Novel Temperate Phages of Salmonella enterica subsp. salamae and subsp. diarizonae and Their Activity against Pathogenic S. enterica subsp. enterica Isolates

Forty strains of Salmonella enterica (S. enterica) subspecies salamae (II), arizonae (IIIa), diarizonae (IIIb), and houtenae (IV) were isolated from human or environmental samples and tested for bacteriophage production. Production of bacteriophages was observed in 15 S. enterica strains (37.5%) belonging to either the subspecies salamae (8 strains) or diarizonae (7 strains). Activity of phages was tested against 52 pathogenic S. enterica subsp. enterica isolates and showed that phages produced by subsp. salamae had broader activity against pathogenic salmonellae compared to phages from the subsp. diarizonae. All 15 phages were analyzed using PCR amplification of phage-specific regions and 9 different amplification profiles were identified. Five phages (SEN1, SEN4, SEN5, SEN22, and SEN34) were completely sequenced and classified as temperate phages. Phages SEN4 and SEN5 were genetically identical, thus representing a single phage type (i.e. SEN4/5). SEN1 and SEN4/5 fit into the group of P2-like phages, while the SEN22 phage showed sequence relatedness to P22-like phages. Interestingly, while phage SEN34 was genetically distantly related to Lambda-like phages (Siphoviridae), it had the morphology of the Myoviridae family. Based on sequence analysis and electron microscopy, phages SEN1 and SEN4/5 were members of the Myoviridae family and phage SEN22 belonged to the Podoviridae family.


Introduction
At present, the Salmonella genus contains two species: S. enterica and S. bongori (V). S. enterica is further subdivided into six subspecies: enterica (I), salamae (II), arizonae (IIIa), diarizonae (IIIb), houtenae (IV), and indica (VI) [1][2][3]. Salmonelloses (infections transferred between humans and animals, mainly by contaminated food) represent an important, global, public a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Identification of lysogenic strains Forty S. enterica strains were tested for bacteriophage production using a cross test method where each strain was tested as a possible producer as well as a possible indicator strain. In addition, four standard E. coli indicators (K12-Row, C6 (φ), B1, and P400), which were available in our laboratory strain collection [20][21], were used to detect phage producers. Briefly, each Salmonella strain was inoculated from a fresh TY broth culture into an agar base layer, 1.5% (w/v), using a sterile needle and then cultivated at 37˚C for 48 hr. The resulting macrocolony was killed using chloroform vapors (30 min) and the plate was overlaid with a top layer of 0.7% (w/v) agar enriched with a suspension of 10 7 indicator bacteria. Phage production was evaluated after overnight cultivation at 37˚C.

Determination of phage titer and inducibility of phages
Phage producers were inoculated into TY broth and incubated overnight. The fresh culture was diluted hundredfold with TY broth and cultivated for an additional 7 hr (37˚C, 200 rpm). Subsequently, the culture was centrifuged at 4000 × g for 15 min to remove bacteria. The supernatant was transferred to a sterile tube and stored at 4˚C with a few drops of chloroform to ensure sterility. To determine the concentration of phage particles, the sterile supernatant was serially diluted 10 times; each dilution was added to a suspension of phage-susceptible bacteria (10 7 bacteria in 3 ml of the melted top layer of 0.7% agar) and spread on agar plates. After overnight cultivation, plaques were counted and the resulting bacteriophage titer was expressed as plaque forming units (PFU)/ml.
In parallel, induction of phage production was tested using the same protocol with a slight modification, which included the addition of mitomycin C (final concentration 0.1 μg/ml) to the culture of lysogenic strains during the exponential growth phase (i.e., after 4 hr of cultivation at 37˚C, 200 rpm); cultivation continued for an additional 3 hr. Phage production was determined using a plaque counting technique. A comparison of phage titers counted from cultures with and without addition of mitomycin C indicated inducibility of phage production.

Activity of phages against clinically important Salmonella isolates
Using the identified phage producers, we examined the activity of phages against a set of 52 clinically important S. enterica subsp. enterica isolates using the cross-test method described above. Clinical isolates belonged to common pathogenic serovars, i.e. Enteritidis (22 isolates), Typhimurium (13 isolates), Choleraesuis (4 isolates), Indiana (3 isolates), Derby (2 isolates), Mikawashima (2 isolates), Agona (1 isolate), Chester (1 isolate), Infantis (1 isolate), Java (1 isolate), Ohio (1 isolate), and San Diego (1 isolate). All these pathogenic strains were collected by The National Reference Laboratory for Salmonella, National Institute of Public Health (NIPH), Prague, Czech Republic, from patients living in the Czech Republic. The characteristics of the pathogenic strains are shown in S1 Table. Analysis of PCR amplification profiles of phages DNA from all phages was isolated using a QIAGEN Lambda Midi Kit (Qiagen, Hilden, Germany) per the protocol recommended by the manufacturer. Total bacteriophage DNA was digested using EcoRI enzyme (New England Biolabs, Ipswich, USA), cloned into the pUC19 cloning vector (New England Biolabs); the resulting clones were sequenced. The resulting sequences were compared with the GenBank database using BLAST to identify similarities to other phages. Subsequently, 12 DNA regions representing different phage genes were selected for primer design. The DNA from all phages was used as a template for PCR amplification (Table 1) and up to 2 genetic regions originating from the same phage were used for further screening. A list of genetic loci and the corresponding primer sequences are shown in Table 1

Preparation of phage stock
Bacteriophage stock was prepared from a single selected plaque [22] that was resuspended in 100 μl of TY broth containing 0.01M CaCl 2 . This phage suspension was added to 0.7% TY agar containing 10 7 phage-susceptible bacteria (i.e., they did not produce their own bacteriophages) and spread on agar plate. After overnight cultivation, the bacteriophage-containing top layer of TY agar, with confluent lysis, was collected into a sterile tube and resuspended in 3 ml of SM buffer (100 mM NaCl, 8 mM MgSO 4 Ã 7H 2 O, 50 mM Tris-Cl, 0.01% (w/v) gelatine) and 0.5 ml of chloroform. The tube was incubated at 37˚C for 20 min and the suspension was centrifuged at 4000 × g for 15 min to remove bacteria and agar residue. The supernatant, containing the phage particles, was transferred to a sterile tube and few drops of chloroform were added.

Isolation of bacteriophage DNA for whole genome sequencing
The bacterial stocks originated from single plaques multiplied in indicator strains (i.e., did not produce their own phages) were used for propagation of phages and isolation of phage DNA (see above). Phages were propagated per previously published protocols [22]. Briefly, 200 ml of TY broth was inoculated with 2 ml of fresh phage indicator culture and cultivated for 4.5 hr Hypothetical protein (37˚C, 250 rpm). Thereafter, 200 μl of 0.02M CaCl 2 and phage particles (% 10 8 ) from our phage stock were added and cultivation continued for an additional 4.5 hr (37˚C, 250 rpm). Finally, the bacterial culture was lysed using chloroform (10 min, 37˚C, 250 rpm). Bacterial nucleic acids in lysates were degraded using RNase A and DNase I (final conc. 1 μg/ml, 45 min, RT) and phage particles were precipitated using NaCl (1M, 1 hr, 4˚C) and PEG 8000 (10%, 16 hr, 4˚C). After centrifugation (11,000 × g, 11 min, 4˚C), the pellet was resuspended in 5 ml of SM buffer and phage particles were purified using isopycnic ultracentrifugation in a CsCl gradient (87,000 × g, 2 hr, 4˚C). The collected bacteriophage suspension was dialyzed with dialysis buffer (10 mM NaCl, 50 mM Tris-Cl, 10 mM MgCl 2 ) for 2 ×1 hr. The dialyzed phage suspension was treated with EDTA (final conc. 20 mM), proteinase K (final conc. 50 μg/ ml) and SDS (final conc. 0.5%) for 1 hr at 56˚C. Phage DNA was isolated by phenol-chloroform, precipitated by isopropanol, and finally diluted in 50 μl of TE buffer.

Electron microscopy
Phage samples for electron microscopy were prepared from phage stocks (originated from a single plaque) according to previously published protocols [22]. Phage suspensions were subsequently precipitated with polyethylene glycol (PEG) and purified using centrifugation in a CsCl gradient [22]. Bacteriophage particles were visualized using a negative staining method. Briefly, phage suspensions (5 μl) were placed on glow discharge activated carbon-coated grids [28]. After an adsorption period (30 s), the unabsorbed liquid material was removed using filter paper. The electron-microscopic grids were then stained with 10 μl of 1% phosphotungstic acid, 2% uranyl acetate, or 2% ammonium molybdate for 10-30 s; excess staining solution was removed using filter paper. Samples were viewed using a MORGAGNI 268D (FEI, Hillsboro, OR, USA) or a Philips CM 100 transmission electron microscope (FEI).

Characterization of non-pathogenic Salmonella enterica strains and identification of lysogenic strains
All non-pathogenic S. enterica strains used in this study (n = 40) were classified into subspecies based on biochemical markers and serotyped using the White-Kaufmann-LeMinor scheme. A total of 16 strains belonged to S. enterica subsp. salamae, 5 strains to subsp. arizonae, 15 strains to subsp. diarizonae, and 4 strains to subsp. houtenae. Strains mostly came from human feces or other human material, and from waste water sludge. In 7 cases, the origin of the strain was unknown (S1 Table).

Analysis of phage activity against pathogenic S. enterica clinical isolates
Activity of all 15 phages from identified phage producers was tested against 52 pathogenic S. enterica subsp. enterica isolates belonging to common pathogenic serovars. The activity spectra of phages are shown in Fig 1. Two strains, i.e. Sen1 and Sen8, produced phages, which lysed the majority of pathogenic isolates, while phages produced by Sen4, Sen5, Sen6, Sen16, Sen34, and Sen35 strains inhibited less than 20% of isolates. Moreover, all pathogenic Salmonella isolates were resistant to phages produced by strains Sen2, Sen14, Sen22, Sen23, Sen24, Sen30, and Sen31. In general, phages from subsp. salamae showed broader activity against pathogenic salmonellae compared to phages from subsp. diarizonae.

Analysis of phage PCR amplification profiles
Variability of identified phages (denoted as "SEN" throughout the manuscript) was analyzed using PCR amplification of several loci obtained during construction of small libraries of phages. Altogether, 12 short fragments (200-300 bp) from different genes were tested in all 15 phages (Fig 2). Phages SEN6, SEN14, SEN16, and SEN30 were negative for all PCR detected DNA regions. Three pairs of phages (i.e., SEN1 and SEN31, SEN4 and SEN5, and SEN22 and SEN24) had identical amplification profiles. The five remaining phages showed a unique spectra of PCR positive reactions. Altogether, nine different amplification profiles were identified. Finally, phages SEN1, SEN4, SEN5, SEN22, and SEN34 were selected for further analysis based on: i) variability in PCR profiles, ii) differences in the activity against pathogenic salmonellae, iii) classification of phage producers into different subspecies, and iv) quality of isolated DNA.

Morphology of phage particles
Phages SEN1, SEN4, and SEN5 from producers belonging to subsp. salamae, and phages SEN22 and SEN34 from producers belonging to subsp. diarizonae were characterized using electron microscopy. A phage suspension obtained from a single plaque was negatively stained and examined with a transmission electron microscope to determine phage morphology. The morphology of phage particles is summarized in Fig 3. Phages SEN1, SEN4, SEN5, and SEN34, shared the same morphology, i.e. a head with icosahedral symmetry and a long contractile tail. Diameters of icosahedral capsids ranged from 54.0-58.3 nm and the length of the tail, including the contractile sheath, ranged from 107.3-148.9 nm. In contrast, phage SEN22 had an icosahedral head (57.8 nm in width; 55 nm in length) with very short noncontractile tail (18.4 nm in width; 15.0 nm in length). Based on previous classifications [29], the first morphological group corresponds to the Myoviridae family, and the second morphological group corresponds to the Podoviridae family.
The main characteristics of phage genomes are shown in Table 3. Genome length of the sequenced phages ranged from 29.7-41.3 kb with G+C content ranging from 47.8-53.4%. Genome lengths were correlated with genome sizes, which were estimated from PFGE (data not shown), indicating that complete genome sequences had been obtained. While the cos site, identical to predicted cohesive ends of phage Psp3 (ggcgtggcggggaaagcat), was found for SEN1, sticky ends of phages SEN4 and SEN5 (ggcgaggcgggggaacgag) were more similar to the cos site of phage P2. In addition, a pac sequence (gaagatttatctgaagtcgtta) identical to that of P22 phage was identified for phage SEN22. Due to unusual SEN34 genome (see below), the packaging-responsible sequences were not identified in the SEN34 genome. All sequenced phages encoded integrase, a characteristic of temperate phages. A BLASTN analysis of whole genomes (Table 3) revealed substantial similarity (ANI: 97.11%) between phage SEN1 and phage PsP3 (a phage belonging to the P2-like group) (Gen-Bank No. AY135486). The genome of phage SEN1 was also very similar (ANI: 81.69%) to the prototype phage P2 (GenBank No. KC618326), especially in the morphogenesis region. Besides the genetic variability found in the region responsible for replication, the lysogenic conversion locus was not present in the SEN1 genome and the region encoding tail fibers differed between SEN1 and P2 phages (Fig 4A). In fact, the tail fiber gene of SEN1 (gp19) was related to another P2-like phage, Salmonella phage RE-2010 (GenBank No. HM770079).
Whole genome sequencing of phages SEN4 and SEN5 revealed that these phages were identical (Table 3), thus representing a single phage type, denoted as phage SEN4/5. Whole genome analysis of SEN4/5, based on BLASTN, found similarities to the prophage sequence within the genome of Enterobacter cloacae complex (GenBank No. CP012162), but not to any known phages (Table 3); however, BLASTP analysis of predicted SEN4/5 proteins showed a distant similarity to several P2-like phages (Fig 4A and S2 Table). Nonetheless, the average nucleotide identity with the P2 phage was calculated to be 0.00%. Thus, SEN4/5 appears to belong to the P2-like phages, but it is clearly different from previously known members.
Both SEN22 and SEN34 phages, isolated from subsp. diarizonae, showed larger genomes compared to subsp. salamae phages ( Table 3). The genome of phage SEN22 was found to be related to several P22-like phages having the greatest similarity (ANI: 94.11%) to Salmonella phage ε34 (GenBank No. NC_011976). The SEN22 genome region encoding the head of the virion was more similar to prototype phage P22 than the region encoding the virion tail and the region responsible for regulation and replication (Fig 4B). In addition, four insertions of mobile elements were identified in the SEN22 genome, including a unique insertion of a homing endonuclease gene between the terminase gene and the portal protein gene.
BLASTN analysis of the SEN34 genome sequence revealed a low degree of similarity to the prophage sequence within the genome of S. enterica subsp. enterica serovar Newport str. CVM 21550 (GenBank No. CP010283), but not to any known phages (Table 3), while the BLASTP analysis of predicted proteins showed that there was a distant similarity between the SEN34 virion assembly genes and two uncharacterized phages including the Burkholderia phage Bups phi1 and Acinetobacter phage Ab105-1phi (S2 Table). Additionally, the region responsible for replication and regulation of SEN34 was similar to several Lambda-like phages (Fig 4B and S2  Table). When all these observations are taken together, SEN34 appears to be a new phage type (ANI with Lambda phage: 0.00%).

Discussion
To date, almost 200 phage types have been identified in the genus Salmonella [30] and if prophages are included, the number is greater than 9,000 [18]. Prophages appear to be very common in Salmonella, as shown by production of 136 functional phages from 173 S. enterica subsp. enterica (serovar Typhimurium) isolates [31]. It has been calculated that there are, on average, 2.8 known prophages present per Salmonella genome [18]. While phages and prophages of subsp. enterica have been intensively studied (reviewed in [18,30,32]), information on phages in other Salmonella subspecies is scarce. In fact, putative prophage sequences were only recently identified in the whole genome sequences of two strains belonging to non-enterica subspecies, i.e. in subsp. arizonae (GenBank: CP000880.1; prophage Sari1; [33]), and subsp. salamae (GenBank: ATFA01000000; [34]). This study focused on phage production within strains S. enterica subspecies salamae, arizonae, diarizonae, and houtenae. While fifteen phage producers were identified in subspecies salamae and diarizonae, no producers were identified in subspecies arizonae and houtenae; however, it is unclear if this difference is due to the relatively low number of investigated strains in the second group (9 out of 40) or due to other factors. Out of 15 identified phages, SEN1, SEN4/5, SEN22, and SEN34 were further characterized. Bacteriophages SEN1, SEN4/5, and SEN34 belonged to the Myoviridae family, while SEN22 showed morphology typical of the Podoviridae family. These findings are in accordance with the fact that the majority of known phages belongs to the Myoviridae family [35].
Analysis of phage genomes revealed that these phages represent novel types. All sequenced phages encoded integrase, a characteristic of temperate phages. While phages SEN1 and SEN22 showed close sequence relatedness to prototype phages P2 and P22, respectively, phages SEN4/5 and SEN34 were quite different from known phages. Our findings are in accordance with a study by Casjens and Grose [18], where SEN4/5 genomic sequences formed a new, well defined, subcluster within the P2-like cluster. Phage SEN34 was similar to several disparate Lambda-like phages, but did not belong to any of the 17 known lambda clusters described by Grose and Casjens [17]. Thus, phage SEN34 is a representative of a new lambda cluster. This is in accordance with a recent comprehensive study of enterobacterial prophages [18], where the authors analyzed the available phage genomes from GenBank database and defined a new lambda phage cluster ("temperate 26") with prototype phage SEN34. Although phage SEN34 is genetically related to Lambda-like phages, morphologically it is a member of the Myoviridae family (in contrast to Lambda phages, which belong to the Siphoviridae family). A similar discrepancy between genomic relatedness and morphology classification has also been shown for the LP65 phage, which had a Myoviridae morphology, however, it had a genome organization similar to the Lambda phage [36].
The additional 10 unsequenced phages identified in this study, likely represent distinct phage types as revealed by (i) differences in amplification profiles of various genomic regions, (ii) analyses of the sources of Salmonella strains, (iii) the spectra of indicator strains, and (iv) inducibility with mitomycin C.
Genomes of temperate phages sequenced in our study showed genome mosaicism, where several parts of the phage genomes were related to several different phages. This is in agreement with previous comparative analyses showing that the morphogenesis region is relatively conserved, while genome mosaicism is prevalent in the early regions of phage genomes [17,[37][38][39][40].
Besides common mosaicism in the early region, genes encoding proteins interacting with the host showed increased genetic diversity. The DNA encoding receptor-binding tailspike protein domain is among the most highly exchanged parts of the tailed phages. It is evolutionarily useful for phages to acquire new receptor specificities by swapping this domain through horizontal gene transfer [33]. While sequenced phages produced by the subsp. salamae showed a similarity to the region encoding tail fibers in the P2-like Salmonella phage RE-2010, no sequence similarity was found for the SEN22 and SEN34 phages.
Phages produced by strains belonging to subsp. salamae showed a broader activity spectra against pathogenic isolates of subsp. enterica compared to phages from subsp. diarizonae. This is in accordance with results from a phylogenetic study of Salmonella subspecies [41], which showed a close relationship between subsp. salamae and subsp. enterica. The observed broad activity of phages SEN1 and SEN8 against pathogenic salmonellae opens up the possibility of finding therapeutic applications for these phages, however, the presence of lysogenic cycles in these phages makes such therapeutic applications rather potential.

Conclusions
In this study, we determined phage production in S. enterica subspecies salamae, arizonae, diarizonae, and houtenae. Out of 15 identified phage producers, five complete phage genomes were determined and four different temperate phages were identified. Phages SEN1 and SEN4/5 clustered with P2-like phages, while phage SEN22 showed sequence relatedness to P22-like phages. Phage SEN34 was distantly related to Lambda-like phages (Siphoviridae), but had a morphology that was characteristic of Myoviridae.
Supporting Information S1