Molecular characterization, structural analysis and determination of host range of a novel bacteriophage LSB-1

The LSB-1 phage has a cystic form of 50 ± 5 nm in diameter, with a short noncontractile tail 60 ± 5 nm long (Fig. 1). It is classified as a member of the Podoviridae family [8‐10].

LSB-1 coliphage nucleic acid was sensitive to Dnase I, but resistant to Rnase A and S1 nuclease (Fig. 2). It was concluded that all the extracts contained double-stranded linear DNA.

The shotgun sequencing and a primer-walking method was used to assemble the whole linear LSB-1 genome of approximately 30 kb. Fifteen major potential ORFs were identified, all of which could be assigned functions based on homology with corresponding genes in the K1F coliphage in a BLAST search (Fig. 3 and 4).

We chose four important proteins, capsid protein (gp10), tail tubular protein (gp12), internal virion protein D (gp16), and endosialidase (gp17), which are present in phage T7 supergroup members, to build phylogenetic trees. As shown in Fig. 5, 6, 7 and 8, trees built using different proteins are broadly similar. Coliphages LSB-1, EcoDs1, and K1F closely clustered towards the end of the group and phages T7, T3, phiYeO3-12 and K11 cluster within another branch of the same group. These data suggest that LSB-1 is most closely related to the K1F-like phages, and should be classified as a new member of the K1F supergroup.

As shown in Fig. 9 and 10, the secondary structure of gp17 protein was composed of 33.10% Alpha helix (Hh), 25.25% Extended strand (Ee) and 30.29% Random coil (Cc).

The PHYRE program generated several possible templates classified as best fit when the gp17 sequence was submitted. All of these templates included the sequences of hydrolase as one of its domains. It also provided the SCOP codes, E-values and estimated precision values for the predicted models. Table 1 shows these parameters provided by different templates.

The endosialidase 3-dimensional structure was chosen to generate coordinates for the Gp17 of LSB-1 based on an E-value of 1.1 and an estimated precision of 70%. Using the PHYRE program, it was difficult to find another closer coordinate template for Gp17 of LSB-1 even though an E-value of less than 2e-6 was considered [11‐13]. Gp17 of LSB-1 included endosialidase amino acid residues from Thr290 to Glu713, with the with the exception of forty-one residues (Ser 465 to Trp475, Tyr497 to Val512, and Lys543 to Leu562). These forty-one residues did not match corresponding sequences within the prediction server. The Gp17 3-dimensional structure model predicted two domains: Domain A with hydrolase activity connected by an intervening unstructured sequence to Domain B. The later domain may have a host-anchoring function. A worm algorithm representation of the model is shown in Fig. 11. Domain A, the larger of the two domains, includes residues Thr290 to Phe528 and corresponds to the N-terminal domain of endosialidase. It comprises 4 α helices and 9 β strands with intermittent unstructured intervening regions. The arrangement is characteristic of crystallized hydrolase. Domain B is composed of 1 α helix and 9 β strands with intermittent unstructured intervening regions. It includes residues Gln542 to Glu713 and corresponds to the C-terminal domain of endosialidase. There were no predicted structural constraints in the connecting region between domains A and B which included the residues of Tyr529 to Asp541.

The molecular surface structure of the Gp17 is presented in Fig. 12. It shows the catalytic pocket in Domain A comprising the catalytic triad, Glut405, Arg415, and Arg487, identical to that found in the endosialidase family of enzymes. Domain A is shown topographically in front of Domain B. The connecting region was difficult to see in this representation because of its opaqueness. Fig. 13 and Fig. 14 show the worm algorithm structures of Gp17and the KIF endosialidase respectively.

Bacteriophage LSB-1 was propagated in liquid culture. The EIEC strain 8401 was infected with phages at a multiplicity of infection of 0.1. Following complete lysis of the host cells, cell debris was removed by centrifugation at 4000 × g for 10 min, and phage particles in the supernatant were concentrated by adding polyethylene glycol 6000 and NaCl to achieve final concentrations of 10% and 1.0 M, respectively. Phage particles were collected by centrifugation at 8000 × g for 10 min. Pellets were re-suspended in 0.01× the original volume of sterile SM buffer (5.8 g sodium chloride, 2 g magnesium sulphate, 100 mg gelatin, 50 mL 1 mol·L-1 Tris (pH 7.5) and 945 mL distilled water). For isopycnic centrifugation, the phage suspension was placed on a cesium chloride gradient stepwise using three solutions whose densities were 1.45, 1.50, and 1.70, respectively. After centrifugation for 60 min at 150,000 × g, the phage band was withdrawn and dialyzed against 10 mM Tris-HCl (pH 7.5) containing 10 mM MgSO. The purified phage (approximately 10¹¹PFU/ml) was stored at 4°C until use.

Seventeen bacterial strains, listed in Table 2, were tested for sensitivity against the isolated phages. The Spot Test [18] was used to determine the host range of the phage. Lytic activity was examined following overnight incubation at 37°C and recorded on a scale as follows: (N) no plaques, (T) turbid plaques, (C) clear plaques. To obtain an accurate estimate of relative phage lytic activity, all host range determinations were carried out simultaneously using a single high-titer stock of purified bacteriophage and all host cells were incubated at 37°C in LB medium.

To examine phage LSB-1 morphology, 50 μl of purified viral stock solution was fixed by addition of 50 μl of 0.5% glutaraldehyde in 4% paraformaldehyde and a drop of this solution was placed on a carbon coated copper grid After waiting 30 min for settlement, excess liquid was removed and the grid was allowed to dry. A drop of 2% phosphotungstic acid was added for 2 min before excess was removed with filter paper before drying and then examination by TEM (Hitachi, model S-800) at an acceleration voltage of 45 kV.

The method used was based upon that described by Sambrook and Russell [19]. Bacteriophage from the concentrated solutions were lysed with the addition of EDTA (final concentration 20 mmol·L-1), proteinase K (final concentration 50 μg·mL-1) and SDS (final concentration 0.5%) and incubation at 56°C for 1 h. The nucleic acid was purified using phenol, phenol/chloroform and chloroform extraction. The final aqueous phase was dialyzed overnight against Tris EDTA buffer (TE). In this method, nucleic acid yield was estimated by agarose gel electrophoresis and comparison of ethidium bromide stain intensity with known DNA standards (Hind III cut lambda phage DNA, New England Biolabs Inc, USA).

The nucleic acid extracts were diluted to a standard concentration of ~20 ng·μL-1. Approximately 250ng of each extract was subjected to digestion with DNase I (Sigma Aldrich), RNase A (Sigma Aldrich) and S1 nuclease (Promega). All reactions were terminated with the addition of EDTA (10 mmol·L-1 final concentration) and analyzed using 0.8% agarose gel electrophoresis at 5 V cm-1.

The coliphage genome was sequenced by the shotgun method. Genomic DNA was sheared by sonication, cloned into pUC18 and sequenced with an ABI 3700 automated DNA sequencer, to give 13-fold coverage of the genome. Sequences were assembled into contigs, and gaps linked using a primer-walking technique (Kaczorowski and Szybalski, 1998) [20]. Potential open reading frames (ORFs) were predicted using ORF Finder http://​www.​ncbi.​nlm.​nih.​gov/​projects/​gorf/​ and manual correction. Translated ORFs were used in a BLAST search against the Swiss-Prot http://​us.​expasy.​org/​tools/​blast and NCBI protein databases http://​npsa-pbil.​ibcp.​fr/​cgi-bin/​npsa_​automat.​pl?​page=​npsa_​sopma.​html. Protein secondary structures were predicted with SOPMA http://​npsa-pbil.​ibcp.​fr/​cgi-bin/​npsa_​automat.​pl?​page=​npsa_​sopma.​html.

Phylogenetic trees were constructed using Mega 4 software following published protocols [21, 22].

The amino acid sequence of endosialidase was determined and compared with the protein structure family databases PDB [23], SCOP [24, 25], and PFAM [26] to identify the most suitable template structure. The eventual template structure was taken from PDB. Three-dimensional models were created using PHYRE http://​www.​sbg.​bio.​ic.​ac.​uk/​~phyre/​ by mapping the coordinates of the template structure with aligned residues of the endosialidase. Computer-generated three-dimensional models were viewed and analyzed using CN3 D 4.1 application software programs obtained from http://​www.​ncbi.​nlm.​nih.​gov/​Structure/​CN3D/​cn3d.​shtml.