The bacterial strains used in this work are described in Supplementary Table S1. Isolation of the phage ϕ80-18 has been described earlier (Zhang and Skurnik, 1994). Both the phage ϕ80-18 and its host strain Y. enterocolitica serotype O:8 strain 8081-c have been deposited to Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH – Leibniz – Institut DSMZ under catalog numbers DSMZ 23253 and DSMZ 23249, respectively. Bacteria and bacteriophages were grown in lysogeny broth (LB, Bertani, 2004) at room temperature (22–25°C RT) unless otherwise indicated.

Yersinia enterocolitica strain 8081-c was grown in LB for 16 h, and 0.1 ml of the culture added to 5 ml of LB. The bacteria were grown aerated at 28°C to exponential phase (OD₆₀₀ = 0.3–0.5), 0.1 ml of a crude phage lysate (4.5 × 10⁷ PFU/ml) was added, and the culture was then incubated overnight at 28°C with shaking. The obtained phage lysate was filter-sterilized using a 0.22 μm Millipore membrane. In addition, a bacteriophage propagation experiment at 4, 28, and 37°C was performed, according to this scheme.

To evaluate the host range of the phage, the infectivity of the membrane-filtered phage lysate (10⁸ PFU/ml) was tested on the bacterial strains listed in Supplementary Table S1, using either the drop-test or plaque formation assay on soft-agar embedded bacteria. The formation of lysis zone or individual plaques was determined after 24 h of incubation. For the efficiency of plating (EOP), the PFU measurements were determined using the double-layer agar method. The EOP was calculated as the ratio between the PFU of the test strain to that of the original host strain Y. enterocolitica serotype O:8 strain 8081 (Stor ID 1258, Supplementary Table S1). The EOP assays were performed in triplicate.

Phage DNA was obtained from high-titer phage preparations as described earlier (Sambrook et al., 2001). Phage DNA was sequenced using the Illumina GAIIx (Genome Analyzer) technology at the FIMM Sequencing unit (Helsinki, Finland). The sequence assembly was done with the NextGene¹ and Staden software packages (Staden et al., 2003). The Artemis genome-browsing and annotation tool (Rutherford et al., 2000) was used for genome annotation. The physical ends of the phage genome and the terminal repeats (approx. 200 bp) of ϕ80-18 could not be identified from the de novo-assembled genomic sequence. To carry out this we used the approach described in details previously (Salem and Skurnik, 2018). Briefly, a 500 bp PCR-amplified fragment of the fliC gene of Y. enterocolitica O:3 was ligated with phosphorylated phage genomic DNA. The ligation mix was then used as a template for PCR using a primer pair of which one primer was fliC–specific and the other primer one of the phage-specific primers predicted to be close to the physical ends of the phage genomes as described (Salem and Skurnik, 2018). The resulting PCR products were purified and sequenced using a fliC –specific nested primed located ca. 200 bp upstream of the ligation junction. The PSI-BLAST (Altschul et al., 1997) and HHPred (Söding et al., 2005) programs were used to identify homologous proteins. Genome identity analysis between different viruses was carried out using StretcherN at EBI (Li et al., 2015). The PHIRE search tool was used to identify phage-encoded RNA polymerase promoters (Lavigne et al., 2004). The sigma-70 specific bacterial promoters and rho-independent terminators were searched using the search tools BPROM and FindTerm, respectively (Solovyev and Salamov, 2011). The annotated genome sequence of phage ϕ80-18 has been deposited into the nucleotide sequence databases under the accession numbers HE956710 and NC_019911.2.

Phage particle proteomes were analyzed by liquid chromatography coupled with mass spectrometry (LC-MS/MS) at the Proteomics Unit, Institute of Biotechnology, University of Helsinki. The phage with a titer >10¹⁰ pfu/mL was used for the analysis. Prior to digestion of proteins to peptides with trypsin, the proteins in the samples were reduced with tris (2-carboxyethyl) phosphine (TCEP) and alkylated with iodoacetamide. Tryptic peptide digests were purified by C18 reversed-phase chromatography columns (Varjosalo et al., 2013) and the mass spectrometry (MS) analysis was performed on an Orbitrap Elite Electron-Transfer Dissociation (ETD) mass spectrometer (Thermo Scientific, Waltham, MA, United States), using Xcalibur version 2.2, coupled to a Thermo Scientific nLC1000 nanoflow High Pressure Liquid Chromatography (HPLC) system. Peak extraction and subsequent protein identification were achieved using Proteome Discoverer 1.4 software (Thermo Scientific). Calibrated peak files were searched against all amino acid sequences of all six open reading frames of ϕ80-18 by a SEQUEST search engine. Error tolerances on the precursor and fragment ions were ±15 ppm and ±0.8 Da, respectively. Hits with at least two identified tryptic peptides were regarded as true hits.

The purified bacteriophage was applied to the surface of formvar carbon-coated copper grids and negatively stained with 2% uranyl acetate for 1 min. The excess of uranyl acetate was then removed from the grids using filter paper and the grids were allowed to air dry for 20 min (Ackermann, 2009). Preparations were visualized using a JEOL JEM-1200 EX 80 kV TEM. The dimensions of the bacteriophages were determined using RADIUS EM Imaging Software.

To determine the thermal stability of phage ϕ80-18, phage samples (4.5 × 10⁷ PFU/ml) were incubated at 4, 25, 40, 50, 60, and 80°C for 2 h. Phage survival was determined from samples collected after 20, 40, 60, 80, 100, and 120 min incubation using the double-layer agar method (Chen et al., 2016; Zhao et al., 2019).

To determine the pH stability of phage ϕ80-18, 200 μl samples of the phage (4.5 × 10⁷ PFU/ml) were incubated under various pH conditions (2, 3, 5, 6, 7, 8, 10, and 12) for 2 h at 28°C. Bacteriophage preparations were mixed with different pH solutions in the volume ratio 1:1. Phage titers in the tubes were determined using the double-layer agar plate method.

Yersinia enterocolitica 8081-c bacteria, grown in 5 ml of LB to an OD₆₀₀ of 0.5, were centrifuged at 12000 × g for 15 min at 4°C, and resuspended in 5 ml of fresh LB medium. The bacteria were then infected with phage ϕ80-18 at a MOI of 0.01, and the phages were allowed to absorb to the bacteria for 5 min at 28°C. To remove the unadsorbed phages the suspension was centrifuged at 14000 × g for 1 min, the bacterial pellet washed twice with fresh LB, and finally resuspended to 5 ml of LB, followed by incubation at 28°C. 100 μl from the sample were withdrawn from the tube every 10 min and the phage titers assayed using the double-layer agar method. The experiment was repeated three times (Zhao et al., 2019).

The phylogeny of phage ϕ80-18 was determined using both the whole genome nucleotide and the RNA polymerase (RNAP) amino acid sequences for the analysis. The genomic sequences of representative Autographivirinae (taxid:542835) phages most closely related to ϕ80-18 (NC_019911.2) were identified using the BLASTN search. The genome-based phylogenetic tree was constructed using the VICTOR web service (Meier-Kolthoff and Göker, 2017) based on the Genome-BLAST Distance Phylogeny (GBDP) method (Meier-Kolthoff et al., 2013) and FastME software. This included 100 pseudo-bootstrap replicates and SPR post-processing (Lefort et al., 2015). The amino acid sequences of the most closely related RNAP proteins were identified using the BLASTP search. The sequences were aligned with MAFFT v7.429 under the L-INS-i strategy (Katoh and Standley, 2013). The best-fit model for tree reconstruction (LG+F+I+G4, chosen according to BIC) was calculated with ModelFinder (Kalyaanamoorthy et al., 2017). The RNAP phylogenetic tree was inferred by maximum likelihood method with IQ-TREE v1.6.11, performing ultrafast bootstrap with 1000 replicates for calculating branch support (Hoang et al., 2018). The phylogenetic trees were visualized with FigTree (Rambaut, 2006) and tanglegram was constructed with Dendroscope (Huson et al., 2007).

Phage ϕ80-18 has a linear double-stranded DNA genome of 42,406 bp with the GC content of 47,64% that is close to that of Y. enterocolitica strain 8081 (47%) (Thomson et al., 2006). Altogether 57 genes were predicted from the sequence, all in the forward strand (Figure 1). No tRNA coding genes were found. The physical ends of the genome contain 325 bp direct repeats (Figure 1). While the function of altogether 29 predicted gene products showed no similarity to any known genes in the databases and remained therefore unassigned, similarity searches by BLASTP (Altschul et al., 1997) and HHPred (Söding et al., 2005) assigned a putative function to 17 gene products. The remaining 11 predicted gene products were identified as phage particle-associated proteins (PPAPs) in the phage particle proteome analysis (Figure 1 and Supplementary Table S2). Altogether 25 PPAPs were detected by LC-MS/MS analysis including those identified as structural proteins such as major capsid (Gp44), phage collar protein (Gp42), scaffolding protein (Gp43) and identified tail component proteins (Gp45, Gp46, and Gp50) as well as the DNA packaging proteins A and B (Gp52 and Gp53). Also the peptidoglycan penetrating lytic murein transglycosylase protein (Gp49) was identified as a PPAP. The catalytic domain of the 1259 residue Gp49 occupies 150 N-terminal residues, thus it is likely that the remaining protein functions as a tape measure protein to determine the length of the tail tube that is extended upon adsorption of the phage particle on host bacteria (Hu et al., 2013). In addition, also the DNA helicase (Gp20), DNA ligase (Gp25), DNA polymerase (Gp28), and 5′-exonuclease (Gp31) proteins were identified as PPAPs, suggesting that they might be injected together with the phage genome into the host cell to facilitate the take-over of the host metabolism. In contrast, the phage-encoded RNA-polymerase and DNA-primase were not PPAPs. For the 56 predicted genes, the initiation codon was ATG and only for the g37 gene encoding the DNA-directed RNA polymerase, it was GTG.

Using the PHIRE search tool, four 25 nt long phage promoters, designated P1 – P4 with a consensus sequence of -TGAT(T/a)(c/g)TCTACCCATATAG(c/t)AA(C/t)(A/t), typical for the Autographiviridae, were identified upstream the g01, g38, g44, and g50 genes (Figure 1 and Supplementary Table S3). These promoters likely regulate the expression of the phage genes during different phases of the infection cycle. In addition, using the BPROM search tool for bacterial sigma-70 type promoters we identified 11 bacterial promoter candidates, designated P5 – P15 (Figure 1 and Supplementary Table S3). While the functionality of these promoters awaits experimental evidence, the highest scores were predicted to P5 located leftmost in the genome and very likely the first one to start transcription upon the injection of the phage genome into the bacterial cell. We did not detect the phage encoded RNA polymerase in the phage particle so it has to be synthetized de novo before the phage promoters can be utilized, therefore, the presence of eleven sigma-70 type promoters scattered around the phage genome will allow transcription of the necessary phage genes, including g37 encoding the phage RNA polymerase. Only one rho-independent terminator was detected by the FindTerm program, located inside the g37 gene encoding the phage RNA polymerase.

While in general, the genomes of many podoviruses can be divided into three regions comprising early, middle and late genes for virus-host interactions, DNA metabolism and virion structure and assembly, respectively (Wang et al., 2016), this classification could not be directly applied to phage ϕ80-18 genome. While the phage RNAP promoters P2, P3, and P4 all seem to direct the transcription of the late genes, only phage RNAP promoter P1 remains for the first half of the genome. Therefore, the sigma-70 type promoters that are scattered around the genome might be involved in the transcription of the early and middle genes. To the latter ones based on functional predictions would belong the genes g18–g37 (Figure 1). Then the predicted early genes are g01–g17, among which are located also genes g09, g12, and g16, that encode PPAPs of unknown function.

The phage ϕ80-18 has been assigned to the Podoviridae family and the Autographivirinae subfamily like the model bacteriophage T7 or T3. BLASTN analysis revealed the highest sequence identity of 98 with 97% coverage (total identity of 93.2%, as determined by the EMBOSS stretcher alignment tool) to another Yersinia phage fHe-Yen3-01 that we recently isolated in Finland (Jun et al., 2018), followed by Pectobacterium phage MA13 (75,5%) and Cronobacter sakazakii phage vB_CskP_GAP227 (73,5%) (Table 1). Whole-genome phylogenetic tree (Figure 2) places f80-18 in well-defined clade, which was defined as Gap227virus (Kajsík et al., 2019) or broader as Ahp1-like subgroup (Wang et al., 2016). This significant phylogenetic association is supported by a tree inferred using single marker, RNA polymerase (RNAP) (Figure 3). Genome alignment of selected phages from this clade (Figure 4) revealed that most of the similarities (local sequence identity ≥ 60%) come from the predicted genes coding for DNA helicase (g20), DNA polymerase (g28), phosphoesterase (g34), RNA polymerase (g37), phage collar (g42), major capsid protein (g44), lytic glycosylase (g49) and DNA packaging protein (g53). The major genomic diversity regions are located to the early gene and the tail fiber protein encoding gene (g50) that score the lowest local identity results (<60%). Notably, the genome of the nearly identical phage fHe-Yen3-01 differs from f80-18 mainly by the absence of the g03 gene. On the other hand, the fHe-Yen3-01 possesses the gene g29 that is not related to any ϕ80-18 genes. The only other major difference between the two phages resides in the N-terminal parts of their respective tail fiber proteins that are only 56% identical, explaining the distinct differences in the host ranges between these phages (Jun et al., 2018).

To characterize the biological properties of ϕ80-18 its one-step growth curve was determined for the host strain Y. enterocolitica 8081-c (Figure 5). The phage seems to grow rather slowly in this host showing an apparent 50 min latent period, and low burst size of 8-10 PFU per infected bacterium. Comparing bacteriophage ϕ80-18 propagation at different temperatures similar effectivity was achieved at temperatures 4°C (4,0 × 10⁸ PFU/ml) and 28°C (8,7 × 10⁷ PFU/ml) and much lower efficiency at 37°C (7,3 × 10⁴ PFU/ml).

In the thermostability test the phage was stable for 2 h between +4 and 50°C, and was slowly inactivated at 60°C the titer dropping one log every 20 min, however, at 80°C it was completely inactivated already after 20 min incubation (Figure 6). The phage tolerated well pH values between 2 and 12 and had apparently optimal pH of 7-8 (Figure 7).

Genome sequence and phylogenetic analysis showed that ϕ80-18 belongs to Podoviridae family of bacteriophages. Transmission Electron Microscopy confirmed that this phage has icosahedral capsid and short non-contractile tail with tail fibers. The dimensions of the phage are 59.0 ± 2.28 nm (n = 14) for capsid vertex to vertex, 59.0 ± 2.9 nm (n = 14) for capsid face to face and 18.3 ± 1.44 nm long tail (Figure 8).

The host range of phage ϕ80-18 was tested using 115 Yersinia strains representing Y. aleksiciae, Y. bercovieri, Y. enterocolitica, Y. frederiksenii, Y. intermedia, Y. kristensenii, Y. mollaretii, Y. nurmii, Y. pekkanenii, Y. ruckeri, and Y. pseudotuberculosis (Supplementary Table S1). Bacteriophage ϕ80-18 was able to infect 16 strains. The host range analysis showed that ϕ80-18 can infect in addition to Y. enterocolitica serotype O:8 strains also strains of serotypes O:4, O:4,32, O:20 and O:21, the latter ones representing similar to serotype O:8 the American pathogenic Y. enterocolitica serotypes. In addition, also strains of the non-pathogenic serotype O:7,8 and two of the bioserotype 1A/O:5 strains were infected by ϕ80-18 (Supplementary Table S1). The LPS O-antigen composed of pentasaccharide repeat units was shown to function as a receptor for phage ϕ80-18 (Zhang et al., 1997), and a very similar structure is present in serotype O:7,8 O-antigen, however, the O-antigen repeat unit structure of O:4,32 shares only the reducing-end sugar, N-acetylgalactosamine, with the O:8 structure (Skurnik and Zhang, 1996). The structures of serotype O:20 and O:21 O-repeat units are not known, however, it is possible that they also contain an O-unit with a reducing-end N-acetylgalactosamine. If so, the phage receptor structure could be composed of the junction between the LPS core and the reducing-end N-acetylgalactosamine of the O-antigen.