Genomic Characterization of the Novel Aeromonas hydrophila Phage Ahp1 Suggests the Derivation of a New Subgroup from phiKMV-Like Family

Aeromonas hydrophila is an opportunistic pathogenic bacterium causing diseases in human and fish. The emergence of multidrug-resistant A. hydrophila isolates has been increasing in recent years. In this study, we have isolated a novel virulent podophage of A. hydrophila, designated as Ahp1, from waste water. Ahp1 has a rapid adsorption (96% adsorbed in 2 min), a latent period of 15 min, and a burst size of 112 PFU per infected cell. At least eighteen Ahp1 virion proteins were visualized in SDS-polyacrylamide gel electrophoresis, with a 36-kDa protein being the predicted major capsid protein. Genome analysis of Ahp1 revealed a linear doubled-stranded DNA genome of 42,167 bp with a G + C content of 58.8%. The genome encodes 46 putative open reading frames, 5 putative phage promoters, and 3 transcriptional terminators. Based on high degrees of similarity in overall genome organization and among most of the corresponding ORFs, as well as phylogenetic relatedness among their DNAP, RNAP and major capsid proteins, we propose a new subgroup, designated Ahp1-like subgroup. This subgroup contains Ahp1 and members previously belonging to phiKMV-like subgroup, phiAS7, phi80-18, GAP227, phiR8-01, and ISAO8. Since Ahp1 has a narrow host range, for effective phage therapy, different phages are needed for preparation of cocktails that are capable of killing the heterogeneous A. hydrophila strains.


Introduction
Aeromonas hydrophila, a gram-negative, rod-shaped, non-spore-forming and facultatively anaerobic bacterium, is widely distributed in aquatic environments, drinking water, chlorinated water supply, and a wide range of food [1][2][3]. It causes various human infections such as bacteremia, pneumonia, endocarditis, empyema, arthritis, biliary tract infections, peritonitis, and skin and soft-tissue infections [4][5][6][7][8]. This species also causes diseases in fish, including Aeromonas septicemia, red sore disease and ulcerative infections mainly affecting carp and catfish [9]. The prevalence of A. hydrophila in Taiwan has been increasing; for example, among 129 patients with soft-tissue infections caused by Aeromonas species in Chi Mei Medical Center in Taiwan during 2009-2011, 77 (59.7%) were identified to be infected by A. hydrophila [10]. Although it has been demonstrated that third-and forth-generation cephalosporins and fluoroquinolones were effective against over 80% of the infections caused by Aeromonas species in Taiwan [8,11,12], the increasing rates of antibiotic resistances have raised the concern in treatment of A. hydrophila infections [13][14][15][16].
Bacteriophages are viruses specifically infecting their bacterial hosts and are estimated to be the most widely distributed and diverse entities in the biosphere. It has been suggested that the activities of bacteriophages are driven forces in maintaining genetic diversity amongst the bacterial community [17]. However, despite the importance of A. hydrophila in causing infections, only a few bacteriophages infecting this bacterium have been described, including characterization of myophages Aeh1, Aeh2, PM2, pAh1-C, pAh6-C, and VTCCBPA6, and filamentous phage PM3 [18][19][20][21][22][23][24][25][26][27], and sequencing of the myophage CC2 [28].
In this study, a lytic podophage infecting A. hydrophila was isolated from waste water, designated as Ahp1, and characterized. Analysis of nucleotide and amino acid sequences revealed that the Ahp1 genome has an organization similar to that of the phiKMV-like phages. However, phylogenetic analysis indicated that Ahp1 is most closely related to phages including Aeromonas salmonicida phage phiAS7, Cronobacter sakazakii phage GAP227, Yersinia enterocolitica phages phi80-18, phiR8-01, and ISAO8. Our analysis thus suggests the clustering of a new subgroup containing these phages, which were previously classified within the phiKMVlike subgroup. To our knowledge, this is the first characterized podophage infecting A. hydrophila.

Bacterial strains, phage and growth conditions
Bacterial strains used in this study are listed in Table 1. Luria Bertani (LB) broth and LB agar (Bacto) at 30°C were used to grow bacteria: A. hydrophila at 30°C, Xanthomonas campestris pv. campestris at 28°C, and the other strains at 37°C. Bacterial growth was monitored turbidimetrically by measuring optical density at 600 nm (OD 600 ), in which an OD unit of 1.0 corresponded to 1.8 × 10 8 CFU/ml. Newly isolated A. hydrophila strains were identified by 16S rDNA sequencing using specific primers [29].

Phage isolation and test for host range
The procedures described previously [31] were used for phage isolation, plaque assay and spot test. To test for host range, spot test was performed by including the bacterial strains separately in the double-layered agar plates and 5 μl of phage lysates (10 7 PFU) were spotted onto the bacterial lawns and dried in a laminar flow for 10 min and incubated for 16-18h. The experiments were repeated 3 times.

One-step growth
Cells of A. hydrophila ATCC 7966 (0.6 U of OD 600 ) were harvested by centrifugation and resuspended in 0.9 ml of fresh LB medium (ca. 10 9 CFU/ml). Ahp1 was added at an MOI of 0.0001 and allowed to adsorb for 30 min at 4°C. The mixture was centrifuged at 12,000 × g for 10 min. The pellets containing infected cells were resuspended in 50 ml of LB and incubated at 30°C. Samples were taken at 5-min intervals (up to 35 min), immediately centrifuged at 12,000 × g for 2 min, then the supernatants were diluted in cold LB medium, followed by determining the phage titers.

DNA techniques
The procedures described previously [31] were used for isolation and restriction enzyme digestion of the phage DNA. Pulsed-field gel electrophoresis (PFGE) was performed as described previously [32], by using a CHEFDRIII System (Bio-Rad Laboratories, Hercules, CA, USA) at 6 V/cm with pulse ramps from 3.5 to 4s for 19.5h for the intact genomic DNA at 9°C in 0.5 × Tris-borate-EDTA buffer, pH 8.0. Midrange I PFG Markers (New England Biolabs) were used as molecular size standards.

Electron microscopy
To observe the phage morphology, 10 μl of Ahp1 suspension (1.0 × 10 12 PFU/ml) was dropped onto the surface of a formvar-coated grid (400 mesh copper grids), let stand for 3 min, stained with 2% uranyl-acetate for 30s, and examined in a Hitachi H-7500 transmission electron microscope operated at 80 kV.

Whole genome sequencing and in silico analysis
The genomic DNA of Ahp1 was sequenced by using Next Generation Sequencing system (Illumina Solexa technology) with end paired method. The genome of Ahp1 was scanned for potential open reading frames (ORFs) with ORF Finder (http://www.ncbi.nlm.nih.gov/projects/gorf/), and GeneMarkS software [33]. Annotation was carried out by comparing translated ORFs in BLASTP (http://blast.ncbi.nlm.nih.gov/ Blast.cgi). The presence of transmembrane domains was verified with TMHMM software [34]. Prokaryotic promoter regions were identified by using the BPROM prediction program on the SoftBerry website (http://www.softberry.com/). Potential phage promoter sites were scanned for using PHIRE software [35]. Palindromic repeat regions were identified by FindTerm program on the SoftBerry website. Putative terminators were defined as palindromic sequences followed by a U-rich stretch and a stable secondary structure (ΔG < −10 kcal/mol). ClustalW was used for multiple alignment which was performed with Molecular Evolutionary Genetics Analysis (MEGA) software 6.0.6 aided by manual adjustments [36]. Phylogenetic analysis was also performed with MEGA by using the neighbor-joining method with 1,000 bootstrap replicates.

Nucleotide sequence accession number
The genome sequence of the Aeromonas hydrophila phage Ahp1 has been deposited in Gen-Bank under accession number KT949345.

Isolation and general properties of Ahp1
Thirteen water samples, including those from sewages, wastewater, and aquariums were screened separately by spot tests on the lawns of four A. hydrophila strains, including ATCC 7966 and three clinical isolates (AH19288, AH60114, and AH300206). One phage was isolated and designated as Ahp1.
To obtain high titer lysate, different conditions were tested. Results showed that infecting a culture of A. hydrophila ATCC 7966 (200 ml of LB medium in a 500 ml flask) at exponential phase (0.8 unit of OD 600 ) with an MOI of 0.0001 caused a complete lysis of the culture within 150 min, resulting in the production of approximately 2.5 × 10 10 PFU/ml of phage progeny. Transmission electron microscopy revealed that Ahp1 possessed an icosahedral head (62 nm in diameter) and a short tail (12.5 nm in length). The morphology was thus similar to a typical member of Podoviridae family (Fig 1). Since no podophage of A. hydrophila has been reported, Ahp1 appears to be the first member of Podoviridae infecting this bacterium.
It has been shown that several lipid-containing phages, such as PRD1, PM2, mycobacteriophage D29 and DS6A, are inactivated by chloroform [37][38][39][40]. In this study, about 10 8 PFU of the phage suspension (100 μl) was mixed with chloroform at concentrations from 1 to 5%, shaken for 5 min, followed by incubation of the mixture at room temperature for 25 min. Results showed that at 5%of chloroform, no infective particle was detectable, indicating that Ahp1 is sensitive to chloroform and suggesting that it may contain structural lipids.
The adsorption rate of Ahp1 on A. hydrophila ATCC 7966 is shown in S1 Fig. Approximately 96% of Ahp1 was adsorbed to the host cells within 2 min and no free phages were detectable in the supernatant at 4 min in our assay conditions, indicating a highly efficient adsorption. To understand the growth, one-step growth curve of Ahp1 on A. hydrophila ATCC 7966 was determined. As shown in S2 Fig, Ahp1 exhibited a latent period of about 15 min, and a short growth period of about 25 min. The average burst size was estimated to be 112 PFU per infected cell.

Ahp1 has a narrow host range
To test for host range, lawns of 42 A. hydrophila strains listed in Table 1 were subjected to spot tests with Ahp1. Results showed that only 6 (14.3%, including ATCC 7966, H6, H10, H23, H30 and H32) strains displayed clearing zones, and the others were resistant to Ahp1. All the susceptible A. hydrophila strains, except ATCC 7966, were environmental isolates.
Bacterial strains belonging to 7 species other than A. hydrophila (Table 1), Acinetobacter baumannii, Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus, Vibrio parahaemolyticus, Vibrio harveyi, and Xanthomonas campestris pv. campestris were also subjected to spot test. Results showed that none of these bacteria were susceptible to Ahp1. These results indicated that Ahp1 has a narrow host range and more phages are needed to form a cocktail for future therapeutic use.

The Ahp1 genome is about 42 kb in size
Several restriction endonucleases were tested and the Ahp1 DNA was found to be cut by EcoRV, HindIII, and EcoRI into 2, 5, and 4 fragments, respectively (data not shown). Digestibility by type II restriction enzymes suggests that Ahp1 has a double-stranded DNA genome. To estimate the Ahp1 genome size, DNA from phage particles was subjected to PFGE. As shown in Fig 2, the genome size of Ahp1 was estimated to be 42 kb, similar to the value estimated by summing up the fragment sizes obtained from restriction digests.

The Ahp1 virion consists of at least 18 proteins
To analyze the virion proteins, purified Ahp1 phage particles were subjected to precast 8-16% gradient polyacrylamide gel (Bio-Rad Laboratories, Hercules, CA, USA, CAT#456-1103) separation following the procedures described previously [31]. As shown in Fig 3, at least 18 protein bands were visualized upon staining the gel with Coomassie brilliant blue. The band with an apparent molecular mass of 36 kDa was the most abundant protein, suggesting that it is the major coat protein of Ahp1.

Sequencing of the Ahp1 genome
The genomic DNA of Ahp1 was extracted from CsCl-purified particles and sequenced by next generation sequencing and primer walking. Results showed that the Ahp1 genome consisted of 42,167 bp, in good agreement with that estimated in PFGE (Fig 2). It had a G + C content of 58.8%, slightly lower than that of the host genome (61.5%). Open reading frame (orfs) prediction suggested 46 orfs, which occupied 92.4% of the genome. These orfs, all located on the same strand, were transcribed in the same direction (Fig 4). All orfs began with AUG, except orf18 and 22 which used GUG and orf35 and 42 which used UUG as the initiation codon (Table 2). Five bacterial promoters (red bent arrows), five phage promoters (black bent arrows), and three factor-independent terminators (black knobs) were predicted (Table 3, Fig 4).

Genome organization
Members of phiKMV-like phages include Pseudomonas aeruginosa phage phiKMV and at least 13 related phages as available from data base. The genome of phiKMV and the related phages are divided into three classes: class I contains early genes, class II encodes proteins that participate in DNA metabolism, and class III contains genes required for virion structure, host lysis, and phage assembly. As shown in Fig 4, organization of the Ahp1 genome was similar to that of phiKMV and the related phages.
It was also noted that in spite of the high degrees of similarity being shared between the homologous ORFs, organization of the phiAS7 genome was different from that of the other similar phages, with its ORF1-ORF20 and ORF22-ORF51 being inverted (S3 Fig). However, when the phiAS7 genome was redrawn by inverting both ORF1-ORF20 and ORF22-ORF51 regions, its gene order became largely the same as that of the other four phages (Fig 4). Our finding indicates that procedures for assembly of the phiAS7 contigs may need to be revisited.

Gene products of Ahp1
Protein products encoded by the Ahp1 class I orfs were either hypothetical or sharing no similarity to those in database (Table 2), similar to the cases in phiKMV-like phages. Roucourt et al. suggest, through yeast two-hybrid experiments [41], that class I genes of phiKMV although most being hypothetical have roles in bacteriophage-host interaction. However, it would be difficult to assign common functions for these Ahp1 ORFs, since they are highly varied in amino acid sequences.
Orf14, the first gene in class II region encoded a potential DnaG-like primase with a PHA02031 (N-terminal) domain (aa 13-69) conserved among phage DNA primase. The protein product of orf15 has domains similar to those of a DNA helicase, which unwinds the DNA duplex during replication initiation: one at aa 188-195 (AXXXXGKT) similar to the phosphate-binding loop (GXXXXGK-T/S) [42] and one at aa 293-298 (IVVFDM) similar to a Mg 2+ -binding site (hhhhDE, where h is a hydrophobic residue) [43]; these domains are also known as Walker A and B motifs, respectively. Orf16 encoded a hypothetical protein, while protein encoded by orf17 found no similar proteins in database. ORF18 was identified as a potential ATP-dependent DNA ligase, with a DNA_ligase_A_M domain (aa 1-205, pfam01068) which included i) an active site motif (aa 6-11, KRDEFR corresponding to K-Y/ A-D-G-X-R) consistently present in ATP-dependent DNA ligases [44] and ii) critical residues (K203 and K205 corresponding to K238 and K240 of phage T7 ligase, respectively) responsible for catalysis and nick recognition [45]. ORF19 was identified as a potential nucleotidyltransferase, containing a NT_ClassII_CCAase domain (aa 23-55, cd05398), which is a CCA-adding enzyme, adding the sequence cytidine(C)-cytidine-adenosine (A) one nucleotide at a time to the 3' end of tRNA in a template-independent reaction [46]. ORF20, a potential DNA polymerase (DNAP), possessed a DNA_polA domain (aa 377-780, pfam00476). Orf21 encoded a hypothetical protein. ORF22, a potential 5'-3' exonuclease, possessed an active site of PIN_53EXO domain (aa 90-156, cd09859) that is conserved in bacterial DNA polymerase   [47]; within the active site was a set of conserved acidic residues (E130, D132, D133, D152, and D154) similar to that essential for binding divalent metal ions required for nuclease activity in Taq DNA polymerase (DNAP) [48]. Orf23 encoded a hypothetical protein. Orf24, encoding a potential endonuclease, contained an Endonuclease_7 domain (aa 23-93, pfam02945). Orf25 encoded a hypothetical protein.
The last gene in class II region was orf28 encoding a potential DNA-dependent RNA polymerase (RNAP) as mentioned above. Alignment of the Ahp1 DNA-dependent RNAP with that of other phages including T7, phiKMV, phi80-18, GAP227 and phiAS7 is shown in Fig 5.   [49][50][51] are conserved in bacteriophage-type RNAP. The invariant D537 (Palm domain) and D812 (Palm domain) in T7 RNAP, the catalytically most critical residues that directly involved in phosphodiester bond formation by coordinating Mg 2+ ions [51,52], are also conserved in Ahp1 (D514 and D754). The conserved K631 and Y639 in the Finger domain of T7 RNAP [50,51], that are important catalytic residues of the active site, were also found in Ahp1 RNAP (K584 and Y592). However, the AT-rich recognition loop and the specificity loop that interact with T7 promoters [53,54] were not conserved in the RNAP of Ahp1 and phiKMV-like phages (Fig 5). Class III region contained genes potentially involved in virion structure and assembly, except orfs 29,30,31,37,38,44, and 46 that encoded hypothetical proteins. Orf32, encoded a potential head portal protein, containing a Head-tail_con domain (aa 2-453, pfam12236) found in bacteria and phages. ORF33 showed 31% similarity to the scaffolding protein of Burkholderia phage JG068 (YP_008853872.1). Amino acid position 11-337 of ORF34 exhibited similarities to domain PHA02004 conserved in major capsid proteins of Pseudomonas phage Bf7 (YP_005098192.1), and Burkholderia phage JG068 (YP_008853873.1). In addition, ORF34 had a calculated molecular mass of 36.9 kDa, similar to that observed for the most abundant band, thought to be the major coat protein, in the SDS-PAGE separation of the Ahp1 virion proteins (Fig 3, Table 2). Amino acid position 2-183 of ORF35 was identified as a potential tail tube protein, which showed similarity to domain PHA00428 conserved in tail tube protein A of Pseudomonas phage Bf7 (YP_005098193.1), and Burkholderia phage JG068 (YP_008853874.1). ORF36 was identified as a potential tail tube protein B which was similar to that of Ralstonia phage RSB1 (YP_002213723.1), and Burkholderia phage JG068 (YP_008853875.1). Notably, such a gene order (head portal protein-scaffolding protein-major   Boldface residues are highly conserved amino acids within known RNAP. Symbols: "*", identical residues in all sequences, ":", highly conserved residues, ".", weakly conserved residues. doi:10.1371/journal.pone.0162060.g005 capsid protein-tail tubular protein A and B) is also observed in all phiKMV-like phages [55]. ORF39, similar to lytic transglycosylase which has a LT_GEWL domain (cd00254) that contains an invariant Glu (E30) [56] for catalysis and a conserved Tyr (Y105) [57], had the LT_GEWL domain being located in aa 30-106 which contained E28 and Y103. ORF40, containing a region (aa 19-127) resembling the Phage_T7_tail domain (pfam03906), was a potential tail fiber protein. Orf41 was identified as a putative holin, because it was small in size with a transmembrane domain, located near the predicted endolysin gene (orf45) [58]; in addition, no other possible protein similar to holin in database was found. ORF42 and ORF43 were predicted to be DNA maturase A and B, respectively, based on their similarity to that of phiAS7 (YP_007007800.1) and Burkholderia phage JG068 (YP_008853883.1). ORF45, containing a region (aa 30-170) similar to the endolysin_autolysin domain (cd00737) was annotated to be the endolysin of Ahp1. Many phages of Gram-negative bacteria encode internally overlapping Rz/Rz1 proteins, with the genes situating immediately downstream of the endolysin gene, to enhance bacterial lysis when the outer membrane is stabilized by divalent cations [59,60]. However, no similar proteins were found in Ahp1 and the closely related phiKMV-like phages, suggesting that the Rz/Rz1 proteins are not used to enhance host lysis and alternative mechanisms may have evolved to enhance bacterial lysis.

Phylogenetic relatedness of Ahp1
As mentioned above, our data of ORF comparison suggested that phiKMV-like phages can be divided into at least two subgroups. To understand the relatedness between Ahp1 and the phiKMV-like phages, phylogenetic analysis was performed using DNAP, RNAP, and major capsid protein of Ahp1 (ORF20, ORF28, and ORF34, respectively) as the sample proteins. The proteins from Autographivirinae subfamily phages including T7-like phages, SP6-like phages, and phiKMV-like phages, each of which encodes its own single-subunit RNA polymerase [61] were also included. As shown in Fig 6, the proteins from Ahp1 were each clustered together with that of phiAS7, phi80-18, GAP227, phiR8-01, and ISAO8 and formed a clade distinct from those of T7-like, SP6-like, and the other phiKMV-like phages, which includes phiKMV and 7 related phages. Taken these together with the results of genomic comparison, we propose to classify Ahp1, phiAS7, phi80-18, GAP227, phiR8-01, and ISAO8 into a new subgroup, designated as Ahp1-like subgroup, within the Autographivirinae subfamily.

Conclusions
In this study, a novel podophage of A. hydrophila, designated Ahp1, has been isolated and characterized. Phylogenetic relatedness among DNAP, RNAP, and major capsid protein suggest that a new subgroup, designated Ahp1-like subgroup, has formed within the Autographivirinae, in addition to T7-like, SP6-like, and phiKMV-like subgroups. Since Ahp1 has a narrow host range, for effective phage therapy, different phages are needed for preparation of effective cocktails that are capable of killing the heterogeneous A. hydrophila strains.