Search tips
Search criteria 


Logo of iaiPermissionsJournals.ASM.orgJournalIAI ArticleJournal InfoAuthorsReviewers
Infect Immun. 2008 February; 76(2): 788–795.
Published online 2007 December 10. doi:  10.1128/IAI.01174-07
PMCID: PMC2223462

A SacB Mutagenesis Strategy Reveals that the Bartonella quintana Variably Expressed Outer Membrane Proteins Are Required for Bloodstream Infection of the Host[down-pointing small open triangle]


Bartonella bacteria adhere to erythrocytes and persistently infect the mammalian bloodstream. We previously identified four highly conserved Bartonella quintana adhesin genes that undergo phase variation during prolonged bloodstream infection. The variably expressed outer membrane proteins (Vomp) encoded by these genes are members of the trimeric autotransporter adhesin family. Each B. quintana Vomp appears to contribute a different adhesion phenotype, likely mediated by the major variable region at the adhesive tip of each Vomp. Although studies document that the Vomp adhesins confer virulence phenotypes in vitro, little is known about in vivo virulence strategies of Bartonella. We sought to determine whether the B. quintana Vomp adhesins are necessary for infection in vivo by using a vomp null mutant. It first was necessary to develop a system to generate in-frame deletions of defined genes by allelic exchange in a wild-type Bartonella background, which had not been achieved previously. We utilized sacB negative selection to generate a targeted, in-frame, markerless deletion of the entire vomp locus in B. quintana. We also recently developed the first animal model for B. quintana infection, and using this model, we demonstrate here that the deletion of the entire vomp locus, but not the deletion of two vomp genes, results in a null mutant strain that is incapable of establishing bloodstream infection in vivo. The Vomp adhesins therefore represent critical virulence factors in vivo, warranting further study. Finally, our allelic exchange strategy provides an important advance in the genetic manipulation of all Bartonella species and, combined with the animal model that recapitulates human disease, will facilitate pathogenesis studies of B. quintana.

Bartonella species are fastidious, gram-negative bacteria that persistently infect the bloodstream of many mammals. The three major Bartonella pathogens infecting humans are Bartonella quintana, B. henselae, and B. bacilliformis. B. quintana is transmitted by the human body louse and causes relapsing fever (“trench fever”), endocarditis, and the highly vascular lesions of bacillary angiomatosis (13). Bacteremia can persist for months, and unsuspected bloodstream infection with Bartonella can be detected in 5 to 14% of asymptomatic individuals in certain geographic regions (6, 27). B. quintana can cause debilitating, even fatal, illness in immunocompromised individuals with cancer, transplanted organs, or AIDS.

Phase and antigenic variation are immune response-evading virulence strategies exploited by microbial pathogens to persist in a host (3). We identified a family of B. quintana proteins that appears to undergo phase variation (28). These surface-localized adhesins, designated Vomp (variably expressed outer membrane proteins), are variably expressed over the course of prolonged bloodstream infection in vivo and are encoded by four highly conserved, tandemly arranged genes (28). These genes, vompD, vompA, vompB, and vompC, are located on a 12.8-kb region of the B. quintana genome. VompA, VompB, and VompC are highly conserved except in the major variable region, located in the N-terminal half of these three Vomp.

The Vomp adhesins are members of a newly recognized group of afimbrial adhesins of gram-negative bacteria known as trimeric autotransporter adhesins (TAA) (9, 15). TAA transport utilizes the type V secretion system, and the most extensively studied TAA is the YadA adhesin of Yersinia enterocolitica. YadA is a multifunctional virulence factor involved in autoaggregation, as well as adherence to epithelial cells, phagocytes, and extracellular matrix proteins, including collagen. We have shown previously that, similar to YadA, VompA is necessary and sufficient to mediate B. quintana autoaggregation and that the heterologous expression of either VompA or VompC in Escherichia coli is sufficient to effect collagen binding (28). Each Vomp appears to contribute a different phenotype: VompA is the major determinant of the autoaggregation phenotype, and VompC contributes most significantly to collagen binding (28). This specificity is likely mediated by the major variable region at the adhesive tip of each Vomp.

Although in vitro studies have documented that YadA, VompA, and VompC confer virulence phenotypes and, in vivo, a Y. enterocolitica yadA mutant is highly attenuated in a mouse infection model (22), little is known about in vivo virulence strategies of Bartonella species. We sought to determine whether the Bartonella Vomp adhesins are necessary for infection in vivo by using a vomp null mutant. First, however, it was necessary to develop a system to generate in-frame deletions of target genes by allelic exchange in a wild-type Bartonella background, which had not been achieved previously. As reported here, we developed a Bartonella mutagenesis approach using sacB negative selection to generate an in-frame, markerless deletion of the entire vomp locus. This strategy created the first targeted, defined deletion mutation in B. quintana. Using a rhesus macaque (Macaca mulatto) animal model (28), we demonstrated that the deletion of the entire vomp locus, but not the deletion of two vomp genes, resulted in an avirulent B. quintana null mutant strain that was incapable of establishing bloodstream infection in the macaque.


Bacterial strains and growth conditions.

B. quintana wild-type strain JK31 (with the entire, four-gene vomp locus intact) was isolated from the blood of an AIDS patient with bacillary angiomatosis. Colonies were isolated directly from the blood of this patient and were frozen after one or two agar passages. All subsequent experiments used B. quintana JK31 streaked from these one- or two-passage frozen stocks. B. quintana strain BQ2-D70 was isolated 70 days postinoculation from the bloodstream of a macaque that was experimentally infected with JK31 (28). B. quintana strains were streaked onto chocolate agar plates, the plates were incubated at 37°C in candle extinction jars, and the strains were harvested after 5 to 7 days (21). E. coli strains were grown in Luria-Bertani medium at 37°C. When required, kanamycin, chloramphenicol, nalidixic acid, or cefazolin was added at a concentration of 50, 35, 20, or 2 μg/ml, respectively. Bacterial strains and plasmids are listed in Table Table11.

Bacterial strains and plasmids

Generation of a nonpolar, in-frame deletion mutation of the vomp locus in B. quintana. (i) Plasmid pJM06 construction.

We employed a two-step mutagenesis method using sacB negative selection of wild-type Bartonella for the first time (11). Primers used for this strategy are listed in Table Table2.2. Plasmids used to generate the mutagenic plasmid pJM06 are shown in Fig. Fig.1A.1A. pJM06 was derived from plasmid pRS14, kindly supplied by C. Dehio (26). The rpsL gene and the sequence homologous to the virB4-flanking regions were excised from pRS14 to generate plasmid pJM02. A 2.6-kb fragment containing sacB from Bacillus subtilis was amplified with primers SacBS-F and SacBS-R from plasmid pJEN34 (generously provided by J. Engel) (12). This PCR product was cloned into TOPO TA (Invitrogen, Carlsbad, CA) and sequenced, and the construct was assayed for SacB function. The sacB fragment was then removed from TOPO TA and cloned into pJM02, creating pJM05 (Fig. (Fig.1A1A).

FIG. 1.
Wild-type B. quintana vomp locus and plasmids utilized in construction of the vomp locus null mutant. (A) Plasmids utilized in the construction of the mutagenic plasmid pJM06, derived from pRS14, are shown. Note that the origin of replication for these ...
Oligonucleotide primers

Plasmid pJM06 was derived by the insertion of a 1.6-kb fragment of the sequences flanking the vomp locus into pJM05. This insert was constructed by PCR “megapriming” from two PCR products amplified from JK31 genomic DNA, similar to the method described by Schulein and Dehio (26). The two PCR products amplified together to create this insert were PCR product 1 (0.85 kb; amplified with primers prJM02 and prJM05 [Fig. [Fig.1B]1B] and including the first 159 bp of the vompD gene at the 5′ end) and PCR product 2 (0.83 kb; generated with primers prJM03 and prJM06 [Fig. [Fig.1B]1B] and including 136 bp of the vompC gene 3′ end). The 1.6-kb PCR product was gel purified and cloned into TOPO TA, sequenced, and cloned into the BamHI site of plasmid pJM05, resulting in pJM06 (Fig. (Fig.1A1A).

(ii) Two-step selection strategy for vomp locus deletion in B. quintana.

Plasmid pJM06 was transferred into wild-type B. quintana JK31 by triparental conjugation. Overnight cultures of two different E. coli parental strains, one containing plasmid pJM06 and the other containing the helper plasmid pRK600, were grown separately in Luria-Bertani media containing kanamycin and chloramphenicol, respectively. The following morning, each of these E. coli cultures was diluted 1:10 and grown to mid-log phase and then centrifuged, washed three times with M199 medium supplemented with 20% fetal calf serum, 2 mM glutamine, and 110 μg of sodium pyruvate/ml (M199S) (21), and resuspended at a final optical density at 600 nm (OD600) of 1.0. Two confluent platefuls of B. quintana cultures were resuspended in 1 ml of M199S, washed once, and resuspended at a final OD600 of 3.0 to 5.0. Fifty microliters of each of the two E. coli parental strains and of the resuspended B. quintana culture were combined and gently mixed on the center of a nonselective chocolate agar plate, and the plate was incubated at 35°C for 5 to 6 h. Transconjugants were then selected on chocolate plates supplemented with kanamycin. The kanamycin-resistant (Kanr) transconjugants resulted from the homologous recombination of the pJM06 suicide plasmid into the B. quintana chromosome at one of the vomp-flanking regions. To verify integration, colony PCR was performed using the primers prJM05 and prJM06 (Fig. (Fig.1B1B).

The Kanr transconjugants from the single-crossover event also contained the sacB gene encoding levansucrase, which is lethal for many gram-negative bacteria when they are grown in the presence of sucrose. This genotype allowed for a subsequent negative selection step, i.e., selection for the loss of the integrated plasmid containing sacB. Transconjugants were grown on chocolate agar containing 5% sucrose to select for the loss of the integrated plasmid through a second homologous crossover event. Colonies that lost the integrated plasmid became sucrose resistant (Sucr) and kanamycin sensitive (Kans), whereas colonies that had undergone spontaneous sacB inactivation became Sucr and Kanr. To distinguish between these two outcomes, colonies were replica plated onto kanamycin plates to exclude colonies that had become Sucr through spontaneous sacB inactivation. Kans Sucr colonies carried either the wild-type allele (reconstitution) or an in-frame deletion of the vomp locus (gene replacement). Colony PCR, using primers prJM05 and prJM06, allowed discrimination between these different outcomes, and one vomp null mutant colony was chosen for further verification and experimental evaluation.

Southern blotting with genomic DNA from the JK31 wild-type and vomp null mutant B. quintana strains.

Genomic DNA from JK31 or the vomp null mutant was digested with EcoRV and subjected to Southern blotting using conventional methods (24), except that probes were labeled with the AlkPhos direct labeling system (Amersham, Piscataway, NJ). Two PCR-amplified probes were used (Fig. (Fig.1B):1B): the conserved probe, consisting of a 446-bp fragment conserved in the 3′ region of all vomp genes (constructed with primers b'consF1 and b'consR1), and the vompD probe, consisting of a 446-bp fragment from the 5′ noncoding region upstream of vompD (constructed with primers VompD_up1 and VompD_up2).

Two-dimensional (2D) sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) separation of the B. quintana TOMP fraction.

Subcellular protein fractions of B. quintana strains were prepared as previously described (17), with the following modifications. Bacteria were harvested from chocolate agar plates, washed twice with phosphate-buffered saline, and pelleted in a microcentrifuge for 2 to 3 min at 4°C. The final pellet was resuspended in 10 mM HEPES buffer. Protease inhibitor cocktail {20 mM AEBSF [4-(2-aminoethyl)-benzenesulfonyl fluoride; Calbiochem]; 1 mg of leupeptin/ml, 0.36 mg of E-64/ml, and 5.6 mg of benzamidine/ml (all from Sigma); and 50 mM EDTA} was added, after which the suspension was incubated on ice for 10 min. Bacterial cells were sonicated with five 1-min bursts while cooling on ice. Cellular debris was pelleted at 4,300 × g for 30 min at 4°C. The supernatant was then centrifuged at 100,000 × g with a Beckman L8-M ultracentrifuge for 1.5 h at 4°C, and the supernatant was discarded. The pellet, comprising the total outer membrane protein (TOMP) fraction, was resuspended in 10 mM HEPES and treated with nuclease (50 mM MgCl2, 100 mM Tris [pH 7.0], 500 μg of RNase [Sigma]/ml, and 1 mg of DNase [Sigma]/ml). The TOMP fraction was then washed twice in 10 mM HEPES and pelleted at 40,000 × g for 30 min at 4°C. The protein concentration was determined, and fractions were frozen at −80°C until use.

2D SDS-PAGE was performed according to the method of O'Farrell (20) (Kendrick Labs, Inc., Madison, WI), as follows: isoelectric focusing was carried out in glass tubes with an inner diameter of 2.0 mm by using 2% ampholines, pH 4 to 8 (BDH; Hoefer Scientific Instruments, San Francisco, CA), for 9,600 V·h. After equilibration for 10 min in buffer O (10% glycerol, 50 mM dithiothreitol, 2.3% SDS, and 0.0625 M Tris, pH 6.8), tube gels were laid on top of 10% acrylamide slab gels (0.75 mm thick) and SDS slab gel electrophoresis was carried out for 4 h at 12.5 mA/gel. Gels were stained with silver stain (19).

Immunoblotting with antibody to the conserved N-terminal domain of B. quintana VompA, VompB, and VompC.

For immunoblotting, whole-cell lysates were prepared as described for subcellular preparations (up to the ultracentrifugation step) and the lysate was separated using SDS-PAGE with 12% acrylamide. Proteins were transferred onto 0.45-μm-pore-size Protran nitrocellulose (Schleicher & Schuell, Keene, NH) and probed with rabbit polyclonal antibody to the N-terminal conserved region in VompA, VompB, and VompC, followed by alkaline phosphatase-conjugated goat anti-rabbit secondary antibody (Zymed, South San Francisco, CA).

Autoaggregation assay of B. quintana JK31 wild-type and isogenic vomp null mutant strains.

Autoaggregation was assayed using the method described by Laird and Cavanaugh (14), with the following modifications. B. quintana was harvested, washed, and resuspended at an OD600 of 1.0 in M199S. A 3-ml aliquot of each B. quintana suspension was added to plastic test tubes, and the tubes were incubated in a CO2-enriched atmosphere at 35°C. To quantify autoaggregation, a 50-μl sample from the top of each culture tube was removed after 1, 2, 6, and 9 h and the OD600 was measured immediately. Samples were assayed in triplicate, and statistical analyses were performed using Student's t test.

Inoculation of rhesus macaques with B. quintana strains to assess the ability of vomp mutants to establish bloodstream infection.

We recently developed an animal model of B. quintana infection in rhesus macaques (Macaca mulatto) that reproduces the prolonged, high-titer bacteremia observed in humans infected with B. quintana (28). No other mammalian host (except humans) is permissive for infection with B. quintana. The animal work was performed at the California National Primate Research Center at the University of California, Davis, which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. The housing and handling of the animals were in accordance with the Guide for the Care and Use of Laboratory Animals (18) and the Animal Welfare Act. The experimental protocol was approved by the University of California, Davis, Animal Use and Care Administrative Advisory Committee.

For this study, the macaque animal model enabled us first to establish the course of infection with wild-type B. quintana in eight animals. Each of the eight macaques was inoculated intradermally with a total dose of the wild-type JK31 strain ranging from 3.5 × 106 to 1.78 × 109 CFU (the mean total dose for the eight animals was 2.5 × 108 CFU). The total dose inoculated into each animal was divided among six different inoculation sites on the animal. Blood from each animal was cultured prior to inoculation and then twice weekly for 4 to 6 weeks after inoculation and weekly for another 2 to 4 months, as previously described (28).

To assess the ability of the isogenic B. quintana vomp null mutant to infect a permissive host, we inoculated two naïve animals intradermally, one with a total of 3.1 × 109 CFU and the other with 5.7 × 109 CFU (mean total dose, 4.4 × 109 CFU) of this null mutant strain, divided among six different sites on each animal. Blood samples from the animals inoculated with the vomp null mutant strain were cultured using the same methods employed for the animals inoculated with the wild-type strain, and the culture results from the two vomp null mutant inoculations were compared statistically with those from the eight wild-type JK31 strain inoculations. We used two-sided Fisher's exact test to assess the statistical significance of the difference in infection rates between B. quintana wild-type and vomp null mutant strains.

In addition, to determine if two vomp genes are sufficient for infection, we inoculated a macaque with strain BQ2-D70, in which only the vompD and vompC genes are present. This BQ2-D70 strain was originally isolated from macaque no. BQ2, which was infected with the wild-type JK31 strain during a previous study. The last culture-positive blood sample from this animal was documented at day 70 postinoculation. During the prolonged, 70-day bloodstream infection, the vompA and vompB genes from the JK31 inoculum strain were deleted in vivo, yielding the strain BQ2-D70. Although no VompC or VompD protein expression in BQ2-D70 can be identified (28), vompC mRNA is detectable by reverse transcriptase PCR (P. Zhang and J. E. Koehler, unpublished data).


A negative selection strategy generates a markerless, nonpolar, in-frame deletion of the vomp locus in a wild-type B. quintana background.

To study the contribution of the Vomp adhesins to B. quintana pathogenesis, i.e., the ability to invade the bloodstream and establish persistent infection in vivo, we developed a two-step mutagenesis strategy for B. quintana by using SacB negative selection (11) to generate a vomp locus null mutant. The expression of the Bacillus subtilis sacB gene in gram-negative bacteria grown in the presence of sucrose is lethal due to the synthesis of levansucrase, thus permitting a two-step selection for the loss of the integrated plasmid containing sacB. Successful deletion of the vomp locus was confirmed by sequencing of the insert and the plasmid-insert junctions in pJM06 prior to conjugative transfer into B. quintana. The deletion of the vomp locus was then confirmed by PCR and sequencing after transfer into the JK31 wild-type strain. Finally, Southern blotting and immunoblotting confirmed the absence of vomp genes and their protein products, respectively, in the vomp null mutant.

Southern blotting with genomic DNA from the vomp null mutant demonstrates the deletion of the vomp genes.

The in-frame vomp deletion was confirmed by Southern blot analysis of EcoRV-digested DNA using two probes, the vompD probe and the conserved probe (Fig. (Fig.2).2). The deletion of the vomp genes resulted in the loss of EcoRV restriction endonuclease sites, generating a larger product in the isogenic vomp null mutant identified with the vompD probe (~6.9 kb for the mutant, compared with ~1.8 kb for the wild type), as shown schematically in Fig. Fig.1B.1B. The conserved probe recognized all four vomp genes in JK31 (on fragments of ~2.1, 3.0, 3.3, and 8.8 kb), but these sequences were deleted in the mutant except for 32 bp also present in the conserved probe sequence, producing only a faint band at ~6.9 kb. Note that the location of the most 3′ EcoRV site was identified using the published genome sequence of the B. quintana Toulouse strain (1); the precise location of this EcoRV site in the B. quintana strain JK31 genome has not yet been determined but is likely to be conserved.

FIG. 2.
The in-frame deletion of the B. quintana vomp locus was confirmed by Southern blotting. The deletion of the vomp locus resulted in the loss of EcoRV restriction endonuclease sites, generating a larger product visualized with the probe for the noncoding ...

2D SDS-PAGE and immunoblotting document the absence of Vomp expression in the isogenic vomp null mutant.

2D SDS-PAGE analysis of the vomp null mutant TOMP fraction revealed that the three proteins previously identified as Vomp adhesins by N-terminal sequencing (28) (Fig. (Fig.3A),3A), in addition to their smaller degradation products, were absent in the vomp null mutant (Fig. (Fig.3B).3B). Immunoblotting of whole-cell lysates from the JK31 wild type and the isogenic vomp null mutant by using antibody to an N-terminal conserved region of VompA, VompB, and VompC confirmed that these proteins were not detectable in the vomp null mutant (Fig. (Fig.3C3C).

FIG. 3.
Vomp expression was not detected in the B. quintana vomp null mutant. (A) VompA, VompB, and VompC proteins were present, each at a mass of ~100 kDa, in the TOMP fraction of the JK31 wild type (bracket and arrow) after separation by 2D SDS-PAGE. ...

The Vomp adhesins are required for autoaggregation.

We have shown previously that Vomp expression in wild-type B. quintana mediates autoaggregation (28). In addition, the heterologous expression of VompA, but not VompC, in E. coli is sufficient to confer an autoaggregative phenotype on nonaggregative E. coli (28). We compared the autoaggregative phenotypes of the wild type JK31 and the isogenic vomp null mutant by using tube suspension assays, each performed in triplicate. The vomp null mutant did not autoaggregate, in contrast to the JK31 wild type (Fig. (Fig.4).4). The OD600 of the wild-type JK31 B. quintana strain was significantly lower than that of the mutant strain (P < 0.001) at the final time point (9 h) by Student's t test.

FIG. 4.
The B. quintana Vomp adhesins mediate autoaggregation. Although the JK31 wild-type strain autoaggregated, the vomp null mutant did not autoaggregate. Each experiment was performed in triplicate, and representative results from one experiment are shown. ...

The presence of all or part of the vomp locus is required to establish B. quintana bloodstream infection in vivo.

The homology between the Vomp adhesins and TAA of other gram-negative bacterial pathogens suggested that the Vomp adhesins have an important role in establishing bloodstream invasion by and persistent infection with B. quintana. To test this hypothesis, we utilized our recently developed animal model of B. quintana infection in the macaque (28). For the present study, it was first necessary to determine the usual course of infection with wild-type B. quintana by inoculating eight animals. Figure Figure55 shows the mean number of B. quintana CFU isolated per milliliter of blood from the eight animals inoculated with wild-type JK31 B. quintana. Every animal inoculated with wild-type B. quintana became infected and developed detectable bloodstream infection by day 14, with a mean peak bacterial burden of 1.62 × 103 CFU/ml and a mean bacteremia duration of 59 days. In contrast, no colonies were recovered from the blood of two naïve animals inoculated with the isogenic B. quintana vomp null mutant at any time point in the 7 weeks following inoculation (Fig. (Fig.5).5). The difference in infection rates (0 of 2 animals for the vomp null mutant versus 8 of 8 for the wild-type strain) has a P value of 0.022 by Fisher's exact test.

FIG. 5.
The B. quintana vomp null mutant was unable to establish bloodstream infection in a rhesus macaque animal model. Eight animals were infected with the B. quintana JK31 wild type, and all developed detectable bloodstream infections by day 14, with a mean ...

To determine if two vomp genes are sufficient for infection, we inoculated a naïve macaque with strain BQ2-D70 (derived in vivo from JK31), in which only the vompD and vompC genes are present in the vomp locus but in which no VompC or VompD protein expression can be identified (28). Unlike the vomp null mutant, which was incapable of invading and establishing infection in the bloodstream, BQ2-D70 inoculation resulted in bloodstream infection, although the peak number of CFU per milliliter was lower than the mean peak number of wild-type JK31 CFU per milliliter and the time to reach this peak was longer (Fig. (Fig.5).5). Thus, although infection was abrogated by the deletion of the entire vomp locus, the presence of only the vompD and vompC genes was sufficient for bloodstream infection with B. quintana in vivo.


B. quintana is an emerging pathogen that has only recently been studied in the laboratory. Few B. quintana virulence factors have been identified or described, and there are few techniques for genetic manipulation. The Vomp adhesins are the first B. quintana virulence factors identified in vivo, and as members of the TAA family, they are expressed on the surfaces of B. quintana cells (28). In most gram-negative bacteria, the TAA is encoded by only one gene (e.g., the Y. enterocolitica yadA gene), and the monomer expressed by this gene then forms homotrimers on the bacterial surface. However, B. quintana Vomp adhesins are encoded by four vomp paralogs, and each adhesin appears to have unique binding specificity. The capacity of each Vomp to bind a different substrate could have evolved through vomp gene duplication to permit B. quintana binding to host cells in the disparate niches that B. quintana must occupy: binding to erythrocytes and endothelial cells in the bloodstream of the mammalian reservoir host and to collagen in the cutaneous lesions of bacillary angiomatosis, as well as to the gastrointestinal epithelium of the body louse vector.

The importance of the TAA in other pathogens, the binding heterogeneity required by B. quintana, and the presence of four Vomp adhesins led us to hypothesize that the Vomp adhesins are necessary for host infection. Surface proteins that are critical virulence factors are frequently targeted by the host immune system early after infection; in response, the bacterium often modifies the expression of these surface proteins by using phase variation. As additional evidence that the Vomp adhesins are important in vivo, these adhesins demonstrate both these properties: the Vomp adhesins are among the most antigenic of the B. quintana outer membrane proteins (5), and the Vomp adhesins undergo phase variation during prolonged host bloodstream infection (28). To test our hypothesis that the Vomp adhesins are necessary for infection in vivo, we used a macaque animal model of infection (28) and quantified the course of infection for eight animals inoculated with B. quintana wild-type strain JK31. All eight animals became bacteremic for a prolonged period (Fig. (Fig.55).

Next, we developed a two-step mutagenesis strategy for B. quintana to create a nonpolar, in-frame, markerless mutation in a fully virulent wild-type background. Such a strategy for Bartonella had never before been achieved, and it overcomes a substantial obstacle to B. quintana research and will facilitate the genetic manipulation of Bartonella species in general. Other Bartonella researchers have used transposon insertion (10) or gene disruption from a single-crossover event (4, 8) to generate mutations in Bartonella species. However, polar effects and reversion to a wild-type genotype (especially in vivo, with selective pressure against the mutation) limit these approaches. Another mutagenesis method employed a spontaneous streptomycin-resistant (Strr) Bartonella isolate (with resistance due to an rpsL mutation) (25, 26), which was used as the parental strain for a two-step targeted allelic replacement. Although this method permitted in-frame deletion, the resulting mutagenized strain retains the parental Strr rpsL mutant genotype. Unfortunately, every spontaneously occurring B. quintana Strr rpsL mutant strain we isolated to use as the parent strain was either avirulent or highly attenuated in our animal model (Zhang and Koehler, unpublished), as has been noted previously for other gram-negative pathogens, e.g., Salmonella spp. (16). It was thus necessary to develop a different approach for generating the vomp null mutant.

We successfully constructed the vomp null mutant by using sacB negative selection (11); the strategy we describe provides an important advance in the genetic manipulation of all Bartonella species. The resulting isogenic vomp null mutant, in a wild-type background, was used to inoculate two naïve macaques. No infection with the vomp null mutant was detectable (Fig. (Fig.5),5), in contrast to infection with the wild-type B. quintana strain. The fact that both vomp null mutant inoculations did not produce infection is unlikely to be due to chance alone (P = 0.022). These data strongly suggest that the vomp null mutant is avirulent or that its ability to infect is severely attenuated in vivo, below the threshold of detection by blood culture.

To determine if two vomp genes are sufficient for infection, we inoculated a macaque with strain BQ2-D70, in which only the vompD and vompC genes are present and no VompC or VompD protein expression can be identified (although vompC mRNA is detectable by reverse transcriptase PCR). Interestingly, we previously found that this BQ2-D70 strain is autoaggregation deficient (28), as we have shown here for the vomp null mutant as well. However, unlike the vomp null mutant, which was avirulent in vivo, we found that BQ2-D70 was virulent in vivo, although the peak number of CFU per milliliter was lower than that of the wild type JK31 and the time to reach this peak was substantially longer (Fig. (Fig.5).5). This result suggests that vompD and/or vompC gene expression was upregulated after animal inoculation or that very low levels of VompD and/or VompC (undetectable by immunoblotting) were sufficient to establish infection with B. quintana. Thus, although infection was prevented by the deletion of the entire vomp locus, the presence of vompD and vompC was sufficient to achieve bloodstream infection.

The function(s) of the Vomp adhesins required for establishing infection in vivo is unknown but likely includes the critical adhesion events observed during B. quintana infection, i.e., binding to erythrocytes and endothelial cells. For Y. enterocolitica, the inactivation of yadA results in a severely attenuated strain that can invade the intestine but cannot disseminate or survive host defenses in a mouse model (22). Furthermore, a Y. enterocolitica mutant with the replacement of two histidyl residues in the adhesin head domain of YadA can be translocated from the intestinal lumen to Peyer's patches but cannot disseminate to the spleen or survive and multiply in extraintestinal tissues (23). The related TAA from Neisseria meningitidis, NadA, is present in three of the four known hypervirulent lineages, suggesting a role for NadA in invasive meningococcal disease, as well (7). The Vomp adhesins appear to be even more critical for virulence than TAA of other pathogens, because they are necessary for the invasion of the normally sterile bloodstream, where the Bartonella can persist in association with erythrocytes. The four paralogous Vomp adhesins could have evolved to provide the great diversity in adhesin specificity that is essential for B. quintana infection and persistence, enabling adhesion to erythrocytes, endothelial cells, and collagen. The identification of the binding specificity and the environmental signals regulating the expression of each Vomp will provide insight into the pathogenicity of the unusual and persistent pathogen B. quintana.


We gratefully acknowledge Christoph Dehio for providing pRS14, Joanne Engel for pJEN34, Peter Bacchetti for assistance with statistical analysis, and Wilhelm von Morgenland, Linda Hirst, and Nicholas Lerche for their assistance with the primate inoculations at the California National Primate Research Center at the University of California, Davis.

This work was supported by NIH R01 AI43703 and R01 AI52813 and a Burroughs Wellcome Fund Clinical Scientist Award in Translational Research (to J.E.K.). J.K.M. received support from a California HIV/AIDS Research Program postdoctoral fellowship award and an NIH T32 AI060537 postdoctoral fellowship award.


Editor: W. A. Petri, Jr.


[down-pointing small open triangle]Published ahead of print on 10 December 2007.


1. Alsmark, C., A. Frank, E. Karlberg, B. Legault, D. Ardell, B. Canback, A. Eriksson, A. Naslund, S. Handley, M. Huvet, B. La Scola, M. Holmberg, and S. Andersson. 2004. The louse-borne human pathogen Bartonella quintana is a genomic derivative of the zoonotic agent Bartonella henselae. Proc. Natl. Acad. Sci. USA 1019716-9721. [PubMed]
2. Reference deleted.
3. Barbour, A. G., and B. I. Restrepo. 2000. Antigenic variation in vector-borne pathogens. Emerg. Infect. Dis. 61-9. [PMC free article] [PubMed]
4. Battisti, J. M., and M. F. Minnick. 1999. Development of a system for genetic manipulation of Bartonella bacilliformis. Appl. Environ. Microbiol. 653441-3448. [PMC free article] [PubMed]
5. Boonjakuakul, J., H. Gerns, Y. Chen, L. Hicks, M. Minnick, S. Dixon, S. Hall, and J. Koehler. 2007. Proteomic and immunoblot analyses of Bartonella quintana total membrane proteins identify antigens recognized by sera from infected patients. Infect. Immun. 752548-2561. [PMC free article] [PubMed]
6. Brouqui, P., B. Lascola, V. Roux, and D. Raoult. 1999. Chronic Bartonella quintana bacteremia in homeless patients. N. Engl. J. Med. 340184-189. [PubMed]
7. Capecchi, B., J. Adu-Bobie, F. Di Marcello, L. Ciucchi, V. Masignani, A. Taddei, R. Rappuoli, M. Pizza, and B. Arico. 2005. Neisseria meningitidis NadA is a new invasin which promotes bacterial adhesion to and penetration into human epithelial cells. Mol. Microbiol. 55687-698. [PubMed]
8. Coleman, S. A., and M. F. Minnick. 2001. Establishing a direct role for the Bartonella bacilliformis invasion-associated locus B (IalB) protein in human erythrocyte parasitism. Infect. Immun. 694373-4381. [PMC free article] [PubMed]
9. Cotter, S., N. Surana, and J. St. Geme III. 2005. Trimeric autotransporters: a distinct subfamily of autotransporter proteins. Trends Microbiol. 13199-205. [PubMed]
10. Dehio, C., and M. Meyer. 1997. Maintenance of broad-host-range incompatibility group P and group Q plasmids and transposition of Tn5 in Bartonella henselae following conjugal plasmid transfer from Escherichia coli. J. Bacteriol. 179538-540. [PMC free article] [PubMed]
11. Foster, P. L. 1991. In vivo mutagenesis. Methods Enzymol. 204114-125. [PMC free article] [PubMed]
12. Garrity-Ryan, L., S. Shafikhani, P. Balachandran, L. Nguyen, J. Oza, T. Jakobsen, J. Sargent, X. Fang, S. Cordwell, M. A. Matthay, and J. N. Engel. 2004. The ADP ribosyltransferase domain of Pseudomonas aeruginosa ExoT contributes to its biological activities. Infect. Immun. 72546-558. [PMC free article] [PubMed]
13. Koehler, J. E. 2000. Bartonella species, p. 339-353. In J. P. Nataro, M. J. Blaser, and S. Cunningham-Rundles (ed.), Persistent bacterial infections. ASM Press, Washington, DC.
14. Laird, W. J., and D. C. Cavanaugh. 1980. Correlation of autoagglutination and virulence of yersiniae. J. Clin. Microbiol. 11430-432. [PMC free article] [PubMed]
15. Linke, D., T. Riess, I. B. Autenrieth, A. Lupas, and V. A. J. Kempf. 2006. Trimeric autotransporter adhesins: variable structure, common function. Trends Microbiol. 14264-270. [PubMed]
16. Maisnier-Patin, S., O. Berg, L. Liljas, and D. Andersson. 2002. Compensatory adaptation to the deleterious effect of antibiotic resistance in Salmonella typhimurium. Mol. Microbiol. 46355-366. [PubMed]
17. Murphy, T. F., K. C. Dudas, J. M. Mylotte, and M. A. Apicella. 1983. A subtyping system for nontypable Haemophilus influenzae based on outer-membrane proteins. J. Infect. Dis. 147838-846. [PubMed]
18. National Research Council. 1996. Guide for the care and use of laboratory animals. National Academy Press, Washington, DC.
19. O'Connell, K. L., and J. T. Stults. 1997. Identification of mouse liver proteins on two-dimensional electrophoresis gels by matrix-assisted laser desorption/ionization mass spectrometry of in situ enzymatic digests. Electrophoresis 18349-359. [PubMed]
20. O'Farrell, P. H. 1975. High resolution two-dimensional electrophoresis of proteins. J. Biol. Chem. 2504007-4021. [PMC free article] [PubMed]
21. Park, S., K. Kelminson, A. Lee, P. P. Zhang, R. Warner, D. Rehkopf, S. Calderwood, and J. Koehler. 2001. Identification, characterization, and functional analysis of a gene encoding the ferric uptake regulation protein in Bartonella species. J. Bacteriol. 1835751-5755. [PMC free article] [PubMed]
22. Pepe, J. C., M. R. Wachtel, E. Wagar, and V. L. Miller. 1995. Pathogenesis of defined invasion mutants of Yersinia enterocolitica in a BALB/c mouse model of infection. Infect. Immun. 634837-4848. [PMC free article] [PubMed]
23. Roggenkamp, A., H. Neuberger, A. Flugel, T. Schmoll, and J. Heesemann. 1995. Substitution of two histidine residues in YadA protein of Yersinia enterocolitica abrogates collagen binding, cell adherence and mouse virulence. Mol. Microbiol. 161207-1219. [PubMed]
24. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.
25. Schmid, M., R. Schulein, M. Dehio, G. Denecker, I. Carena, and C. Dehio. 2004. The VirB type IV secretion system of Bartonella henselae mediates invasion, proinflammatory activation and antiapoptotic protection of endothelial cells. Mol. Microbiol. 5281-92. [PubMed]
26. Schulein, R., and C. Dehio. 2002. The VirB/VirD4 type IV secretion system of Bartonella is essential for establishing intraerythrocytic infection. Mol. Microbiol. 461053-1067. [PubMed]
27. Weinman, D., and H. Pinkerton. 1937. Carrion's disease. IV. Natural sources of bartonella in the endemic zone. Proc. Soc. Exp. Biol. Med. 37596-598.
28. Zhang, P., B. B. Chomel, M. K. Schau, J. S. Goo, S. Droz, K. L. Kelminson, S. S. George, N. W. Lerche, and J. E. Koehler. 2004. A family of variably expressed outer-membrane proteins (Vomp) mediates adhesion and autoaggregation in Bartonella quintana. Proc. Natl. Acad. Sci. USA 10113630-13635. [PubMed]

Articles from Infection and Immunity are provided here courtesy of American Society for Microbiology (ASM)