multiple plaque morphologies. One distinctive plaque was ∼1–2 mm in diameter, hazy and formed a centralized mesa. We termed this phage ΦJB8 (Accession Number: PQ473534). Mesa formation was more obvious at higher titers (Fig. 1A). We applied standard methods to characterize this new phage determining the adsorption co-efficient as 7.25 × 10⁻⁹ mL/min using PAO1 host cells (Fig. 1D). Electron microscopy was challenging as preparations were prone to forming aggregates, but multiple attempts revealed consistent viral particles ∼84 nm (SD = 4 nm, n = 15 particles) in diameter with small tails in each preparation, fitting the classical description of podoviruses (Fig. 1B). Whole genome sequencing revealed a likely complete 47,273 bp genome that was likely circularly permuted. Phageterm could not identify a packaging mechanism. ³⁹ Analysis using BLASTn (discontiguous megablast, default settings) showed the nearest relatives were Pseudomonas phage Bjorn (Accession number: NC_042103.1) with tANI 21% (query coverage 29%, percent identity 72%) and Pseudomonas phage PSA37 with tANI 10% (Accession number: MZ089740, 15% query coverage, 67% percent identity). Given the lack of genetic similarity of the nearest neighbors, ΦJB8 likely represents a member of a new genus of bacteriophage. The host range was quite limited with productive infections seemingly established with EOP >0.5 in 5/25 clinical isolates selected randomly from the TAILΦR library. Some host strains displayed clearing at high concentrations, but no distinctive plaques at lower concentrations, consistent with “lysis from without” activity in 7/25 strains (Fig. 1C).

OPEN IN VIEWER

FIG. 1.

Basic appearance and characteristics of ΦJB8. (A) ΦJB8 has a hazy plaque morphology with formation of a mesa. (B) Negatively stained electron microscopy imaging of ΦJB8 at 40,000X. (C) Host range of ΦJB8 shown by spotting on indicated host strain and displayed as a heatmap showing the Efficiency of Plating (EOP) (D) Adsorption curves of ΦJB8 against PAO1 compared with an LB only control. We display mean and SEM of 3 independent biological replicates.

Genetic analysis

Given low tANI suggesting a novel genus, we pursued further analysis. We employed BLASTn and generated a phylogenetic tree of the nearest neighboring sequences (Fig. 2A) showing the phages ΦJB8, BroderSalsa (Accession number: OL412002) and Bjorn existing in three genetically distant groups that are quite distinct from other Pseudomonas phages. The nearest relative, Bjorn, was discovered in Denmark on a Pseudomonas spp. host in plant compost from Prunus avium.

OPEN IN VIEWER

FIG. 2.

ΦJB8 represents a genetically distinct new genus of bacteriophage. (A) Blastn was used to compare genomes to nearest neighbors and the phylogenetic tree was visualized using NCBI Genome Workbench. (B) VIRIDIC was used to visualize genomic distance between ΦJB8 and its nearest neighbors.

We employed VIRIDIC to visualize the distance between ΦJB8 and the 12 nearest neighbors (Fig. 2B). ³⁴ These data show that while all these neighboring phages possess similar length genomes, the aligned genome fraction (query coverage) and intergenomic similarity (analogous to tANI) between ΦJB8 and other phages is low, even for its nearest neighbor, Bjorn.

Proteomic alignment has been well used in phage biology. ⁴⁰ We used VIPTree to analyze ΦJB8 and its neighboring genomes. ³⁵ We compared ΦJB8 with 132 ICTV reference genomes (MSL40, March 2025) and noted significant divergence from the neighboring genomes. We aligned ΦJB8 to its 2 nearest proteomic neighbors (Bjorn and tf, Fig. 3A). The proteomic comparison was able to align a greater fraction than the nucleotide analysis. Homology was found in numerous proteins, including hypothetical gene products. For those with an annotated function, the closest relationship was found in DNA replication and packaging genes (terminase, DNA polymerase, endonuclease, and DNA primase/helicase) and structural genes (portal protein, virion structural proteins, tail fiber protein, major head protein). We found synteny between genes for which homology could be identified between phages ΦJB8 and Bjorn (Fig. 3B). This is an expected finding given the recombinant, mosaic nature of phage genomes. ⁴¹ Analysis attempts using VContact2 (v.0.9.19, default settings, Prokaryotic Viral RefSeq v.201) classified ΦJB8 as a singleton, concurring with PhaBox (v.2, default parameters), which identified it as a singleton at the genus level.^42,43

OPEN IN VIEWER

FIG. 3.

Bioinformatic comparison of proteomes of ΦJB8 and nearest neighbours. (A) Comparison of ΦJB8 and its nearest relatives Bjornvirus and tf, show typical modular synteny. (B) VIPTree-generated proteomic dendrogram showing proteomic distance of ΦJB8 from 132 nearest ICTV reference genomes (MSL40, March 2025).

Lifestyle of ΦJB8

As the hazy plaques seen when ΦJB8 was plated on PAO1 (Supplementary Fig. S1A) are hallmarks of lysogeny, we investigated whether there were any genetic markers of temperate lifestyle. We used Phage.ai and PhageLeads and neither software predicted lysogeny.^23,44 Given the lack of agreement between morphology and computational prediction, we picked bacteria from ΦJB8-generated mesas (Supplementary Fig. S1A) and streaked them on agar. Subsequently, we picked 100 individual colonies and propagated them on LB agar plates for three serial isolations (Supplementary Fig. S1B). We then streaked each purified isolate on PAO1 top agar lawns with PAO1 ancestor and LB only controls. Neither control generated clearance of the lawn that would suggest phage production. We observed that 10/100 purified colonies showed zones of clearance outside of the streak (Supplementary Fig. S1C). In other iterations of this experiment, we found persistence of ΦJB8 despite 4 passages of the host in 1/20 and 3/50 colonies. Notably, all colonies showing a zone of clearance suggestive of lysogeny on the first streak showed a similar zone on all subsequent streaks. This argues against incidental phage carryover as some colonies would be expected to lose the phage with each passage. These data argue that ΦJB8 persists despite attempts to remove phage by serial streaking, suggesting a lysogenic lifestyle. The variability in frequency of persistence seems most likely due to the underlying stochastic nature of lysogeny as we made every attempt to control for experimental variables (OD600, MOI, incubation times, host strain). We cannot exclude some hidden variable at work here, but the probabilistic nature of lysogeny is well established.^20,45 We followed the methods of Altamirano and Barr and tested the 10 isolates that seem to produce phage and 10 that did not for resistance to ΦJB8. ³⁶ We denote the bacterial strains derived from the ΦJB8 generated mesa JB8.Rx.

We found that all nonproductive strains (JB8.R1-10) were sensitive, with EOP ≈ 1, to ΦJB8 (data not shown) as was the PAO1 ancestor. All productive strains (JB8.R26-86) were resistant to ΦJB8 (Supplementary Fig. S2A). We further tested for production of ΦJB8 in solution. After growing the 10 nonproducers and 10 producers for 6 h in LB media, we tested the supernatants for phage. We found that 9/10 producers generated measurable phage concentrations, while the PAO1 only negative control and all 10 of the nonproducers did not (Supplementary Fig. S2B). All observed plaque morphologies were consistent with ΦJB8. Collectively, these data suggest that ΦJB8 persists within a fraction of ΦJB8-treated Pseudomonas and suggests lysogeny despite the lack of computational prediction.

We were curious as to the mechanism of integration into the host genome, as our analysis did not identify a recombinase or other markers. Thus, we sequenced the 10 nonproducers that had been exposed to ΦJB8 (JB8.R1-10), the 10 ΦJB8 producers (JB8.R26-86) and the PAO1 ancestor (SRA Bioproject: PRJNA1304558). In the PAO1 ancestor and the nonproducers, we did not detect any reads that mapped to ΦJB8. In the ΦJB8 producers, we mapped 14033 reads to the ΦJB8 genome with an average coverage of 43X (compared to 2929X of the host genome) (Fig. 4B). This verifies that ΦJB8 was being passed in those strains. It also suggests that the process of serial streaking was successful at removing any incidental contamination of ΦJB8 in those that were ΦJB8-exposed but not lysogenized (nonproducing strains).

OPEN IN VIEWER

FIG. 4.

WGS shows persistence of the ΦJB8 genome in JB8-resistant cells and a lack of chimeric contigs. (A) Diagram showing likelihood of chimeric contigs depends on episomal maintenance vs integration into the host genome. (B) Number of reads mapping to the ΦJB8 genome present in ΦJB8-exposed strains, ΦJB8-producing strains, and the ancestral PAO1 control. (C) Contigs assembled for each strain were mapped to the host and ΦJB8 genomes and presence of each genome counted.

We assembled contigs from each sequenced strain and in those contigs assembled from the ΦJB8 positive strains we identified 471 host contigs, 27 ΦJB8 contigs and most notably no contigs with contributions from both host and virus (Fig. 4C). This data were surprising as we had anticipated being able to identify prophage sequences that would be evidenced by contigs that are chimeras of both host and phage. Aligning reads from the ΦJB8 positive strains to a reference PAO1 genome (GenBank Accession number: NC_002516.2) showed we had achieved coverage of 2929X (±272) from end to end. This depth of coverage makes it quite likely that we would have been able to identify a prophage. In this case, our data suggest that ΦJB8 is maintained in the host, but does not integrate. This is consistent with an episomal mode of maintenance (Fig. 4A). The low ratio of ΦJB8 to PAO1 host coverage (1:68) most likely reflects the process relatively low carriage rate of the phage secondary to stochastic lysogeny, as mentioned earlier.