Four Novel Caudoviricetes Bacteriophages Isolated from Baltic Sea Water Infect Colonizers of Aurelia aurita

The moon jellyfish Aurelia aurita is associated with a highly diverse microbiota changing with provenance, tissue, and life stage. While the crucial relevance of bacteria to host fitness is well known, bacteriophages have often been neglected. Here, we aimed to isolate virulent phages targeting bacteria that are part of the A. aurita-associated microbiota. Four phages (Pseudomonas phage BSwM KMM1, Citrobacter phages BSwM KMM2–BSwM KMM4) were isolated from the Baltic Sea water column and characterized. Phages KMM2/3/4 infected representatives of Citrobacter, Shigella, and Escherichia (Enterobacteriaceae), whereas KMM1 showed a remarkably broad host range, infecting Gram-negative Pseudomonas as well as Gram-positive Staphylococcus. All phages showed an up to 99% adsorption to host cells within 5 min, short latent periods (around 30 min), large burst sizes (mean of 128 pfu/cell), and high efficiency of plating (EOP > 0.5), demonstrating decent virulence, efficiency, and infectivity. Transmission electron microscopy and viral genome analysis revealed that all phages are novel species and belong to the class of Caudoviricetes harboring a tail and linear double-stranded DNA (formerly known as Siphovirus-like (KMM3) and Myovirus-like (KMM1/2/4) bacteriophages) with genome sizes between 50 and 138 kbp. In the future, these isolates will allow manipulation of the A. aurita-associated microbiota and provide new insights into phage impact on the multicellular host.


Introduction
Bacteriophages, or phages for short, are viruses that infect bacteria. They are widespread in nature and are considered the Earth's most abundant biological entities [1,2]. Phages can be classified based on their replication cycle, morphology, nucleic acid type, and host range. Particularly, their replication mode, such as virulent or temperate, is an essential characteristic [3]. Virulent phages cause lysis of the host bacterium [4], whereas temperate phages integrate their genetic material into the host bacterium's genome and can remain dormant until conditions favor their activation into the lytic cycle [3,5,6]. In the earlier days of phage research, phages were primarily classified based on morphological characteristics. The morphological classification scheme was initially proposed by the International Committee on Taxonomy of Viruses (ICTV) which provided a framework for organizing phages into different families, genera, and species. However, with the advent of molecular biology techniques and the availability of whole-genome sequencing, the focus of phage taxonomy has shifted towards genetic and genomic information [7]. Genomic data, sequence similarities, and phylogenetic analyses assign phages to different taxonomic ranks, including families, subfamilies, genera, and species [8]. Efforts are constantly underway to improve phage tax-

Taxonomic Classification of Bacterial Isolates
Bacteria were enriched and isolated from A. aurita medusae collected in the Baltic Sea as described in the previous study by Weiland-Bräuer et al., 2020 [22]. Additional isolates previously not published were taxonomically classified in the present study. The bacterial isolates were grown, and genomic DNA was isolated from overnight cultures (5 mL) using the Wizard Genomic DNA Purification Kit (Promega GmbH, Walldorf, Germany) according to the manufacturer's instructions. Overall, 16S rRNA genes were PCR-amplified from 50 ng of isolated genomic DNA using the bacterium-specific 16S rRNA gene primer 27F (5 -AGAGTTTGATCCTGGCTCAG-3 ) and the universal primer 1492R (5 -GGTTACCTTGTTACGACTT-3 ) [48] resulting in a 1.5 kb PCR fragment. The fragments were Sanger sequenced at the sequencing facility at the Institute of Clinical Molecular Biology, University of Kiel (IKMB). Sequence analysis was conducted using CodonCode Aligner v. 9.0. Sequence data of full-length 16S rRNA genes were deposited under GenBank accession numbers OQ397638-OQ397653 and OQ398153-OQ398172.

Phage Enrichment, Isolation, and Purification
Water columns were sampled in the Baltic Sea at the Kiel fjord (54.329649, 10.149129) in March 2020 and June 2021. Samples were taken from the surface (<50 cm depth) using a sterile 20 L canister. A total of 50 mL of the seawater samples and 50 mL of MB medium were mixed with 1 mL of an overnight culture of a mixture of potential host bacterial strains (Table 1, column "Use in the study", category "Enrichment/first screening", a total of 55 strains) and placed on a shaking incubator (120 rpm) at 30 • C for 24 h. The mixture was centrifuged at 4000× g for 30 min, and the supernatant passed through a 0.22 µm pore-size polycarbonate syringe filter (Sartorius, Goettingen, Germany) to remove the residual bacterial cells.
The spot test assay, a procedure based on the double-layer plaque technique [49], was used as an initial test to detect virulent phages by plaques on all 55 bacterial strains, separately grown in top-agar on agar plates. Briefly, 10 µL of each filtered mixture was spotted on an MB 1.5% agar plate containing a second solidified layer of 3 mL 0.6% MB top agar mixed with 100 µL of a single bacterial host strain. The plates were incubated overnight at 30 • C. Plaques generated by bacteriophage-induced bacterial lysis were detected the following day. Plaques were exclusively detected for bacterial isolates No. 8 (Pseudomonas sp. MK967010.1), No. 7 (Citrobacter sp., OQ398154), and No. 6 (C. freundii, OQ398153). Those three bacterial isolates were used as bacterial host strains for the following assays to study phage characteristics. Table 1. Bacterial strains used in this study. Bacterial strains were isolated in the study [22]. Isolates are sorted by the last column and phylum level. The column "Use in This study" refers to the use of the strains in the study. If not stated differently, the listed numbers in column "Reference" reflect NCBI Accession Numbers.

Phage Purification, Titration, and Propagation
Original phage plaques were used for further purification of phages. Morphologically distinct plaques were picked from the agar plate using a sterile toothpick and streaked on a freshly prepared double-agar plate with the respective host strain in the top agar. The procedure was repeated three times to ensure pure and single phages. Phage lysate preparations were conducted from top agar plates with approx. 10 5 plaque-forming units per mL (pfu/mL). Phage-containing top agar was collected with a sterile loop and transferred into a 15 mL Falcon tube (Sarstedt, Nümbrecht, Germany). A total of 3 mL of liquid MB was added. The mixture was vortexed and centrifuged at 4000× g for 10 min. The supernatant was filtered through a 0.22 µm polycarbonate syringe filter (Sartorius, Goettingen, Germany) to remove bacterial cells and agar debris. The respective pfu/mL of the resulting phage lysate was determined, and the lysate was stored at 4 • C. Phage stability at 4 • C was analyzed every 2 days for 2 weeks using the double-layer agar method with no significant variance in pfu/mL.
Phage propagation was performed in liquid culture. A total of 1 mL of the respective bacterial host (overnight culture) was inoculated into 48 mL of MB in a 100 mL Erlenmeyer flask and incubated at 30 • C and 120 rpm until OD 600nm = 0.2-0.3 was reached. Then, 1 mL of the phage lysate (10 6 pfu/mL) was added to the cultures, which were further incubated for 3 h at 30 • C with shaking. The culture was transferred to a 50 mL Falcon tube (Sarstedt, Germany) and centrifuged at 4000× g for 10 min. The supernatant was sterile-filtered using a 0.22 µm syringe filter (Sartorius, Goettingen). Phage lysates were stored at 4 • C.

Transmission Electron Microscopy
In total, 50 mL of freshly prepared phage lysate (>10 8 pfu/mL) was ultracentrifuged (Optima XE-100 ultracentrifuge, Beckman Coulter, Brea, CA, USA) at 109,800× g for 30 min. Phage pellets were resuspended in 1 mL of Ultra-pure water (Carl Roth, Karlsruhe, Germany) overnight at 4 • C on a 3D shaker. Subsequently, TEM grids (copper, 400 square mesh, formvar-coated) were glow-discharged for 60 s at a 0.6 mbar air pressure and a 10 mA glow current using a Safematic CCU-010 unit and then incubated with 8 µL of the phage lysate (>10 8 pfu/mL) for 5 min. Grids were washed briefly on six drops of water, stained with 1% uranyl acetate for 10 s, blotted to remove the excess stain, and air dried. Samples were imaged with a Tecnai G2 Spirit BioTwin transmission electron microscope operated with a LaB6 filament at 80 kV, and equipped with an Eagle 4k HS CCD camera, TEM User interface (v. 4.2), and TIA software (v. 2.5) (all FEI/Thermo Fisher Scientific, Hessen, Germany). Open-source Fiji software was used to measure the head width (perpendicular to the vertical axis) and the tail length of phages.

One-Step Growth Curve Analysis
Bacterial host strains were grown overnight in 5 mL MB. A total of 500 µL of the overnight culture was incubated in 50 mL MB at 30 • C until turbidity at 600 nm of T 600 = 0.1-0.2 (10 7 cells/mL) was reached. A total of 10 mL of the bacterial culture was centrifuged at 4000× g for 10 min at 4 • C. The cell pellets were resuspended in 5 mL of an MB medium, and 5 mL of phage lysate (10 7 pfu/mL) was added with a multiplicity of infection (MOI) of 1.0, expressing that one phage particle is exposed to one bacterial host cell [50][51][52]. Initial tests using MOIs of 0.01 and 0.1 did not result in a sufficient latent period and burst size detection. The phage titer was immediately determined by double-agar layer plaque assay (t 0 ) after centrifuging at 4000× g for 10 min to remove the free phage particles before resuspending the samples in 10 mL MB. Phages were allowed to adsorb to the bacterial host cells within 5 min of incubation at 30 • C. Afterwards, the phage titer was again determined by double-agar layer plaque assay (t 5min ) for calculating the adsorption rate and constant k with the following formula [53]: B, initial concentration of bacteria (cells/mL); t, time (min); P, concentration of free phage per ml; P 0 , initial concentration of phage per mL.
Further aliquots were collected in 10 min intervals over a 120 min period, and phage titers were determined. Three independent experiments were performed for each phage.

Efficiency of Plating and Host Range
The host range of isolated bacteriophages was initially determined by the spot assay and verified by the double-layer agar method. A selection of 43 strains belonging to different genera (Table 1, column "Use in the study", category "Host range") was tested. The bacterial strains were individually grown overnight in 5 mL cultures. An aliquot of 100 µL of each culture and 3 mL of the respective culture medium containing 0.6% agar was mixed and poured onto an agar plate. After 15 min at room temperature, to allow the top agar to solidify, 10 µL of the 10-fold serially diluted phage lysate (original 10 9 pfu/mL, diluted in MB) were spotted onto the soft agar. The plates were then incubated at the respective incubation temperature of the host strain (Table 1). Plaques were examined after 16 h of incubation. The Efficiency of Plating (EOP) was calculated by the ratio of the average pfu on a tested host to the average pfu on a corresponding reference (original) host. The variation of EOP values is represented as a heat map using Excel.

Viral DNA Isolation
Genomic DNA was isolated from the phage lysates using a modified phenol-chloroformisoamyl alcohol method [54]. Briefly, 500 µL of phage lysates (10 12 pfu/mL) were treated with 1 U/mL of DNase I and 1 U/mL of RNase A (Thermo Fisher Scientific, Hessen, Germany) in a Reaction Buffer (100 mM Tris-HCl, 25 mM MgCl 2 , 1 mM CaCl 2 ) (Thermo Fisher Scientific, Hessen, Germany) and incubated for1 h at 37 • C to remove external nucleic acids. Afterwards, 0.5 M of EDTA, 40 µL of Proteinase K (20 mg/mL, Thermo Fisher Scientific, Hessen, Germany), 1 M of CaCl 2, and a 20% sodium dodecyl sulfate (Roth, Karlsruhe, Germany) were added before incubating at 37 • C for 2 h. Following this, samples were incubated for a further 2 h at 65 • C. Viral DNA was extracted with an equal volume (vol) of phenol-chloroform-isoamyl alcohol (25:24:1, Roth, Karlsruhe, Germany) and centrifuged at 3000× g for 15 min. The step was repeated, and the supernatant was transferred to phase lock tubes (Quantabio, Hilden, Germany). The aqueous phase was mixed with an equal volume of chloroform and centrifuged at 1600× g for 15 min. The aqueous layer was mixed with 3 M of sodium acetate and 1 vol of isopropanol to precipitate the DNA. After incubation overnight at −20 • C, the DNA was pelleted by centrifugation at 12,600× g for 30 min at 4 • C. The pellet was washed with a 70% ethanol, air dried, and dissolved in Ultra-pure water (Roth, Karlsruhe, Germany). DNA was stored at −20 • C before sequencing. DNA quality and quantity were analyzed using a NanoDrop1000 spectrophotometer and a Qubit double-stranded BR assay kit on a Qubit fluorometer (Thermo Fisher Scientific, Hessen, Germany).

Sequencing, Bioinformatic Analysis, and Annotation of Viral Genomes
Long sequencing reads were obtained using the Oxford Nanopore Technologies Min-ION platform (R9.4.1 flow cell). The MinION sequencing library was prepared according to the manufacturer's guidelines using the SQK-RBK004 Rapid Barcoding Kit. MinION sequencing was performed with MinKNOW v. 22.08.4. The raw sequencing data (fast5 format) were base-called using Guppy v. 6.2.7, and finally, demultiplexing was performed using qcat v.1.1.0. Quality assessment and adapter trimming of the MinION long-read sequences of the four viral genomes was performed using LongQC v1.2.0 [55] and Filtlong v.0.2.1 (https://github.com/rrwick/Filtlong, accessed on 25 October 2022). Filtered sequence reads with an average length > 1000 bps were selected, omitting the worst 5% of reads. Reads per genome were assembled using the assembler Flye v2.9 [56], resulting in complete, single-contig genomes. The number of sequenced reads before and after filtration, the GC content, reads coverage, and N50 values are provided in Table 2. The completeness and contamination of the assembled viral genomes were assessed using CheckV [57]. Viral genomes were annotated using Prokka v.1.14.6 [58] and the "--kindgom Viruses" option. Functional annotation of the four isolated phages was performed using EggNOG mapper v2.1.9 [59] and the eggNOG database v5.0 [60], and the sequence searches were performed using DIAMOND [61]. Putative Auxiliary metabolic genes were additionally annotated (AMG) using DRAM-v with default databases [62], the viral mode of DRAM (Distilled and Refined Annotation of Metabolism). For this purpose, the four isolated genomes were processed through Virsorter2 v2.2.4 [63] with the "--prep-for-dramv" parameter to generate an "affi-contigs.tab" needed by DRAM-v to identify AMGs. Putative AMGs were identified based on the resulting assigned auxiliary score (AMG score ≤3) and metabolic flag (M flag, no V flag, no A flag). COG [64] and KEGG [65] annotations were derived from the EggNOG mapper and DRAM-V results. vConTACT v2.0 [66] with default settings were used to cluster the four viral genomes together with the sequences from the "ProkaryoticViralRefSeq207-Merged" to generate Viral Clusters (VCs) and determine the genus-level taxonomy of the viral genomes.
Relationships between the four isolated phages and reference genomes were analyzed using nucleic acid-based intergenomic similarities calculated with VIRIDIC (Virus Intergenomic Distance Calculator) v1.0r3.6 using default settings [67]. VIRIDIC identifies intergenomic nucleotide similarities between viruses using BLASTN pairwise comparisons and organizes viruses into clusters (genera ≥ 70% similarities and species ≥ 95% similarities). These cut-offs assign viruses into ranks following the ICTV genome identity thresholds. The reference genomes were selected based on the results of the gene-sharing network analysis where the four isolated phages clustered with viruses from the Prokaryotic Viral RefSeq Database. Genome-based phylogeny and classification of the four isolated viruses together with the same reference prokaryotic viruses were performed using the VICTOR web service (Virus Classification and Tree Building Online Resource). VICTOR is a Genome-BLAST Distance Phylogeny (GBDP) method that computes pairwise comparisons of the amino acid sequences (including 100 pseudo-bootstrap replicates) and uses them to infer a balanced minimum evolution tree with branch support via FASTME, including subtree pruning and regrafting postprocessing [68] for each of the formulas D0, D4, and D6, respectively. Trees were rooted at the midpoint [69] and visualized with ggtree [70]. The OPTSIL algorithm [71], the suggested clustering thresholds [72], and an F-value (fraction of linkages necessary for cluster fusion) of 0.5 were used to estimate taxon boundaries at the species, genus, and family levels for prokaryotic viruses [73]. The position and annotation of predicted viral genes in the phage genomes were visualized using Clinker v0.0.27 [74]. Isolated viruses were compared and visualized to the closest related species determined based on intergenomic similarity analysis. Clinker generates global alignments of amino acid sequences based on the BLOSUM62 substitution matrix. A 0.5 identity threshold was used to display the alignments. The complete phage genome sequences (assemblies) are available at NCBI under accession numbers OP902292-OP902295. Raw sequence reads were deposited on the Sequence Read Archive (SRA) under BioProject PRJNA908753 and accession numbers SRR22580853, SRR22580850, SRR22580849, and SRR22580845.

Results
Bacteriophages were isolated from the Baltic Sea water. Four novel phages infecting bacterial colonizers of the Cnidarian moon jellyfish A. aurita were identified and characterized.

Isolation of Bacteriophages from Baltic Seawater Targeting Marine Bacteria Associated with A. aurita
Seawater samples from the Kiel fjord (Baltic Sea) were used for phage enrichments with a pool of 55 different bacteria associated with A. aurita. The bacteria chosen were previously described [22] and represent a diverse set of abundant species associated with A. aurita, possessing varying forms, colors, and colony morphologies (

Plaque and Virion Morphology Assign the Phages to the Class of Caudoviricetes
The identified virulent bacteriophages KMM1 (Pseudomonas phage), KMM2 (Citrobacter freundii phage), KMM3 and KMM4 (Citrobacter sp. phages) formed clear plaques with well-defined boundaries when infecting the respective host-bacterial strain after 16 h of incubation at 30 • C. Notably, infection with KMM3 resulted in a clear center and a turbid surrounding halo ( Figure 1A and Table 3). Lysis plaques were further differentiated by halo size. Phages KMM1, KMM2, and KMM4 generated plaques with a diameter varying between 0.8 mm and 1.2 mm, while KMM3 showed plaques with a diameter of 3 mm ( Figure 1A and Table 3). Seawater samples from the Kiel fjord (Baltic Sea) were used for phage enrichments with a pool of 55 different bacteria associated with A. aurita. The bacteria chosen were previously described [22] and represent a diverse set of abundant species associated with A. aurita, possessing varying forms, colors, and colony morphologies (

Plaque and Virion Morphology Assign the Phages to the Class of Caudoviricetes
The identified virulent bacteriophages KMM1 (Pseudomonas phage), KMM2 (Citrobacter freundii phage), KMM3 and KMM4 (Citrobacter sp. phages) formed clear plaques with well-defined boundaries when infecting the respective host-bacterial strain after 16 h of incubation at 30 °C. Notably, infection with KMM3 resulted in a clear center and a turbid surrounding halo ( Figure 1A and Table 3). Lysis plaques were further differentiated by halo size. Phages KMM1, KMM2, and KMM4 generated plaques with a diameter varying between 0.8 mm and 1.2 mm, while KMM3 showed plaques with a diameter of 3 mm ( Figure 1A and Table 3).   Transmission Electron Microscopy (TEM) images revealed a pre-classification of all phages to the class of Caudoviricetes, characterized by long tails with a collar, base plates with short spikes, six long kinked tail fibers, and isometric heads (Figures 1 and S1). Imaging further indicated that all phages could be assigned to the class of Caudoviricetes. KMM1, KMM2, and KMM4 showed long contractile tails typical for Myovirus-like phages, while KMM3 displayed a long non-contractile tail characteristic for Siphovirus-like phages ( Figure 1, Table 3, and Figure S2). The width of the phage heads of KMM2, KMM3, and KMM4 ranged from 50 ± 1.9 nm to 58.2 ± 2.7 nm. The tail length was similar for KMM2 and KMM4, with an average length of 90.7 ± 4.2 nm, while the tail length of KMM3 was 131.7 ± 9.3 nm. Phage KMM1 was the largest of the isolated phages, with a head width of 75.8 ± 3 nm and a tail length of 176.4 ± 8.4 nm ( Table 3).

Phages KMM1, KMM2, and KMM4 Have a Shorter Lytic Cycle Than Phage KMM3
To assess each phage's capacity for infection, one-step growth curves were conducted with the respective host strains in an MB medium at 30 • C for 120 min in three independent biological replicates ( Figure 2). The adsorption rate and constant k were calculated within 5 min of adsorption time, resulting in 99% of already adsorbed phage particles after 5 min leading to an adsorption constant ranging between 1.08 × 10 −7 and 8.70 × 10 −8 ( Table 4). The calculated values for latent time and burst size are displayed in Table 3. KMM1 and KMM2 each showed an approximately 20 min latent period resulting in a burst size of 55 pfu/cell after 100 min (KMM1), while KMM2 released an average yield of 280 pfu/cell after 110 min. KMM4 infection resulted in a prolonged latent period of 30 min leading to 120 released phages (pfu/cell) after 100 min. KMM3 showed the most extended latent period with 45 min and was characterized by a burst size of 60 pfu/cell, reached after 90 min.

KMM1 Is a Powerful Broad-Host-Range Phage, Whereas KMM2, KMM3, and KMM4 Are Narrow-Host-Range Phages
The KMM1 phage was initially found to infect Pseudomonas sp., while KMM2, KMM3, and KMM4 were shown to infect Citrobacter spp., which are phylogenetically classified in the Pseudomonadaceae and Enterobacteriaceae, respectively. The host range of the phages was determined by spot assays on 43 strains (Table 1, column "Use in the study", category "Host range") belonging to the same genera, Pseudomonas and Citrobacter.
Furthermore, phages were tested against representatives of Shewanellaceae and Rhodobacteraceae of phylum Proteobacteria, Staphylococcaceae and Streptococcaceae of Bacilli, and Chryseobacterium, Olleya, and Maribacter of the abundant class of Flavobacteriia present in the A. aurita-associated microbiota. Bacterial sensitivity to a given bacteriophage was evaluated based on the occurrence of a lysis halo. Additionally, the respective phage efficiency of plating (EOP) was determined with those bacteria showing lysis in the spot tests. EOP for each host bacterium was calculated by comparing it with a score of 10 9 pfu/mL obtained for the original host infection. As shown by the heatmap in Figure 3, KMM1 infects, in addition to the primary host, two strains of Pseudomonas, one Shewanella strain also belonging to Gamma-Proteobacteria, one Sulfitobacter strain belonging to Alpha-Proteobacteria, and 14 strains of the Gram-positive family Staphylococcaceae, two of them with a slightly higher EOP. The phages KMM2, KMM3, and KMM4 showed comparable and narrow host ranges within the genus Citrobacter. However, the observed phage titres and EOP were different as indicated by the color coding dependent on the value (Figure 3). Phages KMM2 and KMM4 were further able to infect the Enterobacteriaceae bacterium Shigella flexneri. In contrast, phage KMM3 infected two Escherichia coli strains of Enterobacteriaceae. The phages infected none of the Flavobacteriia representatives.  The KMM1 phage was initially found to infect Pseudomonas sp., while KMM2, KMM3, and KMM4 were shown to infect Citrobacter spp., which are phylogenetically classified in the Pseudomonadaceae and Enterobacteriaceae, respectively. The host range of the phages was determined by spot assays on 43 strains (Table 1, column "Use in the study", category "Host range") belonging to the same genera, Pseudomonas and Citrobacter.
Furthermore, phages were tested against representatives of Shewanellaceae and Rhodobacteraceae of phylum Proteobacteria, Staphylococcaceae and Streptococcaceae of Bacilli, and Chryseobacterium, Olleya, and Maribacter of the abundant class of Flavobacteriia present in the A. aurita-associated microbiota. Bacterial sensitivity to a given bacteriophage was evaluated based on the occurrence of a lysis halo. Additionally, the respective phage efficiency of plating (EOP) was determined with those bacteria showing lysis in the spot tests. EOP for each host bacterium was calculated by comparing it with a score of 10 9 pfu/mL obtained for the original host infection. As shown by the heatmap in Figure 3, KMM1 infects, in addition to the primary host, two strains of Pseudomonas, one Shewanella strain also belonging to Gamma-Proteobacteria, one Sulfitobacter strain belonging to Alpha-Proteobacteria, and 14 strains of the Gram-positive family Staphylococcaceae, two of them with a slightly higher EOP. The phages KMM2, KMM3, and KMM4 showed comparable and narrow host ranges within the genus Citrobacter. However, the observed phage titres and EOP were different as indicated by the color coding dependent on the value (Figure 3). Phages KMM2 and KMM4 were further able to infect the Enterobacteriaceae bacterium Shigella flexneri. In contrast, phage KMM3 infected two Escherichia coli strains of Enterobacteriaceae. The phages infected none of the Flavobacteriia representatives.

Novel Phage Species Confirmed by Genome Sequencing Analysis
The viral genomes of the highly effective virulent phages were sequenced using Nanopore technology. Complete phage genomes were assembled (NCBI Accession Nos. OP902292-OP902295) from Nanopore long reads of the double-stranded DNA. Table 2 summarizes the key information regarding sequencing, assembly, and annotation. Three of the four viral genomes were assigned "high-quality" (>90% completeness), while phage KMM2 was assigned as "complete" due to the presence of direct terminal repeats (DTR), which may indicate a circular genome. KMM1 has a 137 kbp genome with a GC content of 31%, while KMM2 and KMM4 showed genome sizes of approx. 87 kbp bp with an average GC content of 39%. The KMM3 genome was found to be the smallest, with 49 kbp but with the highest GC content of 43%. In total, 259 putative ORFs were predicted in the genome of phage KMM1, 137 and 138 ORFs were predicted in KMM2 and KMM4 genomes, respectively, and 92 ORFs in the KMM3 genome (Table 2). Phages were clustered into species and higher-order groups to investigate phage genomic diversity and identify closely related groups of phages. Viral Clusters (VCs) and genus-level taxonomy of the four isolated and sequenced viral genomes were generated (Table S1). VICTOR (based on pairwise whole genome distance comparisons) was used to compare 96 previously described viral taxa with the phage genomes of KMM1-KMM4. The results indicated that

Novel Phage Species Confirmed by Genome Sequencing Analysis
The viral genomes of the highly effective virulent phages were sequenced using Nanopore technology. Complete phage genomes were assembled (NCBI Accession Nos. OP902292-OP902295) from Nanopore long reads of the double-stranded DNA. Table 2 summarizes the key information regarding sequencing, assembly, and annotation. Three of the four viral genomes were assigned "high-quality" (>90% completeness), while phage KMM2 was assigned as "complete" due to the presence of direct terminal repeats (DTR), which may indicate a circular genome. KMM1 has a 137 kbp genome with a GC content of 31%, while KMM2 and KMM4 showed genome sizes of approx. 87 kbp bp with an average GC content of 39%. The KMM3 genome was found to be the smallest, with 49 kbp but with the highest GC content of 43%. In total, 259 putative ORFs were predicted in the genome of phage KMM1, 137 and 138 ORFs were predicted in KMM2 and KMM4 genomes, respectively, and 92 ORFs in the KMM3 genome (Table 2). Phages were clustered into species and higher-order groups to investigate phage genomic diversity and identify closely related groups of phages. Viral Clusters (VCs) and genus-level taxonomy of the four isolated and sequenced viral genomes were generated (Table S1). VICTOR (based on pairwise whole genome distance comparisons) was used to compare 96 previously described viral taxa with the phage genomes of KMM1-KMM4. The results indicated that the four identified phages belong to three different ranks within the class Caudoviricetes, as they are grouped into three different clades in the phylogenetic tree ( Figure 4, Table S1). More precisely, based on the latest ICTV classification framework, the four phages are assigned to the phylum Uroviricota. KMM1 assignment was resolved until the family Herelleviridae. KMM2 and KMM4 were classified until the genus level Suspvirus, belonging to the subfamily Ounavirinae. KMM3 belongs to the Drexlerviridae family, Tempevirinae subfamily, and subclusters into the genera Hicfunavirus and Tlsvirus due to gene homologies to both genera. Based on VICTOR and vConTACT2 analysis, the four isolated phages belong to four predicted genera and three species (Figure 4; Table S1). VIRIDIC (Virus Intergenomic Distance Calculator) was used to determine the pairwise intergenomic similarities between the phage genomes characterized in this work compared to reference genomes. The intergenomic analysis provided evidence that the isolated viral genomes are four novel species ( Figure S2, Table S1).  Although experimental results revealed four different phages; genome analyses using VICTOR and VIRIDIC resulted in contrasting statements. Both programs confirmed that KMM1 and KMM3 are novel species. However, those analyses did not determine whether KMM2 and KMM4 were one or two species or whether they were novel. In the next step, genomes were annotated and compared to their best homologs ( Figure 5). Open reading frames (ORFs) were identified, encoding basic phage-related functions, including phage DNA metabolic proteins, phage structural proteins, lysis-related proteins, and hypothetical proteins ( Figure 5 and Table S2). Genome annotations of KMM2 and KMM4 showed minor differences in their direct comparison, such as the length of genes and the presence or absence of specific genes ( Figure 5B), suggesting that these are two different and novel phages, verifying phage genome analysis using VIRIDIC.

Discussion
Viruses are found in all habitats on Earth [75], but their importance is probably most evident in the ocean, where they are considered a source of diversity in genetic variation [76][77][78]. Estimates suggest that phage numbers are tenfold higher than those of bacteria in the ocean, with phage particle estimates of 10 23 , resulting in turnover rates of 10 25 infections and lysis events per second in the ocean impacting nutrient cycling [79,80]. The relative proportions of virulent and temperate phages vary depending on various environmental factors, such as temperature, salinity, and nutrient availability [81,82]. In general, virulent phages tend to dominate in nutrient-rich environments with high microbial diversity and abundance and are thus more prevalent in surface waters, while temperate phages are more prevalent in nutrient-poor environments in deeper (oligotrophic) waters [83]. For instance, in the Baltic Sea, it has previously been demonstrated that virulent viruses are more common in surface waters, whereas lysogeny predominates in deep marine waters [77,84]. Lytic representatives of Siphovirus-like (52%), Myovirus-like (42%), and Podovirus-like (6%) phages of the Caudoviricetes class were consistent in the surface water throughout all seasons within the Baltic Sea [84][85][86][87]. Those phages can have several roles particularly in the microbiomes of marine animals and plants, e.g., to maintain a healthy microbiome and prevent the spread of diseases [24,[88][89][90]. Further, phages appear respon- In summary, four new phages from the Baltic Sea water column were identified and characterized that efficiently and effectively infect Pseudomonas and Citrobacter bacteria, members of the complex microbiota of the moon jellyfish A. aurita. Particularly, phage KMM1 is a highly virulent, efficient phage infecting Gram-negative and Gram-positive bacterial species of genera Pseudomonas and Staphylococcus.

Discussion
Viruses are found in all habitats on Earth [75], but their importance is probably most evident in the ocean, where they are considered a source of diversity in genetic variation [76][77][78]. Estimates suggest that phage numbers are tenfold higher than those of bacteria in the ocean, with phage particle estimates of 10 23 , resulting in turnover rates of 10 25 infections and lysis events per second in the ocean impacting nutrient cycling [79,80]. The relative proportions of virulent and temperate phages vary depending on various environmental factors, such as temperature, salinity, and nutrient availability [81,82]. In general, virulent phages tend to dominate in nutrient-rich environments with high microbial diversity and abundance and are thus more prevalent in surface waters, while temperate phages are more prevalent in nutrient-poor environments in deeper (oligotrophic) waters [83]. For instance, in the Baltic Sea, it has previously been demonstrated that virulent viruses are more common in surface waters, whereas lysogeny predominates in deep marine waters [77,84]. Lytic representatives of Siphovirus-like (52%), Myovirus-like (42%), and Podovirus-like (6%) phages of the Caudoviricetes class were consistent in the surface water throughout all seasons within the Baltic Sea [84][85][86][87]. Those phages can have several roles particularly in the microbiomes of marine animals and plants, e.g., to maintain a healthy microbiome and prevent the spread of diseases [24,[88][89][90]. Further, phages appear responsible for several diseases that harm corals and their symbionts [91][92][93][94]. Besides, phages promote the evolution of new traits or the acquisition of beneficial genes within a microbiome by mediating horizontal gene transfer between bacteria by transduction [95,96]. Lastly, the impact mentioned above on nutrient cycling in marine ecosystems by breaking down bacterial cells and releasing nutrients into the environment can have important implications for marine organisms' overall health and productivity [97]. Although the role of Cnidarian bacterial communities has already been intensively investigated [22,[98][99][100], the impact of (virulent) phages on Cnidarian and particularly on A. aurita's bacterial colonizers has yet only rarely been studied. However, different Hydra species have been shown to harbor a diverse host-associated virome predominated by bacteriophages [39,101,102]. Changes in environmental conditions altered the associated virome, increased viral diversity, and affected the metabolism of the metaorganism [102]. The specificity and dynamics of the virome point to a potential viral involvement in regulating microbial associations in the Hydra metaorganism [101].
In this study, four phages (Pseudomonas phage KMM1, Citrobacter phages KMM2, KMM3, and KMM4) were isolated from the Baltic Sea water column (Kiel fjord) surrounding A. aurita individuals by a cultivation-based approach, infecting previously isolated bacteria, Pseudomonas and Citrobacter, both present in the associated microbiota of A. aurita [21,22]. Both genera, Pseudomonas and Citrobacter, are Gram-negative bacteria widely distributed in marine environments, including seawater, sediments, and marine eukaryotes [103][104][105][106]. Pseudomonas species play a critical role in the marine nitrogen cycle, as they are capable of nitrogen fixation and denitrification [107]. While less is known about Citrobacter-specific ecological roles in marine environments, it can cycle degrading organic matter [108]. Both Pseudomonas and Citrobacter are opportunistic pathogens that can cause infections in marine multicellular organisms [103,104,109]. However, they are likewise considered essential players in maintaining the health and balance of marine ecosystems [110,111]. Consequently, virulent phages infecting those bacteria might disturb the ecosystem and metaorganism homeostasis. Bacteriophages that infect Pseudomonas and Citrobacter have been identified in marine environments, and they may play essential roles in regulating the abundance and activity of these bacteria [82,108,109]. For example, bacteriophages that infect Pseudomonas can control its population size and limit its ability to degrade organic matter, significantly impacting nutrient cycling in marine ecosystems [112,113]. Similarly, bacteriophages that infect Citrobacter can reduce the abundance of this bacterium and thus limit its ability to colonize and persist in marine environments, potentially reducing the risk of animal diseases and human exposure to this pathogen, e.g., through seafood consumption [103,114,115].
Phages KMM1, KMM2, and KMM4 showed a clear, roundish plaque morphology, as previously described for most Caudoviricetes with long contractile tails (formerly known as Myovirus-like phages) [116]. Phage KMM3, on the other hand, showed larger plaques with a clear center surrounded by a turbid halo, commonly referred to as a "bull's eye" plaque [117,118]. The clear halo in the plaque's center represents the phage's lytic activity. The turbid ring surrounding the clear halo is formed by accumulating uninfected or partially infected host bacterial cells. These cells can resist phage infection (acquired resistance, defense systems) or have only been partially infected, potentially based on the aging of the bacterial lawn (non-infective after log phase), associated increases in the size of microcolonies making up the bacterial lawn, or because of less general phenomena such as the lysis inhibition phenotype [51,119]. However, it is important to note that phage plaque morphology can vary depending on the specific phage-host system and experimental conditions [120,121].
KMM1 reflects a broad host range phage capable of infecting a wide range of bacterial hosts, whereas KMM2/3/4 reflect narrow host range phages, infecting only a limited number of bacterial strains or species [80,122]. Remarkably, KMM1 efficiently infects Gram-negative Pseudomonas strains as well as a variety of strains of various Gram-positive Staphylococcus species. Gram-negative and Gram-positive bacteria differ in cell walls and membrane characteristics [123,124]. Gram-positive bacteria have thick multilayered peptidoglycan in their cell envelope, while Gram-negative bacteria's cell walls have only a thin layer of peptidoglycan covered by an outer membrane. Consequently, phages require strain-, species-, or even higher-order-specific characteristics for bacteria recognition, attachment, and lysis. Phages usually use their tail fibers or spikes to recognize and attach to specific receptors on the bacterial cell wall of Gram-negatives, such as lipopolysaccharides (LPS) or outer membrane proteins [125,126]. The tail fibers or spikes bind to these receptors through specific interactions, which can be highly selective. Gram-positive bacteria are likewise precisely recognized and infected by teichoic acids or other surface proteins [127][128][129]. Moreover, phages replicate differently inside their host cells, depending on the bacterial structures. Gram-negative-specific phages enter the host cell by injecting their genetic material directly into the cytoplasm, where it replicates and assembles new phage particles [130]. In contrast, Gram-positive phages usually enter the host cell by binding to receptors on the host surface, replicating and assembling new phage particles in the cytoplasm [131,132]. To the best of our knowledge, only a few characterized phages in the marine environment have such a broad host range as the phage KMM1. However, four characterized phages are already isolated from activated sludge samples, and one phage identified from a freshwater sample capable of infecting both types [133][134][135]. We can only speculate that KMM1 might recognize and bind to conserved structural components of bacterial cells. KMM1 might have evolved in the marine environment due to the frequent presence of Pseudomonas and Staphylococcus, allowing recognition and infection of a wide range of bacterial hosts. Potentially, KMM1 prefers the highly abundant Pseudomonas as a host, but it can also infect Staphylococcus under certain, possibly stressful, environmental conditions that promote Staphylococcus abundance. Genome analysis of KMM1 supports the experimentally collected data, i.e., that it is capable of infecting both Gram-positive and Gram-negative bacteria. In this respect, we identified several features within the KMM1 genome indicative of Gram-negative bacteria infection, such as Pseudomonas. In more detail, we identified YbiA-like superfamily proteins (IPR037238) derived from various Gram-negative bacteria (Table S2), such as E. coli K-12 [136], and Pseudomonas, suggesting the infection of Gram-negative bacteria. Similarly, the invasin/intimin celladhesion domain (IPR008964) was found in the phage genome, common in phages that infect Gram-negative bacteria such as Erwinia [137]. In contrast, we further identified the CHAP (cysteine, histidine-dependent amidohydrolase/peptidase) domain (IPR007921) in the KMM1 genome. This domain has been shown to be specifically responsible for the major catalytic activity of the endolysin in degrading cell wall peptidoglycan of staphylococci, including methicillin-resistant Staphylococcus aureus [138][139][140]. Future studies will reveal more insights into the complex attachment and infection mechanisms of phage KMM1 and disclose the underlying mechanisms of its broad host range.
All phages identified in this study showed effective and efficient lysis of Pseudomonas and Citrobacter by fast and effective binding of the phage to the host cells, short latency periods, and high burst sizes (Tables 3 and 4). Those characteristics are important features affecting natural microbiomes and relevant for potential therapeutic applications. Phage therapy uses intact natural phages or phage compounds to treat bacterial infections [141]. Due to the growing number of antibiotic-resistant bacterial species and the ban on the use of antibiotics in the aquatic environment [78,[142][143][144], the interest in phage therapy particularly for aquaculture increased during the last few decades [145][146][147]. Phage therapy relies on extraordinary qualities of phages, including host specificity, self-replication, wide distribution, and safety [43,141,146,148]. Since phages are a natural way of managing bacterial infections, their usage does not contribute to the development of antibiotic resistance or the deposition of harmful residues in the environment. Finally, phages are versatile since they may be used alone or in cooperation with antibiotics or other therapies to improve their potency against bacterial infections. These features are entirely applicable in aquaculture, where traditional approaches to deal with pathogenic bacteria, such as antibiotics, are impossible [114,115]. Particularly the identified broad host range bacteriophage KMM1 targets various bacteria, increasing its convenience in aquaculture where different bacterial species can cause infections [43]. KMM1's high infectivity rate can quickly and effectively reduce bacterial populations, limiting the spread of infection. This phage's stability, safety, and cost-effective production under different environmental conditions, such as pH and temperature, must be tested for potential use as an effective alternative to antibiotics in controlling bacterial infections in aquaculture.
Lastly, our study demonstrated that phage research methods are still in their infancy, although several benchmarking studies on genomic and seasonal variation and diversity of tailed phages in the Baltic Sea were already published [84,85]. Bioinformatics tools specifically developed explicitly for phages still lag behind similar tools used in bacterial research [149][150][151]. However, some improved tools in recent years include Phaster, a web-based tool for identifying and annotating phage sequences in bacterial genomes and predicting their completeness [152]. VirSorter, VirFinder, and Phage AI implement machine learning to identify viral sequences in metagenomic datasets, distinguish between viral genomes, plasmids, and transposons, and predict phage host ranges and other characteristics [153,154]. In the present study, we used VICTOR for pairwise distance-based comparisons of the whole genome, vConTACT2 to determine the genus-level taxonomy of viral genomes, and VIRIDIC for calculating the intergenomic distance of viruses. The results obtained on species assignment of KMM2 and KMM4 using VICTOR and vCon-TACT2 demonstrate that phage softwares result in contrasting statements and must be improved. Using VICTOR and vConTACT2 did not allow for the differentiation of highly similar phages. Experimental data on differences within the lytic cycle and host range (Figures 2 and 3), in combination with genome annotation, pointed to different species assignments of KMM2 and KMM4 ( Figure 5). However, by improving estimates of phage genome similarity, particularly for distantly related phages, analyzing datasets including thousands of phage genomes, and creating an informative heatmap that includes not only the similarity values but also information about the genome lengths and aligned genome fraction, VIRIDIC finally confirmed taxonomic rank assignments ( Figure S2 and Table S1).
In summary, the present study describes the identification and characterization of four novel bacteriophages. The broad-host-range phage KMM1 infects members of genera Pseudomonas and Staphylococcus. Likely, KMM1 adapted its attachment and infection mechanisms through co-evolution with its bacterial hosts. Phages KMM2-4 infect Citrobacter and close relatives of Enterobacteriaceae, thus possessing a narrow host range. Although all identified phages demonstrated effective and efficient virulent properties relevant for phage application, future studies on the impact of those phages on the native microbiome of the moon jellyfish A. aurita, a model in metaorganism research, are of particular interest. Such studies may provide insights into the complex interdependence of phages and their bacterial hosts and how these relationships affect microbiomes to investigate the impact on the eukaryotic host A. aurita. It is conceivable that the infection of A. aurita with phages KMM1-4 might cause substantial changes in the bacterial community, potentially disrupting the multicellular host's homeostasis. Even assuming that the phages impact A. aurita's microbiome structure, it may be that the eukaryotic host has mechanisms to maintain its homeostasis. Future studies will focus on the characterization of attachment and infection mechanisms, particularly of phage KMM1, and the impact of the identified phages on the microbiome and, consequently, the health of A. aurita.