Isolation and characterisation of novel phages infecting Lactobacillus plantarum and proposal of a new genus, “Silenusvirus”

Bacteria of Lactobacillus sp. are very useful to humans. However, the biology and genomic diversity of their (bacterio)phage enemies remains understudied. Knowledge on Lactobacillus phage diversity should broaden to develop efficient phage control strategies. To this end, organic waste samples were screened for phages against two wine-related Lactobacillus plantarum strains. Isolates were shotgun sequenced and compared against the phage database and each other by phylogenetics and comparative genomics. The new isolates had only three distant relatives from the database, but displayed a high overall degree of genomic similarity amongst them. The latter allowed for the use of one isolate as a representative to conduct transmission electron microscopy and structural protein sequencing, and to study phage adsorption and growth kinetics. The microscopy and proteomics tests confirmed the observed diversity of the new isolates and supported their classification to the family Siphoviridae and the proposal of the new phage genus “Silenusvirus”.

growth kinetics measurements provide a latent period between 19 min (Lactobacillus phage Y1) and 135 min (Lactobacillus phage P2) 12,15 and a burst size between 10.8 ± 0.4 (Lactobacillus phage FAGK2) and 214.5 ± 4 (Lactobacillus phage P2) 15,16 . At the same time, the current taxonomy of Lactobacillus phages has been limited to two families, Herelleviridae and Siphoviridae, although a proposal that extends it to the family Myoviridae has been recently published 13 . The Herelleviridae family hosts just two species members, Lactobacillus virus Lb338-1 and Lactobacillus virus LP65, while Siphoviridae includes the two genera of Lactobacillus phages to have been officially approved by the International Committee on Taxonomy of Viruses (ICTV) 17 . The first genus is called "Cequinquevirus" and contains the species Lactobacillus virus C5, Ld3, Ld17, Ld25A, LLKu and phiLdb 18,19 . The second genus is called "Coetzeevirus" and involves the species Lactobacillus virus phiJL-1, Pediococcus virus clP1 and Lactobacillus virus ATCC 8014-B1 19 .
Here we report the discovery of three new Lactobacillus phages that target industrially relevant strains of L. plantarum. We identify considerable genetic diversity between the newly isolated phages and the Lactobacillus phages of the database, which we ground on morphological, growth kinetics, as well as genomic data. Further, our results highlight the need to revisit and fully characterise prior isolates, to better appreciate the complete spectrum of diversity of Lactobacillus phages. Methods environmental samples, phage assays and bacterial strains. Lactobacillus phages were isolated from organic household waste samples. The samples were collected from two different organic waste treatment plants in Denmark (treatment plants A and B). Initially, the samples were split into two subsamples and processed as detailed in Kyrkou et al. 13 . The resulting filtrates were screened for phages using the double agar overlay method 20 and a top layer of 0.4% w/v agarose. Specifically, efficiency of plating assays were performed against two indicator strains that had earlier been acquired from private collections, L. plantarum L1 (wine fermentation isolate) and L. plantarum MW-1 (grape isolate). Single plaques were resuspended in 0.7 mL SM buffer and later filtered by 0.45-μm pore size PVDF spin filters (Ciro, Florida, USA). Each purified plaque underwent two further reisolationfiltration cycles to ensure pure stock cultures and was stored at 4 °C. For transmission electron microscopy (TEM) and protein sequencing, lysates of 10 10 plaque-forming units (PFUs)/mL) were further purified and concentrated using caesium chloride (CsCl) gradient ultracentrifugation, as described elsewhere 21 . All incubations of phage manipulations were done at 25 °C using De Man, Rogosa and Sharpe (MRS) broth and agar media supplemented with 10 mM CaCl 2 (MRS), whereas indicator strains were grown in MRS at 37 °C. teM analysis and structural protein sequencing. Phage morphology and structural proteins were determined using the CsCl-purified stocks. Micrographs of phage Silenus were generated as in other studies 13,22 .
The mean values and standard deviations of all Silenus virion dimensions were elucidated after inspection of 20-23 phage particles. Structural proteins were sequenced following published protocols 13,22 . Briefly, 100 μL of the CsCl-purified stocks were filtered through an Amicon Ultra filter unit (MWCO 30 kDa) and desalted four times. Phage particles (10 μL) were denaturised in 6 M urea, 5 mM dithiothreitol and 50 mM Tris-HCl (pH 8) and destabilised by freeze-thawing. Proteins were reduced (1 h incubation, 60 °C) and alkylated in 100 mM iodoacetamide and 50 mM ammonium bicarbonate, digested with 0.8 μg trypsin in 50 mM ammonia bicarbonate (40 μL) and diluted in 0.05% trifluoroacetic acid. The resulting peptides were analysed with an Ultimate 3,000 RSLCnano UHPLC system coupled with an analytical column (75 μm × 250 mm, 2 μm C18) and a Q Exactive HF mass spectrometer (ThermoFisher Scientific, Denmark). The twelve most intense ions were selected using MS Orbitrap scans and subsequently MS/MS-fragmented at a normalised collision energy (28) and a resolution of 60,000 (m/z 200). The output data were analysed in Proteome Discoverer 2.2 (ThermoFisher Scientific) and searched against predicted phage proteins by the Sequest HT algorithm.
phage DnA isolation, library construction and sequencing. Phage DNA was extracted from the filtered stock lysates according to a standard phenol/chloroform method 23 . Sequencing libraries were constructed with the Nextera ® XT DNA kit (Illumina Inc., San Diego, California, USA) according to the manufacturer's instructions for library preparations. The library normalisation, pooling and sequencing were done as indicated elsewhere 24 . All phage genomes were sequenced as a part of a flowcell on the Illumina MiSeq platform using the v2, 2 × 250 cycles chemistry.

Bioinformatics analyses.
De novo genome assembling was done with SPAdes (v. 3.5.0) 25 and the assemblies were cross-verified with Unicycler (v. 0.4.3) 26 and CLC Genomic Workbench (v. 9.5.3; CLC bio, Aarhus, Denmark) according to already published methods 13,27 . The annotation pipeline involved automatic protein annotations with RASTtk 28 and GeneMark 29 as a gene caller, followed by manual curations based on the predictions of BLASTp, HHpred 30 and sometimes PfamScan 31 . Lysin amino acid sequences were multiply aligned with Clustal Omega and viewed with MView (v. 1.63) using the default settings 32 . Transmembrane domains were identified with TMHMM 33 . The genome of phage Sabazios was scanned for −1 frameshift slippery sequences near the two lysin genes with FSFinder 34 and all genomes were scanned for tRNA genes with tRNAscan-SE (v. 2.0) 35 .
The analyses for the taxonomic classification of the new phage isolates followed the criteria and analyses scheme recommended by ICTV 36,37 . For each pair of compared query-subject phage genomes, Blastn query cover was multiplied by identity, according to the crude method for estimating nucleotide similarity of ICTV 36 . For more stringent comparisons on the nucleotide level the tool Gegenees 38 was used, after customising the fragment size/sliding step size (50/25) and threshold (0%). Gegenees divides a full genome (query) into fragments, then searches for BLAST "seeds" of each fragment against another genome (reference). The final phylogenomic distance of a query from a reference is the average value of all fragments' BLASTn scores expressed as a percentage of the score each fragment would yield towards itself (at 100% identity). Phylogenetic analyses were conducted for the phages Silenus, Sabazios, Bassarid, and compared to phages that returned a Gegenees score of at least 0.05. Two phylogenetic trees were built with the default pipeline of "One Click mode" (http://phylogeny.Lirmm.fr/) 39 . The first tree was based on the major capsid protein, the second on the large subunit terminase. All-against-all protein homology checks were performed between and within proteomes of the phages studied here and their closest relatives. The CMG-biotools package 40 , which implements the BLASTp algorithm, was chosen for this purpose. Pairs of proteins that aligned for at least 50% of the longest sequence and shared at least 50% of their nucleotides within the aligned region were considered as positive hits. Visualisation of genome alignments among the phages of this study and their two closest phage relatives were done with Easyfig 41 and the BLASTn algorithm. phage adsorption and growth kinetics. Phage adsorption and one-step growth experiments were done at a multiplicity of infection of 0.05. Strain MW-1 was grown to an OD 600 of 3.2, which corresponds to approximately 10 8 colony-forming units (CFUs)/mL for this strain. Immediately after, MW-1 cultures were infected with phage Silenus and incubated for 10 min at 37 °C. This time of infection was recorded as time point zero. The remaining steps of the adsorption assay and the burst size assay followed an already published protocol 22 under shaking conditions and at an incubation temperature of 37 °C. Phage growth kinetics were monitored on four-fold dilutions of the infected MW-1 cultures in triplicate assays. Samples were harvested from each triplicate assay approximately every 5-10 min, serially diluted and plated against a lawn of MW-1, incubated overnight and then examined for plaques. Supernatants of the infected MW-1 cultures just before the dilution step were plated, as well. The total count of PFUs (unadsorbed phages) from these supernatants designated how many plaques should be disregarded when counting infected centers.
Genomic data availability. Assembled and annotated genomes of phages Silenus, Bassarid and Sabazios were uploaded to GenBank under accession numbers MG765278, MG765275 and MH809528, respectively.

Results and Discussion
phage isolation and burst size. Phages Silenus, Sabazios and Bassarid were all isolated from the organic waste samples. Phages Silenus and Sabazios came from treatment plant B and phage Bassarid from treatment plant A. The phages formed plaques of approximately 1 mm following an incubation period of 24 h at 25 °C in MRS broth and agar media (see Supplementary Fig. S1). Lactobacillus phage Bassarid was isolated after infecting the lawn of L. plantarum L1, while the other two phages were active against L. plantarum MW-1. Adsorption and one-step growth curve tests were performed using Silenus as the representative of the three phages, due to the overall high genomic and morphological similarities among the three phages (details in "Phage morphology", "Support for the description of a new Lactobacillus phage genus" below). The adsorption rate for Silenus was 99.7%, the latent period was 45 ± 5 min and the average burst size was 4.86 progeny virions (for raw data and growth curve see Supplementary Table S1 and Supplementary Fig. S2). Although the burst size of Silenus was low, similar burst sizes have been reported for other Lactobacillus phages 16,42,43 . phage morphology. Most virion-associated genes among phages Silenus, Sabazios and Bassarid were highly conserved (please read "Support for the description of a new Lactobacillus phage genus" and genomic synteny comparison results there) and their virion morphologies were similar (see Supplementary Table S2 and Fig. S3). For these reasons, phage Silenus was selected as the representative of the three phages for the morphology-related analyses. Transmission electron micrographs displayed phage particles with an isometric head (diameter: 55.6 ± 3.2 nm), a neck passage with thin whiskers, a non-contractile flexible tail (length counting including the baseplate: 173.6 ± 5.6 nm; width: 12.0 ± 0.5 nm) and a characteristic double-disc baseplate (length: 12.2 ± 1.0 nm; width: 20.2 ± 1.4 nm) which culminates in short flexible appendages with tiny terminal globular structures. Similar flexible globular structures and capsid-tail dimensions have been observed for other Lactobacillus phages but their exact role remains unknown 3,12 . Interestingly, in some Leuconostoc mesenteroides phages similar baseplate appendages could also adsorb to other parts of the tail in an inconsistent manner 44 . These characteristics classify Silenus to the order Caudovirales and the family Siphoviridae (Fig. 1), which is the most widespread taxonomic classification among Lactobacillus phages 12 .
Basic genomic characteristics. Sequencing reads for each phage were assembled into single contigs of high coverage (667.7x-2,063x). The genome statistics of Silenus, Sabazios and Bassarid are presented in Table 1. Their genome sizes render them some of the smallest phages infecting L. plantarum together with phage phiJL-1 and phage ATCC 8014-B1 (accession numbers: NC_006936 and NC_019916), while their G/C content (42.5%) is close to that of their host (44.5%). In the three 37.9-38.8 kbp, double-stranded (ds) DNA phage genomes, all predicted open reading frames (ORFs) were located at the sense strand, which is in accordance with other phages of L. plantarum 45 and the closely-related Pediococcus damnosus 46 . The total number of predicted ORFs was 58-62 for the three phages. Out of this total, specific functions were assigned to 18 coding sequences for phages Bassarid and Silenus and to 19 for phage Sabazios. This translated to a mean of 30.7% of their coding sequences based on the in silico open reading frame predictions ( Table 1).
Description of phage modules. Three different genetic modules could be discerned, which enable the following: DNA packaging, morphogenesis and lysis (see Supplementary Figs. S4-S6). On the whole, such a pattern of genome organisation corresponds to most Lactobacillus phage isolates 3 . Most of the proteins that could be assigned functions to were identified in all three phages.
DNA packaging module proteins. A gene for the small subunit terminase and a gene for the large subunit terminase were recognised in all new phage genomes. Given that identifying these genes was possible, and that large subunit terminases are conserved among related phages 47 , the predicted large subunit proteins were further (2020) 10:8763 | https://doi.org/10.1038/s41598-020-65366-6 www.nature.com/scientificreports www.nature.com/scientificreports/ chosen to interrogate the phylogeny of the three phages. Downstream of the terminase unit, the sequencing coverage increased nearly twofold for all three phages. This may indicate a pac-type headful packaging strategy, where the variable cleaving sites for packaging termination lead to some regions being dual in some phage particles 48 . A pac-type headful packaging was experimentally determined for the distant relatives of the three new phages, Lactobacillus phage ATCC 8014-B1 and Lactobacillus phage phiJL-1 14,45 . For these reasons, the start of the new phages' genomes was arbitrarily set to the coding sequence of a hypothetical protein just before the small subunit terminase. The terminase unit of the new phages was located next to the morphogenesis module, as typically seen in other phages 49 . In all three genomes, the portal protein constituted the junction of the DNA packaging and the morphogenesis module. Indeed, the portal protein is a crucial component of the packaging motor, because it appears to pump the phage genome into the capsid with the aid of the large subunit terminase (ATPase activity) 50 . An HHpred search revealed that the extended packaging (i.e. including the portal protein) region of the three new phages was highly conserved (Fig. 5). Altogether, the interrelated Bacillus phages SF6 and SPP1 matched that region better than phages ATCC 8014-B1 and phiJL-1. This finding supports the pac-type nature of the new phages, since phages SF6 and SPP1 were shown to rely on pac packaging 51,52 .
Morphogenesis module proteins. Predicted proteins of known function within the morphogenesis module of the three phages included proteins of the capsid (a scaffolding protein and a major capsid protein), proteins of the connector (the portal protein and a putative head-to-tail joining protein) and proteins of the tail (a major tail protein, a putative tape measure protein and a tail fiber protein). LC-MS/MS analysis verified that the representative phage Silenus produces two of the aforementioned proteins, one protein of the capsid (major capsid) and one of the connector (putative head-to-tail joining protein), as well as a hypothetical protein (peg. 5; see Supplementary  Table S3). The latter probably belongs to the tail, since its gene is located right downstream of the putative tape measure gene. Two more proteins have been sequenced but whether these constitute virion-associated proteins or the result of contamination is unclear (peg. 27 and peg. 48; Supplementary Table S3). As expected, the scaffolding protein was not traced by the protein sequencing analysis, because it comprises the core of the pre-assembled prohead, which is surrounded by the major capsid protein 53 . The proteins of the capsid and the putative head-to-tail joining protein were quite conserved among the new phages and largely aligned with their orthologues in Lactobacillus phages ATCC 8014-B1 and phiJL-1, and Pediococcus phage clP1 (BLASTp results). Similarly, the major tail and the putative tape measure proteins of Silenus, Sabazios and Bassarid contained domains with homology to the aforementioned phages and primarily to Lactobacillus phage phiJL-1. In the genomes of the new phages, the tape measure protein, whose length defines the length of the tail, had the longest sequence (1,066-1,078 aa) and was located downstream of the major tail protein 54,55 . The predicted tail fiber protein of the  www.nature.com/scientificreports www.nature.com/scientificreports/ new phages is likely to initiate the infection process through the identification and binding to host receptors on the surface of sensitive cells, as described for other phages 56,57 . An interesting feature of the new phages' genomes was that this tail fiber protein and all hypothetical proteins found in the region between the tape measure protein and the lysis module aligned poorly with existing phage records. In average, this region exhibited low nucleotide similarity between phage Bassarid and phages Silenus, Sabazios, as well (BLASTp results; see also genomic synteny results). Considering that tail fibers shape phage host range, it is expected that the three phages' host range differs considerably from that of database phages. Moreover, tail fibers are subject to horizontal gene transfer between phages in a constant manner 58,59 , which could explain why phage Bassarid's host range varies from that of the other two closely-related phages.

Open Reading Frames with Assigned Function/All ORFs Genome Size (bp) G/C Content (%)
Lysis module proteins. A classic phage lysis cassette comprises two types of proteins, a holin and a lysin. Holins permeabilise the cytoplasmic membrane thereby granting access of the cell wall to lysins, which then degrade the cell wall 60 . Usually, holins have two or three transmembrane domains, but the holins of the phages described in this study belonged to the rare one-transmembrane domain group 61 (TMHMM search). Additionally, their best match was ORF147 from phage phiJL-1, that is likely a holin with one transmembrane domain 45 . Studies indicate that Lactobacillus phages only employ two out of the five classes of lysins, muramidases and amidases 62 . All lysins of the phages studied here strongly matched muramidase entries (HHpred search). In this study, noteworthy was the prediction of two slightly overlapping ORFs for lysins within the module of phage Sabazios. According to Clustal Omega alignments, the first amino acid sequence of Sabazios' lysins aligned with high homology to the first 64.8% of Bassarid's and Silenus' lysin sequence. The second amino acid sequence of Sabazios' lysins aligned with a small overlap to the first amono acid sequence of Sabazios' lysins and with high homology to the last 46.5% of Bassarid's and Silenus' lysin sequence (see Supplementary Fig. S7). It is possible that these two lysins are products of a nonsense or frameshift mutation, which would create a pair of non-functional pieces 63 . Alternatively, the two overlapping CDS may produce the subunits of a dimeric lysin, as described for phages CD27L and CTP1L of Clostridium difficile 64 . A third scenario is that these two lysins are products of a programmed ribosomal slippage, which may either adjust the degree of lysin production as a response to conditions in the cell or produce a specific ratio of two different lysin proteins 65 . So far, ribosomal slippage has principally been reported for structural genes of the tail 66 . In the only studied case of slippage for Lactobacillus phages, both the major capsid and the major tail protein of phage A2 are affected by ribosomal slippage 67,68 . However, FSFinder traced no slippery sequence in the overlap region of the two lysin ORFs of phage Sabazios. The fact that all predicted ORFs were found at the same strand implies the absence of genetic switches and thus, given that no other lysogeny-related genes were detected, we presume that the three new phages are most probably virulent.
Other predicted proteins. The existence of other modules could not be confirmed, but some additional proteins in the genomes of Sabazios, Silenus and Bassarid were assigned to a function. A superfamily II, ATP-dependent helicase, predicted in all three phages, may belong to the family of DEAD/DEAH-box containing helicases (BLASTp search result). DEAD/DEAH-box containing helicases participate in RNA metabolism in many, essential ways, such as by regulating gene expression and signalling 69 . These functions could explain why it is likely for DEAD/DEAH-box containing helicase genes to be found within a phage genome. In all cases, the vast majority of proteins in close proximity to the new phages' helicase could not be annotated. Along with the helicase, a gene of the phages Sabazios and Silenus encoded an adenine-specific methyltransferase, which is a DNA modification enzyme. A BLASTp search revealed orthologues of Sabazios' and Silenus' methyltransferases in the genomes of L. plantarum and other Lactobacillus sp. This finding corroborates that these two phages mimic the host genome's methylation pattern as an active strategy to evade restriction by host-driven endonucleases 70 . In phages, methyltransferases are often transferred through horizontal gene transfer. Genes coding for methyltransferases www.nature.com/scientificreports www.nature.com/scientificreports/ are common in phage genomes and have already been reported in many dairy Streptococcus and Lactococcus phages 71,72 . Methyltransferase-coding genes have also been noted in the genomes of some L. plantarum strains (REBASE search; http://rebase.neb.com/rebase/rebase.html) and in the genomes of Lactobacillus phages PL1 PL-1, J-1, P1174 (NCBI Protein database search).
Selfish genetic elements in the genomes of Silenus, Bassarid and Sabazios were represented by one HNH homing endonuclease. One or more HNH endonuclease genes have been found in the genomes of Lactobacillus phages before 13 . Essentially, HNH endonucleases are highly specialised selfish genetic elements that facilitate the mobility of themselves and of those genes to which they pertain, from genome to genome 73 . In some cases, phages that produce HNH endonucleases can even exclude other competing phage species by cleaving their DNA 74 . Due to the position of the new phages' HNH endonuclease genes next to hypothetical genes, the role of these enzymes could not be deduced. We joined the two ORFs framing each HNH endonuclease of the studied phages and performed a BLASTn analyses. None of the resulting BLASTn hits spanned along the joint region suggesting that no gene was spliced by these HNH endonucleases.  www.nature.com/scientificreports www.nature.com/scientificreports/ Support for the description of a new Lactobacillus phage genus. Multiplying BLASTn query cover by identity yielded a limited (<50%, ICTV criterion) overall nucleotide similarity of phages Silenus, Bassarid and Sabazios to other phage records. The highest BLASTn (>50%) similarity scores were obtained from Lactobacillus phage phiJL-1, Lactobacillus phage ATCC 8014-B1 and Pediococcus phage cIP1. Pediococcus phage clP1 has already been reported similar to Lactobacillus phage ATCC 8014-B1 with a high similarity score of 89% 14 . Such a high similarity between phages of Pediococcus and Lactobacillus is now not surprising. More and more evidence of the close phylogenetic relatedness of these bacterial genera is being gathered. Consequently, some species reclassifications between the two genera have been proposed 75,76 . Along with these, a recent phylogenomic study examining the core-and pan-genomes of 174 type strains of Lactobacillus and Pediococcus, placed Pediococcus as an integral part of Lactobacillus 77 . Phages infecting species from the one genus are thus expected to have a host range that expands to the other genus. Because Pediococcus and L. plantarum strains often coexist, the capacity of L. plantarum phages to infect Pediococcus sp. strains is indeed quite likely, as discussed for Carnobacterium sp. strains and other Lactobacillus sp. strains 14 . The three new phages and eight BLASTn genome records that had some level of nucleotide sequence homology to the new phages were afterwards submitted to Gegenees. All-against-all BLASTn comparisons, performed according to the Gegenees software, resulted in the heatplot of Fig. 2. The scored phylogenomic distances supported a separate grouping of the phages of this study from the other eight phages. Nonetheless, diversity within the group of the new phages was also noted, since phage Bassarid was found to be quite distinct from phages Silenus and Sabazios at the nucleotide level (average normalised similarities of 29.9% and 24.7%).
The phylogenetic trees of the major capsid protein and the large subunit terminase protein corroborated the sorting of the new phages into one individual group (Fig. 3a,b, respectively). On the other hand, diversity within the group was reiterated with phage Bassarid showing occasional, yet low variation (Fig. 3b). Consistent with earlier observations was the clustering of the three phages and phages phiJL-1, ATCC 8014-B1 and clP1 into sister groups. Proteome homology tests were done for the new phages and those phages that appeared to be their distant relatives according to the aforementioned tests (i.e. phiJL-1, ATCC 8014-B1 and clP1). Homology between and within the six phages further clarified how diverse these phages are from existing phage records. Specifically, the three phages share between 62.5-92.1% of their proteome (Fig. 4), while the homology to the proteomes of phiJL-1, ATCC 8014-B1 and clP1 is low (<28.4%). Regarding homology within each phage proteome, no paralogous proteins were found for any of the six examined phages (last row of Fig. 4).
Finally, genomic synteny tests with Easyfig visualised the previously manifested homology among the new phages and provided evidence for their conserved genome architecture (Fig. 5). The major differences between phage Bassarid and the other two new phages were localised in the region of the tail-related structural genes, and particularly at those hypothetical ORFs that neighbour the tail fiber gene (see Supplementary Fig. S6). Nonetheless, no important structural dissimilarities between the tails of Silenus, Sabazios and the tail of Bassarid were further highlighted by TEM (see Supplementary Table S2 and Fig. S3). At the same time, it is seen from Fig. 5 that even if phages phiJL-1 and ATCC 8014-B1 showed some conserved gene order against the new phages www.nature.com/scientificreports www.nature.com/scientificreports/ they did score low in nucleotide homology. A comparison map against phage clP1 is available as Supplementary  Fig. S8.
Taken together, all these results suggest that Lactobacillus phages Silenus, Bassarid and Sabazios form a coherent group and are considerably distinct from all other fully-sequenced phages. Therefore, and in agreement with ICTV criteria 36 , we propose the new Lactobacillus phage genus "Silenusvirus". At present, the new genus should consist exclusively of phages Silenus (founder), Bassarid and Sabazios.

conclusions
A few phages of L. plantarum have been isolated over the years 12 , but this study reports the first of these which can infect wine-related strains of L. plantarum. We characterised and propose three new phage species, Lactobacillus virus Sabazios, Lactobacillus virus Bassarid and Lactobacillus virus Silenus, by means of whole-genome sequencing and in silico protein prediction. Furthermore, we assessed the growth parameters of a representative phage and investigated its morphology with TEM. By comparing the phages Sabazios, Bassarid and Silenus to existing phage records, we demonstrated substantial genomic heterogeneity. This heterogeneity was further evident by the fact that we could assign functions to less than one third of the predicted ORFs. Our findings support the creation of the novel Lactobacillus phage genus "Silenusvirus", with phages Silenus, Sabazios and Bassarid as the new members of this proposed genus. The results of this study shed more light on the diversity of L. plantarum phages and their hosts, and could aid towards the development of efficient phage control interventions in the future.

Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files).