The genus Bartonella comprises facultative intracellular bacteria adapted to mammals, including previously recognized and emerging human pathogens. We report the 2,341,328 bp genome sequence of Bartonella grahamii, one of the most prevalent Bartonella species in wild rodents. Comparative genomics revealed that rodent-associated Bartonella species have higher copy numbers of genes for putative host-adaptability factors than the related human-specific pathogens. Many of these gene clusters are located in a highly dynamic region of 461 kb. Using hybridization to a microarray designed for the B. grahamii genome, we observed a massive, putatively phage-derived run-off replication of this region. We also identified a novel gene transfer agent, which packages the bacterial genome, with an over-representation of the amplified DNA, in 14 kb pieces. This is the first observation associating the products of run-off replication with a gene transfer agent. Because of the high concentration of gene clusters for host-adaptation proteins in the amplified region, and since the genes encoding the gene transfer agent and the phage origin are well conserved in Bartonella, we hypothesize that these systems are driven by selection. We propose that the coupling of run-off replication with gene transfer agents promotes diversification and rapid spread of host-adaptability factors, facilitating host shifts in Bartonella.
Emerging infectious diseases represent an increasing human health problem with many examples of disease outbreaks caused by transmissions from animals to humans, such as, most recently, the bird flu virus. Genes involved in virulence and antibiotic resistance are often carried by mobile elements like plasmids and viruses, which mediate transfer between cells at an amazing speed. Rodents represent a major carrier of infectious agents, and it is therefore particularly important to study the gene transfer processes in bacteria that use rodents as their natural host reservoir. We have studied the genome of a bacterium that is naturally adapted to mice and identified many more putative host-interaction genes than were observed in previously recognized human pathogens. Furthermore, most of these genes are located in a segment of about 25% of the genome, which was massively amplified and packaged into viral particles. This is the first demonstration of targeted packaging of a portion of the bacterial chromosome into viral particles, and we propose that this is a novel strategy for increased exchange of genes involved in the infectious process.
Citation: Berglund EC, Frank AC, Calteau A, Vinnere Pettersson O, Granberg F, et al. (2009) Run-Off Replication of Host-Adaptability Genes Is Associated with Gene Transfer Agents in the Genome of Mouse-Infecting Bartonella grahamii. PLoS Genet 5(7): e1000546. doi:10.1371/journal.pgen.1000546
Editor: Ivan Matic, Université Paris Descartes, INSERM U571, France
Received: December 15, 2008; Accepted: June 1, 2009; Published: July 3, 2009
Copyright: © 2009 Berglund et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by grants to SGEA from the Swedish Research Council, the Goran Gustafsson Foundation, the European Union, the Swedish Foundation for Strategic Research, and the Knut and Alice Wallenberg Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Horizontal gene transfer contributes to phenomena such as pathogen emergence and antibiotics resistance, with major implications for human health. However, little is known about the mechanisms and ecological factors that influence the intensity of horizontal gene transfer in bacteria adapted to wild host populations. With about 170 cases of emerging infectious diseases during the last two decades, more than half of which are caused by bacterial infections , a better understanding of the mechanisms of transfer in natural isolates and how the spread of the mobile gene pool depends on environmental variables such as population size, number of infected hosts and transmission dynamics is needed.
Bartonella is a particularly good model system for studies of the mechanisms whereby genes are transferred in wild host populations and the selective forces leading to fixation of the transferred genes in the population. The genus consists of circa 20 described vector-borne species that infect erythrocytes and endothelial cells of mammals , including rodents that are major carriers of infectious disease agents. Four genomes have been sequenced to date, from Bartonella quintana, the agent of trench fever , Bartonella bacilliformis, the agent of Carrion's disease (unpublished), the cat-adapted Bartonella henselae which causes cat-scratch disease during incidental infection in human  and the rat-associated Bartonella tribocorum . With single, circular chromosomes in the 1.4–2.6 Mb range, Bartonella species have the most highly reduced genomes in the bacterial order Rhizobiales of the alpha-proteobacteria.
A recent genomic survey identified many putative host-adaptability genes essential for bloodstream infection of B. tribocorum in rats . The experimentally best characterized such genes in B. quintana, B. henselae and B. tribocorum are those coding for the type IV (VirB, Trw) and type V (adhesins, autotransporters) secretion systems. The VirB system translocates effector proteins into the cytoplasm of endothelial cells, thereby mediating anti-apoptotic activity and angiogenic reprogramming –, whereas the Trw system is required for invasion of erythrocytes  and displays diversifying selection on the pilus proteins . The Bartonella adhesin (BadA) is involved in binding to endothelial cells, autoaggregation, and mediates a proangiogenic response in B. henselae ,. The homologous proteins in B. quintana have essentially the same function, and are required for bloodstream infection, but have been termed Vomps (variably expressed outer membrane proteins) since they have been shown to undergo sequence and expression variation in a monkey model system –. Less is known about the autotransporters, however they are known to be upregulated during infection of endothelial cells  and may be involved in adhesion to host cells .
Several of the genes for secretion systems are located in a dynamic region of the genome, up to several hundred kb in size, which is thought to have originated by the integration of an auxiliary replicon . DNA from this region was highly amplified in B. henselae upon prolonged growth . Microarray hybridizations suggested bi-directional replication starting from a region encoding a few phage genes . A similar phenomenon has also been observed in lambdoid phages of Salmonella , and has been termed escape replication or run-off replication. It has been suggested to arise from defects in prophage excision ,. Despite the replication of large amounts of the bacterial chromosome, the prophage that induced escape replication in Salmonella produced phage particles that only contained phage DNA .
Bacteriophage particles have been identified in several species of Bartonella – and growth experiments have suggested phage induction and lysis of B. henselae cells as they enter stationary phase . While the morphology of the bacteriophage particles differed slightly – phages from B. bacilliformis and Bartonella vinsonii ssp. berkhoffi were tailed, whereas those from B. henselae and B. quintana lacked tails – they all contained 14 kb linear, double-stranded DNA, packaged in a round to icosahedral head of 40–80 nm in size ,,. No one has been able to fully characterize the 14 kb band, however it has been suggested to consist of random or heterogeneous bacterial DNA ,,.
To learn more about the mechanisms and selective forces driving run-off replication, we determined the genome sequence of Bartonella grahamii strain as4aup, and studied the DNA content of bacteriophage particles from this species. B. grahamii has been isolated from several species of mice and voles and is likely to be one of the most prevalent Bartonella species in wild rodents –. It is transmitted by the rodent flea  and has been involved in two reported cases of human disease ,. The sequenced strain was isolated from a wood mouse (Apodemus sylvaticus) in central Sweden . Through its broad host range, B. grahamii has access to a large gene pool. Since our sequenced isolate has undergone minimal cultivation in the laboratory, we expect it to represent a Bartonella genome in nature, with intact accessory DNA.
We demonstrate that the highly dynamic region of the chromosome, that contains many gene clusters for secretion systems, is extensively amplified and packaged into bacteriophage particles. This is the first report that associates run-off replication with bacteriophage particles. We propose that the combination of these two systems promotes diversification and rapid spread of selectively favored host-adaptability genes within and among Bartonella populations, facilitating host shifts.
Comparative Genomics of Bartonella
B. grahamii strain as4aup contains a single circular chromosome of 2,341,328 bp with a predicted set of 1737 genes (Table 1). Additionally, 31 genes were identified on a plasmid of 28,192 bp, called pBGR3 in analogy with two plasmids detected in another strain of B. grahamii . We annotated 236 sequences as pseudogenes, most of which are remnants of full-length B. grahamii genes. The five sequenced Bartonella genomes are largely collinear albeit with inversions across the terminus of replication in all lineages (Figure 1). A segment downstream of and including the second rRNA operon is inverted across ori in B. bacilliformis.
A schematic illustration of the phylogenetic relationship of the five sequenced Bartonella species is shown to the left of a linear representation of their genomes. The total size of each genome is shown within parenthesis. Genes are color-coded based on phylogenetic classifications and annotation. Grey lines between genes indicate orthology. Bt: B. tribocorum, Bg: B. grahamii, Bh: B. henselae, Bq: B. quintana, Bb: B. bacilliformis.
The Bartonella core genome, as defined by the number of predicted orthologs in all five species, consists of 904 genes. In addition to this conserved set of genes, we identified sixteen segments in a total of 732 kb in B. grahamii that are highly variable in comparison to the previously published Bartonella genomes, here called genomic islands (Table 2). Eight of these islands (BgGI 9–16) are located in close proximity to each other in a segment of 461 kb that contains 282 genes. The high density of islands indicates that this region is particularly susceptible to foreign DNA and we will refer to this highly atypical segment of the genome as the chromosomal high plasticity zone.
High fraction of imported genes in the rodent-associated species.
A prediction of the total amount of horizontally transferred genes by phylogenetic analysis (excluding plasmid genes and hypothetical genes solely identified in Bartonella) classified 729 imported genes (35%) in B. tribocorum and 457 such genes (26%) in B. grahamii (Figure 2). The human-specific pathogens B. quintana and B. bacilliformis have the smallest genomes with less than 125 imported genes each (~10%). Only a few genes were initially identified in B. quintana that were absent or pseudogenized in B. henselae . Our analysis shows that these genes have homologs in the rodent-associated Bartonella species; substantiating the hypothesis that B. quintana has evolved by lineage-specific losses with no novel gene acquisitions . B. bacilliformis contains a few imported genes solely identified in this lineage, including 15 copies of a hypothetical gene that is also present in Leptospira interrogans. However, because there are no known close relatives of B. bacilliformis, we cannot infer the direction of evolution (loss or gain) of genes specific to this species.
Diagram showing the number of genes that are predicted as vertically inherited (from closely related alpha-proteobacterial species), imported (horizontally transferred to the ancestor of Bartonella or more recently), Bartonella-specific (only present in Bartonella), and orphans (only present in one species) in the sequenced Bartonella genomes.
A functional dissection of the 457 imported genes in B. grahamii shows that one third code for bacteriophage proteins and most of the remaining for secretion systems, membrane proteins, transport proteins or hypothetical proteins (Table 3). The much higher number of imported genes in the rodent-associated lineage can be attributed both to novel genes and to a higher copy-number of genes that are also present in the other species. Specific to this lineage are for example genes for membrane proteins in the genomic island BgGI 3, which is located in a region where the other three genomes display undisrupted synteny (Figure S1). Also unique to the rodent-associated species are genes for ubiquinol-cytochrome c reductase and a hypothetical protein with similarity to a genomic island gene in Photorhabdus luminescens, which is present in seventeen and nine copies in B. grahamii and B. tribocorum, respectively. Many of the genes solely identified in B. tribocorum are located in a prophage present in four genomic islands.
Expansion-Contraction of the Chromosomal High Plasticity Zone
Sequence similarity to auxiliary replicons.
Our analysis suggests dramatic changes in the mobile gene pool in the rodent- versus the human-associated Bartonella lineages. Because many of the expansion-contraction processes have occurred in the high plasticity zone, we first re-examined the hypothesis that this region was initially derived from the integration of an auxiliary replicon . To this end, we blasted all B. grahamii genes against the genomes of Brucella suis, Ochrobactrum anthropi and Agrobacterium tumefaciens, three related species with one major chromosome and a large auxiliary replicon. We found that top hits (E<1e−10) to the auxiliary replicons were circa 2.5–3 times more frequent in the high plasticity zone, with most other such hits located within or in the vicinity of the genomic islands (Figure 3).
The innermost circle represents the genome of B. grahamii, with prophages in yellow, genomic islands in magenta and the chromosomal high plasticity zone in black. Numbers within the circle represent the genomic islands BgGI 1–16. The remaining circles show on which replicon the top Blast hit of each B. grahamii gene is located in each of Brucella suis (Bs; inner circle), Ochrobactrum anthropi (Oa; middle circle), and Agrobacterium tumefaciens (At; outer circle). The genome positions are shown on the outside, with 100 kb between each line.
A closer inspection of the high plasticity zone revealed two conserved segments, here called core IIa and core IIb, which are present in all sequenced Bartonella genomes (Figure 4). These segments are located between the two rRNA operons in B. bacilliformis, B. henselae and B. quintana, suggesting that this is the ancestral organization. In the rodent-associated species, the segment core IIb has been rearranged to a position immediately upstream of the first rRNA operon. Despite their conservation in Bartonella, the genes in core IIb have top hits to genes on auxiliary replicons, favoring their integration with an auxiliary replicon.
Comparative gene map of the chromosomal high plasticity zone in the five sequenced Bartonella genomes. The locations of T4SS (virB/vbh) and T5SS (autotransporters, fha/hec) are shown. Species abbreviations are as in the legend to Figure 1. The total size of the region in each species is shown within parentheses. Due to the rearrangement around the replication origin in B. bacilliformis, the chromosomal high plasticity zone is divided into two parts, located on different sides of the origin.
Acquisition and evolution of genes for secretion systems.
Located between or near the conserved segments core IIa and core IIb are gene clusters for type IV secretion systems (T4SS) in all species except B. bacilliformis, and type V secretion systems (T5SS) in all species. In B. grahamii, we identified four operons for T4SS, one near core IIb (virB), one at the far end of the high plasticity zone (vbh-1), one downstream of the high plasticity zone (trw), and the fourth being plasmid-encoded (vbh-2). The T5SSs in the genomic islands of B. grahamii include fourteen genes for autotransporters, six copies of the two-component fha/hec operon and five copies of the Bartonella adhesin. The number of genes for T4SSs and T5SSs is highly variable across species, ranging from only eight genes in B. bacilliformis to more than one hundred in B. grahamii (Table 4).
The autotransporters are ubiquitously present in all five Bartonella genomes and their genomic locations suggest that they were ancestrally positioned in between core IIa and core IIb. A phylogeny of the transmembrane domain of the autotransporters revealed two distinct Bartonella clades, both of which contain all species (Figure 5A). B. bacilliformis represents the earliest diverging lineage in both clades, suggesting that the acquisition event(s) extends back to the root of the Bartonella tree. Due to a rearrangement in the rodent-associated lineage, the autotransporter genes in the high plasticity zone of these species are physically separated into two clusters with 5–7 genes in each cluster. In B. grahamii, these genes are located in BgGI 9 and BgGI 12 (Figure 4). Their physical separation corresponds to their phylogenetic divergence, suggesting that each of the two clusters have expanded and contracted independently through a series of tandem gene duplication and/or deletion events.
Phylogenetic trees of (A) autotransporters and (B) filamentous hemagglutinin. Bartonella genes are named with locus_tag (BARBAKC583 abbreviated as BB) and color-coded according to species: B. grahamii in red, B. tribocorum in blue, B. henselae in green, B. quintana in purple and B. bacilliformis in brown. After each B. grahamii gene, the genomic island in which the gene is located is indicated with numbers in circles. Branch lengths are according to maximum likelihood analysis and numbers represent bootstrap support values.
The virB operon is conservatively placed between one cluster of autotransporter genes and core IIb, however, in the rodent-associated species this whole region has been translocated to a position immediately upstream of the first rRNA operon (Figure 4). The virB operon is absent from the B. bacilliformis genome and has been suggested to be acquired in a recent branch of the Bartonella tree . Although homologs to the virB genes are circulating in the Rhizobiales auxiliary gene pool, the gene phylogenies do not match the species phylogeny, supporting horizontal gene acquisition .
Of the genes for secretion systems, the fha/hec gene cluster, which codes for filamentous hemagglutinin (FHA) and a hemolysin activator protein, is most labile in its presence/absence pattern. In the B. grahamii genome fha/hec is present in three copies in BgGI 4 and in one copy in BgGI 10, BgGI 11 and BgGI 14. The operon is located within a segment that is 7-fold repeated in the genome, hereafter referred to as the fha-repeat. The repeat is flanked by integrase and phage genes and varies in length from 10 to 28 kb, with nucleotide sequence identities ranging from 80 to 100%. Phylogenetic analysis of FHA reveals two distinct Bartonella clades (Figure 5B), both of which have their closest relative in other species, suggesting that the fha/hec genes in Bartonella were acquired twice independently. However, unlike the autotransporter genes, the fha/hec genes do not seem to have been located between the rRNA genes in the common ancestor, but were inserted into the high plasticity zone in B. grahamii very recently. Since the fha/hec operon, whenever present, is located in the proximity of phage genes, it might integrate, rearrange and excise at high frequencies.
Bacteriophage Genes and Particles
Since phage related genes contribute much of the variability in Bartonella, we classified the most abundant such genes in B. grahamii into four phage clusters (Figure 6A), the largest of which is duplicated and referred to as prophage Ia (Figure 6B) and prophage Ib. This prophage has a size of circa 45 kb and contains one stretch of genes for capsid and portal proteins with similarity to phage lambda and another stretch encoding baseplate and tail proteins with similarity to phage P2. Many of the genes have homologs in prophages in Vibrio harveyi, Wolbachia pipientis and Ochrobactrum anthrophi, all with some conservation of gene order. The presence of genes in B. bacilliformis with similarity to those in prophage I, its absence in the closely related Brucella spp. and its equidistant relationship to unrelated species suggest that the prophage was acquired in an ancestor of the Bartonella species.
(A) Circular overview of the B. grahamii as4aup genome. The locations of phage clusters (yellow), genomic islands (magenta), the chromosomal high plasticity zone (black) and the gene with similarity to S. thermophilus phage Sfi18 (blue) are shown. Numbers outside the circle represent the phage clusters I-IV. Numbers within the circle represent the genomic islands BgGI 1–16. Detailed comparative gene maps are shown for (B) prophage Ia, (C) phage cluster II, and (D) phage cluster III. The color legend shown in (D) also applies to (B) and (C). Genes identified in mass spectrometry are marked with black dots. The putative origin of replication identified in the microarray analysis is located close to the helicase in (D).
Several shorter clusters of phage genes may represent remnants of other integrated phages. Two such clusters, here called phage cluster II (Figure 6C) and phage cluster III (Figure 6D) are located in the chromosomal high plasticity zone and are well conserved in the Bartonella genomes. Phage cluster II includes genes for lambda-like phage portal and terminase proteins. Paralogs to some of these genes are also present in BgGI 4 and BgGI 5. Phage cluster III encodes a helicase with similarity to bacteriophage replication proteins and an exonuclease; these two conserved genes are located near to a region of mostly non-coding DNA that is variable in size across the Bartonella genomes.
Phage cluster IV contains genes for helicase (dnaB), phage late control protein D (gpD), phage tail X (gpX) and phage tail U (gpU) and is located adjacent to autotransporter genes in BgGI 5, BgGI 9 and BgGI 12. Among the sequenced Bartonella species, the genes in phage cluster IV are only present in B. tribocorum, where they belong to a prophage that is 35 kb in size and is similar to bacteriophage Mu. These results suggest that this phage integrated in the common ancestor of B. grahamii and B. tribocorum. Finally, B. grahamii has 29 copies of a gene with similarity to a gene from S. termophilus phage Sfi18, mostly in clusters of three genes within the fha-repeat.
We observed bacteriophage particles consisting of a round to icosahedral head of 50–70 nm in diameter and a sheathed tail of approximately 100 nm by transmission electron microscopy of bacteriophage preparations from B. grahamii strains af165up (Figure 7) and as4aup (Figure S2A). Many heads lacked tails, while there were few loose tails, suggesting the production of two morphologically different bacteriophage particles. Bacteriophage particles were also observed in B. henselae strain GreekCat-23 (Figure S2B), which was included for comparison, since it has been shown to lack prophage I .
Transmission electron microscopy of bacteriophage particles isolated from B. grahamii strain af165up. Round to icosahedral heads of 50–70 nm were observed both with and without tail, whereas there were few loose tails. The white bar is 100 nm.
We examined the protein profiles of the phage particles from all three strains using SDS-PAGE and mass spectrometry (Table S1). Two proteins from prophage I, corresponding to GpF1 and Gp20 (Figure 6B, genes with black dots), were identified in the B. grahamii strains, suggesting that prophage I is active in both these strains. The dominating protein in all strains was a phage related protein (Bgr_16730/BH14010) from phage cluster II (Figure 6C, genes with black dots). In B. henselae strain GreekCat-23 we also identified two other proteins from phage cluster II, corresponding to BH13960, a putative phage tail protein, and hypothetical protein BH13990 (Figure 6C, genes with black dots). The protein encoded by BH13960 has previously been observed in bacteriophage preparations from B. henselae and referred to as PapA .
Additionally, hemin binding protein A (HbpA), also known as Pap31, was present in bacteriophage preparations from B. henselae and B. grahamii grown on hematin agar plates. However, when B. henselae was cultivated in liquid medium in the absence of hemin, we were unable to identify HbpA among the extracted phage proteins (Table S1). This suggests that HbpA is not part of the bacteriophage, as has been suggested previously , but rather a bacterial outer membrane protein that is strongly expressed in the presence of hemin and co-precipitates with the phage particles.
We extracted bacteriophage DNA from all three strains, and agarose gel electrophoresis revealed two bands of circa 14 and 45 kb in B. grahamii strain af165up (Figure 8) and one band of circa 14 kb in B. grahamii strain as4aup (Figure 8) and B. henselae strain GreekCat-23 (Figure S3). While a 14 kb band has been observed in previous studies, this is to our knowledge the first report of a larger band in Bartonella phage DNA.
Agarose gel electrophoresis of B. grahamii bacteriophage DNA. Lane 1, Marker (Low Range PFG Marker, New England Biolabs); lane 2, phage DNA from B. grahamii af165up; lane 3, phage DNA from B. grahamii as4aup; lane 4, cellular DNA from B. grahamii af165up; lane 5, cellular DNA from B. grahamii as4aup; lane 6, marker. The sizes of the relevant bands of the marker are shown to the left.
Massive Amplification of the Chromosomal High Plasticity Zone
Amplification and packaging of the chromosomal high plasticity zone and prophage I into phage particles.
To examine the gene content of the bacteriophage DNA, we constructed a microarray from 4,438 oligomers of 60 bp in size, which covers 1703 protein-coding genes (96% of the predicted protein-coding genes in B. grahamii), 110 pseudogenes and 663 intergenic regions. We hybridized bacteriophage DNA from B. grahamii strains as4aup and af165up against cellular DNA from the same strain harvested at the same time point. In both strains, we observed an increase in hybridization signal over the chromosomal high plasticity zone and until ori, with the strongest signal in the centre and gradually decreasing in both directions (Figure 9AB). The peak, at which the signal is approximately 30-fold stronger in the phage DNA compared to the cellular DNA, is located close to or within the helicase in phage cluster III (Figure 6D). This site is homologous to the site where the amplification peak was observed in hybridizations of genomic DNA from B. henselae . In strain af165up, we observed a strong amplification of prophage I, which is also visible in strain as4aup but to a much smaller extent.
Results from microarray hybridizations of phage DNA versus cellular DNA of the same strain. The x-axis represents the genome of B. grahamii as4aup and the y-axis the hybridization signal (log2-ratio of phage DNA and cellular DNA). To exclude possible misinterpretation of repeated probes, probes with more than one exact match in the genome were not plotted unless located in the prophage regions. (A) Total phage DNA from B. grahamii as4aup. (B) Total phage DNA from B. grahamii af165up. Phage DNA extracted from (C) the 45 kb band and (D) the 14 kb band in agarose electrophoreses from B. grahamii af165up. Above the x-axis in all graphs is a representation of the genome of B. grahamii as4aup with the same color-coding as in Figure 1. In particular, phage genes are yellow. Grey dots in (C) and (D) show the result of the hybridization of total phage DNA from B. grahamii af165up (Figure 9B).
To specifically assess the content of the 14 kb and the 45 kb bands in strain af165up, we extracted DNA from each band, and hybridized against cellular DNA. The results showed that the hybridization signal from the 45 kb band was concentrated to prophage I (Figure 9C), in agreement with the predicted size of this prophage. Although no 45 kb band was visible in phage DNA extracted from strain as4aup, we believe that prophage I is active also in this strain, albeit to a much lower extent, since proteins encoded by genes in this bacteriophage were identified in mass spectrometry, and this region was slightly over-represented in the total phage DNA. The 14 kb band contained DNA from the entire genome, with an over-representation of the amplified DNA from the chromosomal high plasticity zone (Figure 9D). This result is in agreement with previous descriptions of random or heterogenic DNA, and explains why others have experienced so many difficulties when trying to characterize this DNA.
The prophage and the chromosomal high plasticity zone are amplified and packaged into phage particles during all growth phases.
To examine whether induction of the prophage and run-off replication occur predominantly in dying cells, or possibly is the cause of bacterial death, we collected phage particles from three different time points from a liquid culture of B. grahamii strain af165up, selected to represent exponential growth phase (growing, viable), stationary phase (non-growing, viable) and death phase (non-growing, non-viable) (Figure 10). We extracted phage DNA from each time point to hybridize against chromosomal DNA extracted from the exponential growth phase. To obtain enough DNA for the hybridization experiments, we amplified the phage DNA using whole genome amplification. This procedure will reveal differences in the relative abundance of different genes in the phage versus the chromosomal DNA samples, but does not show differences in the total amount of phage DNA produced at the three growth stages. The results showed clearly that replication of prophage I is initiated already during the exponential growth phase with prophage DNA being over-represented in the phage particles at all three time points (Figure 11). Likewise, the products of run-off replication were present in relatively higher quantities in the isolated phage particles than in the preparations of chromosomal DNA at all time points.
Growth curve of B. grahamii strain af165up in supplemented Schneider's medium. Bacterial growth was determined by measuring the OD600 in triplicates (grey line) and by quantifying the number of viable bacteria, expressed as CFU/ml (black line), at 24-h intervals. Samples were collected at the mid-logarithmic growth phase (day 2), the stationary phase (day 5), and at the end of the death phase (day 11).
Results from microarray hybridizations of DNA extracted from phage particles in (A) exponential phase (B) stationary phase and (C) end of death phase. To exclude possible misinterpretation of repeated probes, probes with more than one exact match in the genome were not plotted unless located in the prophage regions.
In this study, we have resolved the long-standing debate about the content of the Bartonella bacteriophage particles. Altogether, our data suggests the presence of two distinct bacteriophages in B. grahamii, one of which is encoded by prophage I and packages its own DNA. The other bacteriophage, which is encoded by genes in phage cluster II, contains chromosomal DNA with an over-representation of DNA from the high plasticity zone. Bacteriophage particles that package random bacterial DNA rather than their own, with no lysis or other obvious negative consequences for the host cell are called gene transfer agents (GTAs). They are usually encoded by a segment of circa 15 kb that contains phage head and tail genes and they transfer DNA in 4–14 kb pieces –. These characteristics agree well with our observations, suggesting that the bacteriophage packaging the 14 kb band is a GTA. To our knowledge, this is the first demonstration of a GTA that packages chromosomal DNA that has first been partially amplified by run-off replication.
The identification of a GTA encoded by phage cluster II explains the previously puzzling observation of bacteriophage particles containing 14 kb DNA in many Bartonella species, including B. quintana and B. bacilliformis, that lack prophage I. A gene transfer agent, called RcGTA, which packages the genome in 4.5 kb pieces has previously been identified in the alpha-proteobacterial species Rhodobacter capsulatus . Homologous gene clusters are widespread in the alpha-proteobacteria and the congruence in phylogenetic relationships inferred from rRNA and RcGTA genes suggests vertical transmission from a single GTA-containing ancestor . However, no such genes could be identified in any of the Bartonella genomes, although most other members of the Rhizobiales contain partial and/or rearranged RcGTA-like gene clusters. The genes in phage cluster II also show no similarity to other known GTAs, suggesting that the Bartonella GTA is of a novel kind.
Despite the absence of sequence similarity between these two GTAs in the alpha-proteobacteria, their overall organization is similar, including genes coding for terminase, portal protein, and putative capsid and tail proteins. We identified gene clusters homologous to the Bartonella GTA in three recently sequenced genomes from the Rhizobiales species Azorhizobium caulinodans, Methylobacterium radiotolerans and Rhodopseudomonas palustris (Table S2), all of which also contain the RcGTA-like gene cluster. The scattered occurrence of the Bartonella GTA gene clusters in the alpha-proteobacteria indicates horizontal transmission.
Could run-off replication be initiated from a plasmid replication initiation site inherited from a once self-replicating megaplasmid? Such a phenomenon is believed to have occurred in the archaea, where multiple replication initiation sites have been demonstrated ,, and suggested to have arisen from the capture of an extra-chromosomal plasmid-derived element . Arguing against a plasmid-derived origin in Bartonella is that in the region close to the peak of the amplification we observed no sequence similarity to the repABC genes, which drives the replication and contains the ori in the megaplasmids of the Rhizobiales . These genes are also not present anywhere else in the amplified region, providing no evidence for a megaplasmid-derived origin of replication.
It is more likely that replication starts from an origin derived from an inactivated prophage, as does escape replication in Salmonella. In this species, run-off replication can be induced in the laboratory by inactivating the integrase and/or excisionase genes , showing the ease with which a prophage can lose control of its own replication machinery. Capture of bacteriophage replication machineries has also been observed in mitochondria, where the nuclear genes for DNA polymerase and DNA primase-helicase (Twinkle) are derived from T7 bacteriophages . It is interesting to note that the helicase that is located at the peak of run-off replication in Bartonella is homologous to Twinkle, supporting the hypothesis that the extra origin of replication is phage-derived. Homologous genes are present also in several other bacterial genomes, raising the possibility that capture of bacteriophage systems for bacterial replication may occur more frequently than recognized until now. However, no genes for the T7-like DNA or RNA polymerases could be found in the Bartonella genomes, suggesting that replication and transcription of the chromosomal high plasticity zone is not simply driven by a T7-like bacteriophage.
An important question is whether run-off replication and GTA-production are two independent phenomena, resulting from an error in the phage replication machinery and a degrading prophage, or whether the systems driving these processes co-evolve under selection and control of the bacterial genome. Several lines of evidence argue in favor of selection, particularly the conservation of phage cluster III, containing the origin of run-off replication, and phage cluster II, encoding the GTA, in all sequenced Bartonella genomes. Normally, prophages degrade rapidly as observed for prophage I, which has been lost independently from B. quintana and individual B. henselae strains ,. This is in analogy to the maintenance and vertical inheritance of the RcGTA-like gene cluster among several alpha-proteobacterial species, which has been attributed to selection .
What could be the selective advantage of linking run-off replication with a gene transfer agent? One advantage that we can think of is gene diversification by recombination and rapid spread of new gene variants. Because many genes for secretion systems are located in the amplified region, host adaptation and host switches could select for linkage of the two systems. Indeed, rapid diversification and horizontal transfer of host-adaptability genes within Bartonella populations have been demonstrated for the trw gene cluster . Rodents in particular represent a large and genetically highly divergent host population, and multiple variants of surface proteins could help escape the immune system and promote interactions with a diverse set of host cell molecules. Under this hypothesis, it may be no coincidence that the rodent-associated Bartonella genomes contain the highest number of genes for secretion systems.
If run-off replication and production of GTAs are driven by selection, what are the cellular mechanisms and conditions responsible for regulating these processes? Our observation that the Bartonella GTAs containing amplified DNA are present already during exponential growth phase suggests that these genes are not induced to escape stressed or dying cells, nor that the GTAs are the cause of lysis and cell death. The relatively higher abundance of the products of run-off replication observed previously at later growth phase in B. henselae  is probably due to a gradual accumulation of amplified DNA with time. We can also exclude the possibility that run-off replication is regulated by prophage I, in trans, since B. henselae strain GreekCat-23, which lacks this prophage, still induces run-off replication . Although the Bartonella GTA may very well be under the control of bacterial genes, the regulatory circuits involved are probably different from the growth-dependent production of RcGTA by histidyl-aspartyl signaling proteins  as well as by quorum sensing through long-chain acyl-homoserine lactones .
An exciting avenue for future work is to investigate how expression of the Bartonella GTA is regulated and whether run-off replication is regulated by the same control systems. It would also be interesting to determine where and when run-off replication and bacteriophage production occur within the rodent host and its fleas and whether different genome variants are generated during the infectious process. The availability of both the B. grahamii and the mouse genome also provides a technical platform that will enable future studies on the co-evolution of bacterial and host genes in natural mouse populations.
Materials and Methods
Bacterial Strains and Culture Conditions
Bartonella grahamii strains as4aup and af165up were isolated from a wood mouse and a yellow-necked mouse, respectively, captured in the vicinity of Uppsala in central Sweden . Bartonella henselae strain GreekCat-23 was collected from a cat in the Thessaloniki area in Greece by Aphrodite Tea. The Bartonella strains were routinely grown on hematin agar plates in a humidified 5% CO2 incubator at 35°C. For growth in liquid culture, plate-grown bacteria were inoculated into Schneider's insect medium (Sigma) supplemented with 10% FBS and 5% sucrose as described by Riess et al. , additionally supplemented with 25 mM HEPES to keep the pH near 7. Numbers of viable bacteria were determined as colony-forming units (CFU) by plating 10-fold serial dilutions.
Whole Genome Sequencing
Starting from single colonies, B. grahamii strain as4aup was cultured on hematin agar plates for five days. Genomic DNA was randomly sheared by nebulization and 1–3 kb sized fragments were recovered. The extracted fragments were cloned into a modified M13 vector using the ‘double adaptor’ method . After 7–8 hours propagation in E. coli, ssM13 DNA was prepared for direct sequencing using Multi Screen MABCN1250 filter plates from Millipore combined with sodium-perchlorate lysis. Direct sequencing of the inserts was performed using DYEnamic ET Terminator Cycle Sequencing Kit (Amersham) on the MegaBACE 1000 DNA Analysis System according to the manufacturer's instructions. Cycle sequencing reaction cleanup was performed with the AutoSeq 96 Dye Terminator Clean-up Kit from Amersham.
A total of 31,166 shotgun sequences were obtained from the M13 library of B. grahamii. After assembly using the Phred and Phrap software ,, physical gaps were closed by short and long-range PCR using primers from contig ends. Long-range PCR products were sheared by nebulization and end repaired with DNA Terminator End Repair Kit from Lucigen and blunt-end cloned into pcSmartHCKan vectors using the Clonesmart cloning kit and E. coli XL2-Blue Ultracompetent Cells from Stratagene. Recombinant colonies were grown in 50 µl 2xYT/kanamycin at 37°C overnight without shaking. 1 µl of the culture was used for amplification with TempliPhi Amplification Kit (GE-Healthcare). Both ends of the inserts were sequenced as described above. All ambiguous sites were manually edited with Consed  by re-sequencing of PCR products when necessary. A number of polymorphic sites in the badA region remain unresolved, these may well represent authentic polymorphisms since this gene is known to evolve very rapidly but we cannot exclude the possibility that there should be an additional badA gene.
Southern Blotting and Assembly Verification
Bacteria were grown on hematin agar plates for 10 days. Prior to DNA extraction, the cells were suspended in TNE buffer (10 mM TRIS pH 8.0, 150 mM NaCl and 1 mM EDTA) and centrifuged; washes were repeated twice. DNA was isolated in agarose plugs, made using 2% SeaPlaque GTG agarose (Cambrex Bio Science, ME, USA) in 0.5× TBE buffer as described in , with minor modifications. One mm thick slices of DNA-containing plugs were digested with 10 U of NotI, AscI, RsrII, SbfI (New England Biolabs) and SgfI (Promega) restriction endonucleases separately, after 30 min pre-equilibration in TBE buffer on ice and subsequent 30 min pre-equilibration with an appropriate restriction buffer. Restriction was performed overnight at recommended conditions. The DNA fragments were separated in 0.9% PFGE-grade agarose (SeaKem Gold; Cambrex Bio Science) in 0.5× TBE buffer in the GenNavigator System apparatus (Amersham Biosciences) at 14°C and 5.6 V/cm, for a total of 65 hours. The total run was separated in six phases, using switch times ramped from 5 to 150 s. The sizes of the fragments were estimated using PFGE l-ladder and Yeast Chromosome PFG marker (New England Biolabs). Thirty-seven DNA fragments serving as probes were amplified, purified and labeled as described earlier . Southern blotting, hybridization and signal detection were performed as described earlier .
NotI restriction digest was used to verify the assembly of the B. grahamii genome. DNA was digested and separated as described above, using 1.1% low melting point SeaPlaque agarose for casting the gel. The gel was run and bands were excised as suggested by the protocol extraction of high molecular weight DNA from gel . The blocks of gel containing individual bands of interest were excised and the agarose was digested using ß-agarase I (New England Biolabs) according to the manufacturer's instructions. From the reaction mix, DNA was extracted with equal volume of phenol: chloroform: isoamyl alcohol (25:24:1). The aqueous phase was cleaned up with equal volume of chloroform: isoamyl alcohol (24:1), and DNA was precipitated with 2 volumes ice-cold isopropanol at −20°C overnight, washed with 70% ethanol twice and re-dissolved in TE buffer pH 7.5. 1 µg DNA from each of the two smallest NotI bands was hybridized against total genomic DNA to a microarray (see below), to see which part of the genome the band represented.
Gene Prediction and Annotation
Curation and annotation of the genome were done with the annotation platform GenDB . Protein-coding genes were predicted with GLIMMER  and CRITICA . Similarity searches were performed against several databases, including GenBank (nt and nr), UNIPROT , KEGG  and COG . Protein domains were identified with InterProScan  and the InterPro database . SignalP , helix-turn-helix , and TMHMM  were used. RNA genes were identified with tRNA-Scan-SE  and ARAGORN .
Gene Classification, Definition of Genomic Islands, and Phylogenetic Analysis
Orthologs between Bartonella species were predicted as reciprocal best Blastp  hits, excluding genes with top hits in the same species. Paralogs were predicted using blastclust (http://www.ncbi.nlm.nih.gov/Web/Newsltr/Spring04/blastlab.html), requiring 50% similarity over 40% of the length. Phage genes were defined based on Blast hits and literature.
Genes were defined as orphans if there was no convincing Blast hit in the GenBank nr database. Genes were defined as Bartonella-specific if the only hits were to other Bartonella species. For remaining genes, all orthologs and paralogs along with Blast hits from a selection of 27 completely sequenced alpha-proteobacterial genomes were used for phylogenetic analysis. To get a broad representation of species we further selected top Blastp hits up to a total of 40 genes (E<1e−10), excluding all other alpha-proteobacterial hits. Protein sequences were aligned with clustalW , and phylogenetic trees were inferred with PHYML , using the JTT model of amino acid substitution, fixed proportion of variable sites, one substitution rate category, estimated gamma distribution parameter and 100 bootstrap replicates. To get a rough estimation of the number of imported and vertically inherited genes, we inspected all trees with less than 18 Blast hits to the selected alpha-proteobacterial genomes manually. Imported genes were defined as those for which there was no indication of vertical inheritance from the ancestor of Bartonella and its close relatives. The genes for Type IV secretion systems were classified as imported despite being common in the alpha-proteobacteria, since previous analyses suggest a recent acquisition of these genes . Gene classifications were transferred from B. grahamii to the other Bartonella species using best Blast hit (E<0.001). Genes in the other species without homologs in B. grahamii were manually classified.
Genomic islands were manually located in B. grahamii based on a synteny comparison with the previously published genomes (B. tribocorum, B. henselae and B. quintana). A genomic island was defined as a contiguous segment larger than 10 kb containing orphans, non-orthologous genes, or long non-coding regions. Island boundaries were extended to include adjacent Bartonella-specific genes or imported genes. For simplicity, boundaries of islands were always defined as the boundaries of the adjacent non-island genes.
Protein sequences for FHA and autotransporters were aligned with Kalign  and phylogenetic trees were inferred with RAxML , substitution model PROTMIXWAG, 100 bootstrap replicates. Since the autotransporter genes are of very variable length, only the conserved beta-domain, identified with HMMPfam in InterProScan , was used for phylogenetic inference.
Isolation of Phage Particles, Extraction of Genomic and Phage DNA, and Mass Spectrometry
Bacterial cells were either harvested after 5 to 15 days of culture on hematin agar plates, or at selected time points during growth in liquid culture. While samples from liquid culture were collected directly, plate-grown bacteria were either overlaid with 5 ml Brain-Heart Infusion Broth (Difco) the day prior to collection or re-suspended in SM buffer. After centrifugation at ca. 4000×g for 15 min at 4°C, supernatants were collected, filtered through 0.2 µm filters and enriched using polyethylene glycol as described in . Phage pellets were re-suspended in 30 µl TE buffer pH 8.0 or PBS.
Extraction of DNA from the bacterial pellets was performed using the AquaPure Genomic DNA Kit (Bio-Rad) according to the manufacturer's instructions. Phage DNA from plate cultures was extracted as described in . Phage DNA from liquid culture was extracted in combination with amplification using the REPLI-g Mini kit (Qiagen) according to the manufacturer's protocol for DNA amplification from blood or cells. Concentrations were determined by measuring absorbance using a NanoDrop ND-1000.
Size selection of phage DNA fragments from non-amplified preparations was carried out by separation in 0.7% (w/v) low melting point agarose (SeaKem Plaque, Cambrex) gels containing ethidium bromide, in 1× TAE buffer at 4°C at 6 v/cm overnight. The fragments were visualized under near-UV light (λ = 360 nm) and excised with a sterile scalpel blade. Two volumes of ß-agarase I buffer (New England Biolabs) amended with 100 mM NaCl, 30 µM spermine and 70 µM spermidine  were added to each slice, followed by equilibration on ice three times for 30 min. Gel slices were melted at 65°C for 15 min, cooled to 42°C for 10 min and digested with 2 U of ß-agarase I (New England Biolabs) for one hour. ß-agarase was deactivated at 95°C for 5 minutes and the DNA was extracted by equal volume of phenol-chloroform-isoamyl alcohol (25:24:1) as described in . DNA from the 14 kb band of B. grahamii af165up was amplified using the REPLI-g Mini kit.
The protein components of the bacteriophage particles were separated by SDS-PAGE (12%) after being heat-denaturized at 95°C for 5 minutes. Mass spectrometry was performed for each band. Proteins were identified based on computational searches in the NCBI non-redundant (nr) database and a local database consisting of the B. grahamii protein sequences.
Microarray Design and Hybridization
We designed 4438 60-mer oligonucleotides, representing 1703 protein-coding genes, 110 pseudogenes and 663 spacers with OligoArray2.1 . A probe was considered to belong to a gene if at least half the probe was located within the predicted borders of the gene. The probes were spotted in three replicates on Ultra-GAPS-coated slides (Cornings, Inc.). Slides were cross-linked at 250 mJ/cm2, and prehybridized at 42°C for 45–60 min in 3X SSC, 0.1% SDS and 0.1 mg/ml bovine serum albumin. Labeling of genomic DNA was performed as in . Hybridizations were performed at 42°C overnight in hybridization solution containing 5X SSC, 0.1% SDS, 0.1 mg/ml sonicated salmon sperm DNA and 20% formamide.
Analysis of Microarray Data
Scanning and image analysis were performed with a GenePix 4100A scanner and the GenePix 5.1 software (Axon Instruments, Molecular Devices) as described previously . Phage and bacterial DNA are referred to as Ch1 and Ch2 respectively, and spots meeting the following criteria were removed from further analysis: Spots flagged as bad or not found during quantification, spots with more than 5% saturated pixels in either channel, Ch1 spot median below 3 times the Ch1 background median, less than 95% of pixels having a Ch1 intensity higher than background intensity plus one standard deviation, less than 90% of pixels having a Ch1 intensity higher than background intensity plus two standard deviations or spots having less than 70 pixels. M values were computed as log2(Ch1/Ch2).
The genome sequence of B. grahamii has been deposited in GenBank under the accession numbers CP001562 for the chromosome and CP001563 for pBGR3. The microarray data have been deposited in the ArrayExpress database of the European Bioinformatics Institute under the accession numbers A-MEXP-1576 for the array design and E-MEXP-2149 for the experimental data.
Comparative gene map of the genomic region containing BgGI 1.
(0.27 MB PDF)
Electron micrographs of bacteriophage particles isolated from cultivations with B. grahamii and B. henselae. (A) Phage particles isolated from B. grahamii as4aup were observed both with and without tails. (B) Phage particles isolated from B. henselae GreekCat-23 without visible tails. The black bars are 100 nm.
(0.34 MB PDF)
PFGE migration of undigested DNA from bacteriophage particles isolated from B. henselae GreekCat-23. The arrow indicates a 14 kb DNA band. Lane 1, low-range PFGE marker; lane 2, empty; lane 3, lambda mix marker; lane4, GreekCat-23 bacteriophage DNA.
(0.13 MB PDF)
Proteins identified in mass spectrometry of bacteriophage preparations from B. grahamii and B. henselae.
(0.06 MB PDF)
Rhizobiales species with homologs to the genes in phage cluster II.
(0.07 MB PDF)
We thank Kirsten Ellegaard at the Department of Molecular Evolution, Uppsala University, for aid with whole-genome amplification of the extracted phage DNA; Leif Ljung at the Department of Medical Cell Biology, Uppsala University, for assistance with electron microscopy; and Åke Engström at the Department of Medical Biochemistry and Microbiology, Uppsala University, for assistance with mass spectrometry. We thank Annelie Waldén and Max Käller at the Royal Institute of Technology, Stockholm, Sweden, for microarray printing.
Conceived and designed the experiments: SGEA. Performed the experiments: ECB OVP FG ASE KN. Analyzed the data: ECB ACF AC SGEA. Contributed reagents/materials/analysis tools: MH HL. Wrote the paper: ECB ACF SGEA.
- 1. Jones KE, Patel NG, Levy MA, Storeygard A, Balk D, et al. (2008) Global trends in emerging infectious diseases. Nature 451: 990–3.
- 2. Dehio C (2001) Bartonella interactions with endothelial cells and erythrocytes. Trends Microbiol 9: 279–85.
- 3. Alsmark CM, Frank AC, Karlberg EO, Legault B, Ardell DH, et al. (2004) The louse-borne human pathogen Bartonella quintana is a genomic derivative of the zoonotic agent Bartonella henselae. Proc Natl Acad Sci U S A 101: 9716–21.
- 4. Saenz HL, Engel P, Stoeckli MC, Lanz C, Raddatz G, et al. (2007) Genomic analysis of Bartonella identifies type IV secretion systems as host adaptability factors. Nat Genet 39: 1469–76.
- 5. Schulein R, Dehio C (2002) The VirB/VirD4 type IV secretion system of Bartonella is essential for establishing intraerythrocytic infection. Mol Microbiol 46: 1053–67.
- 6. Schulein R, Guye P, Rhomberg TA, Schmid MC, Schröder G, et al. (2005) A bipartite signal mediates the transfer of type IV secretion substrates of Bartonella henselae into human cells. Proc Natl Acad Sci U S A 102: 856–61.
- 7. Schmid MC, Scheidegger F, Dehio M, Balmelle-Devaux N, Schulein R, et al. (2006) A translocated bacterial protein protects vascular endothelial cells from apoptosis. PLoS Pathog 2: e115. doi:10.1371/journal.ppat.0020115.
- 8. Seubert A, Hiestand R, de la Cruz F, Dehio C (2003) A bacterial conjugation machinery recruited for pathogenesis. Mol Microbiol 49: 1253–66.
- 9. Nystedt B, Frank AC, Thollesson M, Andersson SGE (2008) Diversifying selection and concerted evolution of a type IV secretion system in Bartonella. Mol Biol Evol 25: 287–300.
- 10. Kaiser PO, Riess T, Wagner CL, Linke D, Lupas AN, et al. (2008) The head of Bartonella adhesin A is crucial for host cell interaction of Bartonella henselae. Cell Microbiol 10: 2223–34.
- 11. Riess T, Andersson SGE, Lupas A, Schaller M, Schäfer A, et al. (2004) Bartonella adhesin a mediates a proangiogenic host cell response. J Exp Med 200: 1267–78.
- 12. Zhang P, Chomel BB, Schau MK, Goo JS, Droz S, et al. (2004) A family of variably expressed outer-membrane proteins (Vomp) mediates adhesion and autoaggregation in Bartonella quintana. Proc Natl Acad Sci U S A 101: 13630–5.
- 13. Schulte B, Linke D, Klumpp S, Schaller M, Riess T, et al. (2006) Bartonella quintana variably expressed outer membrane proteins mediate vascular endothelial growth factor secretion but not host cell adherence. Infect Immun 74: 5003–13.
- 14. MacKichan JK, Gerns HL, Chen Y, Zhang P, Koehler JE (2008) A SacB mutagenesis strategy reveals that the Bartonella quintana variably expressed outer membrane proteins are required for bloodstream infection of the host. Infect Immun 76: 788–95.
- 15. Seubert A, Schulein R, Dehio C (2002) Bacterial persistence within erythrocytes: a unique pathogenic strategy of Bartonella spp. Int J Med Microbiol 291: 555–60.
- 16. Litwin CM, Rawlins ML, Swenson EM (2007) Characterization of an immunogenic outer membrane autotransporter protein, Arp, of Bartonella henselae. Infect Immun 75: 5255–63.
- 17. Lindroos H, Vinnere O, Mira A, Repsilber D, Näslund K, et al. (2006) Genome rearrangements, deletions, and amplifications in the natural population of Bartonella henselae. J Bacteriol 188: 7426–39.
- 18. Frye JG, Porwollik S, Blackmer F, Cheng P, McClelland M (2005) Host gene expression changes and DNA amplification during temperate phage induction. J Bacteriol 187: 1485–92.
- 19. Fukasawa T, Hirai K, Segawa T, Obonai K (1978) Regional replication of the bacterial chromosome induced by derepression of prophage lambda. IV. Escape synthesis of gal operon in phage 82. Mol Gen Genet 167: 83–93.
- 20. Umemori E, Sasaki Y, Amano K, Amano Y (1992) A phage in Bartonella bacilliformis. Microbiol Immunol 36: 731–6.
- 21. Anderson B, Goldsmith C, Johnson A, Padmalayam I, Baumstark B (1994) Bacteriophage-like particle of Rochalimaea henselae. Mol Microbiol 13: 67–73.
- 22. Barbian KD, Minnick MF (2000) A bacteriophage-like particle from Bartonella bacilliformis. Microbiology 146 (Pt 3): 599–609.
- 23. Chenoweth MR, Somerville GA, Krause DC, O'Reilly KL, Gherardini FC (2004) Growth characteristics of Bartonella henselae in a novel liquid medium: primary isolation, growth-phase-dependent phage induction, and metabolic studies. Appl Environ Microbiol 70: 656–63.
- 24. Maggi RG, Breitschwerdt EB (2005) Isolation of bacteriophages from Bartonella vinsonii subsp. berkhoffii and the characterization of Pap31 gene sequences from bacterial and phage DNA. J Mol Microbiol Biotechnol 9: 44–51.
- 25. Holmberg M, Mills JN, McGill S, Benjamin G, Ellis BA (2003) Bartonella infection in sylvatic small mammals of central Sweden. Epidemiol Infect 130: 149–57.
- 26. Jardine C, Appleyard G, Kosoy MY, McColl D, Chirino-Trejo M, et al. (2005) Rodent-associated Bartonella in Saskatchewan, Canada. Vector Borne Zoonotic Dis 5: 402–9.
- 27. Telfer S, Begon M, Bennett M, Bown KJ, Burthe S, et al. (2007) Contrasting dynamics of Bartonella spp. in cyclic field vole populations: the impact of vector and host dynamics. Parasitology 134: 413–25.
- 28. Inoue K, Maruyama S, Kabeya H, Yamada N, Ohashi N, et al. (2008) Prevalence and genetic diversity of Bartonella species isolated from wild rodents in Japan. Appl Environ Microbiol 74: 5086–92.
- 29. Bown KJ, Bennet M, Begon M (2004) Flea-borne Bartonella grahamii and Bartonella taylorii in bank voles. Emerg Infect Dis 10: 684–7.
- 30. Kerkhoff FT, Bergmans AM, van Der Zee A, Rothova A (1999) Demonstration of Bartonella grahamii DNA in ocular fluids of a patient with neuroretinitis. J Clin Microbiol 37: 4034–8.
- 31. Serratrice J, Rolain J, Granel B, Ene N, Conrath J, et al. (2003) Bilateral retinal artery branch occlusions revealing Bartonella grahamii infection. Rev Med Interne 24: 629–30.
- 32. Seubert A, Falch C, Birtles RJ, Schulein R, Dehio C (2003) Characterization of the cryptic plasmid pBGR1 from Bartonella grahamii and construction of a versatile Escherichia coli-Bartonella spp. shuttle cloning vector. Plasmid 49: 44–52.
- 33. Frank AC, Alsmark CM, Thollesson M, Andersson SGE (2005) Functional divergence and horizontal transfer of type IV secretion systems. Mol Biol Evol 22: 1325–36.
- 34. Anderson B, Scotchlas D, Jones D, Johnson A, Tzianabos T, et al. (1997) Analysis of 36-kilodalton protein (PapA) associated with the bacteriophage particle of Bartonella henselae. DNA Cell Biol 16: 1223–9.
- 35. Bowers TJ, Sweger D, Jue D, Anderson B (1998) Isolation, sequencing and expression of the gene encoding a major protein from the backteriophage associated with Bartonella henselae. Gene 206: 49–52.
- 36. Lang AS, Beatty JT (2000) Genetic analysis of a bacterial genetic exchange element: the gene transfer agent of Rhodobacter capsulatus. Proc Natl Acad Sci U S A 97: 859–64.
- 37. Matson EG, Thompson MG, Humphrey SB, Zuerner RL, Stanton TB (2005) Identification of genes of VSH-1, a prophage-like gene transfer agent of Brachyspira hyodysenteriae. J Bacteriol 187: 5885–92.
- 38. Rapp BJ, Wall JD (1987) Genetic transfer in Desulfovibrio desulfuricans. Proc Natl Acad Sci U S A 84: 9128–9130.
- 39. Bertani G (1999) Transduction-like gene transfer in the methanogen Methanococcus voltae. J Bacteriol 181: 2992–3002.
- 40. Anderson LM, Bott KF (1985) DNA packaging by the Bacillus subtilis defective bacteriophage PBSX. J Virol 54: 773–80.
- 41. Lang AS, Beatty JT (2007) Importance of widespread gene transfer agent genes in alpha-proteobacteria. Trends Microbiol 15: 54–62.
- 42. Lundgren M, Andersson A, Chen L, Nilsson P, Bernander R (2004) Three replication origins in Sulfolobus species: synchronous initiation of chromosome replication and asynchronous termination. Proc Natl Acad Sci U S A 101: 7046–51.
- 43. Robinson NP, Dionne I, Lundgren M, Marsh VL, Bernander R, et al. (2004) Identification of two origins of replication in the single chromosome of the archaeon Sulfolobus solfataricus. Cell 116: 25–38.
- 44. Robinson NP, Bell SD (2007) Extrachromosomal element capture and the evolution of multiple replication origins in archaeal chromosomes. Proc Natl Acad Sci U S A 104: 5806–11.
- 45. Cevallos MA, Cervantes-Rivera R, Gutiérrez-Ríos RM (2008) The repABC plasmid family. Plasmid 60: 19–37.
- 46. Shutt TE, Gray MW (2006) Bacteriophage origins of mitochondrial replication and transcription proteins. Trends Genet 22: 90–5.
- 47. Schaefer AL, Taylor TA, Beatty JT, Greenberg EP (2002) Long-chain acyl-homoserine lactone quorum-sensing regulation of Rhodobacter capsulatus gene transfer agent production. J Bacteriol 184: 6515–21.
- 48. Riess T, Dietrich F, Schmidt KV, Kaiser PO, Schwarz H, et al. (2008) Analysis of a novel insect cell culture medium-based growth medium for Bartonella species. Appl Environ Microbiol 74: 5224–7.
- 49. Andersson B, Wentland MA, Ricafrente JY, Liu W, Gibbs RA (1996) A “double adaptor” method for improved shotgun library construction. Anal Biochem 236: 107–13.
- 50. Ewing B, Hillier L, Wendl MC, Green P (1998) Base-calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res 8: 175–85.
- 51. Ewing B, Green P (1998) Base-calling of automated sequencer traces using phred. II. Error probabilities. Genome Res 8: 186–94.
- 52. Gordon D, Abajian C, Green P (1998) Consed: a graphical tool for sequence finishing. Genome Res 8: 195–202.
- 53. Cornillot E, Croux C, Soucaille P (1997) Physical and genetic map of the Clostridium acetobutylicum ATCC 824 chromosome. J Bacteriol 179: 7426–34.
- 54. Sambrook J, Russell DW (2001) Molecular Cloning: A Laboratory Manual. 3rd ed. Cold Spring Harbor, NY: CSHL Press.
- 55. Meyer F, Goesmann A, McHardy AC, Bartels D, Bekel T, et al. (2003) GenDB–an open source genome annotation system for prokaryote genomes. Nucleic Acids Res 31: 2187–95.
- 56. Delcher AL, Harmon D, Kasif S, White O, Salzberg SL (1999) Improved microbial gene identification with GLIMMER. Nucleic Acids Res 27: 4636–41.
- 57. Badger JH, Olsen GJ (1999) CRITICA: coding region identification tool invoking comparative analysis. Mol Biol Evol 16: 512–24.
- 58. Wu CH, Apweiler R, Bairoch A, Natale DA, Barker WC, et al. (2006) The Universal Protein Resource (UniProt): an expanding universe of protein information. Nucleic Acids Res 34: D187–91.
- 59. Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M (2004) The KEGG resource for deciphering the genome. Nucleic Acids Res 32: D277–80.
- 60. Tatusov RL, Galperin MY, Natale DA, Koonin EV (2000) The COG database: a tool for genome-scale analysis of protein functions and evolution. Nucleic Acids Res 28: 33–6.
- 61. Quevillon E, Silventoinen V, Pillai S, Harte N, Mulder N, et al. (2005) InterProScan: protein domains identifier. Nucleic Acids Res 33: W116–20.
- 62. Mulder NJ, Apweiler R, Attwood TK, Bairoch A, Bateman A, et al. (2005) InterPro, progress and status in 2005. Nucleic Acids Res 33: D201–5.
- 63. Bendtsen JD, Nielsen H, von Heijne G, Brunak S (2004) Improved prediction of signal peptides: SignalP 3.0. J Mol Biol 340: 783–95.
- 64. Dodd IB, Egan JB (1990) Improved detection of helix-turn-helix DNA-binding motifs in protein sequences. Nucleic Acids Res 18: 5019–26.
- 65. Krogh A, Larsson B, von Heijne G, Sonnhammer EL (2001) Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J Mol Biol 305: 567–80.
- 66. Lowe TM, Eddy SR (1997) tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Res 25: 955–64.
- 67. Laslett D, Canback B (2004) ARAGORN, a program to detect tRNA genes and tmRNA genes in nucleotide sequences. Nucleic Acids Res 32: 11–6.
- 68. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. J Mol Biol 215: 403–10.
- 69. Chenna R, Sugawara H, Koike T, Lopez R, Gibson TJ, et al. (2003) Multiple sequence alignment with the Clustal series of programs. Nucleic Acids Res 31: 3497–500.
- 70. Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 52: 696–704.
- 71. Lassmann T, Sonnhammer ELL (2005) Kalign–an accurate and fast multiple sequence alignment algorithm. BMC Bioinformatics 6: 298.
- 72. Stamatakis A (2006) RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22: 2688–90.
- 73. Zdobnov EM, Apweiler R (2001) InterProScan–an integration platform for the signature-recognition methods in InterPro. Bioinformatics 17: 847–8.
- 74. Rouillard J, Zuker M, Gulari E (2003) OligoArray 2.0: design of oligonucleotide probes for DNA microarrays using a thermodynamic approach. Nucleic Acids Res 31: 3057–62.
- 75. Lindroos HL, Mira A, Repsilber D, Vinnere O, Näslund K, et al. (2005) Characterization of the genome composition of Bartonella koehlerae by microarray comparative genomic hybridization profiling. J Bacteriol 187: 6155–65.