|Home | About | Journals | Submit | Contact Us | Français|
We report the plaque propagation and genomic analysis of Xfas53, a temperate phage of Xylella fastidiosa. Xfas53 was isolated from supernatants of X. fastidiosa strain 53 and forms plaques on the sequenced strain Temecula. Xfas53 forms short-tailed virions, morphologically similar to podophage P22. The 36.7-kb genome is predicted to encode 45 proteins. The Xfas53 terminase and structural genes are related at a protein and gene order level to P22. The left arm of the Xfas53 genome has over 90% nucleotide identity to multiple prophage elements of the sequenced X. fastidiosa strains. This arm encodes proteins involved in DNA metabolism, integration, and lysogenic control. In contrast to Xfas53, each of these prophages encodes head and DNA packaging proteins related to the siphophage lambda and tail morphogenesis proteins related to those of myophage P2. Therefore, it appears that Xfas53 was formed by recombination between a widespread family of X. fastidiosa P2-related prophage elements and a podophage distantly related to phage P22. The lysis cassette of Xfas53 is predicted to encode a pinholin, a signal anchor and release (SAR) endolysin, and Rz and Rz1 equivalents. The holin gene encodes a pinholin and appears to be subject to an unprecedented degree of negative regulation at both the level of expression, with rho-independent transcriptional termination and RNA structure-dependent translational repression, and the level of holin function, with two upstream translational starts predicted to encode antiholin products. A notable feature of Xfas53 and related prophages is the presence of 220- to 390-nucleotide degenerate tandem direct repeats encoding putative DNA binding proteins. Additionally, each phage encodes at least two BroN domain-containing proteins possibly involved in lysogenic control. Xfas53 exhibits unusually slow adsorption kinetics, possibly an adaptation to the confined niche of its slow-growing host.
The gammaproteobacterium Xylella fastidiosa is an insect-transmitted, xylem-inhabiting bacterium and the causal agent of several plant diseases, most notably Pierce's disease of grapes (PD) and citrus variegated chlorosis (CVC) (12). Because of the economic importance of this pathogen, it was the subject of early genomic analysis; the genome of the CVC strain 9a5c was the first completed for a plant pathogen (53). Five additional X. fastidiosa genome assemblies are available: a PD strain (Temecula), an oleander leaf scorch (OLS) strain (Ann1), and three almond leaf scorch (ALS)-associated strains (M12, M23, and Dixon) (3, 4, 14, 53). One conclusion from these analyses is that phage may have played a significant role in the evolution of X. fastidiosa (19, 44, 53). Phage-related sequences make up between 9% and 15% of the Xylella genomes, and 47 phage elements were identified in four genomes (18). Many strain-specific differences are located in prophage (18). Most of the prophage elements appeared to be the result of independent insertion events, implying that active phages are frequently encountered. However, most insertions occur in a common 900-kb region and were associated with large genomic rearrangements (18). Whether the prophages are directly responsible for rearrangements or whether sites prone to recombination are also prone to phage integration is not known.
Despite the abundant evidence of their presence, there are no reports of successful plaque propagation of X. fastidiosa phage. Presumptive podophage particles were observed by electron microscopy in ultrathin sections of strain Temecula (13). The majority of prophage sequences were reported to be structural genes for siphophage assembly, and putative siphophage were also observed in an unidentified strain (18). Like many bacteria, the polylysogenic status of X. fastidiosa implies that each strain may produce more than one type of phage. Therefore, it is not possible to conclusively correlate an image of a virion produced by a lysogen with a specific prophage region based on morphological descriptions alone, nor is it possible to draw any conclusion about whether any particular observed virion is actually viable. Phages that can be propagated and plaque purified on X. fastidiosa are needed to further our understanding of X. fastidiosa phage biology. The work presented here describes the propagation, purification, and genomic analysis of X. fastidiosa phage Xfas53.
PW broth medium as modified by Sherald et al. (52), designated PW-M, was used for growth of Xylella isolates, except that the final bovine serum albumin content was 0.3% as modified by Hill and Purcell (27). For solid medium (PW-MG) and soft gel, the PW-M broth was amended with 9 g/liter and 4.5 g/liter, respectively, of Gelrite Gellan gum (Sigma). Alternative solidifying agents tested included Gelzan CM (Sigma), Phytagel (Sigma), and plant cell culture-tested (PCCT) agar (Sigma). Cultures were grown at 28°C, and liquid cultures were monitored at λ = 600 nm using sidearm flasks. California X. fastidiosa isolates included in the study were Temecula, Ann1, and Dixon (25). Texas isolates included one each from Plantanus occidentalis (XF1), Iva annua (XF18), Helianthus annuus (XF5), Ratibida columnifera (XF37), Ambrosia psilostachya (XF23), Vitis aestivalis (XF39), and Vitis mustangensis (XF41); three isolates from Ambrosia trifida var. texana (XF16, -40, and -43); two from Nerium oleander (XF93 and -95); and 15 from Vitis vinifera (XF48, -50, -52, -53, -54, -56, -58, -59, -60, -66, -67, -70, -71, -76, and -78). All isolates were single colony purified by the streak isolation method and stored at −80°C after amending PW-M broth cultures to a final concentration of 20% (vol/vol) glycerol. Isolates were confirmed at the species and subspecies level using PCR analysis as previously described (26).
Cultures were grown in PW-M broth at 28°C for 7 to 10 days. Cells were removed by centrifugation (10,000 × g, 15 min at 5°C) in a J2-21 centrifuge (Beckman Coulter). The supernatant was filtered through a 0.22-μm filter (Supor; Pall Life Sciences). Five- to seven-day cultures of X. fastidiosa isolates were used to make indicator suspensions in PW-M broth (A600 = 0.5). Soft gel overlays used to survey bacterial supernatants for phage activity were made by mixing 100 μl of the bacterial suspension with 7 ml of molten PW-M soft gel, pouring the mixture on PW-MG plates, and allowing it to solidify and dry. Each supernatant was serially diluted in phage buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 8 mM MgSO4), and soft gel overlays were spotted with 10 μl of each dilution, allowed to dry, and incubated at 28°C for 5 to 7 days. Positive supernatants showing either plaques or cleared zones were serially diluted in phage buffer and then plated as described above, except that the phage dilutions were directly mixed with the indicator suspension before addition of molten soft gel. Individual plaques were excised from the overlay and suspended in phage buffer, and their titers were determined. This procedure was repeated twice to obtain a single plaque isolate for each phage. High-titer lysates (1010 PFU/ml) were prepared by harvesting overlays of plates exhibiting confluent lysis with 5 ml of phage buffer, macerating the soft gel overlay, clearing the lysate by centrifugation (10,000 × g, 15 min at 4°C), and filter sterilizing it through a 0.22-μm filter. Lysates were stored at 4°C.
Filter-sterilized phage suspensions were concentrated by centrifugation (90,000 × g for 2.5 h at 5°C) using a type 60Ti rotor in a Beckman L8-70 M ultracentrifuge. Pellets were resuspended in phage buffer and treated with DNase I and RNase A (Sigma) at a final concentration of 1 μg/ml at 25°C for 2 h. CsCl was added to the phage suspension at a final concentration of 0.75 g/ml and centrifuged (300,000 × g for 18 h at 5°C) in a VTi 65.2 rotor. The visible phage band was extracted using an 18-gauge syringe needle and dialyzed against phage buffer amended to 1 M NaCl overnight at 4°C and twice for 4 h at 25°C against phage buffer to obtain a suspension of 1011 PFU/ml. The CsCl-purified phage was stored at 4°C.
The kinetics of phage adsorption was determined by infecting X. fastidiosa strain Temecula suspended in PW-M broth at a multiplicity of infection of ~0.01. Samples were taken at 2-h intervals, and titers were determined after removal of cells by filtration of the sample through a 0.2-μm filter using strain Temecula as the indicator. The rate of phage particle disappearance is defined as dP/dt = −kBP, where B is the concentration of bacteria, P is the concentration of free phage at any time (t), and k is the adsorption constant in ml cell−1 min−1 (50a).
Three-day cultures (A600 = 0.15) were washed by centrifugation (10,000 × g, 15 min at 5°C), resuspended in 0.85% NaCl, and exposed to 254-nm UV radiation (360 μW/cm2) for 0, 5, 7, and 10 s in a sterile petri dish (5 ml/dish). After exposure, cells were collected by centrifugation, resuspended in PW-M broth, and incubated at 28°C with shaking (200 rpm). One-milliliter samples were removed 1, 2 and 3 days post-UV exposure and filter sterilized, and their titers were determined to determine PFU.
Mitomycin C inductions were performed by adding mitomycin C (Sigma) to a final concentration of 0 to 2.0 μg/ml to growing cultures and incubating them in the dark at 28°C for 48 h with shaking (200 rpm). After 48 h, the titers of filter-sterilized samples were determined using the indicator strain Temecula.
Heat induction experiments were conducted by exposing 3-day cultures grown in PW-M broth at 28°C with shaking (200 rpm) (A600 = 0.15) to 42°C for 0, 15, 30, and 60 min in a shaking water bath. Cultures were then incubated at 28°C with shaking (200 rpm) for an additional 4 days and filter sterilized, and their titers were determined.
Electron microscopy of CsCl-purified phage (3.7 × 1011 PFU/ml) was performed by diluting the phage with phage buffer and applying 5 μl onto a freshly glow-discharged Formvar-carbon-coated grid for 1 min. Grids were then washed briefly on deionized water drops and stained with 2% (wt/vol) aqueous uranyl acetate. For transmission electron microscopy (TEM) of phage with bacteria, a bacterial suspension (8 × 108 CFU/ml) was mixed with dilutions of purified phage and incubated at 25°C for 30 min. Five microliters of the mixtures was then applied onto freshly glow-discharged Formvar-carbon-coated grids and left to adhere for 1 min. Grids were then washed briefly on deionized water drops and stained with 2% (wt/vol) aqueous sodium phosphotungstate, pH 7.0. Specimens were observed on a JEOL 1200EX transmission electron microscope operating at an acceleration voltage of 100 kV. Images were recorded at calibrated magnifications on Kodak 4489 film.
DNA was isolated from CsCl-purified phage suspensions as described previously (54). Genomic DNA was sized by pulsed-field gel electrophoresis. Library preparation, sequencing, sequence assembly, and analyses were performed essentially as described previously (54, 56, 57). The genomic DNA library was prepared in the pSmart-LCKan vector (Lucigen). Sequencher (Gene Codes Corporation) was used for sequence assembly and editing. Protein coding regions were predicted using Genemark (http://opal.biology.gatech.edu/GeneMark/gmhmm2_prok.cgi) and manually edited in Artemis (http://www.sanger.ac.uk/Software/Artemis/) using the phage genome annotation tool ArtAnnoPipe (http://athena.bioc.uvic.ca/node/541) (42, 50). Dot plots were generated using JDotter (6). Predicted proteins were compared to proteins in the GenBank database using BLAST (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi). Conserved domains, lipoprotein processing signals, and transmembrane domains (TMDs) were identified with InterProScan (http://www.ebi.ac.uk/Tools/webservices/services/interproscan), LipoP (http://www.cbs.dtu.dk/services/LipoP/), and TMHMM (http://www.cbs.dtu.dk/services/TMHMM/), respectively.
SDS-PAGE was carried out on CsCl-purified phage particles along with Precision Plus protein standard (Bio-Rad) and stained with Coomassie brilliant blue R250 (37).
The sequence of phage Xfas53 has been entered as GenBank accession number GQ421471.
The growth of X. fastidiosa lawns and phage plaque formation were found to be sensitive to the solidifying agent used in the medium. X. fastidiosa failed to form uniform lawns in standard Bacto Agar. Alternative solidifying agents were tested, including Gelrite, Gelzan, and PCCT agar. Plaque formation was observed in overlays made with Gelrite at 0.45% (Fig. (Fig.1A)1A) or PCCT agar at 0.75%. Very small plaques were observed with Gelzan and Phytagel (data not shown). Experiments described here were performed with Gelrite (Gellan gum) overlays. Supernatants from liquid cultures of 30 individual X. fastidiosa isolates were spotted on overlays seeded with each isolate. Supernatants from four of these (isolates 48, 53, 58, and 60) developed plaques on overlays of strain Temecula following a 5- to 6-day incubation. Titers in these uninduced culture supernatants ranged from 1 × 103 to 5 × 103 PFU/ml.
The phage produced by lysogenic X. fastidiosa strain 53 was chosen for further study. Individual plaques were excised and plaque purified twice to obtain a clonal isolate, designated phage vB_XfaP_Xfas53 (Xfas53) (Fig. (Fig.1A).1A). After a 5- to 6-day incubation, Xfas53 formed clear plaques on overlays of strain Temecula. An additional 5- to 6-day incubation resulted in the development of turbidity in the plaques, suggestive of lysogen formation; however, viable bacteria could not be rescued from the plates to confirm this (data not shown). Phage Xfas53 failed to form plaques on overlays of strain 53, as would be expected for a superinfection-resistant lysogen. Transmission electron microscopy of purified phage revealed that phage Xfas53 possesses a 55-nm head and a short, 12-nm-diameter, noncontractile tail, typical of podophage morphology (Fig. (Fig.1B).1B). Electron micrographs of concentrated supernatants revealed that the phage in the other three positive supernatants also possessed podophage morphology (data not shown). The adsorption rate constant of phage Xfas53 for host X. fastidiosa Temecula was determined from three replicate experiments to be (6.8 ± 1.5) × 10−12 ml cell−1 min−1 (Fig. (Fig.22).
Prophage induction experiments of X. fastidiosa strain 53 using either mitomycin C, UV, or heat shock treatment did not result in an increase in phage production above that observed in uninduced cultures as determined by phage titer or culture density. Mitomycin C treatments above 1.5 μg/ml resulted in a complete loss of cell viability without inducing the production of virions, as indicated by the absence of changes in culture turbidity following treatment. In heat induction experiments, there was no increase in the titer above that observed in the control culture. The exposure of cells to 42°C for 30 and 60 min resulted in loss of cell viability, as measured by absorbance. Examination by electron microscopy of a 100-fold-concentrated supernatant of uninduced strain Temecula, grown under the same conditions in which strain 53 produced phage Xfas53, showed no evidence of phage particles.
Xfas53 genomic DNA migrated as a single 36-kb band in pulsed-field gel electrophoresis (data not shown). Clones from a shotgun library were sequenced to eightfold coverage, and the sequence reads were assembled into a single contig of 36,674 bp. The genome has an average GC content of 57%, which is slightly higher than the 53% GC content of Xylella. The circular assembly was opened upstream of the predicted integrase (designated gene01). The longest region not predicted to encode a protein is 820 nucleotides between gene23 and gene24. A GC skew plot (not shown) indicated that this gap either corresponds to a replication origin or is possibly a relic of the recombinatorial origin of the right and left arms of the phage (see below).
A total of 45 protein coding genes were predicted to be encoded by the Xfas53 genome (Fig. (Fig.3;3; Table Table11 ). Xfas53 genes are organized into two transcription units with the first 18 genes transcribed from the reverse strand and all but three of the remaining 27 genes transcribed from the forward strand (Fig. (Fig.3).3). Genes at the presumptive divergent promoter region included some with similarity to proteins implicated in lysogenic control.
Xfas53 is a temperate phage, and among the predicted genes are those expected to be responsible for phage integration and regulation of lysogeny. gene01 is predicted to encode an integrase, as gp01 has a conserved catalytic C-terminal integrase domain and sequence similarity to the archetypical integrase from phage lambda. There are four or five Xfas53 gp1 homologues exhibiting over 90% identity in each of the X. fastidiosa genomic assemblies. Xylella strains are known to possess multiple, closely related integrases (18). For example, strain Temecula possesses five integrases (locus tags PD1019, PD0764, PD1196, PD0384, and PD1139) that are over 90% identical to Xfas53 gp1. These, along with the Xfas53 integrase, cluster with the largest Xylella integrase groups (18). The most closely related phage-encoded protein is only 52% similar, encoded by Pseudomonas phage F10 (36).
gp17 was identified as a lambda cI-like repressor based on the presence of a conserved C-terminal LexA-family serine peptidase domain (PFAM PF00717 peptidase_S24). This signature belongs to proteins of the MEROPS peptidase family, which includes numerous prophage repressor proteins, including Salmonella (SE1 and ST104) and Pseudomonas (F10, MP22, DMS3, and D3) phages. The D3 cI homologue was shown to exhibit binding specificity for the D3 operator sequences (21, 34). Xfas53 also encodes two BroN domain-containing proteins, gp08 and gp22 (29). BroN domains are DNA-interacting domains found in proteins whose specific function is conferred by a number of different C-terminal domains. A third protein, gp21, possesses a COG3617 antirepressor domain. The significance of these proteins is discussed in more detail below.
gene04, gene06, and gene07 encode proteins with conserved domains indicative of a DNA helicase, nuclease, and DNA polymerase, respectively. Highly similar homologues of these three genes, as well as gene09 and gene10, are present in syntenic clusters in several otherwise unrelated podophages. These include Burkholderia phage BcepC6B, Acyrthosiphon pisum endosymbiont phage APSE-1, and Bordetella phages BPP-1, BMP-1, and BIP-1. This linkage suggests that gp09 and gp10 also function in DNA replication. gene23 encodes an 845-amino-acid (aa) residue protein with an amino terminus that aligns only with highly related prophage-encoded proteins in the Xylella genome entries and BCPG_00731 from Burkholderia cenocepacia PC184 (data not shown). The carboxy-terminal half of gp23 contains both a D5_N domain and a P4 primase C-terminal domain. These latter two domains are frequently associated with N-terminal primase domains (30). The lack of a detectable primase active site conserved domain at the amino terminus of gp23, however, indicates that this protein cannot be conclusively annotated as a primase without experimental evidence.
The Xfas53 lysis cassette consists of gene25, gene26, gene27, and gene28 and is organized similarly to the archetypical lambda lysis cassette SRRzRz1. Of these, only gene26, encoding the endolysin (Lyz), was identified based on amino acid sequence similarity. The Xfas53 Lyz is related to members of the canonical T4 lysozyme family, with a glycoside hydrolase domain containing an E-8 aa-D/C-5 aa-T catalytic triad that flanks a substrate-binding cleft (Fig. (Fig.4A).4A). However, Xfas53 Lyz has a predicted TMD at the N terminus (residues 12 to 30), indicating that it belongs to the signal anchor and release (SAR) endolysin subgroup (65). SAR endolysins are secreted by the host sec system to the periplasm, where they accumulate during the latent period in an inactive form tethered to the membrane by the SAR domain. Like other SAR domains and unlike normal TMDs, the SAR sequence of Xfas53 Lyz is only weakly hydrophobic, with only 10 of its 19 positions occupied by strongly hydrophobic amino acid residues, a disproportion that is required for the ability to exit the bilayer (58). When the membrane is permeabilized (by the holin) and thus deenergized, the SAR domain exits the bilayer, refolds to its active conformation, and attacks the peptidoglycan (Fig. (Fig.4B).4B). In the membrane-tethered form, Glu31 of the catalytic triad of Xfas53 Lyz is predicted to be juxtaposed to the periplasmic surface of the cytoplasmic membrane, making it impossible to form a functional active site (Fig. (Fig.4A).4A). Xfas53 Lyz thus belongs to a class of SAR endolysins which rely on the steric hindrance of the membrane to prevent premature activation of the endolysin, rather than the well-documented covalent control based on disulfide bond isomerization observed in the SAR endolysin of coliphage P1 (64).
Xfas53 gene25, or hol, is predicted to encode the holin. In addition to its position preceding the endolysin gene, hol has other hallmarks of a holin gene (62). First, Xfas53 Hol is the right size, spanning 115 residues, and has an extremely hydrophilic, cytoplasmic C-terminal domain. Second, the Xfas53 Hol possesses two TMDs, the most hydrophobic of which is TMD2 (residues 73 to 92) (Fig. (Fig.4C).4C). Another TMD, not predicted by typical TMD search algorithms, is present from residues 55 to 70 and has characteristics of a SAR domain, with a high percentage (10 out of 16) of weakly hydrophobic or polar residues (65). The presence of a SAR domain followed by a typical TMD suggests that Xfas53 Hol is a pinholin, analogous to the holin of phage 21, S21. Pinholins are a recently discovered class of holins that make small holes in the host membrane (47). These holes are sufficient to depolarize the membrane and allow exit of SAR endolysins but are not large enough to allow escape of canonical cytoplasmic endolysins. Even more so than with other holin genes, the structure of gene25 suggests multiple levels of negative regulation at the levels of transcription, translation, and function. There are three possible start codons in Xfas53 hol, all of which have appropriately placed Shine-Dalgarno sequences. Starts at codon 1, 16, and 35 would produce 115 (Hol115)-, 90 (Hol90)-, and 70 (Hol70)-amino-acid products, respectively. The two longer products would have N-terminal extensions with either 1 or 2 extra positively charged residues compared to Hol70. The prototype pinholin, S21, exhibited dynamic membrane topology, in that TMD1 is required to exit the bilayer before pinhole formation by TMD2 can proceed. N-terminal extensions with positively charged residues were shown to inhibit escape of TMD1 from the bilayer and not only retarded hole formation but also conferred negative-dominant antiholin character (47). Thus, both Hol115 and Hol90 are predicted to be antiholins that would specifically inhibit Hol70. In addition, the hol mRNA would contain a strongly predicted RNA stem-loop structure overlapping the ribosome-binding site for the start codon at 35 (Fig. (Fig.4C).4C). This mirrors similar stem-loops that reduce holin translation, in favor of antiholin synthesis, in lambda and other phages (62). Finally, the same stem-loop structure is followed by an oligopyrimidine stretch, strongly suggesting that it constitutes a rho-independent transcriptional terminator.
The spanin component genes, gene27 (Rz) and gene28 (Rz1), were easily identified by their characteristic gene architecture, in which Rz1 is a reading frame encoding a short (45-aa mature length), proline-rich (14/45 residues) outer membrane (OM) lipoprotein entirely embedded out of frame within Rz (55). Like other Rz proteins, Xfas53 Rz is a type II (N-in, C-out) cytoplasmic membrane protein with a highly charged periplasmic domain predicted to have significant α-helical character. The Rz and Rz1 proteins form the spanin complex which spans the periplasm during the latent period (2). When the cell wall is degraded by the endolysin, the complex is thought to undergo a conformational change and cause disruption of the OM, possibly by fusing it with the cytoplasmic membrane.
The right arm of the Xfas53 genome possesses a virion morphogenesis cassette with protein and gene order similarity to podophage P22, APSE-1, Shigella flexneri phage Sf6, and Thalassomonas phage BA3 (Fig. (Fig.3A)3A) (10, 20, 59, 60). As expected, the Xfas53 virion morphogenesis genes exhibit a mosaic relationship with these phages (10). Individual similarities were not high. For example, the homologues from phage BA3, the most closely related phage, ranged from 44% to 53% in similarity. Because phage P22 has been extensively studied in terms of capsid morphogenesis, DNA packaging, and DNA injection apparatus assembly, it is useful to make comparisons with Xfas53, even when individual gene similarities are low (32). Nine Xfas53-encoded proteins could be confidently correlated with P22 equivalents. These include portal, scaffold, major head protein, the three DNA injection proteins, the large terminase subunit (TerL), and two of the three P22 tail accessory factors.
Xfas53 encodes homologues of P22 gp1 (portal), gp8 (prohead scaffold), and gp5 (major capsid); therefore, Xfas53 gp32, gp33, and gp34 probably function as the portal, scaffold, and major head protein, respectively (Fig. (Fig.3A;3A; Table Table1).1). The predicted molecular mass of the gp34 major head protein is 43 kDa. When proteins from purified Xfas53 virions were analyzed by SDS-PAGE, a single major band of approximately 43 kDa was observed (data not shown). In P22, three proteins, gp7, gp20, and gp16, are packaged inside the mature capsid and function in DNA injection. Homologues of these proteins were assigned to gp38, gp39, and gp40. Although Xfas53 gp37 is significantly related to P22 gp7, the P22 gp20 and P22 gp16 equivalents are more similar to those from phage Sf6 (10).
DNA packaging is the next step of P22 virion morphogenesis (9, 31). The putative TerL homologue is gp31, which is highly similar (60%) to P22 gene2-encoded TerL. Several lines of evidence suggest that Xfas53 utilizes a head-full packaging mechanism, similar to P22 (11), including the lack of defined termini in the sequence assembly, the failure of the genomic DNA to form multimers during agarose gel electrophoresis, and the presence of a pac fragment following restriction digest analysis of genomic DNA (not shown). The lack of an identifiable TerS subunit was not surprising, given the diversity of TerS proteins. The size and genetic location of gene30 near the pac site make it a likely terS candidate (63).
The most significant differences between Xfas53 and P22 virion morphogenesis were found in the head completion genes (corresponding to P22 gp4, gp10, and gp26) and tailspike gene (corresponding to P22 gp9) (45). Association of P22 gp10 with gp4 is a prerequisite for the attachment of gp26 (45). P22 gp26 forms the central tail needle, which penetrates the host envelope during the DNA injection process, and the gp9 tailspike, which functions in two critical roles: recognition of the host O antigen and hydrolysis of the lipopolysaccharide (45). Xfas53 encodes recognizable homologues of P22 gp4 and gp10 (Xfas53 gp35 and gp36, respectively) but not homologues of either P22 gp26 or gp9. This is not unexpected given that gp9 and gp26 equivalents vary even among the closely related P22-like phages (10). While Xfas53 gp35 is similar in size to P22 gp4, the P22 gp10 equivalent, Xfas53 gp36, is larger (642 residues compared to 472) due primarily to the presence of a carboxy-terminal extension. One possibility is that Xfas53 gp36 is a functional fusion between P22 gp10 and gp26 equivalents. gp36 is predicted to adopt considerable extended conformation (data not shown). Alternatively, due to gene position and size, it is possible that either gp44 or gp45 encodes tailspike or needle equivalents. Structural prediction programs suggest that Xfas53 gp45 might have a tendency to form beta wrap structures, similar to P22 gp26 and gp9 (data not shown).
Embedded within the virion morphogenesis cassette and encoded on the complementary strand are three small open reading frames, gene41, gene42, and gene43, encoding proteins of 84, 116, and 81 residues, respectively (Fig. (Fig.3A).3A). Both gp41 and gp42 are strongly predicted to be anchored in the inner membrane by one and two TMDs, respectively (data not shown). gp42 is significantly related to numerous bacterial inner membrane proteins of unknown function. Many phages, including P22, T4, and P1, possess superinfection-exclusion mechanisms that utilize small, poorly conserved, inner membrane or periplasmic proteins to prevent the entry of DNA from superinfecting phage (28, 40, 41, 43, 48, 59). Based on the gene location, protein size, and predicted membrane association, it is tempting to speculate that Xfas53 gp41, gp42, and gp43 are involved in superinfection exclusion despite the complete lack of primary structural similarity.
Prophages related to Xfas53 are present in each of the draft (Ann1 and Dixon) and complete (Temecula, M23, M12, and 9a5c) X. fastidiosa genome entries (Fig. (Fig.3B3B and and5).5). These include previously identified prophage elements Xfp1 and Xfp2 (in strain 9a5c); Xpd1 (in strain Temecula); Xop2, Xop3, and Xop9 (in strain Ann1); and Xap1 and Xap5 (in strain Dixon) (7, 8, 18, 53, 61). Related prophage elements were also identified in the unpublished but complete X. fastidiosa genome assemblies from strains M12 (located between nucleotides 1207697 and 1254095 of GenBank entry GI:167964044) and strain M23 (located between nucleotides 1297360 and 1344391 of GenBank entry GI:182630682). These phages can be divided into three groups based on alignments with each other: Xfas53 and Xop2; Xop9; and Xfp1, Xpf2, Xpd1, Xap1, Xap5, Xop3, M12, and M23 (Fig. (Fig.5).5). Xop2 and Xfas53 align over their entire length, indicating that Xop2 is expected to produce a similar, P22-related podophage. Xop2 is shorter than Xfas53, due primarily to deletions in the replication and regulation module, resulting in the loss of several genes, and therefore may not be functional. The most significant difference in gene content between Xop2 and Xfas53 is in the lysis cassette. The Xop2 lysis cassette has an atypical gene order in that the endolysin (XfasoDRAFT_4319) is encoded by the gene preceding the holin (XfasoDRAFT_4320), which is followed by the genes for Rz (XfasoDRAFT_4321) and Rz1 (not annotated). Similarly arranged lysis cassettes are present in Xfp1 and related phages (see below). Pairwise nucleotide alignments between Xfas53 and the prophage elements reveal that they all have left arms related to that of Xfas53, from gene01 (integrase) to gene23 (primase) (Fig. (Fig.5).5). In Xop9, this includes coding regions from the integrase (XfasoDRAFT_3154) to the gp22 BroN domain protein (XfasoDRAFT_3171) (Fig. (Fig.55 and data not shown). The right half of Xop9 is not related to any of the other phages (Fig. (Fig.5)5) and has virion structural genes encoding proteins distantly related to those from Actinobacillus myophage Aaphi23 (reference 49 and data not shown).
Eight of the prophage elements (Xfp1, Xfp2, Xpd1, Xop3, Xap1, Xap5, M12, and M23) were found to be over 90% identical to each other over their entire length (Fig. (Fig.33 and and5).5). M12 has a significant inverted region and may be defective. For ease of reference, this group of prophage elements will be referred to as “Xfp1-like” prophage elements. The similarity between Xfas53 and the Xfp1-like prophage elements is restricted to the left half of the genome, encompassing the gp01 integrase equivalent to gp23. Immediately following gene23, there is an ~800-bp noncoding region (Fig. (Fig.3).3). DNA alignments suggest that recombination occurred in the distal portion of gene23 (data not shown).
The right half of the Xfp1-like prophages, encoding the lysis and virion morphogenesis proteins, is unrelated to Xfas53 and is organized like lambdoid prophages. However, the tail assembly genes are more closely related to P2, suggesting that the progenitor of these phages was a hybrid between a P22-like phage and a P2-like phage (7, 8). The tail assembly cluster of these prophages encodes homologues of P2 D, X, U, FII, FI, I, J, W, and V. Virions produced by these elements would be expected to demonstrate contractile-tailed (myophage) morphology. At 624 residues, the predicted major capsid protein of these phages (corresponding to PD1106 in Xpd1) is larger than that from typical phages. The major capsid proteins from several phages, for example, Stenotrophomonas phage S1, are synthesized from similarly sized precursors (23). Like S1 and lactococcal bacteriophage c2, the Xpd1 major capsid protein contains a prohead protease domain, suggesting that the protein is self-processed and could also contain a scaffolding domain (15, 23). The Xfp1-like phages possess homologues of lambda nu1, A, and B, encoding the lambda TerS, TerL, and portal proteins, respectively. As in Xop2, the lysis cassette of the Xfp1-like prophage elements has an atypical gene arrangement, with the endolysin gene preceding the holin gene. The endolysin (XF0707 in Xfp1) is a glycoside hydrolase and is followed by XF0708, a homologue of the class I (3-TMD) holin from Burkholderia phage Bcep781 (56). The holin gene is followed by the Rz equivalent (XF0709 in Xfp1). Either the Rz1 equivalent in each of these prophages is not annotated (in Xfp1, Xfp2, and Xpd1) in the current version of the bacterial genome, or the incorrect start codon is annotated (in M12 and M23), resulting in the predicted protein being truncated from the amino terminus. Rz1 equivalent genes are frequently annotated incorrectly, due to their unusual gene arrangement (55). Interestingly, while the Rz equivalent of Xpd1 is unrelated to known Rz equivalents, the Rz1 equivalent is a homologue of PRD1 P37, which has been demonstrated to be an authentic Rz1 (35).
The region of Xfas53 encoding gp20 and gp21 has an unusual organization, being composed of 2.7 copies of a 377-nucleotide direct tandem repeat (Fig. (Fig.6).6). These repeats are 87% identical. Variations of this repeat are present in each of the related Xylella prophages (Fig. (Fig.6).6). The prophages in M23 and Temecula each contain 4.7 copies of the repeat unit. In Xfp1, Xfp2, and M12, there is only a single, partial, and more degenerate copy of the repeat unit. The Xfas53 repeats are 87% identical with 5% of the differences attributed to mismatches and 7% of the differences attributed to insertions or deletions (indels). Each of the repeat units has an ATG served by a strong Shine-Dalgarno sequence predicted to serve in translational initiation. This ATG is located in the 3′ half of the repeat, such that the nucleotide sequence of the repeat is staggered in relation to the predicted coding region. The last repeat is truncated at this ATG, which serves as the start codon for the gene22 equivalents. For Xfp2, the single-repeat-related sequence starts at this ATG and a fusion between the gene21 and gene22 equivalents is predicted to be expressed (XF2506). Five versions of the repeat-encoded protein are predicted to be encoded by Xpd1 and M23.
The repeat-encoded proteins are not related to proteins of known function, although most contain an amino-terminal conserved domain (COG3617) annotated as a prophage antirepressor (Fig. (Fig.6).6). Because of the indels, the amino acid sequences of the repeat-encoded proteins are less conserved than the nucleotide sequences. gp22 and the Xfp1-like prophage element equivalents each contain this same amino-terminal COG3617 conserved domain, which also encompasses a BroN (pfam02498) domain (Table (Table22 ). Additionally, gp8 and the Xfp1-like prophage element equivalents contain either two or three copies of the COG3617/BroN domains (Table (Table2).2). Not surprisingly, all of these proteins share some similarity to each other, due to the presence of the common COG3617 domain. Because the BroN domain defines a DNA binding domain found in numerous regulatory elements, it may be that these proteins are also regulatory elements that interact directly with DNA.
Despite the extensive evidence of the involvement of phage in the genome diversity of X. fastidiosa and previous reports of the visualization of phage particles, this is the first report of the successful propagation of phage on X. fastidiosa. The extended time that it takes to prepare the plating culture (5 to 6 days) and the long incubation time required to visualize plaques (5 to 6 days) add to the challenge of manipulating Xylella phage. Xylella most commonly exists as an asymptomatic endophyte, showing disease symptoms on relatively few susceptible hosts (12). Xylem fluids are the most nutrient poor of all plant tissues, and Xylella bacteria are slow growing. Interestingly, the rate of adsorption of phage Xfas53 to strain Temecula was found to be extraordinarily low, with an experimentally determined adsorption rate constant of 6.8 × 10−12 ml cell−1 min−1. This value is the lowest reported for a phage with its plaque-permissive host, 100- to 1,000-fold lower than those reported for other well-studied phages such as lambda PaPa (which lacks its side tail fibers) (24) at 1.3 × 10−9 ml cell−1 min−1 and a reconstructed wild-type (wt) lambda phage at 9.9 × 10−9 ml cell−1 min−1 (51). In the case of phage P22, a phage whose virion structural proteins are related to those of Xfas53, this constant has been reported as 3.4 × 10−9 ml cell−1 min−1 with sensitive Salmonella hosts (39). The majority of models suggest that a high adsorption rate is a trait of improved fitness (1, 51). Given the unique ecology of X. fastidiosa, a xylem-limited plant pathogen, there may be benefits to slow adsorption. Growth is limited to individual xylem vessels unless connecting pit membranes are breached (16, 46). The transport characteristics of phage in xylem are unknown, but it is unlikely that phage virions can breach intact xylem pit membrane, considering that even small plant viruses require active transport through pit membranes (46). Pit membranes are constructed of plant cell wall material, and X. fastidiosa is able to traffic between vessels due to the production of plant cell wall-degrading enzymes (12). If X. fastidiosa phages require physiologically active host cells to facilitate movement through the xylem, slow adsorption might be advantageous in a situation where complete elimination of host bacteria in a vessel would result in reduced overall phage fecundity. Another effect of rapid phage adsorption in a xylem vessel would be an increased chance of progeny phage loss due to superinfection in the confined, host-limited space. Similar confined, host-limited space would also be confronted in the insect mouth parts where Xylella can reproduce. Recent experiments using double-layer agar plates as a model for reduced diffusion in biofilms suggest that while high adsorption rates are advantageous early in phage association with a biofilm, low adsorption rates are strongly selected for within the biofilm (22). It was proposed that low phage adsorption rates may increase overall phage fecundity in a biofilm-like environment by permitting longer periods of phage diffusion, increasing the effective size of the potential host pool, and reducing the probability of the phage adsorbing to an already infected cell or cell debris. There are conditions under which a reduced adsorption rate might be beneficial even to planktonic bacteria. Cyanophage AS-1, for example, exhibits light-dependent changes in adsorption kinetics (33). An interesting model was proposed suggesting that the lack of adsorption in the dark would result in phage remaining as free particles during the period when the host cells are not metabolically active, reducing the chances of the phage being grazed by nocturnal predators of bacteria (17).
A feature of the evolution of Xfas53 and related prophages is the expansion of BroN (pfam02498) and COG3617 domain-containing proteins. Bro (baculovirus repeat open reading frame) domain proteins are widespread in Baculovirus and other large DNA viruses, including phages (5). Bro domain proteins typically have at least two conserved domains with the amino-terminal domain, BroN, contributing the DNA-interacting domain and the specialized function being conferred by various C-terminal domains (29). Because the COG3617 regions include all of the BroN domains, it is tempting to speculate that COG3617 corresponds to a more distant lineage of BroN proteins also involved in DNA binding. BroN domain proteins often have C-terminal domains of various functions including endonucleases, protein-interacting domains, and additional DNA-interacting domains, as well as multiple conserved domains of unknown function (29). The evolution of Bro domain proteins is characterized by extensive lineage-specific domain expansion followed by domain shuffling. The tandem arrays that generate the multiple COG3617-containing proteins appear to be involved in expansion of COG3617 domain-containing proteins. The widespread occurrence of these proteins in X. fastidiosa prophage strongly suggests that they may play a pivotal and global role in X. fastidiosa phage biology. Increased expression of the Xfp1 Bro domain proteins was detected following heat shock treatment (18). We have conducted UV and mitomycin C induction experiments (using conditions which induce prophages of Escherichia coli, Burkholderia spp., and Pseudomonas spp.) with several X. fastidiosa lysogens, including the Xfas53 lysogen, without evidence of increased phage production or culture lysis (data not shown). Heat induction experiments with strain 53 did not show an increase in titer above that observed for the uninduced culture. The absence of lysogenic induction, despite the constitutive release of a basal level of virions in liquid culture, is puzzling. The Xfas53 cI repressor homologue appears to be a typical LexA family phage repressor, with conserved serine and lysine residues critical for protease activity. One possibility is that UV and mitomycin C treatments do not induce the SOS response required for prophage induction. Nevertheless, Xylella does encode homologues of E. coli RecA, LexA, and exonuclease V and components of the RecBCD complex known to participate in the repair of double-strand breaks produced during DNA replication (53, 61). Our understanding of prophage induction based on E. coli phage biology may not be applicable to X. fastidiosa phage systems.
Where does the obviously hybrid relationship of Xfas53 to the Xylella prophages fit into our current understanding of phage evolution? The capacity of phage to form hybrids has been known for some time. Hybrid phages are not particularly common, although the relationship between siphophage lambda and podophage P22 has long been recognized (38). The abundance of hybrid phages in Xylella implies that specific conditions are present that promote hybrid phage formation. X. fastidiosa strains carry multiple copies of closely related prophages and multiple combinations of hybrid phages, a situation that can serve only to promote large-scale genomic rearrangements as well as the formation of new hybrid phage permutations (18). It was suggested that the larger number of genomic rearrangements of Xf-CVC and Xf-ALS than of Xf-PD and Xf-OLS may be the result of the presence of duplicated prophage elements in the former two strains, thus providing indirect evidence of the impact of prophage in reorganization of the X. fastidiosa genome (18). The Xfp1-like prophages themselves are of an ancient hybrid origin, having head assembly proteins related to lambda and tail assembly proteins related to those from P2 (8, 18). The extremely high DNA identity between Xfas53 and these myophage elements, however, indicates that this hybrid is of a very recent origin. One possibility is that these recombination events frequently generate nonfunctional prophages, which would also account for the lack of prophage induction. The isolation and propagation of Xfas53, a functional phage of recent hybrid origin, demonstrate that at least some of these combinations are in fact capable of forming viable virions.
We are grateful to Steven E. Lindow and Donald A. Cooksey, for providing Xylella strains. We thank Marie-Anne Van Sluys for providing us with the assembled Ann1 and Dixon prophage DNA sequences.
This work was supported by grants from the Texas Pierce's Disease Research and Education Program (to C.F.G.) and grant EF 0523951 from the National Science Foundation (to R.Y.) and funding from Texas AgriLife Research.
Published ahead of print on 6 November 2009.