Sixty-seven phages were isolated using standard enrichment techniques from both treated sewage effluent and river water between 2013 and 2015 using D. solani MK10 as the host organism. Transmission electron microscopy (TEM) showed two different morphological groups, a selection of which are shown in Figure 1 alongside the previously characterized phage LIMEstone1 (Adriaenssens et al., 2012b).

Of 24 phages imaged, all possessed an icosahedral head and a tail, placing them in the order Caudovirales. Three possessed short tails, characteristic of the family Podoviridae (such as ϕXF28 in the last panel of Figure 1) whilst the rest possessed longer contractile tails and belong to the family Myoviridae. The 21 Myoviridae members did not appear to possess the tail fibers characteristic of the family. Instead, short tail spikes were observed, (first three panels of Figure 1), and these are generally associated with the family Podoviridae.

Other Dickeya phages with similar morphology have been described and were shown to be efficient generalized transducing phages (Matilla et al., 2014). Due to the transduction capability of certain phages shown by Matilla et al., the 67 newly-isolated phages were also tested for ability to affect horizontal gene transfer. Of these, 51 (including the 21 phages with the non-classical morphology) proved capable of transducing chromosomal markers. Twelve of the isolates, all of which had the non-classical morphology, were also tested for generalized transduction, and proved capable of transferring plasmids between Dickeya species. The results of transduction of the plasmid pBR322 by three of these phages are shown in Figure 2.

Based on the bacterial strains tested, LIMEstone1 is capable of forming plaques on strains of D. solani only (Adriaenssens et al., 2012b). The phages isolated during this study were tested against a variety of Dickeya strains, listed in Table 1, to determine their host range. The majority of the phages presented here exhibited the same host range as LIMEstone1 and were only capable of forming plaques on D. solani strains but not isolated representatives of other Dickeya species used in this study. However, eight of the phage isolates had a wider host range extending to species such as Dickeya dieffenbachiae, Dickeya paradisiaca, and Dickeya zeae (Table 2).

Two key genes known to be conserved between these phages, those for DNA polymerase (DNAP) and tail spike protein 1 (TSP1), were sequenced for several of the newly-isolated phages. These nucleotide sequences were then compared to those of LIMEstone1 as shown in Table 3. All of the phages in Table 3 were able to form plaques on D. solani. The corresponding amino acid sequences were compared between these phages and phylogenetic trees were created as shown in Figure 3 (DNAP) and Figure 4 (TSP1). These show that, in agreement with Table 3, the DNAP genes of ϕXF4 and ϕXF11 grouped together away from the other phages with a branch length of 0.053. The other phages formed two clusters that differed in a single amino acid. The TSP1 genes formed two distinct clusters, with the genes from ϕXF16 and ϕJA1 forming their own cluster with a branch length of 0.092, whereas the other phages all had identical TSP1 genes. The final 20 amino acids were trimmed from the TSP1 genes as the sequencing data for some of the phages was insufficient for tree construction.

ϕXF4 and ϕJA15 were isolated over a year apart yet they shared the same host range and PCR amplification and preliminary sequence analysis showed 100% DNA identity in the TSP1 genes, although they differ in their DNAP genes. The full genomes of both phages were then sequenced. Both consist of circular double-stranded DNA of 151,519 and 153,757 bp, respectively. The genome of ϕXF4 has a G+C content of 49.4% and contained 185 predicted genes with lengths ranging between 116 and 4,838 nucleotides, as shown in Figure 5. ϕJA15 has a G+C content of 49.2% and contained 188 predicted genes of lengths between 122 and 4,838 nucleotides, as shown in Figure 6. Full annotation tables for the two genomes can be found in Tables S1, S2, respectively.

The genomes of the two new phages shared 97% DNA identity, with the main areas of difference being regions encoding endonucleases and hypothetical proteins. A comparison of both of these phages with the previously published phage LIMEstone1, showed over 95% DNA identity, with the major areas of difference consisting of genes thought to be involved in DNA replication (such as homing endonucleases and polymerases) as well as introns located throughout all three genomes. These areas of difference are highlighted in Figures 5, 6.

All bacterial strains used in this study are listed in Table 1. Dickeya species were routinely grown at 30°C in Luria broth (LB) or on LB agar plates (1.5%, wt/vol, agar). Phages were stored at 4°C in phage buffer (10 mM Tris-HCl, pH 7.4, 10 mM MgSO₄, 0.01%, wt/vol, gelatin) over a few drops of NaHCO₃⁻ saturated chloroform.

Treated sewage effluent was collected from a sewage treatment plant in Cambridge, United Kingdom (Matilla and Salmond, 2014). River water was collected from multiple sites along the River Cam. Samples were filter sterilized before 5 mL of the sample was added to 2x LB along with 500 μL of an overnight culture of D. solani MK10. This mixture was incubated overnight in a 250 mL flask at 30°C with shaking at 250 rpm. One milliliter of the enriched sample was mixed with 100 μL of chloroform (saturated with sodium hydrogen carbonate) and vortexed to lyse bacterial cells. The sample was centrifuged at 16,000 x g for 4 min and 10 μL of a serial dilution series of the supernatant was mixed with 200 μL of an overnight bacterial culture and 4 ml of LB top agar. This mixture was poured as an overlay on an LBA plate and incubated overnight at 30°C. Single phage plaques were picked with a sterile toothpick, placed into 100 μL phage buffer, and shaken with 40 μL of chloroform to kill any bacteria. The phages obtained were plaque purified three times. High-titre phage lysates were then obtained as described previously (Petty et al., 2006). Briefly, 10-fold serial dilutions of the phage were incubated overnight in an agar overlay as already described. Those plates exhibiting confluent lysis (seen as a mosaic-like effect in which the plaques are close to merging) were used for lysate preparation. The top agar was removed from the plate, vortexed with chloroform before sedimentation at 2,200 x g for 20 min at 4°C. The supernatant was removed and vortexed with a few drops of chloroform to produce the final lysate.

High-titre lysates for transmission electron microscopy were obtained as described previously (Petty et al., 2006) using 0.35% (w/v) LB agarose instead of 0.35% (w/v) LB agar overlays. Ten microliters of high-titer phage lysates were adsorbed onto 400-mesh copper grids with holey carbon support films (Agar Scientific, Stansted, United Kingdom) for 30 min. The copper grids were discharged in a Quorum/Emitech K100X system (Quorum, Ringmer, United Kingdom) prior to use. After 1 min, excess phage suspension was removed with filter paper and phage samples were negatively stained by placing the grids for 5 min in 10 μL of 2% phosphotungstic acid (PTA) neutralized with sodium hydroxide, or with 10 μL of 2% uranyl acetate for 2 min. The grids were then blotted on filter paper to remove the excess solution and allowed to air dry for 10 min. Phages were examined by transmission electron microscopy in the Multi-Imaging Centre (Department of Physiology, Development and Neuroscience, University of Cambridge) using an FEI Tecnai G2 transmission electron microscope (FEI, OR, USA). The accelerating voltage was 120.0 kV, and images were captured with anAMT XR60B digital camera running Deben software.

The host range of isolated phages was determined by plating out 10-fold serial dilutions of the phage lysates, onto agar overlays containing the six species of Dickeya listed in Table 1. To avoid potential confusion with “lysis from without,” only phages that produced lysis at low dilution and individual plaques were considered as being able to infect the host.

To test for transduction, phage lysates were generated on donor bacterial strains carrying the desired plasmid or chromosomal marker as already described. Transduction was performed by mixing phage lysate with an overnight culture of the recipient host to achieve a multiplicity of infection of 0.1, meaning that for each phage there were 10 bacterial cells. The mixture was left on the lab bench at room temperature for 20 min, followed by incubation on a rotary wheel at 30°C for 30 min. The infected culture was then centrifuged and the bacterial pellets washed with LB twice to eliminate any remaining non-adsorbed phage. The bacterial pellets were resuspended in 1 mL LB and 100 μL aliquots were spread onto LBA plates with drug selection for the chromosomal or plasmid marker. Appropriate standard controls, which were routinely negative, were used to score for any spontaneous resistance of the recipient strain. One hundred microliters of the phage lysate was also spread onto LBA plates to confirm lysate sterility.

Genomic DNA was extracted using Phase Lock Gel tubes (5 Prime, Hamburg, Germany) following manufacturer's instructions for isolation of Lambda DNA. DNA Polymerase (DNAP) and Tail Spike Protein 1 (TSP1) genes were amplified using Phusion polymerase (ThermoScientific, MA, USA) following standard protocols. TSP1 was amplified using the primers oJA1 and oJA2, and DNAP by the primers oJA14 and oJA15, listed in Table 1. PCR products were sequenced by GATC Biotech AG (Konstanz, Germany). Sequences were compared using NCBI Blast and the Artemis Comparison Tool (Carver et al., 2005). Phylogenetic trees were constructed using MEGA 7.0.26 (Kumar et al., 2016).

ϕXF4 was sequenced on the Illumina Bench Top MiSeq Sequencer (Illumina, CA, USA) at the DNA Sequencing Facility, Department of Biochemistry, University of Cambridge, UK. The resulting 138,803 reads were trimmed, quality assessed and assembled using Geneious 7.1.5 (Biomatters Ltd.), leading to higher than 100x coverage of the full genome. Gaps or single nucleotide polymorphisms were further filled or verified by Sanger sequencing to produce one contig. ϕJA15 was sequenced on the Illumina MiSeq Sequencer at MicrobesNG (Birmingham, UK). The 454,086 reads were trimmed using Trimmomatic (Bolger et al., 2014), assessed for quality using BWA-MEM (Li, 2013) and assembled using SPAdes 3.7.1 (Bankevich et al., 2012) with standard settings, leading to higher than 100x coverage of the full genome and producing one contig. Annotation of both genomes was performed using Prokka 1.11 (Seemann, 2014) using standard settings and LIMEstone1 (NC_019925.1) as a scaffold. Genome maps were generated using Circos (Krzywinski et al., 2009). Genomes were deposited in Genbank using Sequin (NCBI) and are available under accession numbers KY942057 (XF4) and KY942056 (JA15). Genomes were compared using NCBI Blast and the Artemis Comparison Tool (Carver et al., 2005).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.