FepA- and TonB-Dependent Bacteriophage H8: Receptor Binding and Genomic Sequence

Bacterial strains, plasmids, and media. E. coli strains were grown in LB (61) medium with or without appropriate antibiotics. RWB18-60 (5) and KDF541 (81) are recA entA fepA derivatives of AB1515 that were isolated by spontaneous resistance to colicin B. GUC12 (32-34), KDF571 (81), and KDO23 are spontaneous tonB isolates that were isolated by resistance to colicins. We also utilized a precise, in-frame chromosomal deletion of the entire fepA (OKN3) and tonB (OKN1) structural genes (58) in E. coli strain BN1071 (46). For the pUC18 derivatives pITS449 (5) and pITS944 (68), ampicillin was added to a concentration of 100 μg/ml. For the pHSG575 (38) derivatives pITS11 and pITS23 (89), chloramphenicol was added to a concentration of 20 μg/ml. In experiments that measured FepA expression, cells were grown in morpholinepropanesulfonic acid (MOPS) minimal medium (66) to mid-log or late log phase.

Isolation of bacteriophage H8.The phage H8 was isolated from Salmonella enterica serovar Enteritidis in Poland and adapted to the S. enterica serovar Enteritidis propagation strain 64/M for phage typing. The first report of the phage typing system was documented during a phage typing colloquium in Wernigerode, Germany, in 1975 (50), and the scheme was published in 1977 (49). In 1985 the Polish S. enterica serovar Enteritidis phages were analyzed by the Hungarian phage typing scheme (52). We used this latter scheme for serovar Enteritidis during routine phage typing. To establish the type of H8 we screened isogenic S. enterica serovar Enteritidis strains lacking the catecholate receptors FepA, IroN, and Cir. We saw the loss of H8 phage lysis in Salmonella strains carrying fepA::Tn10dTc (WR1425) and in FepA-deficient E. coli strains (H1875 and H1876).

Phage infection assays.Bacteria were grown in 5 ml of LB broth at 37°C overnight. A phage stock suspension (∼10¹⁰ PFU per ml) was serially diluted in LB broth, and 10-μl aliquots were mixed with 50 μl of bacterial culture and incubated for 2 min at room temperature. The infected cell suspension was mixed with molten top agar and plated on LB plates. After incubation at 37°C for 16 h, the phage plaques were counted. For analysis of mutant FepA proteins, susceptibility to H8 was expressed as a percentage of infectivity relative to an isogenic host strain that expressed wild-type FepA. Expression levels for all the mutant FepA proteins, as well as their proper folding and assembly into the OM, were previously determined by quantitative immunoblotting and flow cytometry, respectively, with anti-FepA monoclonal antibodies (4, 16, 68, 89, 90). We verified their expression levels in the phage susceptibility assays, using wild-type FepA expressed from the same plasmid as a positive control (data not shown).

Phage infection competition experiments.KDF541/pITS449 was grown overnight at 37°C in LB medium plus 100 μg/ml ampicillin. FeEnt was added to 2 × 10⁸ cells in 0.1 ml of LB broth to a concentration of 40 uM, and the suspension was incubated for 2 min at room temperature. A total of 10 μl of phage suspension (10⁶ PFU/ml) was added, the mixtures were incubated for 15 min at 37° and diluted with 1 ml of LB broth, and the cells were pelleted in a microcentrifuge at 14,000 rpm for 1 min. Cell pellets were resuspended in 100 μl of LB broth mixed with 2.5 ml of molten top agar and plated on LB agar. Phage plaques were counted after 16 h at 37°C.

Phage binding competition experiments.A total of 10⁵ PFU of H8 phage in 0.5 ml of LB medium plus 10 uM CaCl₂, either without FeEnt or containing twofold serial dilutions of FeEnt (0.04 to 20 μM), was mixed with 10⁸ cells of OKN3/pITS23 in 0.5 ml of LB medium plus 10 μM CaCl₂. The samples were incubated in a 37°C water bath for 40 min and centrifuged at 8,000 × g for 5 min at 4°C. Aliquots of the supernatant were diluted and plated on a lawn of OKN3/pITS23 to determine the number of unbound PFU.

Western immunoblotting.Bacteria were grown in LB broth overnight, subcultured (1%) in MOPS medium, and shaken at 37°C for 5.5 h, to mid-log phase. A total of 5 × 10⁷ cells were suspended in sodium dodecyl sulfate-polyacrylamide gel electrophoresis sample buffer and boiled for 5 min, and the proteins in the lysate were resolved on polyacrylamide slab gels (3). The proteins were transferred to nitrocellulose membranes and incubated for 1 h with anti-FepA monoclonal antibody 45 (65), diluted 0.1% in 5% skim milk. The nitrocellulose was washed five times with tap water and incubated with goat anti-mouse immunoglobulin G-alkaline phosphatase (0.1%; Sigma-Aldrich) in 5% skim milk for 1 h. The membranes were washed five times with tap water and developed with nitroblue tetrazolium and bromochloroindolyl phosphate (65).

Preparation of bacteriophage H8 DNA.One plaque of phage H8 grown on strain KDF541/pITS449 was picked with a sterile toothpick and resuspended in 500 μl of LB both. One hundred microliters of the phage suspension was diluted to 1 ml with 0.01 M MgCl₂-0.01 M CaCl₂ and used to inoculate 2 × 10⁸ cells of KDF541/pITS449 in 50 μl of LB broth. After 15 min at 37°C, the suspension was diluted to 50 ml with LB broth and shaken overnight at 37°. The resulting lysate was clarified by centrifugation in sterile 30-ml Corex tubes for 20 min at 3,000 rpm. The supernatant was centrifuged for 1 h at 45,000 rpm in a 70 Ti rotor. The phage pellet was resuspended in 180 μl of 50 mM Tris-Cl, pH 8.0, and chilled on ice. After two phenol extractions and a chloroform extraction, the DNA was precipitated with two volumes of ice-cold ethanol and resuspended in 50 μl of 10 mM Tris-Cl-1 mM EDTA, pH 8.

Genomic sequencing.The detailed procedures for random shotgun cloning, fluorescent-based DNA sequencing, and subsequent analysis were previously described (6, 19, 77, 78). Fifty-microgram portions of purified phage DNA were randomly sheared and made blunt ended (6, 77, 83). After kinase treatment and gel purification, fragments in the 1- to 3-kb range were ligated into SmaI-cut bacterial alkaline phosphatase-treated pUC18 (Pharmacia), and the ligation mixture was transformed by electroporation into E. coli strain XL1 Blue MRF′ (Stratagene). A random library of approximately 1,200 colonies was picked from each transformation and grown in Terrific broth (83) supplemented with 100 μg of ampicillin for 14 h at 37°C with shaking at 250 rpm. The cells were harvested, and their plasmids were isolated by a cleared lysate-based protocol (6).

Sequencing reactions (19, 77) were performed using the Amersham ET terminator sequencing reaction mixes. The reactions were incubated for 60 cycles in a Perkin-Elmer Cetus DNA Thermocycler 9600 under the cycle conditions recommended by the manufacturer. Any unincorporated dye terminators were removed by ethanol precipitation at room temperature, and after the fluorescent-labeled nested fragment sets in double-distilled water were dissolved, they were resolved by electrophoresis on an ABI 3700 Capillary DNA Sequencer. After base calling with the ABI analysis software, the analyzed data were transferred to a Sun workstation cluster and assembled using Phred and Phrap (24, 25). Overlapping sequences and contigs were analyzed using Consed (31). Gap closure and proofreading were performed using either custom primer walking or using PCR amplification of the region corresponding to the gap in the sequence, followed by sequencing directly using the amplification or nested primers or by subcloning into pUC18 and cycle sequencing with the universal pUC primers (6, 19, 77, 78). In some instances, additional synthetic custom primers were necessary to obtain at least threefold coverage for each base. The mean factor of coverage over the genome was 15. The resulting phage sequence was analyzed on Sun workstations using Artemis (79).

We discovered putative novel genes by first using Artemis to call open reading frames (ORFs) greater than 100 bp in the ordered and oriented single H8 contig. We then removed ORFs that overlapped annotated genes that were previously identified by BLASTp (2) analysis against T5 and those that overlapped gaps in the assembly. We next used the BLAST algorithm to compare the remaining sequences against the GenBank database (E value of 10⁻⁶), seeking relationships to known genes. Genes without homology were classified as putative novel genes. We then ran the entire assembly through the ab initio FgenesV and GenemarkS programs, and both of these programs identified nine genes. Analysis of these putative novel genes by PSORT and TMHMM resulted in their predicted cellular localization.

Nucleotide sequence accession number.The sequence of phage H8 was deposited into GenBank database under accession number AC171169.

Morphology of bacteriophage H8.H8 was isolated from a collection of serovar Enteritidis bacteriophages that originated in Poland in 1975. The virus was selected for its ability to infect fepA⁺ but not fepA strains of serovar Enteritidis and E. coli (Tables 1 and 2). Electron microscopic depictions showed an icosahedral head and long tail with fibers that closely resembled the morphology of bacteriophage T5 (Fig. 1). Both H8 and T5 tail assemblies contain an elongated, pointed spike, presumably comprised by their pore-forming tail proteins.

• Open in new tab
• Download powerpoint

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 1 with permission of the publisher).

FepA specificity and TonB dependence of bacteriophage H8.H8 infected only E. coli strains carrying a functional fepA allele (Table 1). Strains with a defective fepA structural gene, like the S. enterica strains W1334 and WR1425, the spontaneous colicin B-resistant E. coli strain RWB18, and the genetically engineered E. coli strain OKN3 (containing a precise, in-frame deletion of fepA) were completely resistant to infection by H8 and did not produce plaques when exposed to the phage. The host range of bacteriophage H8 included several gram-negative bacterial species (Table 2). Besides Escherichia and Salmonella, the virus infected Shigella, Citrobacter, and Serratia in liquid or solid LB medium (Tables 1 and 2). On the other hand, with the exception of the Salmonella, we also identified strains of these same genera that were H8 resistant. Furthermore, H8 failed to propagate on all the strains of Enterobacter, Klebsiella, Morganella, Proteus, and Yersinia that we tested.

Experiments with plasmids also demonstrated the FepA specificity of H8. In the E. coli fepA strains KDF541, RWB18-60, or OKN3, plasmid-encoded expression of FepA restored H8 plaque-forming ability to equivalent or higher levels than those conferred by strain BN1071, which expresses wild-type FepA from its chromosome (Table 1). E. coli also produces other ferric catecholate receptors, including Fiu, IroN, and Cir. No bacteriophages are known to utilize these OM proteins, nor do they play a role in H8 penetration (Table 1). Because of its close relationship to bacteriophage T5 (see following), we evaluated the possibility that H8 infection might take place through FhuA. But the presence or absence of FhuA had no impact on the susceptibility of cells to H8 infection. The genetically engineered strain OKN73, which contains precise deletions of both fhuA and fepA, did not acquire sensitivity to H8 when transformed with a plasmid that expressed wild-type FhuA (pITS11 [89]). It did, however, become sensitive to H8 when it harbored the fepA ⁺ plasmid pITS23 (data not shown).

H8 infection was TonB dependent. E. coli strains GUC12, KDO23, and OKN1 are fepA ⁺ tonB derivatives of C600 and BN1071; all three were unable to propagate the phage (Table 3). Likewise, we saw the TonB dependency of H8 in S. enterica (SR1001) and serovar Enteritidis (WR1529). As is the case for TonB, although their exact physiological functions are unknown, the products of the exbB and exbD loci are needed for normal function of TonB-dependent OM transport systems (9, 29). In addition, the ExbB and ExbD proteins bear structural similarity to, and are functionally interchangeable with, TolQ and TolR (10). Our experiments reiterated this interchangeability with regard to infection by H8: the exbB mutation alone in S. enterica WR1893 and E. coli HE1 did not cause phage resistance. But as seen before for other ligands (10), exbB and tolQR were ostensibly interchangeable with respect to H8 infection. That is, only the exbB tolQ and exbB tolR double mutants (E. coli strains HE2 and HE10) were resistant to H8.

Effect of FepA expression levels and the presence of FeEnt.The titer of H8 lysates varied with the concentration of FepA in the E. coli OM (Fig. 2). The number of PFU on each host strain fluctuated with the FepA expression level. The same phage lysate gave the most PFU when FepA was expressed from the low-copy-number vector pITS23 (derived from pHSG575; 3 to 5 copies per cell) (38, 96); fewer plaques resulted when FepA originated from a single chromosomal copy (BN1071) (46). Relative to BN1071, FepA expression increased 1.5-fold in OKN3/pITS23, and the number of H8 PFU was 1.2-fold higher on the latter strain. We also tested H8 plating efficiency on bacteria expressing FepA from the high-copy-number vectors pITS449 and pT944 (pUC18 derivatives) (68). Because these constructs contain a truncated promoter region (5), they produce less FepA than OKN3/pITS23, which has a native, full promoter (89). H8 plating efficiency was lowest on these strains, probably as a result of physiological differences caused by the higher plasmid load (see Discussion).

• Open in new tab
• Download powerpoint

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.

We compared the susceptibility of E. coli to H8 in the absence and presence the presumably competitive ligand FeEnt. Preincubation of E. coli OKN3/pITS23 with FeEnt (10 uM) reduced the plating efficiency of phage H8 to 9% of its value in the absence of the ferric siderophore (Table 3), suggesting that the phage shares common binding sites with FeEnt on FepA. Bacteriophage adsorption assays in the presence of FeEnt verified this supposition: increasing concentrations of the ferric siderophore strongly inhibited the binding of H8, with a 50% inhibitory concentration (IC₅₀) of 98 nM and complete inhibition above 300 nM (Fig. 2).

H8 susceptibility of bacteria expressing FepA-FhuA chimeras and FepA loop deletion mutants.Both the N and C domains of FepA are necessary for H8 infection. H8 was unable to infect strains expressing FepA mutant proteins without the N-terminal globular domain (a deletion of residues 1 to 150 [Δ1-150]) (89) even at a high multiplicity of infection (Table 3). Nor did the production of hybrid receptor proteins that exchanged the N domain of FepA into the C domain of FhuA, or vice versa (89), confer phage sensitivity.

FepA contains 11 external loops linking the 22 strands of its β-barrel, and two loops that originate from the N domain. H8 infectivity was strongly impaired by deletions (68) that eliminated or affected these loops (Table 3 and Fig. 3). Susceptibility conferred by the deletion alleles fell into three categories: I, no infectivity and complete resistance to H8 infection (deletions of loops L2, L4, L7, L9, and L11, yielding ΔL2, ΔL4, ΔL9 and ΔL11 mutants); II, less than 15% of wild-type susceptibility (ΔL5, ΔL8, and ΔL10); III, 16 to 50% of wild-type susceptibility (ΔNL1, ΔNL2, and ΔL3). Among the ferric siderophore, bacteriocins, and bacteriophage, the FepA loop deletions most significantly impaired interactions with the phage. All 11 loop deletions affected the interaction of FepA with H8, and five of them rendered the bacteria fully resistant to infection. Loop deletion mutations may cause unexpected, potentially global changes in OM protein structure, and as a result we cautiously interpret their negative phenotypes with regard to H8. However, it is relevant in this sense that the loop deletion mutants are generally functional, albeit at reduced levels, in the transport of FeEnt and susceptibility to colicins B and D. Furthermore, with regard to H8, the loop deletions fell into different categories than those previously observed for FeEnt or the colicins (68). For instance, several of the mutants that were completely resistant to phage infection (ΔL2, ΔL4, ΔL9, and ΔL11) transported FeEnt and were susceptible to killing by both colicins. Only the ΔL7 mutant rendered FepA nonfunctional for all four of the ligands. Elimination of either N domain loop (ΔNL1 or ΔNL2 mutant) had a lesser effect on H8 infection than surface loop deletions, suggesting a secondary role in H8 adsorption.

• Open in new tab
• Download powerpoint

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.

Effect of Ala substitution mutations on phage infectivity.FepA is adapted to the chemical determinants of FeEnt, resulting in subnanomolar affinity of the ligand-binding equilibrium The anionic, catecholate iron complex associates with basic and aromatic amino acids in the receptor protein that were previously categorized with regard to their position in its vestibule and temporal priority in the binding process (4, 16, 67).

We surveyed a collection of 50 Ala single, double, and triple substitution mutations in these regions of FepA to determine their impact on H8 infectivity. Some were originally generated and analyzed on pUC plasmids carrying the fepA structural gene (pITS449 and pT944; both plasmids express wild-type FepA, but the latter is genetically engineered to introduce convenient restriction sites) (16, 67). More recent constructions were made on the low-copy-number plasmid pHSG575 (4, 89, 90). Because we did not desire to reclone all the mutants into the low-copy-number vector, we instead related H8 susceptibility to appropriate negative (RWB18-60 or KDF541) and positive (e.g., RWB18-60 or KDF541/pITS449 for pUC clones; RWB18-60 or KDF541/pITS23 for pHSG575 clones) controls and report H8 susceptibility of the mutants as a percentage of wild-type activity.

Like the effects of loop deletions, the effects of Ala substitutions on H8 susceptibility roughly fell into similar categories. However, the effects of substitution mutations on OM protein function were less severe than those of deletions, necessitating the addition of another category, IV, with <51 to 80% of wild-type susceptibility. Among 16 substitution mutations for aromatic amino acids in the loop extremities, only three produced significant reductions in phage infection efficiency (F329A, Y478A, and Y553A) (Table 4). Similarly, among 15 substitutions of Ala for Lys in the loop extremities, only two (K328A and K483A) reduced the efficiency of H8 infection, in both cases to a level that was about 10% of wild-type FepA. On the other hand, deeper within the receptor's vestibule, residues that participate in binding and/or transport of the ferric siderophore and colicins were also essential to productive interactions with the phage. Two arginines (R286 and R316) participate in the binding and transport of FeEnt (67), and these same residues were also critical to phage infection (Table 4). Likewise, an Ala substitution for Y260, deep within the B2 region, reinforced the finding that the ferric siderophore, colicins, and phage utilize common determinants in this region of the receptor protein: Y260A reduced FeEnt transport affinity 100-fold; colicin B killing, 10-fold; and H8 infectivity, 50-fold. To summarize these data, the elimination of aromatic or basic amino acids in the outermost (B1) regions of FepA, which function in the initial adsorption of FeEnt, was generally less detrimental to phage infection than the alteration of residues deep within the vestibule (B2 region), which are thought to participate in the attainment of binding equilibrium, prior to ligand uptake (Fig. 3).

The H8 genome.The morphological relationship of the H8 genome to T5 and dependence on FepA and TonB led us to determine the genomic sequence of bacteriophage H8 (NCBI accession number AC171169). The chromosome displayed a close structural and sequence relationship to that of T5, a double-stranded DNA virus in the order Caudovirales and the family Siphoviridae. T5 infection of E. coli through FhuA is TonB independent, whereas two other phage in this group, φ80 and T1, utilize FhuA in a TonB-dependent manner. Two more TonB-independent members of the family, BF23 and λ, penetrate through the OM proteins BtuB and LamB, respectively. All these phages possess long noncontractile tails and isometric or prolate capsids. Electron micrographs of H8 (Fig. 1) were consistent with this morphology. The H8 chromosome contained 104.4 kb, within which we identified 143 ORFs (Table 5), including six loci that encode tRNA (for M, I, T, G, S, and R) in the region between 20 and 25 kb. The majority of the translated proteins from the potential genes were conspicuously homologous to known or predicted proteins of other bacteriophages: predominantly T5, but also BF23 and RB16, in the family Siphoviridae and Felix 01 in Caudovirales. A total of 120 predicted H8 proteins were most closely related to homologs in T5, and overall these were 84.2% identical and 88.7% similar to proteins encoded by the T5 genome (Table 5 and Fig. 4 and 5). We also found nine potentially unique gene products in H8 that originated from ORFs at 1264, 2916, 2913, 32459, 43128, 70336, 96735, 97635, and 103573 bp on the phage chromosome, encoding proteins from 34 to 241 amino acids. The unique genes were analyzed by PSORT, which predicted five cytoplasmic and four inner membrane proteins that ranged from 4 to 27 kDa (Table 6).

• Open in new tab
• Download powerpoint

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.

• Open in new tab
• Download powerpoint

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.

Overall, the H8 genome is closely related to that of T5 (Fig. 4 and 5), which is arranged according to its life cycle. We refer the reader to the annotated T5 genome (99), which well agrees with extensive genetic data compiled about the phage over the past 50 years. Two striking features of the T5 chromosome appear again in H8: terminally redundant DNA sequences that facilitate a two-step DNA transfer mechanism and the presence of large tracts of genes that can be deleted (76, 85). We did not characterize the DNA transfer process of H8, but its chromosome contains homologous DNA to the FST genes of T5 that encode proteins and enzymes which facilitate its two-step DNA injection process. Although these appear in the H8 annotation only at the right end of the chromosome, this distribution is probably an artifact of the assembly of the sequencing data. If the H8 chromosome contains identical 6-kb DNA sequences at its extremities, as T5 does, then the shotgun sequencing approach will not differentiate them, and results from these regions will collapse together as one sequence. However, the coverage of shotgun reads from the putative repetitive region was significantly higher than from any other portion of H8 DNA, and the likely explanation is the existence of the same 6-kb sequence at both ends of the H8 chromosome. Thus, our experiments suggest the presence of terminally repetitive DNA at the ends of the H8 chromosome, exactly the same as in the T5 chromosome. Secondly, H8 contains two deletions relative to T5 that span almost 10 kb. These gaps in H8 sequence correspond to deletable portions of the T5 chromosome (23.5 to 42.5 kb) (59, 99) that encode 24 tRNAs and 35 other ORFs, including two HNH-homing endonucleases and another putative endonuclease. The dispensable nature of this region in T5 concurs with its absence in H8. In this sense H8 is analogous to a T5 deletion mutant with a novel host range. Both phages also contain numerous other gaps, relative to each other, that eliminate nonessential genes (Fig. 4 and 5).

In the absence of genetic or physiological data from the newly identified virus and in view of its close relationship to T5, we relied on sequence homologies to bacteriophage proteins as the basis of its genomic annotation (Fig. 5). Among 135 potential proteins encoded in its genome, 49 (36%) were functionally annotated, mainly as enzymes involved in phage DNA replication and repair, nucleotide metabolism, or structural proteins. Despite high overall homology between H8 and T5, their tail proteins, which initiate the infectious process, were noticeably less conserved. The overall level of identity and similarity in the tail protein region of the chromosome was 63% and 74%, respectively, which is less than the level seen in other regions (84% and 89%, respectively). One functionally important tail protein, the H8 homolog of ltf (long tail fiber protein; ORF 109), was 50% truncated relative to that of T5, and the remaining polypeptide was only 24% identical and 29% similar to the corresponding portion of Ltf. Similarly, the putative receptor binding protein of H8 that recognizes and adsorbs to FepA was most like the comparable tail protein of BF23 (27% identical and 42% similar) rather than that of T5. These variations provide a structural explanation for the different host range of bacteriophage H8.

In total, four phage/receptor systems were of most interest to our experiments: H8/FepA, T5, T1/FhuA, and BF23/BtuB. Besides the genome of H8, reported herein, full or partial genomic sequences of other LGP-specific bacteriophage are known, including those of the TonB-dependent phage T1 and the TonB-independent phages T5 and BF23. Sequence data for BF23 is incomplete, but the genes that encode some of its tail proteins are known and sequenced (62, 63). Blast analyses of H8 DNA identified ORFs at 94.7 kb and 82.8 kb that encode homologous proteins to the experimentally demonstrated receptor binding proteins of BF23 (Hrs) (48) and T5 (pb5, encoded by oad) (39, 63) and the pore-forming tail proteins of T5 (pb2) (7, 28), respectively. The receptor binding protein (encoded by rbp) of H8 was one of the least conserved in the genome, relative to its T5 homolog (26% identity; 41% similarity). Sequence comparisons between Rbp, Hrs, and pb5 revealed five regions of local identity and similarity distributed along their length. The most conserved region was at their N termini, where the first 45 residues of the three proteins were 49% identical and 91% similar (Fig. 6); the other four regions had identity and similarity scores of 17% and 81%, 21% and 89%, 33% and 71%, and 50% and 90%, respectively. Besides these conserved regions, we observed two variable regions (Fig. 6) that may function in specific adsorption of the three viruses. In the case of H8, these regions contain a preponderance of acidic residues, and one of them (residues 138 to 213) likely participates in adsorption of the phage tail to basic residues in the receptor protein (see Discussion) (67, 90).

• Open in new tab
• Download powerpoint

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.

The tail pore-forming protein of bacteriophage T5, pb2, was identified by in vitro demonstrations of its channel activity (28) and its FhuA-dependent penetration of lipid bilayers (7). We observed a homologous H8 protein that was 66% identical and 72% similar to T5 Pb2. On the basis of this high identity or homology, it is virtually certain that the H8 gene also encodes a protein that participates in pore formation and DNA transfer. Hence we named it for tail pore formation (tpf). Except in its central portion, the H8 protein was closely related to its T5 homolog (aside from the central 370 amino acids, showing 79% identity) (Fig. 6). The central region (residues 380 to 750) was 36% identical to pb2, a value that still implies a comparable structural fold. It was noteworthy that in the more variable central region, the TonB-dependent phage H8 Tpf protein contained the sequence GEGIPVGLA, which has identity and similarity to the consensus TonB box region near the N termini of siderophore receptor proteins. The analogous pb2 protein of the TonB-independent phage T5 does not contain a similar sequence; a gap occurs in the alignment of Tpf and pb2 at this site.