Journal of Bacteriology, March 2004, p . 1503-1517, Vol . 186, No . 5

Genomic and Genetic Analysis of Bordetella Bacteriophages Encoding Reverse Transcriptase-Mediated Tropism-Switching Cassettes

Minghsun Liu,¹ Mari Gingery,¹ Sergei R . Doulatov,¹ Yichin Liu,^2, Asher Hodes,¹ Stephen Baker,³ Paul Davis,³ Mark Simmonds,³ Carol Churcher,³ Karen Mungall,³ Michael A . Quail,³ Andrew Preston,⁴ Eric T . Harvill,^1, Duncan J . Maskell,⁴ Frederick A . Eiserling,¹ Julian Parkhill,³ and Jeff F . Miller^1*

Department of Microbiology, Immunology, and Molecular Genetics, University of California, Los Angeles, Los Angeles, California 90095,¹ Department of Chemistry, Yale University, New Haven, Connecticut 06520,² The Sanger Institute, The Wellcome Trust Genome Campus, Hixton, Cambridge, United Kingdom,³ Centre for Veterinary Science, Department of Clinical Veterinary Medicine, University of Cambridge, Cambridge CB3 OES, United Kingdom⁴

Received 5 August 2003/ Accepted 3 November 2003

 
Liu et al . recently described a group of related temperate bacteriophages that infect Bordetella subspecies and undergo a unique template-dependent,reverse transcriptase-mediated tropism switching phenomenon[Liu et al., Science 295: 2091-2094, 2002] . Tropism switchingresults from the introduction of single nucleotide substitutionsat defined locations in the VR1 [variable region 1] segmentof the mtd [major tropism determinant] gene, which determinesspecificity for receptors on host bacteria . In this report,we describe the complete nucleotide sequences of the 42.5- to 42.7-kb double-stranded DNA genomes of three related phage isolates and characterize two additional regions of variability . Forty-nine coding sequences were identified . Of these coding sequences, bbp36 contained VR2 [variable region 2], which is highly dynamic and consists of a variable number of identical 19-bp repeats separated by one of three 5-bp spacers, and bpm encodes a DNA adenine methylase with unusual site specificity and a homopolymer tract that functions as a hotspot for frameshift mutations. Morphological and sequence analysis suggests that these Bordetella phage are genetic hybrids of P22 and T7 family genomes, lending further support to the idea that regions encoding protein domains, single genes, or blocks of genes are readily exchanged between bacterial and phage genomes . Bordetella bacteriophages are capable of transducing genetic markers in vitro, and by using animal models, we demonstrated that lysogenic conversion can take placein the mouse respiratory tract during infection.

 
Parasite adaptation to dynamic host characteristics is a commontheme in biology . We recently identified a unique mechanismof adaptation that governs the interactions between a groupof bacterial pathogens belonging to the Bordetella genus anda family of bacteriophages that infect them [21] . As pathogens of numerous mammalian species, Bordetella spp . undergo major changes in gene expression as they transition through their infectious cycles [9] . As part of their adaptive strategy, Bordetellaphages use a novel mechanism to evolve new ligands that allowthe use of alternative surface receptors for host cell entry.

Bordetella pertussis, Bordetella parapertussis, and Bordetellabronchiseptica are highly related, gram-negative coccobacillithat infect respiratory epithelial surfaces in humans and othermammals [25] . In response to a variety of environmental signals,these subspecies modulate virulence gene expression throughthe BvgAS signal transduction system, which controls a spectrumof gene expression states . BvgAS signaling occurs through amultistep phosphorelay involving the BvgS transmembrane sensorkinase and the BvgA response regulator [41, 42] . When the systemis active [Bvg⁺ phase], expression of virulence factors suchas adhesins, toxins, and a type III secretion system is induced.When BvgAS is inactive [Bvg^- phase], an alternative set of genesare expressed, including motility and urease genes in B . bronchisepticaand virulence-repressed genes in B . pertussis [8].

BPP-1 is a temperate bacteriophage initially found in a clinical isolate of B . bronchiseptica that displays a marked tropism for Bvg⁺ phase B . pertussis, B . parapertussis, and B . bronchiseptica[21] . The primary receptor for BPP-1 is pertactin, an outermembrane autotransporter protein that is only expressed in Bvg⁺ phase Bordetella spp . At a frequency of approximately 10^-6,BPP-1 gives rise to two classes of tropic variants . One class,designated BMP [Bvg minus-tropic phage], has an acquired tropismfor Bvg^- phase bacteria . The second class, designated BIP [Bvgindiscriminate phage], can infect both Bvg⁺ and Bvg^- phase B. bronchiseptica with equal efficiency . We showed that the tropism determinant mapped to a 134-bp sequence, VR1, located at the3' end of the mtd locus [21] . Further examination demonstratedthat VR1 undergoes site-specific sequence alterations at positionscorresponding to adenine residues in a closely related repeat,the template repeat [TR], which is located downstream of VR1 in a noncoding region.

On the basis of our initial genetic analysis, we hypothesizethat tropism switching involves the production of a TR-containingRNA intermediate followed by reverse transcription by the productof the phage-encoded reverse transcriptase [Brt] and subsequentintegration of a mutagenized cDNA copy of TR at VR1 . The mtd,VR1, TR, and brt loci comprise a novel "evolution cassette"that functions to generate diversity in ligand-receptor interactions.The extent of diversity appears to be vast, as the variabilitysystem is theoretically capable of generating nearly 10¹² polypeptide sequences at the C terminus of Mtd [21].

To better understand the biology of Bordetella phages, we obtained the complete nucleotide sequences of BPP-1, BMP-1, and BIP-1as part of the Bordetella genome sequencing project . We also carried out genetic and molecular analyses on a second regionof variability within the bpp36 locus and on a unique phage methylase encoded by bpm . We demonstrated that all phage types could be used to transduce genetic markers between differentstrains of Bordetella and, using animal models of B . bronchiseptica colonization, we determined that in vivo lysogenic conversion could take place in the respiratory tract during infection.

 
Bacterial strains, phage, plasmids, and media. Table 1 lists the bacterial strains, phages, and plasmids used in this study . B . bronchiseptica and Escherichia coli were maintainedon standard Luria-Bertani [LB] broth and LB agar as describedpreviously [2] . For allelic exchange, sucrose-sensitive cointegrantswere grown in modified LB broth containing 10% sucrose withno NaCl [LBS] . Antibiotics were routinely used at the followingconcentrations: kanamycin, 50 µg/ml; ampicillin, 100 µg/ml;streptomycin, 60 µg/ml; rifampin, 20 µg/ml; andchloramphenicol, 20 µg/ml . Bordet-Gengou [BG] agar supplemented with 7.5% defibrinated sheep blood was used for routine growth of B . bronchiseptica . When appropriate, media were supplemented with 40 µg of 5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside [X-Gal] per ml.

 
Nucleic acid manipulation. Standard methods were used for preparation of genomic and plasmidDNA, restriction enzyme digestions, agarose gel electrophoresis,DNA ligations, and other DNA manipulations [5, 35] . To determine whether noncovalent circularization through base-pairing occurs in the phage genome, PstI-digested phage DNA was heated to 65°C for 30 min and immediately cooled on ice before being loaded onto a 0.5% agarose gel . Restriction enzymes and Taq polymerase were purchased from New England Biolabs, Promega, Roche, or Stratagene and used according to the manufacturer's instructions. Bordetella phage DNA was prepared according to the Qiagen lambda DNA minipreparations from plate lysates [Qiagen], except thecolumn purification step was omitted.

Phage lysates. For B . bronchiseptica, LB broth and agar were routinely usedfor phage propagation . Plate lysates were prepared by usingthe soft-agar overlay method [1] . Briefly, 50 µl of anovernight B . bronchiseptica culture was added to 2.5 ml of 0.7%top agar kept molten at 42 to 46°C . Phage lysate was thenadded in sufficient quantity to cause confluent lysis within24 h . Phage particles were eluted from the lawn by adding 4 ml of SM buffer [35] to the plate and incubating the plate at4°C for 3 to 5 h . Resuspended lysates were passed through sterile syringe filters [Corning] [pore size, 0.2 µm] afterward . The phage titer was then determined by serial dilution with RB53 as the tester strain.

Transduction. Phage lysates from donor strains were added to 300 to 500 µlof recipient strains at a multiplicity of infection of 0.1 to0.01 and incubated at 37°C for 1 to 2 h . Cells were spun down at 12,000 x g for 5 min, washed twice in 0.85% saline,resuspended in 100 µl of 0.85% saline, and plated on BG[no blood] plates containing kanamycin . For each transduction,a control was performed by plating 100 µl of the correspondinglysate onto BG plates containing the corresponding antibiotics.

Electron microscopy. Electron microscopy was performed by one of two methods . Inthe first, concentrated phage samples were diluted 30-fold intoa volatile buffer [20 mM ammonium acetate, pH 7.4], then appliedto single-carbon support films mounted on freshly cleaved micaas follows . Carbon films were floated onto 200 µl of sample,followed by washes in 20 mM ammonium acetate, pH 7.4, then in distilled H₂O, for 2 min each . Samples were stained by floating the carbon onto 0.2% uranyl acetate for 30 s, then picking the film up onto a copper grid, and blotting off excess stain . Inthe second, carbon-coated Parlodion support films mounted ongrids were made hydrophilic immediately before use by high-voltagealternating current glow discharge . Samples were applied directlyonto grids and allowed to adhere for 2 min . The grids were rinsedwith 3 drops of distilled water, negatively stained with 1%uranyl acetate, and blotted dry with filter paper . Specimenswere examined in a Hitachi H-7000 electron microscope at anaccelerating voltage of 75 kV.

Sequence determination. Phage DNA was sonicated and size-fractionated on agarose gels.Plasmid libraries were generated in pUC18 with insert sizesof 1.4 to 2.0 kb . Each clone was sequenced once from each endwith ABI Big-Dye terminator chemistry on an ABI 3700 capillarysequencing machines . The final sequences were generated from705 sequencing reads, giving 7.4-fold total coverage [BMP-1];875 sequencing reads, giving 8.8-fold total coverage [BIP-1];and 722 sequencing reads, giving 7.0-fold total coverage [BPP-1].All repeats were bridged by clone end read-pairs or end-sequencedPCR products to confirm the assembly.

Plasmid rescue. To identify the phage integration site, a ColE1-based plasmidwith a gentamicin resistance cassette, oriT, and a 1.4-kb insertcontaining the complete cI repressor coding sequence was transferredinto lysogenized B . bronchiseptica strains ML6401 and ML6403to create two cointegrants . Since the phage genomes did notcontain any BamHI site while the vector backbone contained oneBamHI site, genomic DNA preparations from the two cointegrantswere digested with BamHI . The digested fragments were then self-ligatedand transformed into competent E . coli XL1 cells . Gentamicin-resistant transformants were recovered . Two gentamicin-resistant plasmids, pML83-B101 from the ML6401 cointegrant and pML83-B301 from theML6403 cointegrant, were chosen for further analysis.

Bacterial conjugation. All B . bronchiseptica conjugations were carried out by triparentalmatings with the mobilizing strain DH5[pRK2013] [11] . Growthof Escherichia coli donors was inhibited with plates supplementedwith streptomycin . B . bronchiseptica transformants or cointegrantswere selected on the basis of their resistance to chloramphenicol.

PstI site protection by Bpm. For expressing bpm from the broad-host-range plasmid pBBR1MCS,a PCR fragment was amplified from the BPP-1 genome with twoprimers, BpmHF [5'-AGCAAGCTTGCGCAAGCGTGGTCATCG-3'] and BpmBR[5'-AGCGGATCCCCGGTCAGATCAAATCGG-3'] . The PCR fragment was amplifiedwith Pfu Turbo [Stratagene] with pML68-16 as the template andcloned into the BamHI and HindIII sites of pBBR1MCS, downstreamof the lac promoter P_lac, to make pML93-bpm106 . To test PstIsensitivity, complementary oligonucleotides were ordered thatcontained the sequence shown in Table 3 [Invitrogen] . For eachcomplementing pair, one oligonucleotide would contain a GATCoverhang and the other would contain an AGCT overhang . The annealingwas done by mixing 20 µl of each oligonucleotide [100µM in water] and heating to 100°C for 10 min, followedby cooling in room temperature . The annealed fragments werethen cloned into the BamHI and HindIII sites of pBluescriptII KS+ . The finished constructs were then transformed by themselvesor cotransformed with pML93-bpm106 into E . coli DH5 . The plasmidswere then purified by miniprep and analyzed by restriction digestions.

 
Construction of bpm and bbp36 mutants by allelic exchange. The constructs for introducing in-frame deletions into bpm andbbp36 were made by overlap PCR with three primers each [20].For bpm, the primers were META5 [5'-AGCGGATCCCAGCCCACCATCTG-3'],DMETB3 [5'-AGCGGATCCGCGTCGGTCGTGTCC-3'], and MET-L was the linkerprimer [5'-CGCTCATCGCCATGCCGAGCGGCATCGAGCGCGAACC-3'] . The finalPCR fragment was first cloned into the EcoRI site of pBluescriptII KS+, then excised with KpnI and XbaI for cloning into pRE112to make pML83-112M3A.

For bbp36, the three primers were 36K-F [5'-AGCGGTACCATGTCCCTCGAAGCAGC-3'], 36X-R [5'-AGCTCTAGAATCGCCGCCCACGTGTC-3'], and the linker primer, 36-L [5'-CGAACTTGTGGTTGTCGTCGAGGCTCATGGCGTCAGCCCTCCATCGCC-3'].The PCR fragment was cloned into the KpnI and XbaI sites of pRE112 to make pML89-D36-6 . Genomic DNA from ML6401 was usedas the template in all reactions . The allelic exchange vectorswere introduced into B . bronchiseptica strains ML6401, ML6403,and ML6405 via triparental mating . Sucrose-sensitive, chloramphenicol-resistant cointegrants were inoculated into LB overnight and then plated on LB agar plates supplemented with 10% sucrose . The resulting colonies were then screened for sensitivity to chloramphenicoland the in-frame deletion by PCR with primers flanking the targetregion . Multiple lysogens carrying the expected mutation inphage genes were then selected for further analysis.

High-pressure liquid chromatography and mass spectrometry analyses of phage DNA nucleoside composition. We used the method described by Magrini et al . with slight modificationsto analyze phage DNA composition [24] . After hydrolysis and dephosphorylation, the free nucleosides were analyzed by reverse-phase high-pressure liquid chromatography with an analytical RP Microsorb-MV 300 Å C18 column [Varian, Walnut Creek, Calif.] on a Rainin Dynamax SD-200 solvent delivery system with a Rainin DynamaxPDA-2 diode array detector . The elution profiles were monitoredat both 215 nm and 254 nm; 50 µl of each sample was loadedonto the column in a 50 mM KH₂PO₄ [pH 5.8] solvent containing5% [vol/vol] methanol . The nucleosides were eluded at flow rateof 1 ml/min in an initial 5 to 10% methanol gradient over 20min, followed by a 10 to 65% methanol gradient over 20 min.The collected fraction containing the peak at Rt = 29 min wassubjected to electrospray mass spectrometry performed by theYale Cancer Center Mass Spectroscopy Resource and the HowardHughes Medical Institute Biopolymer Laboratory/W . M . Keck FoundationBiotechnology Resource Laboratory . Nucleoside standards [adenosine,N⁶-methyladenosine, and 5-methylcytosine dissolved in 1 mM deferoxaminemesylate-20 mM sodium acetate at pH 5] were run at the beginningand end of each set of analysis runs.

PCR conditions. Taq polymerase was used unless specified otherwise . The reactionmixture contained 1x Taq assay buffer B [Promega], 2 mM MgCl₂,5% dimethyl sulfoxide, 1 µM each primer, and 200 µMdeoxynucleoside triphosphate mix . PCR cycling conditions wereas follows: initial denaturing at 95°C for 5 min, denaturingat 94°C for 1 min, annealing at 55°C unless specifiedotherwise, and extension at 72°C for 1 min per kb of expectedPCR product . The cycle was repeated 29 more times and concludedwith a 5-min final extension step.

Animal experiments. Animals experiments were carried out as previously described[3, 12, 26] . Briefly, C57BL/6 mice and Wistar rats were obtained from Charles River Laboratories [Wilmington, Mass.] . Inocula were grown at 37°C in Stainer-Scholte broth and normalizedby optical density at 600 nm . Rats and mice lightly sedatedwith halothane were given a dose consisting of 0.5 x 10⁶ to1.0 x 10⁶ bacteria in 50 µl of phosphate-buffered saline.Colonization of the nasal cavity and a portion of the trachea[0.5 cm for mice and 1 cm for rats] was quantified by homogenizingeach tissue in 200 µl of phosphate-buffered saline, platingaliquots onto BG blood agar, and counting the colonies after2 days of incubation at 37°C.

Bioinformatics. Artemis software was used to collate data and facilitate annotation[http://www.sanger.ac.uk/Software/Artemis/] [33] . Phage DNAsequences were compared with the EMBL/GenBank entries by BlastNand BlastX [4] . Potential coding sequences were identified withcodon usage and positional base preference methods, and thepredicted protein sequences were searched against a nonredundantprotein database with WUBlastP and FastA . Inverted repeats wereidentified with the Emboss applications [http://www.uk.embnet.org/Software/EMBOSS/Apps]. Sequences from the Bordetella sequencing projects are available from the Sanger Center web site [http://www.sanger.ac.uk/Projects/Microbes/]. Bacterial signal peptides were predicted with the SignalP program [29] . The sequences and annotations have been submitted to theEMBL and GenBank databases . Motifs are described by accession numbers from the Pfam and InterPro motif databases [prefixesPF and IPR, respectively] . Rho-independent transcriptional terminators were identified with the TransTerm algorithm [10].

Nucleotide sequence accession number. The genome sequences determined here have been deposited in EMBL/GenBank under accession number AY029185.

 
Phage derivation and morphological analysis. BPP-1 was originally isolated from B . bronchiseptica RB30, arabbit strain, following UV induction . BMP-1 was derived fromBPP-1 after repeated rounds of passage on RB54 and selectionfor tropism switching . BIP-1 was isolated from B . bronchisepticaBB3464, a cat strain, also following UV induction . BPP-1 andBIP-1 readily infect B . pertussis, B . parapertussis, and B. bronchiseptica strains in the Bvg⁺ phase, while BMP-1 only forms plaques on Bvg^- phase B . bronchiseptica.

Figure 1 shows the morphology of BPP-1 . On the basis of structuralcharacteristics, it belongs to the Podoviridae family of phageswith isometric heads and short noncontractile tails, generallysimilar in appearance to phages T7 and P22 . Phage particles have an icosahedral capsid 60 nm in diameter, a short tubular tail with a decorating collar, and six tail fibers with unusual, bilobed globular ends . A remarkable overall hexagonal symmetry,akin to that of a snowflake, is prevalent in the structure ofthe capsid, the base plate, and the tail fibers . BPP-1 particlesare considerably more stable in solution and show greater infectivitythan either BMP-1 or BIP-1, presumably due to greater stabilityof pertactin-tropic Mtd . For this reason, BMP-1 and BIP-1 couldnot be sufficiently concentrated to obtain high-resolution electron microscopy images . At lower resolution, the morphologies ofthese phage were indistinguishable from that of BPP-1 [datanot shown].

 
Determination of phage DNA sequences and attachment sites [attB and attP]. The complete nucleotide sequences of BPP-1, BIP-1, and BMP-1were independently determined to be 42,493 bp, 42,638 bp, and42,663 bp, respectively . The length differences were due to a variable number of tandem repeats within the VR2 region of bbp36, and the three phage sequences are identical except for changes at VR1, VR2, two single nucleotide polymorphisms within mtd, and single nucleotide insertions and deletions within a homopolymer tract "G-string" located within bpm [see below]. The overall base composition is 65.4% GC for all three genomesand is similar to that of the host bacterium [66% GC for B. bronchiseptica].

In assembling the sequences, the lack of abrupt stops or discontinuities in the template suggested that the genomes of BPP-1, BIP-1,and BMP-1 are circular . Analysis of DNA from purified phageyielded restriction fragments corresponding in size to thosethat would be expected from a circular genetic map, and partialdenaturation failed to reveal evidence for cohesive ends [datanot shown] . Since BPP-1, BIP-1, and BMP-1 are tailed phages,the packaged genomes are likely to be linear with overlappingpermutations.

Plasmid rescue was used to clone the phage integration sites. Figure 2 shows the organization of two resulting plasmids, pML83-B102and pML83-B301, which were derived from RB50 lysogens containingBPP-1 and BIP-1, respectively . Sequence analysis indicated thatboth plasmids include one of the two junctions, attL, wherethe Bordetella phage had integrated into the B . bronchisepticagenome . The B . bronchiseptica sequence matched the region containingthe single his tRNA locus, and examination of the phage sequencerevealed that it contains a 27-bp sequence that is identicalto the 3' end of the his tRNA gene . The last 27 bp of the geneare duplicated when the Bordetella phage integrates into thegenome, and this sequence therefore comprises the attP core.As a result, the his tRNA gene is not disrupted . Since the phagegenetic map is circular, we numbered the Bordetella phage genomesequence starting with position 1 of the 27-bp attP core.

 
Identification and analysis of phage coding sequences. We identified 49 putative coding sequences by analyzing bothstrands of the phage genome for open reading frames encodingproteins 50 amino acids or longer that also contained plausibletranslational control signals . All of the predicted coding sequencesencode proteins larger than 7 kDa . Most of the predicted codingsequences are named with the prefix Bbp [for Bordetella bronchisepticaphage] . Ribosome-binding sites were identified when possible.ATG is the start codon in all except five of the predicted codingsequences . Of the five exceptions, four use GTG and one is predictedto use TTG as its start codon . Similar to lambdoid phages, thegenome is organized in a major leftward unit of expression,which includes bbp1 through bbp31, and a rightward unit of expression,containing cI through bbp50 . A stem-loop structure at the 3'end of bpp50 is predicted to form a rho-independent transcription terminator which may prevent the extension of transcriptionfrom prophage sequences into adjacent host loci [Fig . 2 and 3a] . Specific information regarding each coding sequence islisted in Table 2, and a schematic representation of the phagegenome is shown in Fig . 3 . Particularly noteworthy featuresof the genome are briefly described below.

 
[i] Tropism switching cassette region: bbp4, bbp7, brt, TR, and mtd. Bbp4 is a small predicted protein containing a region [aminoacids 53 to 72] with 45 to 60% identity to sequences in ribosomalmaturases [matK] from a number of flowering plants [e.g., Agapetesschortechinii] . This segment is part of a longer sequence patternfor type II intron maturases [Pfam PF01348] which are involvedin group II intron splicing . The proximity of bbp4 gene to brt,mtd, and the TR element raises the possibility that it is partof the phage tropism-switching module.

Brt [Bordetella reverse transcriptase] contains a region [amino acids 72 to 265] that matches the Pfam entry PF00078 rvt [reverse transcriptase] with an E-value of 3e-14 . Using a His-6-tagged derivative of Brt, we previously demonstrated that the proteindoes indeed have reverse transcriptase activity [21] . Deletion and site-directed mutagenesis experiments also showed that the reverse transcriptase domain is required for the Bordetella phage to undergo tropism switching.

As shown in Fig . 3B, located within the 278-bp intergenic regionbetween brt and bbp7 is the 134-bp TR sequence . Depending onthe particular phage isolate, TR sequences are 81 to 99% identicalto the closely linked VR1 sequence . Differences between TR andVR1 occur at positions in VR1 corresponding to adenine residues in TR, of which there are 23 . The sequence of TR is invariant, it is required for tropism switching, and synonymous substitution experiments indicate that TR acts as a template in the DNA diversity-generating process [21] . Although the predicted product of bbp7 has nosignificant matches in the database, its location is intriguing.The ATG start codon is 30 bp downstream from the stop codonof mtd, and the bbp7 stop codon lies immediately upstream ofthe beginning of TR . It is therefore possible that bbp7 playsan as yet undiscovered role in tropism switching.

The mtd gene contains the 134-bp VR1 segment that has been shown to be the receptor tropism determinant [21] . Two single nucleotidepolymorphisms located outside of VR1 in mtd were identifiedwhen the three phage genomes were compared . These two nucleotides,located at bp 605 and 652, have not been observed to undergovariation associated with tropism switching . Instead, they differedbased on which lysogenic B . bronchiseptica isolate the particularphage was derived from . Although the polymorphisms result inamino acid substitutions, they appear to have no effect on hosttropism . Preliminary results indicate that Mtd binds to phage receptors on the Bordetella cell surface [Doulatov et al., unpublisheddata], and experiments to determine the precise location ofMtd in mature phage particles are currently under way.

[ii] bbp9 through bbp21: phage structural genes. The region encompassing bbp9 to bbp21 is predicted to encode phage structural and assembly-related proteins . Three coding sequences in this region are likely to encode products withenzymatic activity . Bbp9 contains a region that is similar tothe E . coli eliminase, ElmA, which depolymerizes capsular polysaccharide[18] and Bbp11 contains a soluble lytic transglycosylase motif,which is commonly found in phage structural proteins with murinehydrolytic activity, and they appear to facilitate penetrationof the peptidoglycan layer during cell entry [19, 34] . Bbp18is highly similar over its central region [amino acids 67 to 172] to the corresponding segment of Salmonella enterica serovar Typhimurium phage LT2 endoprotease . Phage proteases are typically involved in cleavage of structural proteins during assembly, suggesting that Bbp18 may be a phage assembly-related protease. Although most expected structural components of BPP-1 and itsfamily members are difficult to predict on the basis of sequencesimilarity alone, Bbp12 displays weak similarity to tail fiberproteins from other tailed phages, and Bbp21 is predicted toencode the head-tail connector.

[iii] bbp25 and bbp26: DNA packaging. Bbp25 is similar over the central region to terminase-like proteinsin Mesorhizobium loti and the archaeon Methanosarcina acetivorans strain C2A . It is also similar over a shorter region to large terminase subunits from two archaeophages, psiM2 [Methanobacterium, E = 1e-06] and psiM100 [Methanothermobacter, E = 1e-06] . There is also a predicted ATP/GTP binding site motif near the N terminus, characteristic of large phage terminases . Bbp26 is most similar over its central region [amino acids 26 to 161] to the centralregion [amino acids 11 to 132] of enterobacteriophage HK620small terminase subunit [140 amino acids; E = 3e-05] . bbp25and bbp26 are therefore likely to encode the large and smallterminase subunits, respectively, which form part of the DNApackaging machinery.

[iv] bbp29 and bpm. Bbp29 appears to be a two-domain protein involved in DNA replication.The N-terminal half [amino acids 21 to 316] has greatest similarityto RepA from the cyanobacterial Synechococcus sp . strain PCC7942plasmid pUH24, which encodes an essential replication protein[43] . The C-terminal half is highly similar to primase and helicase proteins from a number of phages and contains an ATP/GTP binding motif.

Bpm [for Bordetella phage methylase] is highly similar to a number of methylases from bacteria, archaea, and viruses . Themost similar proteins in the database are site-specific DNA methyltransferases from two Xanthomonas species . In addition, highly similar proteins are found in the Streptomyces coelicolor genome, the F plasmid of E . coli K-12, Yersinia pestis plasmidpMT1, E . coli virulence plasmid pO157 [E < 7e-20 for all],and many other bacteria and plasmids . Related proteins are foundin numerous bacteriophages and archaeophages, the most similar of which is an adenine methyltransferase from an archaeal halophilic virus, Ch1 [E = 3e-14] . Bpm contains several motifs, includinga DNA methylase N-4/N-6 family signature, an S-adenosyl-L-methionine binding domain motif, and two motifs shared by all adenine methylases, an N-terminal Asp-Pro-Pro-Tyr motif and a C-terminal Phe-X-Gly-X-Gly [FXGXG] sequence [15, 40] . One common feature of this familyof methylases is that they are usually not part of restriction-modificationsystems . In some species, such as Sinorhizobium meliloti andCaulobacter crescentus, they are essential for viability [47].A functional analysis of Bpm is described below.

[v] bbp31 and the cI lysis repressor. bbp31 is located directly adjacent to the cI repressor homologand is transcribed in the opposite direction [Fig . 3C] . Bbp31 is highly similar to phage APSE-1 protein P2, a Cro repressor homolog, and it contains a predicted helix-turn-helix DNA-binding motif . cI is a homolog of lambda cI-like repressors from a variety of phages, including 434, P22, HK97, and lambda, and it also contains a predicted helix-turn-helix DNA-binding motif . Totest the predicted repressor function of the protein, we constructedin-frame deletions in the cI loci of BPP-1 and BIP-1 . Plaquesproduced by the cI mutants were less turbid than wild-type plaques,and cI mutants were unable to form lysogens . Expression of thecI gene alone on a broad-host-range vector, pMMB207, in sensitiveB . bronchiseptica strains was sufficient to confer resistanceto phage lysis [data not shown] . These observations are consistentwith the hypothesis that cI is the lysis repressor and together point to cI and Bbp31 as controlling the lysis-lysogeny switch.

[vi] bbp36. Bbp36 has sequence similarity to ice nucleation proteins ofseveral bacterial plant pathogens, such as Xanthomonas campestris,Pseudomonas syringae, and Erwinia uredovora [36, 46, 48] . Allmembers of this class of proteins, including Bbp36, containimperfect repeats of a consensus octapeptide . The nonrepetitiveN-terminal regions of the Bbp36 protein and the ice nucleationproteins show the highest similarity, but they are also similarat the nonrepetitive portions of their C termini . SignalP analysissuggests that Bbp36 carries a signal peptide . This raises thepossibility that, like ice nucleation proteins, Bbp36 may beexported to the cell surface during the lysogenic phase . Thebbp36 gene contains the second major region of variability,VR2, which is described in detail below.

[vii] bbp42 and bbp47. bbp42 is predicted to encode a multidomain DNA polymerase withhigh sequence similarities to DNA polymerases from numerousbacteria and phages . Amino acid motifs found in Bbp42 includea 3'-5' exonuclease motif, a class II aminotransferase motif,a DNA-directed DNA polymerase domain, and an N-6 adenine-specificDNA methylase segment . Bbp47 is highly similar to a number ofhelicase-like proteins found in phages, bacteria, archaea, andeukaryotes . It contains a DEAD/DEAH box helicase motif and apredicted ATP/GTP binding site . The Bbp42 and Bbp47 proteinsare likely to constitute the phage DNA replication machinery.

[viii] bbp49 and bbp50: lysogeny genes. The last functional module in the right arm is predicted toencode two proteins involved in excision and integration . Bbp49displays weak sequence similarity to several phage excisionases,and Bbp50 is highly similar to numerous integrase proteins.Bbp50 contains a phage integrase motif of the tyrosine site-specificrecombinase family.

Syntenic regions. As shown in Fig . 4, several phage and prophage genomes thatcontain regions with similar coding sequences in the same orderas in the Bordetella phage genome were identified, implyingevolutionary relatedness . On the left arm, a short region ofpartial synteny was found with a 30-kb unstable genetic elementin Legionella pneumophila which is apparently of phage originand is responsible for phase-variable expression of a virulence-associatedlipopolysaccharide [23] . The right arm of the phage genome displayspartial synteny with nine loci encoded by APSE-1, a Podoviridaephage that infects a secondary endosymbiont of the pea aphidAcyrthosiphon pisum [44] . This genomic similarity includes thedivergently expressed bbp31 and cI loci . Partially overlappingsyntenic regions were also found with three cryptic prophagesin Xylella fastidiosa 9a5c, a bacterial citrus pathogen [37],and Staphylococcus aureus phage 12.

 
Variable region 2. Sequence comparisons between BPP-1, BIP-1, and BMP-1 revealeda striking heterogeneity near the 3' end of bbp36 . This region,which was designated VR2, consists of a series of identical19-bp repeats separated by one of three 5-bp spacers [Fig . 5A].To investigate a possible relationship between VR2 and tropismswitching, we determined the VR2 sequences of multiple phageisolates which were derived from different parents and representall three host tropism types [Fig. 5B] . Each VR2 contained avariable number [n] of the 19-bp cassettes separated by n -1 spacers . Although a diversity of patterns were observed, inno instance was the bbp36 reading frame altered by variationin VR2.

 
As shown in Fig . 5B, in several cases we observed identical VR2 sequences in phages with different host tropisms . This was in contrast to previous results with VR1 [21] . Given the lack of correspondence between VR2 variability and tropism switching, we introduced an in-frame deletion into bbp36 to directly measure the effects on phage infectivity, specificity, or tropism switching. BPP-1bbp36 and BMP-1bbp36 produced viable phages that retainedtheir parental specificity and were fully capable of tropismswitching [Table 3] . Although VR2 varies at high frequency,its function is not related to host specificity . The patternof VR2 variability can be accounted for by a slipped-strandmispairing mechanism . Slipped-strand mispairing occurs as aresult of slippage of DNA polymerase during replication of highlyrepetitive templates, resulting in occasional insertion or deletionof repeat units [6] . Sampling of randomly selected BPP-1 progenysuggested that at least 10% carried a different VR2 sequence.This high frequency is remarkable, considering the apparentlack of selective pressure for bbp36 function or variability.

Bpm encodes a DNA adenine methylase with novel site specificity. Examination of the bpm gene from BPP-1, BIP-1, and BMP-1 revealed a variable stretch of G residues located 13 bp upstream of the highly conserved FXGXG motif . The BPP-1 sequence contained eightG's, BIP-1 contained nine, and BMP-1 contained ten . In bothBIP-1 and BMP-1, the additional guanosine residues result inframeshift mutations . Homopolymer tracts such as the G-stringsequence in bpm are associated with an increased frequency offrameshift mutations [20, 39] and are sometimes used as mechanismsto promote phase variability [14] . The fact that both BMP-1 and BIP-1 contained frameshift mutations suggested that they occur frequently during routine passage and/or are associatedwith tropism switching.

The first hint that bpm may encode a functional methylase came from analyzing the Bordetella phage genome by restriction endonucleasedigestion . We found that BPP-1 DNA was resistant to PstI, whileBMP-1 and BIP-1 DNA was not . To determine if protection fromPstI digestion correlated with expression of bpm, an in-framedeletion was introduced into the bpm locus in BPP-1 . Phage DNApurified from the bpm mutant was no longer resistant to PstIdigestion [data not shown] . DNA purified from BPP-1 and BIP-1phage particles was subjected to hydrolysis and dephosphorylationto produce nucleosides for analysis by reversed-phase high-pressureliquid chromatography . As shown in Fig . 6, a peak at R_t = 29 min, which corresponds to the expected peak for N⁶-methyladenine, was present in BPP-1 DNA and greatly diminished in DNA prepared from BIP-1 . No peak corresponding to 5-methylcytosine was detectedin either sample . The peak corresponding to N⁶-methyladenine was collected and subjected to electrospray mass spectroscopy analysis . The result confirmed the identity of the peak as N⁶-methyladenine [expected mass, 265.12; actual mass, 266.14] . Taken together, the results support the conclusion that Bpm is a DNA adenine methylase.

 
An EcoRI fragment from BPP-1 containing the bpm coding sequence plus 8 kb of surrounding sequence was cloned into pUC19 to obtain pML68-16 . Of the four PstI sites [three in the insert and one in the multiple cloning site of the plasmid], only the siteon the vector backbone was susceptible to PstI digestion . This raised the possibility that Bpm does not methylate all PstI sites . Cotransformation of pML68-16 with several other plasmids bearing PstI sites from a variety of sources further corroborated the observation that only a subset of available PstI sites were protected by Bpm . Since all six PstI sites on the Bordetella phage genome were apparently methylated, sequences surrounding these sites were compared [Table 4] . The alignment, centered around each PstI site, revealed additional sequence features that were shared by all of the protected sites.

 
A series of constructs containing modified sequences were madewith synthetic oligonucleotides, and plasmids containing themwere cotransformed with a broad-host-range plasmid expressingbpm . The constructs were purified and assayed for resistanceto PstI digestion . The results are shown in Table 4 and are consistent with the conclusion that in addition to the coreCTGCAG sequence, the AG dinucleotide located 6 bp upstream ofthe PstI site is also required for adenine methylation . MostDNA methylase enzymes identified to date have recognition sequencesthat are contiguous and range from 4 to 8 bp [http://rebase.neb.com]. Bpm, on the other hand, is not only a relatively small DNA methylase [predicted size, 27.4 kDa] but it also appears to have a nonpalindromic recognition site that stretches over 14 bp . Three of the six CTGCAG sites on the Bordetella phage genome contain the required AG dinucleotide on only one strand, raising the possibility that these PstI-resistant sites are hemimethylated.

The BPP-1bpm mutant did not reveal any qualitative or quantitativedefect in plaquing or tropism switching compared to wild-typeBPP-1 [Table 3] . Analysis of the available B . pertussis, B. parapertussis, and B . bronchiseptica genomic sequences suggests the presence of several restriction-modification systems [http://www.sanger.ac.uk/Projects/Microbes/]. One hypothesis is that bpm protects phage DNA from host restriction.The increased specificity conferred by the AG dinucleotide wouldresult in modification of only a subset of host PstI sites.

Generalized transduction. To facilitate genetic analysis of Bordetella subspecies, wetested the ability of BPP-1cI to serve as a generalized transducingphage . Two B . bronchiseptica mTn5-lacZ1 [Km^r] transposon mutantsin Bvg-regulated genes were used as donor strains for theseexperiments . One strain carried a transposon insertion in fhaB,the structural gene for filamentous hemagglutinin, which isexpressed in the Bvg⁺ phase . The other strain carried a transposon insertion in wbmD, which is part of the lipopolysaccharide biosynthetic locus and is expressed in the Bvg^- phase . Wild-type B . bronchisepticaRB50 was used as the recipient strain . The overall transductionfrequency was approximately 10^-7 transductions per PFU . Furthermore,it was confirmed that the ß-galactosidase activityof RB50-derived transductants grown under Bvg⁺ or Bvg^- conditionsmatched well with those measured in the donor strains [datanot shown] . Similar results were obtained with TnphoA B . pertussisdonor strains and B . pertussis 18323 as the recipient.

Since the fhaB locus is proximal to bvgAS, we tested for cotransductionof Km^r and bvgAS markers from RB54 [bvgS] to RB50 with BIP-1cI and vice versa . Approximately 88% of the Km^r transductants were also Bvg⁺ when RB54 was the recipient and 80% of the transductantswere Bvg^- when RB50 was the recipient . These experiments demonstratethat BPP-1 and BIP-1 can be used as tools for generalized transduction.

In vivo lysogenic conversion. Since BPP-1 uses the Bvg⁺ phase protein pertactin as a receptor,we tested whether in vivo lysogenic conversion could occur inthe mouse respiratory tract . Equal numbers of RB30 [lysogenicfor BPP-1] and RB50 marked with gentamicin resistance [RB50Gm]were coinoculated [5.5 x 10⁴ CFU/animal] . Bordetella organismswere recovered from the nasal septum and the trachea of eachanimal 25 days postinoculation . Gentamicin-resistant and -sensitivecolonies were counted, and the recovered RB50Gm colonies werecharacterized to determine if they had acquired resistance to BPP-1 [Fig . 7] . To confirm that BPP-1 resistance was due tolysogeny, the presence of phage sequences was verified either by PCR detection of the cI gene or by production of infectious phage particles . In four of five mice, there were detectable levels of RB50Gm lysogenized with BPP-1, indicating that invivo transmission of BPP-1 could indeed take place during respiratory tract infection, as has been demonstrated for other phages,such as CTX [45] . In several animals, recovery of RB50 was lower than that of RB30, possibly due to phage killing.

 
BPP-1, BIP-1, and BMP-1 are among the first phages that infect bacteria from the beta subdivision of the proteobacteria tobe completely sequenced . Our analysis indicates that BPP-1 isa novel genomic hybrid, combining characteristics of lambda-likegenome organization with the presence of T7-like structuralgenes [Table 2, see below] . The overall gene organization is modular and shows a high degree of mosaicism, as demonstratedby multiple segments with similarity to genes from diverse bacteriaand phages . The most notable feature of these phages is thepresence of a unique, template-dependent, reverse transcriptase-mediated tropism-switching mechanism encoded in an "evolution cassette"on the left arm of the phage genome . Preliminary results suggestthat similar modules exist in other phage and bacterial genomes[S . Doulatov et al., unpublished data].

The tail and capsid morphology of BPP-1 groups it with Podoviridae according to the classification used by the International Committee on the Taxonomy of Viruses . The International Committee on the Taxonomy of Viruses phage classifications, based primarily ontail morphology, have recently been questioned due to the lackof correlation with genome characteristics and evolutionaryrelatedness [31] . Based on genome and proteome features, it was suggested that the Podoviridae are more accurately segregated into several groups, in which short-tailed P22 clusters with long-tailed lambdoid phages due to their genetic similarities,while short-tailed T7-like phages form a separate group . Thereis no known genetic relationship between P22 and T7, which haveentirely different lifestyles [temperate versus lytic] and transcriptional control mechanisms.

The organization of the BPP-1 genome is distinctly lambdoid,with two major clusters [left and right arms] that differ accordingto the direction of transcription . Structural and assembly proteinsappear to be encoded on the left arm and DNA metabolism functionson the right, demarcated by a lambda-like divergent expressionregion encoding cI and Cro homologs . This contrasts with theunidirectional organization of genes in T7-like phages . Similaritiesin genome organization are found with phage APSE-1 . Like BPP-1,APSE-1 is a short-tailed phage with a lambda-like genome [44]. Synteny between regions of the BPP-1 and APSE-1 genomes, most strikingly in the segment containing bbp38 through bbp47, is detectable even at the DNA level and indicates their close relationship. This is remarkable considering that their bacterial hosts occupy very different niches, the mammalian respiratory tract for B. bronchiseptica versus intercellular and intracellular locations within the pea aphid A . pisum for the endosymbiotic host of APSE-1 [44] . Furthermore, the APSE-1 host is a member of theEnterobacteriaceae, which is phylogenetically distant from thebordetellae . The same syntenic region shows similarities to several other phage genomes [Fig . 4] . This implies a commonancestry for these phages and suggests that they have lambda-likemechanisms of DNA metabolism.

Structural features of BPP-1 indicate commonalities with T7-like phages . The capsid diameter is identical to that of T7 [60 nm[13]], but larger than APSE-1 [45 to 55 nm [44]] . The BPP-1tail is also similar in shape to the tail of T7 [Fig . 1] . BPP-1 has some structural features that are absent from T7, most notably the bilobed, globular structures at the tips of the tail fibers. However, similar tail fiber ends have been described for some capsule-specific T7-family members [e.g., E . coli strain K-235 1.2 [16] and Klebsiella phage K11 [32]] . These tail fibers have capsule-lytic hydrolase activity, most likely localized at thetips . A similar function may also reside in the globular endsof BPP-1 tail fibers, perhaps involving the bbp9 gene product.

Several polypeptides with similarity to T7-like phage proteinsare predicted to be encoded in the BPP-1 structural gene region.The position of bbp11 in the genome and its murein transglycosylase sequence motif suggest significant homology with transglycosylases in T7-like phages, which participate in creating a passage through the peptidoglycan layer to allow DNA entry during infection[19] . Sequence similarities between Bbp21 and head-tail connectorproteins from several T7-like phages provide further evidencefor conservation of structural features . Sequence similaritycould indicate regions of Bbp21 that interact with other conservedstructural proteins and/or with DNA, since the connector isthe portal for DNA . Interestingly, the most highly conservedsequences in BPP-1 are those predicted to encode proteins thatinteract with DNA [i.e., helicase, DNA polymerase, methylase,cI and Cro repressors, integrase, large and small terminases,head-tail connector].

The majority of Bordetella phage proteins predicted by our analysis lack strong similarities to proteins in the GenBank database. One possible explanation is that relatively few phages thatinfect bacteria that are phylogenetically related to Bordetellahave been analyzed in detail . The hybrid architecture of theBPP-1 genome supports emerging views of bacteriophage phylogenyand evolution [30] . Phages such as BPP-1, P22, and APSE-1, witha lambda-like genome and a short-tail structural gene cassette,suggest a "braided" rather than a vertical lineage for tailedphages . These hybrids support the idea that regions encodingprotein domains, single genes, or blocks of genes are readilyexchanged between bacterial and phage genomes . The likelihoodthat more hybrid phage genomes exist suggests that segregationof characteristics is not as limited as previously thought,and a combinatorial continuum of variety may exist among phages.

Perhaps the most remarkable characteristic of the Bordetella phages analyzed here is their propensity to undergo targeted DNA sequence variation . VR1, as part of a larger "diversity generating cassette," allows the phages to undergo host tropism switching [21] . This ability has an obvious evolutionary advantage,as it confers an expanded host range . VR2 appears to undergoa significantly different type of variation, likely mediated by slipped-strand mispairing . Although the advantage conferred by VR2 variability remains to be determined, it is intriguingthat the product is a predicted secreted protein with similaritiesto surface proteins on other gram-negative bacteria . Finally,the homopolymeric tract in bpm causes inactivation of the BpmDNA adenine methylase upon acquisition of frameshift mutations,which also appear to occur at high frequency . Neither the functionalrole of the Bpm methylase nor the significance of phase variationis currently known.

The sequence analysis reported here, along with previous studies[21], suggests numerous applications for BPP-1 derivatives,gene products, and genetic elements . Phage-encoded proteins,including holins and lysins, have recently been used as effectiveantimicrobial agents [7, 22, 28], and several products [Bbp9, Bbp11, and Bbp18] encoded in the Bordetella phage genome are predicted to have antimicrobial activities . The completed sequences allowed the construction of cI mutants, which can be used asgeneralized transducing phages for transferring markers betweenB . pertussis, B . parapertussis, and B . bronchiseptica . The identificationof attachment sites and integration genes could facilitate thedevelopment of single-copy genomic integration vectors for theBordetella genus and possibly other related bacteria, and theunusual site specificity of Bpm may be useful for introducingstrand-specific methylation of adenine residues at defined locations.

Perhaps the most interesting potential applications of these Bordetella phages derive from their ability to switch tropism. This could, for example, provide a significant advantage fortheir use in phage therapy [38] . Bordetella infections are confinedto respiratory epithelial surfaces, which should be accessibleto therapeutically administered phages . Phage variants arisingvia the tropism-switching mechanism encoded on the left arm of the genome could potentially overcome mutations in receptor proteins that would otherwise confer resistance to infection. Finally, further characterization of Brt, TR, and other cis- and trans-acting elements that promote variability in VR1 could lead to the development of novel genetic systems for evolving desired functional attributes in heterologous proteins of interest.

 
We thank members of the J . F . Miller laboratory for constructive input throughout the course of this project.

M.L . was supported by a research fellowship from the AmericanLung Association and training grant GM-08042 to the UCLA-CalTechMedical Scientist Training Program from the NIH . A.H . is a predoctoral trainee recipient of Microbial Pathogenesis Training Grant 2-T32-AI-07323. This work was supported by NIH grant AI38417 [J.F.M.] . The sequencing of the BPP-1, BMP-1, and BIP-1 genomes was supported by the Wellcome Trust.

 
* Corresponding author . Mailing address: Department of Microbiology, Immunology and Molecular Genetics, 10833 Le Conte Ave., UCLA School of Medicine, Los Angeles, CA 90095 . Phone: [310] 206-7926 . Fax: [310] 267-2774 . E-mail: jfmiller@ucla.edu.

Present address: Center for Neurologic Diseases, Brigham andWomen's Hospital and Department of Neurology, Harvard MedicalSchool, Cambridge, MA 02139.

Present address: Department of Veterinary Science, The Pennsylvania State University, University Park, PA 16802.

1. Adams, M . H. 1959 . Bacteriophages . Interscience Publishers, Inc., New York, N.Y.
2. Akerley, B . J., P . A . Cotter, and J . F . Miller. 1995 . Ectopic expression of the flagellar regulon alters development of the Bordetella-host interaction . Cell 80:611-620.
3. Akerley, B . J., D . M . Monack, S . Falkow, and J . F . Miller. 1992 . The bvgAS locus negatively controls motility and synthesis of flagella in Bordetella bronchiseptica . J . Bacteriol . 174:980-990.
4. Altschul, S . F., T . L . Madden, A . A . Schaffer, J . Zhang, Z . Zhang, W . Miller, and D . J . Lipman. 1997 . Gapped Blast and PSI-Blast: a new generation of protein database search programs . Nucleic Acids Res . 25:3389-3402 .
5. Ausubel, F . M., R . Brent, R . E . Kingston, and D . D . Moore. 1995 . Short protocols in molecular biology . John Wiley & Sons, Inc., New York, N.Y.
6. Belkum, A., S . Scherer, L . Alphen, and H . Verbrugh. 1998 . Short-sequence DNA repeats in prokaryotic genomes . Microbiol . Mol . Biol . Rev. 62:275-293 .
7. Bernhardt, T . G., I . N . Wang, D . K . Struck, and R . Young. 2001 . A protein antibiotic in the phage Qbeta virion: diversity in lysis targets . Science 292:2326-2329 .
8. Cotter, P . A., and V . J . DiRita. 2000 . Bacterial virulence gene regulation: an evolutionary perspective . Annu . Rev . Microbiol. 54:519-565.
9. Cotter, P . A., and J . F . Miller. 2000 . Bordetella, p . 619-674 . In E . Groisman [ed.], Principles of bacterial pathogenesis. Academic Press, San Diego, Calif.
10. Ermolaeva, M . D., H . G . Khalak, O . White, H . O . Smith, and S . L . Salzberg. 2000 . Prediction of transcription terminators in bacterial genomes . J . Mol . Biol . 301:27-33.
11. Figurski, D . H., and D . R . Helinski. 1979 . Replication of an origin-containing derivative of plasmid RK2 dependent on a plasmid function provided in trans . Proc . Natl . Acad . Sci . USA 76:1648-1652.
12. Harvill, E . T., P . A . Cotter, M . H . Yuk, and J . F . Miller. 1999 . Probing the function of Bordetella bronchiseptica adenylate cyclase toxin by manipulating host immunity . Infect Immun . 67:1493-1500 .
13. Hausmann, R., and M . Messerschmid. 1988 . Inhibition of gene expression of T7-related phages by prophage P1 . Mol . Gen . Genet . 212:543-547.
14. Henderson, I . R., P . Owen, and J . P . Nataro. 1999 . Molecular switches—the ON and OFF of bacterial phase variation . Mol . Microbiol . 33:919-932.
15. Kaszubska, W., C . Aiken, C . D . O'Connor, and R . I . Gumport. 1989 . Purification, cloning and sequence analysis of RsrI DNA methyltransferase: lack of homology between two enzymes, RsrI and EcoRI, that methylate the same nucleotide in identical recognition sequences . Nucleic Acids Res . 17:10403-10425.
16. Kwiatkowski, B., B . Boschek, H . Thiele, and S . Stirm. 1982 . Endo-N-acetylneuraminidase associated with bacteriophage particles . J . Virol . 43:697-704.
17. Landt, O., H . P . Grunert, and U . Hahn. 1990 . A general method for rapid site-directed mutagenesis with the polymerase chain reaction . Gene 96:125-128.
18. Legoux, R., P . Lelong, C . Jourde, C . Feuillerat, J . Capdevielle, V . Sure, et al. 1996 . N-Acetyl-heparosan lyase of Escherichia coli K5: gene cloning and expression . J . Bacteriol . 178:7260-7264.
19. Lehnherr, H., A . M . Hansen, and T . Ilyina. 1998 . Penetration of the bacterial cell wall: a family of lytic transglycosylases in bacteriophages and conjugative plasmids . Mol . Microbiol . 30:454-457.
20. Levinson, G., and G . A . Gutman. 1987 . Slipped-strand mispairing: a major mechanism for DNA sequence evolution . Mol . Biol . Evol . 4:203-221.
21. Liu, M., R . Deora, S . R . Doulatov, M . Gingery, F . A . Eiserling, A . Preston, J . Duncan, R . W . Simons, P . A . Cotter, J . Parkhill, and J . F . Miller. 2002 . Reverse transcriptase-mediated tropism switching in Bordetella bacteriophage . Science 295:2091-2094 .
22. Loeffler, J . M., D . Nelson, and V . A . Fischetti. 2001 . Rapid killing of Streptococcus pneumoniae with a bacteriophage cell wall hydrolase . Science 294:2170-2172 .
23. Luneberg, E., B . Mayer, N . Daryab, O . Kooistra, U . Zahringer, M . Rohde, et al. 2001 . Chromosomal insertion and excision of a 30 kb unstable genetic element is responsible for phase variation of lipopolysaccharide and other virulence determinants in Legionella pneumophila . Mol . Microbiol . 39:1259-1271.
24. Magrini, V., D . Salmi, D . Thomas, S . K . Herbert, P . L . Hartzell, and P . Youderian. 1997 . Temperate Myxococcus xanthus phage Mx8 encodes a DNA adenine methylase, Mox . J . Bacteriol . 179:4254-4263.
25. Mattoo, S., A . K . Foreman-Wykert, P . A . Cotter, and J . F . Miller. 2001 . Mechanisms of Bordetella pathogenesis . Front . Biosci . 6:E168-E186.
26. Mattoo, S., J . F . Miller, and P . A . Cotter. 2000 . Role of Bordetella bronchiseptica fimbriae in tracheal colonization and development of a humoral immune response . Infect . Immun . 68:2024-2033 .
27. Moak, M., and I . J . Molineux. 2000 . Role of the Gp16 lytic transglycosylase motif in bacteriophage T7 virions at the initiation of infection . Mol . Microbiol . 37:345-355.
28. Nelson, D., L . Loomis, and V . A . Fischetti. 2001 . Prevention and elimination of upper respiratory colonization of mice by group A streptococci by with a bacteriophage lytic enzyme . Proc . Natl . Acad . Sci . USA 98:4107-4112 .
29. Nielsen, H., J . Engelbrecht, S . Brunak, and G . von Heijne. 1997 . Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites . Protein Eng . 10:1-6.
30. Pedulla, M . L., M . E . Ford, J . M . Houtz, T . Karthikeyan, C . Wadsworth, J . A . Lewis, D . Jacobs-Sera, J . Falbo, J . Gross, N . R . Pannunzio, W . Brucker, V . Kumer, J . Kandasamy, L . Keenan, S . Bardarov, J . Kriakov, J . G . Lawrence, W . R . J . Jacobs, R . W . Hendrix, and G . F . Hatfull. 2003 . Origins of highly mosaic mycobacteriophage genomes . Cell 113:171-182.
31. Rohwer, F., and R . Edwards. 2002 . The Phage Proteomic Tree: a genome-based taxonomy for phage . J . Bacteriol . 184:4529-4535 .
32. Rudolph, C., E . Freund-Molbert, and S . Stirm. 1975 . Fragments of Klebsiella bacteriophage no . 11 . Virology 64:236-246.
33. Rutherford, K., J . Parkhill, J . Crook, T . Horsnell, P . Rice, M . A . Rajandream, and B . Barrell. 2000 . Artemis: sequence visualization and annotation . Bioinformatics 16:944-945.
34. Rydman, P . S., and D . H . Bamford. 2000 . Bacteriophage PRD1 DNA entry uses a viral membrane-associated transglycosylase activity . Mol . Microbiol . 37:356-363.
35. Sambrook, J., T . Maniatis, and E . F . Fritsch. 1989 . Molecular cloning: a laboratory manual . Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
36. Schmid, D., D . Pridmore, G . Capitani, R . Battistutta, J . R . Neeser, and A . Jann. 1997 . Molecular organisation of the ice nucleation protein InaV from Pseudomonas syringae . FEBS Lett . 414:590-594.
37. Simpson, A . J., F . C . Reinach, P . Arruda, F . A . Abreu, M . Acencio, R . Alvarenga, et al. 2000 . The genome sequence of the plant pathogen Xylella fastidiosa . The Xylella fastidiosa Consortium of the Organization for Nucleotide Sequencing and Analysis . Nature 406:151-157.
38. Summers, W . C. 2001 . Bacteriophage therapy . Annu . Rev . Microbiol . 55:437-451.
39. Tautz, D., M . Trick, and G . A . Dover. 1986 . Cryptic simplicity in DNA is a major source of genetic variation . Nature 322:652-656.
40. Trautner, T . A., B . Pawlek, U . Gunthert, U . Canosi, S . Jentsch, and M . Freund. 1980 . Restriction and modification in Bacillus subtilis: identification of a gene in the temperate phage SP beta coding for a BsuR specific modification methyltransferase . Mol . Gen . Genet . 180:361-367.
41. Uhl, M . A., and J . F . Miller. 1996a . Central role of the BvgS receiver as a phosphorylated intermediate in a complex two-component phosphorelay . J . Biol . Chem . 271:33176-33180 .
42. Uhl, M . A., and J . F . Miller. 1996b . Integration of multiple domains in a two-component sensor protein: the Bordetella pertussis BvgAS phosphorelay . EMBO J . 15:1028-1036.
43. Van der Plas, J., R . Oosterhoff-Teertstra, M . Borrias, and P . Weisbeek. 1992 . Identification of replication and stability functions in the complete nucleotide sequence of plasmid pUH24 from the cyanobacterium Synechococcus sp . PCC 7942 . Mol . Microbiol . 6:653-664.
44. van der Wilk, F., A . M . Dullemans, M . Verbeek, and J . F . van den Heuvel. 1999 . Isolation and characterization of APSE-1, a bacteriophage infecting the secondary endosymbiont of Acyrthosiphon pisum . Virology 262:104-113.
45. Waldor, M . K., and J . J . Mekalanos. 1996 . Lysogenic conversion by a filamentous phage encoding cholera toxin . Science 272:1910-1914.
46. Warren, G., and L . Corotto. 1989 . The consensus sequence of ice nucleation proteins from Erwinia herbicola, Pseudomonas fluorescens and Pseudomonas syringae . Gene 85:239-242.
47. Wright, R., C . Stephens, and L . Shapiro. 1997 . The CcrM DNA methyltransferase is widespread in the alpha subdivision of proteobacteria, and its essential functions are conserved in Rhizobium meliloti and Caulobacter crescentus . J . Bacteriol . 179:5869-5877.
48. Zhao, J . L., and C . S . Orser. 1990 . Conserved repetition in the ice nucleation gene inaX from Xanthomonas campestris pv . translucens . Mol . Gen . Genet . 223:163-166.

Free Online Full-text Article

What Is Amino Acid?, What Is Staphylococcus Aureus?, What Is Molecular Microbiology?, What Is Cell Biology?, What Is Salmonella?, e, Microbe, c, Microbes, o, Microorganisms, o, Bacteria, s, Bacterium, n, Escherichia coli, e, Streptococcal, a, Escherichia coli, r, Pseudomonas, e, Streptococcal, a, Yeasts, i, Pasteurella, a, Bacillus, o, Staphylococcus aureus, r, Kluyveromyces, r, Escherichia coli, o, Streptomycin, n, Escherichia coli, s, Microbial, c, Microbial, a, Cell suspensions, i, Cell suspensions, r, Candida albicans, n,