Identification and Molecular Characterisation of a Novel Mu-Like Bacteriophage, SfMu, of Shigella flexneri

S. flexneri is the leading cause of bacillary dysentery in the developing countries. Several temperate phages originating from this host have been characterised. However, all S. flexneri phages known to date are lambdoid phages, which have the ability to confer the O-antigen modification of their host. In this study, we report the isolation and characterisation of a novel Mu-like phage from a serotype 4a strain of S. flexneri. The genome of phage SfMu is composed of 37,146 bp and is predicted to contain 55 open reading frames (orfs). Comparative genome analysis of phage SfMu with Mu and other Mu-like phages revealed that SfMu is closely related to phage Mu, sharing >90% identity with majority of its proteins. Moreover, investigation of phage SfMu receptor on the surface of the host cell revealed that the O-antigen of the host serves as the receptor for the adsorption of phage SfMu. This study also demonstrates pervasiveness of SfMu phage in S. flexneri, by identifying complete SfMu prophage strains of serotype X and Y, and remnants of SfMu in strains belonging to 4 other serotypes, thereby indicating that transposable phages in S. flexneri are not uncommon. The findings of this study contribute an advance in our current knowledge of S. flexneri phages and will also play a key role in understanding the evolution of S. flexneri.


Introduction
A large number of mobile DNA elements play an important role in the evolution of bacterial genomes. These elements can translocate from site to site within and between bacterial genomes, generating insertional mutations and various types of genome rearrangements, thereby creating an important source of genetic variation. These transposing elements include IS element, composite transposons and some bacteriophages which propagate via transposition [1,2].
Enterobacteria phage Mu of Escherichia coli is the best studied transposable phage. It is known to randomly integrate into the host chromosome during both lytic and lysogenic developments often leading to mutations in the host [3,4]. In addition, phage Mu carries heterogeneous host DNA sequences on the ends of the packaged genome which results in transduction of variable amounts of host DNA [5]. These properties make bacteriophage Mu an important tool for genetic research.
Genus Shigella belongs to the family Enterobacteriaceae, and is the causative agent of shigellosis or bacterial dysentery. It is responsible for more than 165 million cases annually, of which 1.1 million results in death. Among the four known species of Shigella, S. flexneri is the primary cause of endemic shigellosis prevalent in the developing countries, and is the most frequently isolated species world-wide [15]. Based on the structure of the lipopolysaccharide O-antigen, S. flexneri is divided into 19 serotypes [16]. To date, seven temperate bacteriophages originating from various serotypes of S. flexneri have been isolated. All of these carry the O-antigen modification genes and mediate serotype conversion by integration into the host chromosome. Moreover, based on their genome organization, they are known to be members of the lambdoid family of phages [17][18][19][20][21][22].
In this study, we report the isolation and characterisation of a novel Mu-like bacteriophage, SfMu, from a wild type serotype 4a strain of S. flexneri. Comparative genome analysis revealed high level of proteome correlation between bacteriophage SfMu and Mu. Analysis of SfMu cell wall receptor indicates that the phage SfMu uses O-antigen of the LPS as its receptor. Additionally, this study also identified cryptic and complete SfMu prophage in various serotypes of S. flexneri, suggesting that Mu-like phages are not uncommon in S. flexneri.

Bacterial strains, bacteriophage and media
Bacteriophage SfMu was induced from the serotype 4a strain of S. flexneri, SFL2241, using UV irradiation protocol described by Adam et al [23]. Bacteriophage stocks were prepared by propagating the induced lysate on serotype Y strain (SFL124) [24], and precipitating phage using the polyethylene glycol, as described in Sambrook et al [25].
Luria-Bertani (LB) broth or LB agar was used for culturing of bacteria and NZCYM was employed for propagation of the phage.

Electron microscopy
Phage particles were allowed to absorb onto the carbon-coated copper grid before they were negatively stained with 2% phosphotungstenic acid (pH 7.0) and examined under a Hitachi H7000 transmission electron microscope.

Plaque assay modified with Antibiotic (PAMA)
Phage SfMu plaque assays were performed using the PAMA technique described by Santos et al [26]. Briefly, 0.1 ml of phage lysate was mixed with 0.1 ml of stationary phase culture of the host bacterium, and incubated at 37°C for 20 min. The suspension was then mixed with 3 ml of LB top agar containing 10% glycerol and 0.5 mg/l ampicillin, and poured onto a LB agar plate supplemented with 10% glycerol and 0.5 mg/l ampicillin. Plates were allowed to dry and then incubated overnight at 37°C.

DNA techniques
Bacterial genomic DNA was isolated using GE Healthcare genomic DNA isolation kit (GE Healthcare), according to the manufacturer's instructions. Restriction enzymes were obtained from New England Biolabs (NEB) and used according to the manufacturer's directions. PCR amplification was performed using the PfuUltra II Fusion HS DNA Polymerase (Stratagene). When necessary, the PCR products were purified by using the Wizard SV Gel and PCR Clean Up System (Promega). Sequencing of the purified products was performed using Big Dye Terminator v3.1 Cycle Sequencing Kit and were run at the Biomedical Resources Facility, John Curtin School of Medical Research, Australian National University.
Multiplex PCR amplifications were performed in 20 μl volumes with 0.125 μM of each primer, 2 Units Taq DNA polymerase, 2.25 mM MgCl 2 , 1 X Taq buffer, 50 ng DNA, and 200 μM of dNTP mix. DNA amplifications were performed under the following conditions: initial denaturation at 95°C for 2 min, followed by 35 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 3 min. The final extension was carried out at 72°C for 5 min. PCR products were analysed by electrophoresis using 0.7% agarose gels.
For Southern hybridization, 1 μg of genomic DNA was digested with EcoRV restriction enzyme and subjected to electrophoresis on a 0.7% agarose gel. Capillary blotting was then used to transfer DNA onto a positively charged nylon membrane [25]. The membrane was hybridized using EcoRV digested SfMu phage DNA probe prepared using PCR DIG labeling mix (Roche). Pre-hybridization, hybridization, washes and detection (CDP-star) were performed as recommended by Roche.

Phage inactivation
LPS was isolated from dried bacterial cells using the phenol water extraction method [27]. Phage inactivation assay was performed using the protocol described by Sandulache et al [28] with the following modifications. 0.1 ml of LPS (5 to 100 μg) was mixed with 0.1 ml of phage lysate (1x10 4 PFU/ml in LB) and incubated at 37°C for 1 hr. 0.2 ml of stationary phase culture of the indicator strain (SFL124) was then added to the mixture. Following a further incubation for 20 min at 37°C, the mixture was plated using the PAMA technique described above. Plaques were counted after incubating the plates overnight at 37°C.

Phage genome sequencing, annotation and analysis
Complete sequence of phage SfMu was determined by sequencing the genome of its host strain SFL2241 using 250 bp paired end, Miseq, Illumina sequencing, at Ramaciotti Centre, University of New South Wales. The reads generated were de-novo assembled into contigs using Velvet [29]. The gap between the contigs containing phage SfMu genome was closed by amplifying the desired region and sequencing the purified PCR product as described above.
The open reading frames were identified using CLC Main Workbench (Ver 5.5.1, CLC Bio) and NCBI ORF finder program. All predicted orfs were corroborated by the inspection of the Shine-Dalgarno sequence and by homology searches against GenBank using the BlastP algorithm. The tRNAscan-SE program was used to search for tRNA genes [30], and the Rho-independent terminators were identified using ARLOND terminator finding program [31].

Accession Number
The nucleotide sequence of SfMu phage reported in this article has been deposited in the Gen-Bank database as accession number KP010268.

Isolation of SfMu
Bacteriophage SfMu was isolated from the serotype 4a strain, SFL2241. Cell lysis was observed after induction (phage yield: 1x10 4 -10 5 PFU/ml) and on propagating the induced lysate on SFL124 (serotype Y, indicator strain). However, phage SfMu plaques were only visible when the lysates were plated onto the indicator strain using the PAMA technique [26]. The plaques obtained had a clear pin point morphology with well-defined boundaries (Fig 1).
In order to confirm that SfMu is a prophage of SFL2241, Southern hybridization experiments using DIG-labelled SfMu as a probe were performed on EcoRV digested genomic DNA of SFL2241. Results confirmed the presence of phage genome integrated into the host chromosome (S1 Fig).

Morphology
Electron micrographs (EM) of negatively stained SfMu virions reveal an icosahedral head and a contractile tail (Fig 1). These characteristic features are typical of group 'A' phages of family Myoviridae and order Caudovirale, according to the morphological classification of Bradley [34]. The average particle had a head size of approx. 50 nm and a tail length of approx. 100 nm, which is comparable to that of bacteriophage Mu (head size 54 nm and tail 100 nm).

Genome features of bacteriophage SfMu
The complete genome sequence of bacteriophage SfMu was determined by sequencing the DNA of its host (SFL2241), using Illumina (Miseq). The data generated contained approx. 4,733,416 reads with avg read length of 250 bases, which were de-novo assembled into 1500 contigs, using Velvet. The accuracy of contigs was verified by mapping the reads back to the contigs. SfMu phage genome was found to be distributed between 2 contigs. The gap between the two contigs was then closed by targeted PCR and subsequent sequencing to yield a single contiguous sequence of phage SfMu genome totalling 37,146 bp in length. Phage SfMu genome has a G+C content of 51.9% which is comparable to its host (approx. 51%).
Analysis of phage SfMu genome revealed 55 putative open reading frames with a plausible Shine-Dalgarno sequence (Fig 2). Out of 55, only 4 transcribed leftwards and 51 transcribed rightwards. In addition, 49 of the predicted orfs initiated from ATG start, while 6 others used a GTG start. Phage SfMu genome is densely packed with the coding sequences occupying 94.2% of the genome. Several overlapping genes are present in phage SfMu genome, but minimal overlap of the start and stop codons was observed. The maximum overlap of 62 bp was detected between orfs15-16.
Proteins encoded by the predicted orfs were scanned for homologues using BlastP. Based on the similarities, possible functions were assigned to 29 ORFs and the other 26 ORFs showed similarity to uncharacterized proteins (S1 Table). Phage SfMu genome was also analysed for regulatory sequences and was found to contain six putative rho-independent transcription terminators (Fig 2). However, no tRNA genes were identified. To get an insight into the extent of similarity between phage SfMu and phage Mu or D108, SfMu encoded proteins were compared to their counterparts in phage Mu and D108. However, as phage D108 has been studied in less detail, most of the discussion in this report is based on phage Mu homologues. As shown in Fig 2, out of 55, 46 and 44 proteins of phage SfMu were >90% identical to their cognates in phage Mu and D108, respectively. Additionally, the arrangement of genes in phage SfMu was similar to bacteriophage Mu, and was divided into functional modules present in the following order from the left end: immunity, integration and transposition, lysis, head assembly and tail assembly. Moreover, based on the level of transcription during phage Mu's lytic cycle, phage SfMu genome can also be subdivided into early, middle and late regions.
The left end of the phage SfMu genome encodes the regulatory proteins, C-repressor and Ner, which are involved in the regulation of the lytic and lysogenic developments of the phage. While phage SfMu C-repressor and Ner proteins shared 89% and 98% identity to their counterparts in phage D108, identity of only 24% and 47% was obtained with the phage Mu equivalents (Fig 2). Since the C-repressor in phage Mu is required for the establishment and maintenance of lysogeny [35], the difference in the repressor proteins of phage Mu and SfMu agrees with the fact that unlike phage Mu lysogens which can only be induced by heat [36], SfMu lysogens were induced by UV. Two other proteins in the early region were found to be different to their cognates in phage Mu. These were ORF6 and ORF9, proteins of unknown function, which shared 77% and 87% identity, respectively, with their equivalents in both phages Mu and D108. Additionally, ORF7 and 15 of phage SfMu showed <50% identity to their equivalents in D108, however, were identical to the phage Mu homologues.
ORF11 and 14 of phage SfMu are two proteins which have no homologues in phages Mu or D108, thus making these proteins unique to phage SfMu. However, bacteriophages Mu and D108 also contain two proteins (encoded by orfs 12 and 19 in phage Mu, and orfs 12 and 18 in phage D108) which do not share homology with any of the proteins in phage SfMu. As the two unique proteins of phages SfMu and Mu or D108 belong to the same region of the phage genome, they may be divergent homologues having conserved function. However, it is equally possible that they have different functions and each provides some benefit to the host.
The majority of the structural genes in the late region appear to be conserved between the three phages. The only exceptions are ORFs36-37 of phage SfMu, which share 50% and 70% identity, respectively, with their counterparts in both phages Mu and D108 (Fig 2). orf36-37 are located between the head-gene module and the tail-gene module. Studies in phage Mu have revealed that the mutants defective in orf36 produced full heads, unattached tails, and served as tail donors in in vitro complementation assays [37,38], thus suggesting that ORF36 protein might be involved in maturation of heads to allow joining of the tails. In addition, the mutants defective in orf37 were shown to produce abnormally long tails and served as head donors in the complementation assay, suggesting ORF37's involvement in tail formation or stabilization [37,38]. Although ORF36-37 of phage SfMu were different to their phage Mu counterparts, results of BlastP analysis revealed that homologues of these two proteins were present on a cryptic Mu-like prophage in S. boydii.
Proteins encoded by orf49-52 of phage SfMu show homology to phage Mu G region. Genes S, U, U' and S' span the G segment in phage Mu. The proteins encoded by these genes are responsible for the tail fiber biosynthesis and assembly, and confer the host-range specificity to the phage [39,40]. Moreover, this segment lies next to the gin gene which encodes for a protein that promotes inversion of the G segment and therefore determines which pair of the tail fiber biosynthesis and assembly genes is expressed: S and U or S' and U' [41]. Based on the orientation of the G region, two types of phage particles are produced, Mu G(+) and Mu G(-), which differ in their host range. Phage with the G in (+) orientation (expressing S and U genes) are capable of infecting E. coli K12 [28], while those with the G(-) orientation (expressing U' and S' genes) infect E. coli C, Enterobacter cloacae, Serratia marcescens, Citrobacter freundii, Erwinia and Shigella sonnei [42]. The 'G-segment' of SfMu phage matches with that of phage Mu in G (-) orientation i.e. SfMu ORF49, 50, 51, and 52 show 96%, 97%, 90% and 65% identity to phage Mu S', U', U, and S proteins, respectively.

Host Range of bacteriophage SfMu
In order to examine the sensitivity of phage SfMu, strains of E. coli K-12 and 12 serotypes of S. flexneri were tested by the PAMA method. Using the SfMu lysate produced after induction, phage SfMu was able to infect only serotypes Y and 3b strains of S. flexneri. No plaques were observed on E. coli K-12 strain or other serotypes of S. flexneri. The sensitivity to various serotypes of S. flexneri was also determined using SfMu lysate prepared upon infection on serotype Y strain. Interestingly, the host range of phage SfMu remained the same under both conditions. In bacteriophage Mu, lysates obtained upon induction are known to contain both types of phage particles, Mu G(+) and Mu G(-), in equal ratios. Moreover, in the lysates grown by infection, the majority of the phage particles have 'G' in one orientation [36]. As phage SfMu genome contains phage Mu like G segment, it was expected that the lysate obtained upon induction will have two kinds of phage particles (with G in either orientation) infecting different host, thus will show a broader host range than the lysate produced upon induction (containing one kind of phage particles).
The reason for the same host range observed with either lysates of phage SfMu could be because of no inversion in the G segment of phage SfMu. Like phage Mu, SfMu genome contains 34 bp inverted repeats flanking the G segment and is the target for Gin-mediated site specific recombination. Analysis of phage SfMu IR repeats revealed a 1 bp mutation on one of the IR sequences. Since this base change is present in only one of the sequences and it lies next to the two central bases which serve as a possible crossover site in the IR of phage Mu [43,44], it could result in the reduced rate or abolishment of the G segment inversion. However, it is also possible that the same host range is observed because phage SfMu particles in the G(-) orientation do not infect S. flexneri strains and are infectious for different species of bacteria which were not used in our analysis.
As mentioned earlier, the tail fiber proteins of phage Mu determine its host range specificity. Since ORF49 of phage SfMu is >95% identical to phage Mu S' protein, we tested if phage Mu G(-) was able to infect S. flexneri like phage SfMu. Bacteriophage Mu was induced from its E. coli lysogen and the lysate obtained was tested for sensitivity against the 12 serotypes of S. flexneri. However, no plaques were obtained on any of the serotypes. Since phage Mu induced lysate contains both Mu G(+) and G(-) particles [36], it can be concluded that neither Mu G(+) nor G(-) infects S. flexneri. This suggests that although phage SfMu ORF49 and Mu S' share a high sequence similarity, the structure of the two tail fibers is diverse enough to confer host specificity. The results of pairwise alignment of SfMu ORF49 with the S' protein of bacteriophage Mu revealed 18 amino acid changes in ORF49 of SfMu phage (S3 Fig). Moreover, the first 147 amino acids of the two proteins were conserved, and most of the changes identified were in the carboxy terminal part of the protein. Out of 18, 7 amino acids were replaced by amino acids with similar properties, 5 polar and 2 nonpolar amino acids were replaced by a charged amino acid, 1 negatively charged amino acid was replaced by a nonpolar amino acid, while 2 nonpolar and 1 negatively charged amino acids were replaced by polar amino acids. The conserved N-terminal portion of the two tail fibers suggests that this part of the protein interacts with the phage tail, while the variable C-terminal is responsible for host recognition. This is similar to other Myoviridae phages like T4-like phages [45,46], Lactococcus lactis phages (sk1, TP901-1, and bIL170) [47], Pseudomonas phages (Pap1 and JG004) [48], and bacteriophage lambda [49], where the C-terminal domain of the tail fiber protein is responsible for binding to the host receptor.

Cell wall receptor of bacteriophage SfMu
Since bacteriophage Mu uses distal part of the LPS core as its receptor [28,42], we sought to investigate if the primary receptor of SfMu is also located in the LPS. Plaque assays using the PAMA technique were performed using a mutant serotype Y strain lacking O-antigen biosynthesis genes (SFL1195, S2 Table) and several recombinant serotype Y strains carrying a plasmid with different O-antigen modification genes of S. flexneri (S2 Table). Plaques were only obtained in control serotype Y strain and in serotype Y strain carrying O-acetyltransferase gene (converting serotype Y to 3b, SFL1899), which was consistent with the phage SfMu host range results. This indicates that modification of the O-antigen or deletion of the O-antigen biosynthesis genes results in failure of adsorption of phage SfMu to the host cell, thus suggesting that phage SfMu uses O-antigen as receptor.
Results of the plaque assay were further confirmed by performing phage inactivation assay using LPS purified from S. flexneri serotype Y and 3b strains. As shown in Fig 3,

Prevalence of SfMu phage in S. flexneri
Since S. flexneri is not a recognised host for transposable phages, we investigated the prevalence of bacteriophage SfMu in various serotypes of S. flexneri. A total of 194 wild type S. flexneri strains tested, comprising isolates from various geographical regions, and represented 15 serotypes of S. flexneri. All the strains were first screened by performing multiplex PCR using 3 pairs of primers. The primer sets were chosen to target 3 different regions of phage SfMu: c repressor-ner region (900 bp), middle operon regulator-lysozyme (1.2 kb) and gin-mom region (1.5 kb). As shown in S3 Table, all three products of expected sizes were only obtained for the phage SfMu host strain (positive control). However, only the 1.2 kb fragment was obtained in 38 other strains belonging to serotypes 1a, 3a, 3b, 6, Y, Yv, Xv and X. Sequencing of the 1.2 kb PCR product from five these strains revealed that they contained the expected sequence, thus indicating the presence of cryptic SfMu phage in their genome.
The results of multiplex PCR were further validated by performing Southern hybridization. EcoRV digested genomic DNA of 34 out of 38 PCR positive strains were probed with DIG-labelled EcoRV-digested SfMu DNA. As expected, most of the phage fragments were present in the phage SfMu host (positive control) indicating presence of the phage in the host chromosome (S1 Fig). Additionally, the hybridization pattern of 6 other strains (lanes marked with red asterisk, S1 Fig), belonging to serotype X or Y (originating from Vietnam), was very similar to that of SfMu host, thus suggesting that a complete copy of the SfMu prophage might also be present in these strains. However, the absence of the 900 bp and 1.5 kb PCR products (in multiplex PCR) in these strains suggests that the prophage in their genomes might be related to phage SfMu, but is not identical.
Two bands (4.5 kb and 3.7 kb) appeared to be common in most of the strains, indicating that these strains might contain a remnant of the SfMu prophage in their genomes (S1 Fig). However, except in the SfMu lysogen or the 6 strains containing complete SfMu-like prophage, the 3.7 kb band appears to be a little smaller in size (approx. 3.5 kb), which could be because of restriction site polymorphism.
Taken together, these results indicate that Mu-like phages are not uncommon in S. flexneri. Moreover, identification of SfMu like prophage in different serotypes of S. flexneri also suggests that this phage might have a role in transferring diverse sets of virulence genes and other genetic traits from one serotype to another.

Conclusions
This is the first report on the isolation of a transposable phage, SfMu, from S. flexneri. The complete genome of phage SfMu was sequenced, characterised, and compared with bacteriophage Mu and other Mu-like phages. Results of comparative analysis revealed that SfMu is the first Mu-like phage isolated from a different genus yet closely related to phage Mu. Further analysis of phage SfMu cell wall receptor revealed that, bacteriophage SfMu recognises LPS Oantigen as its primary receptor for adsorption. This study also identifies presence of complete or cryptic SfMu prophage in various serotypes of S. flexneri. Further studies on Mu-like phages of S. flexneri will be useful in understanding the role of these phages in evolution, cellular lifestyle, and pathogenicity of S. flexneri.