Isolation and characterization of the first phage infecting ecologically important marine bacteria Erythrobacter

All of the bacterial strains used in this study are listed in Table 1. E. litoralis DSM 8509 and other strains were cultivated at 30 °C in RO medium, an artificial seawater medium containing 1 g/L yeast extract, 1 g/L tryptone, and 1 g/L sodium acetate at pH 7.5 [24].

Phage vB_EliS-R6L was isolated from seawater obtained in March 2014 off the coast of Xiamen, China (118°04′ E, 24°31′ N), using standard virus enrichment and double-layer agar methods. Briefly, E. litoralis DSM 8509 (100 mL) was co-cultured with a pre-filtered (0.22-μm membrane filter; Millipore, USA) seawater sample (20 mL) for 24 h at 30 °C. The culture was filtered again and serially diluted to determine phage activity using a double-layer agar method [25]. A single plaque was collected from the plate using a sterile pipette (Fisher, Canada) and then purified four successive times using the double-layer agar method. Following purification, stock cultures of the phage were prepared using sodium chloride-magnesium sulfate (SM) buffer (100 mM NaCl, 50 mM Tris, 10 mM MgSO₄, and 0.01% gelatin, pH 7.5) supplemented with several drops of chloroform and stored at 4 °C and −80 °C.

For TEM analysis, 1 L of E. litoralis DSM 8509 culture (OD600 = 0.5) was inoculated with the phage at a multiplicity of infection of 10 and cultivated for 24 h at 30 °C. The mixture was centrifuged at 6000×g for 10 min, and the upper aqueous phase was filtered through a 0.22-μm membrane and precipitated with 10% (w/v) dissolved polyethylene glycol 8000 (containing 1 M NaCl). After >8 h at 4 °C, the mixture was centrifuged at 10,000×g for 50 min at 4 °C, and the pellet was gently resuspended in 5 mL of SM buffer. The phages were then purified by CsCl gradient ultra-centrifugation (gradient-density: 1.5 g/mL, 200,000×g, 24 h, 4 °C; Optima L-100 XP Ultracentrifuge, Beckman Coulter). The purified phage particles were collected and dialyzed twice in SM buffer; 20 μL of suspension was added dropwise onto a copper grid and negatively stained with 2% aqueous uranyl acetate for 10 min. Transmission electron micrographs were obtained using a JEM-2100HC transmission electron microscope (JEOL, Japan) at an accelerating voltage of 120 kV. The phage size was calculated from at least 20 particles.

To determine whether phage vB_EliS-R6L contains lipids, its sensitivity to chloroform was examined as described previously [26]. Briefly, 500 μL of the phage suspension (~10⁹ plaque forming units (PFU)/mL) were mixed with 5 μL, 50 μL, or 500 μL of chloroform, vigorously shaken for 2 min, and then incubated at 30 °C for 30 min. The samples were immediately diluted and plated for phage titration using double-layer agar plates inoculated with E. litoralis DSM 8509.

To investigate the host range of phage vB_EliS-R6L, plaque assays were performed on 27 marine bacterial strains, including 21 Erythrobacter strains, two Citromicrobium strains, and one each of the genera Roseobacter, Dinoroseobacter, Lutibacterium, and Halomonas (Table 1). The host range was determined by adding 5 μL of a diluted phage suspension (~10⁷ PFU/mL) dropwise onto the surface of double-layer agar plates inoculated with the bacterial strain of interest. The plates were incubated at 30 °C for up to 7 days, and plaque formation was assessed repeatedly during this period. The efficiency of plating (EOP) of susceptible strains was quantified by calculating the ratio of the PFU obtained with each phage-susceptible strain to the PFU obtained with E. litoralis DSM 8509. All assays were carried out in triplicate.

One-step growth curve experiments were performed as previously described [25, 27]. Briefly, mid-exponential phase E. litoralis DSM 8509 (optical density at 600 nm = 0.3–0.5, 100 mL) was inoculated with phage at a multiplicity of infection of 0.01 and allowed to adsorb for 10 min at 30 °C. The mixture was then centrifuged at 6000×g for 10 min to remove non-absorbed phage in the supernatant; the pelleted cells were resuspended in 100 mL of RO medium, followed by incubation at 30 °C. Two sets of duplicate samples were removed at 20-min intervals for 6 h, and chloroform (1% final concentration) was added to the second set to release the intracellular phage. The two samples were then diluted and immediately plated for phage titration using the double-layer agar plate method. Another set of cultures without phage inoculation served as the blank control. Samples for optical density (OD₆₀₀) measurements from both the treated and untreated cultures were removed at the 20-min intervals for 6 h and at 1-h intervals for the next 4 h. The PFU of each sample was calculated by counting the plaques on the bacterial lawn. The assay was performed in triplicate.

To investigate the thermal stability of the phage, 1 mL of phage vB_EliS-R6L (~10 ⁷ PFU/mL) with SM buffer was incubated for 2 h at 30 °C, 40 °C, 42.5 °C, 45 °C, 50 °C, 60 °C, 70 °C, 75 °C, or 80 °C, after which the phage suspensions were immediately cooled to 4 °C for activity estimation. To evaluate the stability of the phage at different pH levels, RO medium was adjusted to pH 1–14 with sterile 5 M HCl or NaOH solution and then filtered through a 0.22-μm membrane filter (Millipore, USA). Additionally, 1 mL of a phage suspension (~10 ⁷ PFU/mL) prefiltered through a 0.22-μm membrane filter was incubated at 30 °C for 24 h in 9 mL RO medium of different pHs. Phage activity was determined using the double-layer agar method with RO medium (pH 7.5) at 30 °C and assessed by calculating changes in PFU following exposure to the different temperatures and pH levels. All assays were performed in triplicate.

To investigate the effect of temperature on phage infection, 5 μL of a phage suspension (~10 ⁹ PFU/mL) was added dropwise onto double-layer agar plates containing E. litoralis DSM 8509 and incubated at 15 °C, 20 °C, 25 °C, 27.5 °C, 30 °C, 35 °C, and 40 °C for 7 days. To investigate the effect of pH on infection, the pH of RO medium was adjusted with 5 M HCl (pH 4–5), 0.2 M Na₂HPO₄/ NaH₂PO₄ (pH 6–8) or 0.1 M NaHCO₃/Na₂CO₃ (pH 9–11); after autoclaving, the pH was checked with pH test paper and readjusted if necessary. Next, 5 μL of a phage suspension (~10 ⁹ PFU/mL) was added dropwise onto double-layer agar plates inoculated with host cells at different pH values. The plates were incubated at 30 °C for up to 7 days. All assays were performed in triplicate.

To investigate whether the phage can integrate onto the genome of its host, 10 μL of a phage suspension (~10 ⁹ PFU/mL) was added dropwise onto double-layer agar plates inoculated with E. litoralis DSM 8509; the center portion within the plaques was carefully pipetted out and inoculated onto a new plate. After two rounds of isolation and purification, 40 randomly selected bacterial colonies were chosen for colony polymerase chain reaction (PCR) using two pairs of primers, designed according to phage genome annotation, targeting ORF 91 (Major capsid protein) (forward primer 5′ –GCTGACCACCAAGCAGATGA - 3′, reverse primer 5′ - CGGAACGAGGCTATCCCAC - 3′, 521 bp) and ORF 100 (Terminase) (forward primer 5′ - TCATGTGGCAGGCTTGGG - 3′, reverse primer 5′ - GGGTCGGTCCAGTCTTTCG - 3′, 549 bp).

Using the same sample preparation utilized for TEM analysis, 1 mL of a phage suspension was purified by CsCl density-gradient centrifugation, followed by dialysis. To remove free DNA and RNA, the sample was then digested at 37 °C for 1 h with DNase I and RNase A (Takara) at final concentrations of 1 μg/mL. The solution was incubated with proteinase K and sodium dodecyl sulfate at final concentrations of 100 μg/mL and 1% (w/v), respectively, at 55 °C for 2 h. After incubation, the solution was extracted twice with phenol:chloroform:isoamyl alcohol (25:24:1) and once with chloroform:isoamyl alcohol (24:1), after which the solutions were precipitated with sodium acetate and precooled ethanol at final concentrations of 1/10 and 1/1 (v/v), respectively. After overnight incubation at −20 °C, DNA was collected by centrifugation and successively washed twice with precooled 70% and 100% ethanol. The genomic DNA of vB_EliS-R6L was sequenced using the Illumina Miseq platform to generate 2 × 251 bp paired-end reads. The reads were assembled using CLC Genomics Workbench software (18,777 × coverage).

The GeneMarkS online server (http://​exon.​gatech.​edu/​Genemark/​genemarks.​cgi), Glimmer 3.0 (http://​ccb.​jhu.​edu/​software/​glimmer/​index.​shtml), and the ORF Finder online server (https://​www.​ncbi.​nlm.​nih.​gov/​orffinder/​) were used to identify putative open reading frames (ORFs). Genes were annotated using BLAST searches against the NCBI non-redundant (nr) protein database, with a cut-off of E-value ≤10⁻⁵. A temperate and/or lytic lifestyle was predicted using the phage classification toolset (PHACTS) online prediction program (http://​www.​phantome.​org/​PHACTS/​index.​php). Methylase ORFs were searched using the REBASE^R online program (http://​rebase.​neb.​com/​rebase/​rebase.​html).

The amino acid sequences of endolysin protein (ORF 74), major capsid protein (ORF 91), portal protein (ORF 98), and terminase (ORF 100) from phage vB_EliS-R6L were used to construct neighbor-joining phylogenetic trees with MEGA 6.06 and 200 bootstrap replications. For use in the phylogenetic analysis, the amino acid sequences of these four proteins from closely related phages were retrieved from GenBank.

To explore the geographic distribution of vB_EliS-R6L-like phages, the amino acid sequences of the phage ORFs were employed as queries to search against metagenomic databases of the Jiulong River Estuary (JRE), Xiamen, China [28], the Pacific Ocean Virome (POV) and Global Ocean Survey (GOS) (http://​data.​imicrobe.​us/) using tBLASTn at a cut-off of E-value ≤10⁻⁵, an alignment value ≥30 and a score value ≥40. The count abundance of each read was normalized by dividing by the number of total reads in the database and the size of the gene product [20].

The genome sequence of phage vB_EliS-R6L was deposited in the GenBank database under accession number KY006853.

To the best of our knowledge, vB_EliS-R6L is the first phage isolated from the ecologically important marine bacteria of the genus Erythrobacter. vB_EliS-R6L forms small, clear, round (1–4 mm diameter) plaques on a bacterial lawn (Fig. 1). After treatment with different concentrations of chloroform (i.e., 1%, 10%, and 100% (v/v)), the phage showed survival rates of 94.7 ± 4.9%, 83.1 ± 2.5%, and 81.9 ± 1.9%, respectively, indicating that vB_EliS-R6L may not sensitive to chloroform or contain lipids. TEM micrographs revealed that it belongs to the siphovirus family, with an icosahedral capsid 75.9 ± 2.2 nm in diameter and a characteristically long tail of 165.6 ± 2.3 nm (Fig. 1).

Of all the strains tested, phage vB_EliS-R6L could only infect E. litoralis DSM 8509 and E. longus DSM 6997, the only strains for each species that could be obtained from public culture collections. Of the commonly isolated three-tailed phage families (Myoviridae, Siphoviridae, and Podoviridae), Myoviridae phages have a broader host range than species of the other two families. Therefore, it was not unexpected that a narrow host range was observed for phage vB_EliS-R6L. Based on whole-genome comparison, Zheng et al. (2016) reported that Erythrobacter strains cluster into three groups, with strains DSM 8509, DSM 6997, and JL 475 belonging to the same group. These three strains share high 16S rRNA gene identity (> 97%) but can be discriminated by average nucleotide identity analysis [11]. Integrative and conjugative element analysis showed that DSM 8509 and DSM 6997 cluster closely together and away from JL 475, suggesting asynchronous evolution. This may account for the ability of phage vB_EliS-R6L to infect DSM 8509 and 6997 but not JL 475. In addition, previous studies have suggested that the number of tRNAs can be positively correlated with host range due to compensation for different codon usage patterns in host bacteria [29]. No tRNAs were identified in the phage vB_EliS-R6L genome using tRNAscan-SE (1.3.1) software [30], which may also account for its relatively narrow host range. In lysogenic/lytic assays, 17.5% (7/40) of bacterial isolates from the center portion of plaques showed positive PCR amplification using primers specific for phage ORFs. This finding suggests that vB_EliS-R6L may integrate into its host cell and possibly enter into a lysogenic life cycle, which is consistent with our bioinformatic analysis (see below).

According to one-step growth curve experiments, the eclipse and latent periods of phage infection occurred at 2 h 40 min and 3 h post-infection, respectively (Fig. 2). The burst size was ~86 PFU/cell, similar to the latent period and burst size of most phages infecting Roseobacter species, ranging from <1–6 h and 27–1500 PFU/cell, respectively [23, 31‐33]. Stability assessment showed that over 60% of vB_EliS-R6L phage remained active at temperatures up to 50 °C (2 h treatment) and that <1% remained active at temperatures >70 °C (Fig. 3). In addition, the phage survival rate was greater than 77% after 24 h at pH 6, 7, or 8 (Fig. 3). Although vB_EliS-R6L retained some activity after 24 h at pH 3 (39%) and pH 12 (10%), activity was lost below pH 2 or above pH 13. An infection condition test showed that phage vB_EliS-R6L could infect E. litoralis DSM 8509 and form clear plaques on plates within 2 days at 25 °C ~ 35 °C. Visible plaques appeared on plates after 4 days at 15 °C and 20 °C, whereas no clear plaques were visible at 40 °C after 7 days of incubation. In addition, plaques were observed in the infection test within a pH range of 7–10. These data showed phage vB_EliS-R6L particles to be stable, with broad temperature and pH tolerance compared to most isolated phages [34], characteristics that might offer more survival opportunities in the diverse marine environment. However, phage vB_EliS-R6L was only able to successfully proliferate within a relatively narrow range of conditions (i.e., < 40 °C, pH 7–10). Unsuccessful infection might be a consequence of thermal/chemical alterations to the phage structure or host receptors [35, 36], and further investigation is needed.

The complete dsDNA genome of phage vB_EliS-R6L is 65,675 bp in size (GenBank accession no. KY006853). The overall G + C content is 66.5%, similar to that of its host (i.e., 65.2%, GenBank accession no. NZ_CP017057). A total of 108 ORFs were identified (Table 2), and identity of the predicted coding sequences with sequences available in GenBank is low (26–77% at the amino acid level). Homologous sequences in the NCBI non-redundant protein database were found for 58 gene products; however, only 29 had predicted functions (Table 2), 19 of which have been assigned to known functional domain categories. In total, 27 ORFs are homologous with previously identified bacteriophage genes, and 15 are homologs of proteins from siphophage-infecting Alphaproteobacteria. Overall, as suggested by the low degree of coverage (< 3%) of the entire genome sequence identified by BLASTn analysis, the vB_EliS-R6L genome is largely unique compared with other published phage genomes.

Eight genes were found to encode proteins related to DNA metabolism. In addition to DNA modification methylase (ORF 1) and DNA methylase (ORF 67), phage vB_EliS-R6L encodes another three methylase proteins, including a methyltransferase (ORF 34), a type I restriction-modification (R-M) system methyltransferase subunit-like protein (ORF 40), and a cytosine-specific methyltransferase (ORF 55). The identities range from 31 to 56% (46% on average). Four of the five ORFs are predicted to contain a single domain, including a CcrM-like domain (ORF 1, with recognition site of GANTC), two SAM methyltransferase domains (COMT-like) (ORF 34 and 40, a versatile enzyme with various target molecules), and a Dam subfamily domain (ORF 67, with a recognition site of GATC). Methyltransferases are ubiquitous in prokaryotic genomes, and these enzymes are often associated with a cognate restriction endonuclease, forming an R-M system that protects bacterial cells from invasion by foreign DNA such as phages. Approximately 20% of annotated phage genomes encode methylases, and it is proposed that they may help the phage overcome R-M and other phage-targeted resistance systems in the host and prolong the effectiveness of infection [37]. As predicted by REBESE software, one R-M pair was recognized in the genome of E. litoralis DSM 8509 (with the recognition site CCGGAG), and five pairs were found for E. longus DSM 6997 (two of which have recognition sites GGCGCC and CGATCG; the other three have no recognition sites). Those R-M recognition sites indicate 37 potential cleavage sites (23 for CCGGAG, 14 for CGATCG) in the genome of phage vB_EliS-R6L. The predicted recognition site GATC of ORF 67 in the phage genome agrees with the R-M sites of DSM 6997, demonstrating the potential to overcome the host R-M system. Previous studies also found that phage T4 encodes a DAM methylase that targets GATC sites, protecting the phage DNA from an R-M system that recognizes this sequence [38]. Based on REBASE searches, 1051 homologs matched with the five methylase proteins, suggesting that R6L-like methylases are widespread, which may enhance infectivity and evasion of the host R-M system. Phage vB_EliS-R6L may represent a good model for exploitation of phage methylases and marine host-phage interactions. Moreover, Dziewit et al. (2014) suggested that methylases may account for differences in the methylation state and induce host transcriptional changes that are essential for the phage life cycle [39].

Twelve ORFs are predicted to encode proteins involved in the structure and assembly of virions, nine of which are homologous to genes from Pseudomonas (Gammaproteobacteria) and/or Roseobacter isometric siphophages [21, 40, 41]. A further four conjunctive ORFs with unknown functions also exhibit homology to these phage types. This is consistent with the results of the phylogenetic trees generated using major capsid protein and portal protein amino acid sequences (Fig. 4). However, it is noteworthy that except for these 13 ORFs, no other ORFs of vB_EliS-R6L show a high degree of homology to Pseudomonas or Roseobacter isometric phage sequences. It therefore appears that genes associated with the structural architecture of phage vB_EliS-R6L are relatively conserved and may have evolved independently from other genes in the genome. Moreover, the low protein identity predicted between phage vB_EliS-R6L and those homologies (26–77%, 41% on average), as well as clearly distant phylogenetic relationships (Fig. 4), suggest that phage vB_EliS-R6L exchanged genetic material with those closely related phages prior to a distinct evolutionary path.

One putative endolysin gene (ORF 74) and one molecular chaperone (ORF 9) were identified in the genome of vB_EliS-R6L, sharing 50% and 41% amino acid identity, respectively, with the corresponding proteins of the Caulobacter phage Sansa [42]. Most tailed phages achieve lysis via consecutive use two essential proteins, endolysin and holin (which control the length of the infective cycle). Endolysins are phage-encoded enzymes that degrade bacterial peptidoglycan. ORF 74 is predicted to contain one domain: a 176-aa region near the C-terminus that shows homology to proteins of the lysozyme-like superfamily. Although Caulobacter phage Sansa contains a lysis cassette (a holin/anti-holin pair and an endolysin) [42], none of the ORFs identified in phage vB_EliS-R6L exhibit homology to holin proteins. This may be the result of the limited number of holin protein sequences in databases [43, 44]. In addition, ORF 9 is predicted to contain one 49-aa domain homologous to chaperone J, which assists in translation.

Three ORFs are predicted to code for an acyl carrier protein (ORF 32), a nucleoside triphosphate hydrolase (ORF 41), and a phosphoadenosine phosphosulfate reductase (ORF 7). The acyl carrier protein in bacteria is responsible for fatty acid biosynthesis, requiring 4′-phosphopantetheine as a covalently attached cofactor. Acyl carrier protein homologs have also been identified in several other phages [45], though their function remains unclear. ORF 41 of phage vB_EliS-R6L is predicted to include a 292-aa P-loop domain of nucleoside triphosphate hydrolases, which hydrolyze the beta-gamma phosphate bond of a bound nucleoside triphosphate, providing energy for viral metabolism. ORF 7 shows 49% identity to phosphoadenosine phosphosulfate reductases, which have been identified in phages such as Lactobacillus phage AQ113 (GenBank accession no. HE956704) [46], Mycobacterium phage Baka (GenBank accession no. JF937090) [47], and Pseudoalteromonas phage PHS3 (GenBank accession no. KX912252, unpublished). Phosphoadenosine phosphosulfate reductases are thought to be involved in sulfate activation for cysteine biosynthesis. However, no studies have investigated the relationship between the activity of these enzymes and phage metabolism [46, 47].

Based on NCBI BLAST gene annotation results, phage vB_EliS-R6L shares 12 similar ORFs (E-value <10^{– 5}) with the Caulobacter phage Sansa, and another 5 pairs with an E-value <10^{– 3} were found [42] (Table 2). The 12 homologous ORFs include 3 involved in DNA metabolism, 3 structural proteins, 1 methylase, 1 endolysin, 1 nucleoside triphosphate hydrolase, 1 molecular chaperone and 2 proteins of unknown function. However, the identities of the 12 pairs are not high (ranging from 23 to 61%; 34% on average), providing further evidence for the novelty of phage vB_EliS-R6L.

Endolysin protein (ORF 74), major capsid protein (ORF 91), portal protein (ORF 98), and terminase (ORF 100) were chosen for phylogenetic tree construction (Fig. 4). With the exception of the tree based on the terminase protein, in all cases, vB_EliS-R6L clusters with virulent bacteriophages, such as the Caulobacter phage Sansa, roseophages, and Pseudomonas phages. However, the clearly distant phylogenetic relationships with other phages suggest that vB_EliS-R6L is a novel phage. In the terminase-based tree, phage vB_EliS-R6L is located near prophages from Agrobacterium rhizogenes and Rhizobium freirei, agreeing with the BLASTp analysis. In addition, the phage life style predicted by the PHACTS algorithm indicated that it may be a temperate phage. However, no integrase, repressor, or other genes related to the SOS response [48] were identified in the genome of phage vB_EliS-R6L.

Metagenomic analysis indicated that vB_EliS-R6L-like phages are widespread in the examined environmental samples (Fig. 5). Across all metagenomic samples (JRE, POV and GOS), 7138 reads were successfully assigned and detected at rates of 10^{– 9} to 10^{– 7} per amino acid pair in the databases. The greatest matches were found in JRE (1.13 × 10^{– 7} per pair), from which phage vB_EliS-R6L was isolated, followed by POV (1.94 × 10^{– 8} per pair) and GOS (1.89 × 10^{– 8} per pair) coastal samples. This is in agreement with the general distribution of Erythrobacter in the costal environment [1, 2, 49]. Forty-five ORFs were matched to homologs in the databases. The most relative abundant distribution was for ORF 49 (single-stranded DNA-binding protein, with function of DNA replication/repair, 5.28 × 10^{– 9} per pair), ORF 54 (hypothetical protein, 3.05 × 10^{– 9} per pair), ORF 32 (acyl carrier protein, 3.18 × 10^{– 9} per pair) and ORF 100 (terminase, 5.07 × 10^{– 10} per pair). Although the homologs of some ORFs (e.g., 9, 37, 45, 76, 81, 93 and 96) were only found in the JRE virome and/or the POV and GOS coastal samples, the hits for the most matched ORFs covered all three databases. This result suggests that vB_EliS-R6L is a previously unknown phage group that is widely distributed in the marine environment and that it could serve as a good reference for the taxonomic binning of marine viromes in the future.