INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

In 1952, Zinder and Lederberg demonstrated the transfer (generalized transduction) of genetic material between Salmonella enterica serovar Typhimurium (here referred to as serovar Typhimurium) mutants involving a phage intermediary (80). The temperate phage vector, originally called PLT 22, is now commonly referred to as P22 and has continued to be the virus of choice for investigating the genetics of this bacterium.

Morphologically P22 is a member of the virus family Podoviridae, which encompasses viruses with short, noncontractile tails (1). P22 binds to the lipopolysaccharide (LPS) O side chains of serovar Typhimurium or to Escherichia coli strains expressing the serovar Typhimurium rfb cluster (46) via the virion tailspike proteins (65). The latter possess endorhamnosidase activity, which digests the O antigen, permitting passage of the phage through the LPS barrier to the surface of the outer membrane, where tight binding occurs. The linear double-stranded viral DNA enters the host cell, and circularization occurs, mediated by the phage-encoded protein Erf and the host proteins RecA and gyrase (50, 67). The resulting covalent closed and superhelical molecule is the substrate for integration into the host chromosome. The phage integration site, attP22 or ataA, maps within thrW, a gene for threonyl tRNA2 (53), with site-specific recombination being catalyzed, as it is with coliphage , by integration host factor (IHF) and integrase (Int) (40, 62). Upon induction, specialized transducing particles may arise, carrying genes adjacent to the att site, as well as a generalized transducing particle carrying only host DNA.

In the lysogenic state, P22 expresses three different systems that may interfere with superinfection by homologous phages. These are immunity conferred by the prophage repressor (c2), superinfection exclusion mediated by the sieA and sieB genes, and serotype conversion. The presence of the C2 protein represses the replication of homoimmune phage genomes, while the sie genes appear to function in preventing phage DNA injection (29, 49). Lysogenization by P22 also results in the addition of an -linked glucosyl residue to the 6 position of galactose moieties in the LPS O-antigenic tetrameric repeat. This results in a change in serotype from 4,[5],12 to 1,4,[5],12 and prevents the binding of P22 and other serovar Typhimurium phages, a phenomenon known as lysogenic conversion (35, 51). Mutational analysis by Young and his associates determined that gene a1 was responsible for the expression of antigen 1 and showed its relative position in the P22 genome, adjacent to the phage attachment site (attP) (78). Preliminary evidence suggests that this region is highly homologous to the conversion-att-int region from Shigella flexneri bacteriophage SfV (32).

Early transcription events mimic those observed in coliphage -infected cells. Transcription is initiated from two promoters, P_L and P_R, that flank the repressor (c2) gene. The early proteins are 24, a  N homologue which functions as a transcriptional antiterminator, and Cro, which functions to inhibit transcription from P_RM and generally down-regulate transcription from P_L and P_R, thereby favoring lytic development. Another early transcript is initiated from P_ant in the unique immI region, giving rise to an antirepressor, Ant, which functions to inhibit c2 repressor function. Late gene expression is regulated, as it is in coliphage , in an antitermination-dependent mechanism involving gp23, a Q homologue (48). The late genes include a holin (gp13), a lysozyme homologue (gp19), and the genes involved in morphogenesis. The last have been extensively studied (45), revealing that, unlike the situation with  phage morphogenesis, a unique scaffolding protein (gp8) is involved in the formation of a morphogenic core together with portal protein (gp1) and pilot proteins (gp16, -20, and -7). The virus surface is composed almost exclusively of a single protein (gp5). The scaffold is reutilized in subsequent rounds of capsid assembly. In contrast,  uses the product of a gene, Nu3, to play a transient role as a core or scaffold protein, which is subsequently cleaved, and the  protein coat is composed of two main proteins, gpD and gpE.

In lytic development, DNA replication is initiated from an origin (Ori) located within gene 18 (7) in a region which shows superficial similarity to that of coliphage  (gpO-gpP) with the exception that P22 contains a primase (gp18) and a helicase (gp12) (33). Replication requires additional host (55) and viral (75) proteins, leading to the formation of concatemeric molecules (15), perhaps as a result of rolling-circle replication (48). Another aspect distinguishing P22 from  is that DNA packaging in P22 proceeds from a unique site (pac) located within gene 3 on the concatemeric substrate, resulting in the head-full packaging of a limited series of terminally redundant, circularly permuted genomes. P22 packages about 43.4 kb of DNA (11) that has terminal 1.7-kb direct repeats (48) and is 5 to 8% circularly permuted. More recent molecular studies by panová indicated that the terminal redundancy is 0.9 kb (2.2%) (64). In the case of coliphage lambda, concatemeric DNA is cut by terminase at specific sites and packaged. The latter results in unique ends with cohesive extended 5' termini rather than the blunt-ended, terminally repetitious molecules observed with P22.

Many studies have suggested that P22, in spite of its morphology, is a member of the lambdoid family. The layout of its genes is very similar to that of other lambdoid phages, viable -P22 hybrids have been formed in vivo, and of the 36 known P22 genes, 23 are believed to have  analogues (44). These facts confirm the observation of Casjens et al. (19) that "an important feature of the lambdoid phage is that its structure and function are more highly conserved than are actual gene sequences" (48). Another way of looking at this group of phages is that they are a mosaic built up of modules or cassettes (47), and while conserved patterns which suggest familial relationships exist, the overall picture suggests that considerable intervirus or virus-host recombination has occurred, often between viruses infecting distant bacterial groups (26).

Largely because of the morphological difference between  and P22, the latter has been proposed recently as the type virus for a new genus which includes phages L (14), ES18 (56), LP7 (37), 34 (34), and APSE-1 (72). Schicklmaier and Schmieger used complementation and hybridization to demonstrate sequence similarity between ES18 and P22. Limited DNA sequence data from the att-int region to gene 15 has confirmed this (56). Yet even these phages are morphologically different, and the genome size of ES18 is 46.15 kb as opposed to the value of 41.8 kb for P22. This illustrates the problem of establishing relationships based upon limited data.

P22 has been extensively studied, with current emphasis on capsid morphogenesis (45, 69-71), tailspike protein-ligand interactions (59, 65), elucidation of regulatory circuits (23, 54), and transductional analysis (12, 43, 46). P22-Mu hybrid phages have been constructed carrying Mu termini and an internal fragment containing the P22 pac site (57, 77). These insert randomly into the serovar Typhimurium chromosome, package DNA adjacent to the integration site, and have proved extremely useful in chromosomal mapping. In spite of its historical and current importance, the complete genome sequence of P22 has not been reported; rather, a large number of partial sequences are to be found in GenBank. In this report, we have taken those sequences, aligned them, and extended the sequence. The similarity between P22 proteins and those of other bacteriophages was investigated in order to suggest phylogenetic relationships between different phages.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacteria, bacteriophage, and plasmid vector. The LT2 wild-type strain of serovar Typhimurium was obtained from N. L. Martin (Queen's University, Kingston, Ontario, Canada). TOP10 cells [genotype, F mcrA (mrr-hsdRMS-mcrBC) 80lacZM15 lacX74 deoR recA1 araD139 (ara-leu)7697 galU galK rpsL (Str^r) endA1 nupG] (Invitrogen) were used for the recombinant DNA techniques. Bacteriophage P22 was obtained from H.-W. Ackermann (Laval University, Québec, Canada).

Media. Bacteria were grown in Luria-Bertani broth (LB; Difco Laboratories) or on LBA plates (LB with 1.5% [wt/vol] agar). For phage titrations, 3-ml overlays were prepared using LB containing 0.6% (wt/vol) agar. The titers of the phage preparations were determined using the agar overlay technique of Adams (2).

Purification of P22. A culture of serovar Typhimurium LT2 was grown at 37°C overnight, and 5-ml samples were inoculated into four 2-liter flasks, each containing 500 ml of LB. The flasks were incubated at 37°C with shaking at 180 rpm, and the optical density at 650 nm was periodically monitored. When the optical density reached 0.25, P22 was added to a multiplicity of infection of 5. Following 6.5 h of incubation, 20 ml of chloroform was added to each flask. The phage were separated from the cell debris by centrifuging the flasks at 10,000 × g for 10 min at 4°C, and the clarified lysate was retained. The phage were precipitated using 10% (wt/vol) polyethylene glycol (76), harvested by centrifugation at 10,000 × g for 15 min at 4°C, and resuspended in 20 ml of SM buffer [5.8 g of NaCl, 2 g of MgSO₄ · 7H₂O, 50 ml of 1 M Tris · Cl (pH 7.5), 5 ml of 2% gelatin solution per liter] with 10% (vol/vol) Triton X-100 (52). Solid CsCl was added to the crude resuspended phage to a concentration of 0.5 g/ml, and the mixture was layered on a CsCl step gradient prepared as described by Sambrook et al. (52). The tubes were centrifuged at 60,000 × g at 4°C for 2 h in the Beckman L8-70 ultracentrifuge with a SW28.1 rotor. The phage were further purified using a CsCl equilibrium gradient. The material from the CsCl step gradient was added to a type 75T Beckman Quick-Seal centrifuge tube, and the volume was topped off with a CsCl solution with a density of 1.5 g/ml. The tube was sealed and centrifuged using a Type 75Ti rotor at 104,000 × g at 4°C for 24 h. The phage were then removed from the tube with a syringe.

Isolation of P22 DNA. The following were added to the dialyzed phage stock: EDTA to a final concentration of 20 mM, proteinase K (Boehringer Mannheim) to 50 µg/ml, and, sodium dodecyl sulfate to 0.5% (wt/vol). The mixture was incubated at 53°C for 1 h. The lysate was deproteinized by shaking it with phenol-chloroform-isoamyl alcohol (25:24:1 [vol/vol/vol]; Fisher Scientific) followed by centrifugation at 16,000 × g for 25 min. The aqueous layer was extracted once more with phenol-chloroform-isoamyl alcohol and then again with chloroform. The final aqueous layer was dialyzed as previously outlined.

DNA sequencing. The DNA primers used for sequencing were designed by examining the regions near the end of the contiguous sequence or near the conflict in the sequence. Potential oligonucleotide primers were analyzed for melting temperature and secondary structures using Net Primer (Premier Biosoft International), and were synthesized by Cortec DNA Service Laboratories. Fluorescent dye dideoxy chain-terminating DNA sequencing was carried out at Cortec using an Applied Biosystems 373XL automated sequencer. Primer walking, using amplification conditions optimized for sequencing lambda clones, was used to determine the sequence directly from the P22 genomic DNA.

Sequence assembly and analysis. The Applied Biosystems sequence data was collected, stripped of poor-quality data, and assembled into contigs using Seqman II (DNASTAR Inc.). Open reading frames (ORFs) were analyzed using ORF Finder at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/gorf/gorf.html) and WebGeneMark.HMM (42) (http://genemark.biology.gatech.edu/GeneMark/whmm.cgi). In addition, the Find ORF feature of SeqEdit (DNASTAR) was employed to manually scan the sequence for potential genes. A compendium of online tools (http://www.queensu.ca/micr/faculty/kropinski/online.html) was employed in the analysis of the putative genes. Proteins translated at ORF Finder or "translate tool" (http://www.expasy.ch/tools/dna.html) were scanned for homologues by using BLASTP (5, 6) against the nonredundant GenBank protein database (http://www.ncbi.nlm.nih.gov/blast/blast.cgi). Their molecular masses and isoelectric points were determined online at ProtParam tools (http://www.expasy.ch/tools/protparam.html). Where homologues were identified, the sequences were compared using Clustal W (68) at the European Molecular Biology Laboratory-European Bioinformatics Institute (http://www.ebi.ac.uk/clustalw/). In addition, ALIGN at Genestream (Institute de Génétique Humaine) at its website (http://www2.igh.cnrs.fr/bin/align-guess.cgi) was employed to compare two sequences. Proteins were also scanned against the Prosite (9, 30) and Protein Families (Pfam) (10) databases for conserved motifs at the Swiss Institute for Experimental Cancer Research ProfileScan server (http://www.ch.embnet.org/software/PFSCAN_form.html). To predict transmembrane proteins, two online programs were employed, TMPred (31) at the European Molecular Biology networkSwiss node (http://www.ch.embnet.org/software/TMPRED form.html) and TMHMM (63) at the Center for Biological Sequence Analysis at The Technical University of Denmark (http://www.cbs.dtu.dk/services/TMHMM-1.0/).

Cloning and serotype conversion. Two PCR primers (CCAAACCACTTAGCAATCAGC and AGCGCTAATTAAACCTAACAACTATGG) were designed to flank the gtrABC cassette. These were used together with Taq DNA polymerase to amplify the gtrABC genes and upstream sequence. The amplicon was ligated into the pCRII-TOPO vector and transformed into the TOP10 chemically competent cells (Invitrogen). The cells were then recovered in SOC medium (52), and after 1 h of incubation at 37°C, aliquots were plated onto LBA plates containing ampicillin (100 µg/ml) and X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) (40 µg/ml). Plasmid DNA was isolated using the alkaline lysis technique (52) and electroporated into serovar Typhimurium LT2. Ampicillin-resistant clones were tested for the ability to agglutinate in anti-O1,2,12 serum (Difco Laboratories).

Nucleotide sequence accession number. The nucleotide sequence described in this manuscript has been deposited with GenBank and has been assigned accession no. AF217253.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Sequence assembly and analysis. Twenty-four P22 sequences were retrieved from GenBank and assembled, using SeqMan, into four contigs ranging from 0.6 to 24.6 kb. These amalgamated sequences revealed 17 discrepancies. Using primer walking, we have corrected these errors, linked the contigs, and extended the assembly to the ends of the unique sequence. Sequencing from the integrase (int) gene leftward resulted in sequence that was identical to sequence derived from sequencing rightward from gene 9 (tailspike protein) (Fig. 1). This region extends to over 800 bp, the limit of our sequencing reactions, and corresponds to the terminally redundant ends of the genome. To circumvent the problems in presenting the circularly permuted, terminally redundant genome, the map (Fig. 1) was opened adjacent to a 15-bp stem-loop structure (AATAAAAATGGGTGTaaACACCCATTTTTATT [bases in the loop are shown in lowercase]) with a calculated G of 17.1 kcal/mol located downstream of the tailspike protein gene. The unique genomic sequence is 41.7-kb, which is remarkably similar to the value of 41.6 kb calculated by Chisholm and colleagues on the basis of restriction endonuclease digestion (20). The DNA has an overall moles percent GC content of 47.1, which is somewhat less than the published value of 50 (38).

View larger version (25K): [in this window] [in a new window]   FIG. 1.   ORFs and sequence similarity regions in P22 DNA. The ORFs for which no genetic designation had been made previously are labeled based upon the number of amino acid residues (e.g., ORF-80 would encode a protein with 80 amino acids). The striped and open boxes correspond to regions which have nucleic acid sequence similar to S. paratyphi A or coliphage  DNA, respectively.

Sequence similarity between P22 and other phage and Salmonella DNA. P22 DNA shares 13.5% sequence similarity with that of phage lambda DNA as shown by hybridization experiments (61). Using the BLAST2 algorithm of Tatusova and Madden, we identified five >300-bp regions in lambda DNA which shared >85% sequence identity with the P22 sequence (66). These are indicated in Fig. 1. The regions of greatest DNA sequence similarity correspond to genes ninB, ninG, and 23. This is also verified at the amino acid level.

P22 ORF analysis. Many of the P22 ORFs were previously identified (48). We reanalyzed the sequence data, updating the positions of the previously identified ORFs and making limited corrections. For example, ORF67 was previously referred to as ORF87 due to a mistake in the DNA sequence. The sequence was reanalyzed using more modern algorithms, revealing a total of 65 ORFs, 29 (45%) of which showed no similarity to those of any other protein in the GenBank protein database. We were very cautious in defining what constituted an ORF, relying on the work of previous workers or the presence of a clearly recognizable ribosome-binding site. This manuscript will concentrate on those genes that have been identified by the current authors. A complete list of P22 ORFs is displayed in Table 2.

(i) GtrA. This 363-bp ORF (45.4 mol% GC) encodes a 13.5-kDa protein with a calculated pI of 9.4. As predicted by TMHMM analysis, this small protein has four transmembrane domains, with the amino and carboxy termini of the protein arrangement on the cytoplasmic side of the inner membrane. The protein shows strong sequence homology to the serotype conversion proteins from the S. flexneri phages SfV, SfII, and SfX, which, like P22, carry out glucosylation of the O antigen. Morphologically, these phages belong to three different virus families: Podoviridae (SfV), Myoviridae (SfII), and Inoviridae (SfX). GtrA also showed homology to the products of defective prophages in S. flexneri (orf1, 77% identity) (3) and E. coli (hypothetical gene b2350, 79% identity). In addition, we have been able to identify homologues in the incomplete Salmonella genomes. P22 GtrA shares 93% sequence identity with proteins in S. typhi, serovar Typhimurium LT2, and S. paratyphi A. Guan et al. have proposed that this highly conserved group of proteins function as flippases, translocating glucosylated undecaprenyl phosphate from the cytoplasmic face to the periplasmic face of the inner membrane in gram-negative cells (25).

(ii) GtrB. This 933-bp ORF (41.8 mol% GC) encodes a protein with a mass of 35,130 Da and a pI of 8.8. TMHMM revealed two transmembrane domains in the latter two-thirds of the protein, suggesting that both the amino and carboxy termini are cytoplasmic. This protein also was found to exhibit considerable sequence similarity to proteins from Shigella phages SfII, SfV, and SfX (Table 2). In addition, it has 86% sequence identity to a hypothetical 34.6-kDa protein (YFDH_ECOLI; GenBank accession no. P77293) associated with a defective prophage in the E. coli genome. The latter protein is defined as a dolichol-phosphate mannosyl transferase (EC 2.4.1.83; also known as dolichol-phosphate mannose synthase). In S. flexneri, two proteins, GenBank accession no. AAF09026.1 and AAC39272.1, are 87% identical to P22 GtrB. MEME/MAST analyses (8) revealed additional homologues to putative sugar transferases from Synechocystis (P74505), Bacillus (YKCC_BACSU, YKNOT_BACSU), and Streptomyces (CAA20162). Based upon the analysis of the phage SfX conversion genes, GtrB is probably a bactoprenol glucosyl transferase, catalyzing the transfer of glucose from an activated nucleotide intermediate to bactoprenol phosphate (4, 25).

(iii) GtrC. This 1,458-bp ORF begins with a GTG and is preceded by a TAAGG sequence 9 bp upstream which resembles the consensus for a ribosome-binding site (TAAGGAGGT). The gene would encode a protein of 485 amino acids with a mass and pI of 55,233 Da and 8.7, respectively. Interestingly, the guanine-cytosine content, 31.9 mol%, is considerably less than that of the bulk DNA. BLASTP analysis failed to reveal any related sequences in the protein databases. Examination of the unpublished Salmonella genome sequences showed that, in each case, the gtrAB cassette was followed by a third gene encoding a large protein with multiple transmembrane domains. The product of the third gene in S. paratyphi A was shown to be a GtrC homologue exhibiting 98% sequence identity with the phage P22 protein. This suggests that either this bacterium harbors a phage closely related to P22, as suggested by the DNA similarity data, or that the gtrABC cassette originated from a defective prophage in S. paratyphi. It is worth noting that the base composition of the third gene in the conversion cassettes identified in Shigella and E. coli are all relatively low in GC content (4).

(iv) 23. The antitermination protein involved in controlling late transcription is defined by gene 23. This protein is highly homologous to a group of proteins from phage including lambda, HK022, and PS34, suggesting similar types of regulation of late-gene expression in these morphologically different viruses.

Cloning putative conversion genes. Using Martin Reese's Promoter Prediction by Neural Network program (http://www.fruitfly.org/seq_tools/promoter.html), a sequence (TTGATCGGTAACAACGATCAATTAACATGCATTA [promoter 35 and 10 consensus sequences shown in boldface and underlined]) with similarity to sigma70 promoters in E. coli was found 138 bases upstream of the gtrA gene. Using PCR, we amplified the gtrABC genes plus the putative promoter and ligated the amplicon into the pCRII-TOPO vector, which was subsequently transformed into E. coli TOP10 cells. Ampicillin-resistant clones were isolated, and the orientation of the insert relative to the lac promoter was determined by restriction endonuclease digestion. DNA from clones representing both orientations of the gtrABC cluster were electroporated into serovar Typhimurium LT2, and recombinant clones were tested for agglutination with anti-serogroup A sera. All agglutinated, confirming that (i) the amplified segment contained its own promoter and (ii) it encoded the genes necessary for complete seroconversion.

Evolutionary considerations. The phylogeny of phages has been discussed in two excellent reviews by Campbell (18) and Casjens et al. (19). Relationships have been hypothesized based on similar morphology, conservation of gene arrangement, ability to recombine, cross hybridization patterns, and sequence identity. Phage evolution can be thought of in terms of the recombinational exchange of gene modules or cassettes (19), and examination of the P22 protein data (Table 2 and Fig. 2) shows clear evidence that this type of evolutionary pathway occurred during P22 evolution. The acquisition of the xis-int-gtrA-gtrB cassette is such a case, where S. flexneri phage SfV also shares this module and similar morphology.

View larger version (30K): [in this window] [in a new window]   FIG. 2.   Diagrammatic representation of the potential phylogenetic significance of similar proteins in P22 and other phages. The outermost level indicates those proteins that share >30% sequence identity. The genes on the map line correspond to proteins which share 60% identity, while the inner arrows refer to proteins which share 90% sequence identity. A single color is employed where the P22 protein has only a single homologue. In cases where similarity is to two proteins, the arrowhead and shaft are differently colored. Where homology is shared with 3 different phage proteins, the gene is represented in grey.