FepA- and TonB-dependent bacteriophage H8: receptor binding and genomic sequence

Abstract

H8 is derived from a collection of Salmonella enterica serotype Enteritidis bacteriophage. Its morphology and genomic structure closely resemble those of bacteriophage T5 in the family Siphoviridae. H8 infected S. enterica serotypes Enteritidis and Typhimurium and Escherichia coli by initial adsorption to the outer membrane protein FepA. Ferric enterobactin inhibited H8 binding to E. coli FepA (50% inhibition concentration, 98 nM), and other ferric catecholate receptors (Fiu, Cir, and IroN) did not participate in phage adsorption. H8 infection was TonB dependent, but exbB mutations in Salmonella or E. coli did not prevent infection; only exbB tolQ or exbB tolR double mutants were resistant to H8. Experiments with deletion and substitution mutants showed that the receptor-phage interaction first involves residues distributed over the protein's outer surface and then narrows to the same charged (R316) or aromatic (Y260) residues that participate in the binding and transport of ferric enterobactin and colicins B and D. These data rationalize the multifunctionality of FepA: toxic ligands like bacteriocins and phage penetrate the outer membrane by parasitizing residues in FepA that are adapted to the transport of the natural ligand, ferric enterobactin. DNA sequence determinations revealed the complete H8 genome of 104.4 kb. A total of 120 of its 143 predicted open reading frames (ORFS) were homologous to ORFS in T5, at a level of 84% identity and 89% similarity. As in T5, the H8 structural genes clustered on the chromosome according to their function in the phage life cycle. The T5 genome contains a large section of DNA that can be deleted and that is absent in H8: compared to T5, H8 contains a 9,000-bp deletion in the early region of its chromosome, and nine potentially unique gene products. Sequence analyses of the tail proteins of phages in the same family showed that relative to pb5 (Oad) of T5 and Hrs of BF23, the FepA-binding protein (Rbp) of H8 contains unique acidic and aromatic residues. These side chains may promote binding to basic and aromatic residues in FepA that normally function in the adsorption of ferric enterobactin. Furthermore, a predicted H8 tail protein showed extensive identity and similarity to pb2 of T5, suggesting that it also functions in pore formation through the cell envelope. The variable region of this protein contains a potential TonB box, intimating that it participates in the TonB-dependent stage of the phage infection process.

FIG. 1.

Bacteriophage H8 morphology. H8 particles were observed by transmission electron microscopy at a magnification of 100,000. The inset at the bottom left shows T5, observed by metal-shadowed transmission electron microscopy at magnification 93,150. (Reprinted from reference with permission of the publisher).

FIG. 2.

(Top) H8 susceptibility and FepA expression level. BN1071 expresses FepA from its wild-type chromosomal structural gene. The fepA strain OKN3 was transformed with pITS23 or pT944, both of which carry fepA⁺ alleles and produce different amounts of wild-type FepA in the OM. The bacteria were grown in MOPS medium to late log phase, and lysates from 5 ×10⁷ cells were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western immunoblotting with ant-FepA monoclonal antibody 45 (65) and ¹²⁵I-labeled protein A. The intensities of the FepA bands in the four strains were determined by image analysis on a Storm Scanner (Molecular Dynamics) and related to those produced by a set of standards with purified FepA. The same bacteria were plated on LB agar, and the number of PFU was determined. The experiment was repeated three times; the mean standard deviation of the PFU determinations was 10.6%. (Bottom) Inhibition of H8 binding by FeEnt. E. coli strain OKN3/pITS23 was grown in LB broth and exposed to H8 (10⁵ PFU) in the absence or presence of increasing concentrations of FeEnt. After 45 min at 37°C, the mixtures were centrifuged, and the number of phage in the supernatant (PFU_FREE) was determined by serial dilution and plaque assays. PFU_BOUND was calculated as PFU_TOTAL − PFU_FREE, and percent bound in the presence of FeEnt was calculated as PFU_{BOUND (+FeEnt)}/ PFU_{BOUND (−FeEnt)} × 100. Data were analyzed by the IC₅₀ algorithm of Grafit 5.013 (Erithacus Ltd., Surrey, United Kingdom), which yielded an IC₅₀ value of 0.098 μM for FeEnt.

FIG. 3.

Analysis of FepA mutants. (Left) The space-filling model shows a view looking down onto the surface of the protein from the exterior. The N-terminal domain is colored cyan, and the TonB box region is black. Other colored regions or residues indicate sites that affected H8 infection. Residues removed by class I loop deletions (ΔL4, ΔL7, ΔL9, and ΔL11), class II (ΔL5, ΔL8, and Δ10), and class III (ΔNL1, ΔNL2, and ΔL3) are colored red, yellow, and green, respectively. Site-directed Ala substitution mutations for individual basic, aromatic, and acidic residues are colored blue, magenta, and purple, respectively. (Right) In the ribbon model the protein was rotated −90° along its x axis to show the location of individual substitution mutations within the loops. The figure also depicts the B1 and B2 regions of the FepA surface vestibule, which participate in the initial and secondary stages of ligand binding (16). The bacteriophage utilizes sites that are broadly distributed across the outer, B1, region of the receptor protein, but single substitutions in the inner, B2, region also impair H8 infection, as well as FeEnt transport and colicin B/D killing.

FIG. 4.

Overall comparison of H8 and T5 genomic structure. Alignment of the annotated genomes was made by ACT4 (http://www.sanger.ac.uk). The annotated T5 genomic sequence was obtained from NCBI (http://www.ncbi.nlm.nih.gov/genomes), accession number NC005859. Homology between the DNA sequences is displayed as vertical bars of graded color between the genomes of H8 (accession no. AC171169) and T5 from a minimum identity value of 70% (white) to a maximum identity of 100% (red). The figure also depicts the location of genes on the positive (top) or negative (bottom) strands of the bacteriophage chromosomes. For both genomes, ORFs are indicated by colored boxes according to their functional categories as previously described for T5 (99): DNA replication and repair, red; nucleotide metabolism, magenta; host interaction, yellow; other enzymes, green; structural proteins, blue; unknown function, white. The genes encoding the receptor-binding and pore-forming and tail proteins are colored orange.

FIG. 5.

Comparison of ORFs in the H8 and T5 genomes. Alignment of the annotated genomes was performed as described in the legend of Fig. 4. Pre-early, early, and late regions of the T5 genome are marked by red, blue, and yellow underlines, respectively; deletable regions are further underlined with cyan. For both genomes, genes and their transcriptional directions are indicated in colored boxes with arrows indicating the directions of transcription. Genes are colored according to their functional categories using the scheme that was previously described for T5 (99): DNA replication and repair, red; nucleotide metabolism, magenta; host interaction, yellow; other enzymes, green; structural proteins, blue; unknown function, white. ORFs encoding the receptor-binding and pore-forming and tail proteins are colored orange. Gaps in the H8 chromosome relative to that of T5 are shown as blue stippled boxes; gaps in the T5 chromosome relative to that of H8 are shown as black stippled boxes.

FIG. 6.

Analysis of the putative receptor-binding and pore-forming proteins of bacteriophage H8. (Top) CLUSTAL W alignment of the tail receptor-binding proteins of bacteriophages T5 (protein pb5 or Oad) and BF23 (Hrs) with the putative receptor binding protein (Rbp) of H8 illustrates strong homology in five regions (boxed in red), with identical (marked with a star below) and similar (BLOSUM 62 matrix; marked with a colon or dot below) residues in the sequences colored red. In contrast to these conserved regions, the alignment also shows two variable regions (boxed in black). In the case of H8, a preponderance of acidic residues (cyan; basic residues are highlighted in green) and aromatic residues (highlighted in yellow) exist in the upstream variable domain (H8 residues 138 to 213) that may participate in adsorption to basic residues within FepA (16, 68, 90) (see Discussion). Charged and aromatic amino acids that are unique to H8 are listed below the alignment. (Bottom) The alignment of the tail pore-forming protein of T5 with its H8 homolog (ORF 114, encoded by tpf) reveals strong homology between the two sequences (66% identity; 72% similarity), with conspicuous identities (highlighted in blue) and similarities (light blue) along their lengths. The alignment has most homology near the N and C termini; the relatedness weakens in the central region. The enlarged elements of sequence boxed in red illustrate the three regions of most significant homology among the proteins; the region boxed in black illustrates the greater variability of the central portion. The H8 protein contains the sequence GEGIPVGLA, which bears similarity to the consensus TonB box regions near the N termini of siderophore receptor proteins. FepA contains the TonB box sequence DDTIVVTAA. The tabular comparison of TonB boxes illustrates the variability that occurs in such regions, despite the fact that they all presumably physically interact with the single protein, TonB. The top four proteins, from Pseudomonas aeruginosa (PaePfeA), S. enterica serovar Typhimurium (StyIroN and StyFepA), and E. coli (EcoFepA), are orthologs that transport FeEnt. The next six proteins (EcoCir, EcoFecA, EcoBtuB, EcoIutA, EcoFhuA, and EcoFhuE) are E. coli LGP paralogs. These 10 proteins, as well as relevant regions of colicin B (EcoColB) and H8 Tpf, were aligned by the PILEUP algorithm (GCG, Madison, WI). Residues highlighted in yellow are conserved (either identical or similar; tabulated for each position below the below the alignment) in the consensus TonB box sequence. The column at right lists for each individual protein the number of identical or similar residues to the consensus core (ETIVV) or full (DETIVVTAA) TonB box consensus sequence, respectively.