Search tips
Search criteria 


Logo of iaiPermissionsJournals.ASM.orgJournalIAI ArticleJournal InfoAuthorsReviewers
Infect Immun. 2012 March; 80(3): 929–942.
PMCID: PMC3294634

Hemin Binding Protein C Is Found in Outer Membrane Vesicles and Protects Bartonella henselae against Toxic Concentrations of Hemin

B. A. McCormick, Editor


Bartonella species are Gram-negative, emerging bacterial pathogens found in two distinct environments. In the gut of the obligately hematophagous arthropod vector, bartonellae are exposed to concentrations of heme that are toxic to other bacteria. In the bloodstream of the mammalian host, access to heme and iron is severely restricted. Bartonellae have unusually high requirements for heme, which is their only utilizable source of iron. Although heme is essential for Bartonella survival, little is known about genes involved in heme acquisition and detoxification. We developed a strategy for high-efficiency transposon mutagenesis to screen for genes in B. henselae heme binding and uptake pathways. We identified a B. henselae transposon mutant that constitutively expresses the hemin binding protein C (hbpC) gene. In the wild-type strain, transcription of B. henselae hbpC was upregulated at arthropod temperature (28°C), compared to mammalian temperature (37°C). In the mutant strain, temperature-dependent regulation was absent. We demonstrated that HbpC binds hemin and localizes to the B. henselae outer membrane and outer membrane vesicles. Overexpression of hbpC in B. henselae increased resistance to heme toxicity, implicating HbpC in protection of B. henselae from the toxic levels of heme present in the gut of the arthropod vector. Experimental inoculation of cats with B. henselae strains demonstrated that both constitutive expression and deletion of hbpC affect the ability of B. henselae to infect the cat host. Modulation of hbpC expression appears to be a strategy employed by B. henselae to survive in the arthropod vector and the mammalian host.


Bartonella henselae is a blood-borne, Gram-negative bacterial pathogen for which the natural reservoir host is the domestic cat. B. henselae infection in the cat is characterized by invasion of erythrocytes and persistent, asymptomatic bacteremia (12). B. henselae occupies a second niche: the gastrointestinal tract of its arthropod vector, the cat flea (Ctenocephalides felis) (13). Human B. henselae infections are epidemiologically linked to exposure to B. henselae-infected cats and fleas (24, 25, 59). In healthy humans, B. henselae causes a zoonotic infection known as cat scratch disease, typically resulting in swollen lymph nodes near the scratch site of B. henselae inoculation. In immunocompromised patients, B. henselae can cause vasculoproliferative lesions (bacillary angiomatosis), as well as a relapsing bloodstream infection that can persist for months (23, 24).

Iron is essential for viability and pathogenicity in bacteria. However, free iron is sequestered in the bloodstream and tissues of eukaryotic hosts to prevent or attenuate infection by pathogenic bacteria (45). Pathogenic bacteria therefore have evolved strategies to acquire iron, including release of siderophores to scavenge free iron or expression of outer membrane transporters that recognize iron-containing proteins found in the host (53). Heme, a small molecule present either free or bound to hemoglobin in the bloodstream of mammals, is an alternative source of iron within the host (56). Heme consists of a porphyrin ring containing an Fe2+ ion and is called hemin when the bound iron is in the Fe3+ state. Heme also is used as a prosthetic group in important eukaryotic and bacterial proteins, such as cytochrome c, hemoglobin, and catalases (14).

Bartonella species are unusual because they require heme or hemoglobin as their sole iron source, and bartonellae are unable to grow on defined media containing iron salts or the mammalian iron-containing compounds lactoferrin and transferrin (44). The published genome sequences of B. henselae and the closely related strain B. quintana do not reveal genes with homology to the gene encoding ferrochelatase, the enzyme that catalyzes the terminal step in heme biosynthesis (1). These observations indicate that B. henselae and B. quintana must acquire exogenous heme not only as a source of iron but also as a prosthetic group for bacterial enzymes and proteins.

Bartonellae are exposed to very different concentrations of heme in the two distinct niches they occupy. In the mammalian host, where heme is stringently sequestered, Bartonella must employ strategies for heme scavenging. B. quintana and B. henselae genomes encode a family of five outer membrane proteins (OMP) called hemin binding proteins (Hbp) (1, 32). HbpA initially was identified as a phage-associated protein of 31 kDa (Pap31) in the cell-free supernatant of B. henselae (8), although subsequent association of HbpA/Pap31 with phage has not been reported. The HbpA proteins in B. quintana and B. henselae bind heme in vitro (10, 11), but the mechanisms of heme binding and uptake are unknown.

In the gut of the obligately hematophagous arthropod vector, bartonellae tolerate the unusually high, toxic levels of heme present and can grow in levels of up to 1 to 2 mM hemin in vitro (33, 34, 44). The iron atom within heme can generate toxic reactive oxygen species that can lead to lipid peroxidation, DNA damage, and other detrimental effects on bacterial cells (19, 46). Therefore, bacteria that utilize heme as an iron source must maintain mechanisms to metabolize or sequester heme and ameliorate its toxic effects. The mechanisms by which Bartonella tolerates the unusually high concentrations of heme found in the arthropod gut are not known.

In this study, we first sought to identify genes involved in heme acquisition and detoxification in B. henselae. We used a mariner-based transposon (26) to perform a forward genetic screen to identify mutants with perturbed accumulation of hemin in B. henselae. We identified one mutant strain in which the hbpC gene is constitutively transcribed (hbpC*). Analysis of the hbpC* transposon mutant strain and an isogenic strain in which hbpC was deleted enabled us to examine the role of HbpC in hemin binding, expression of other genes in the hbp family, survival in media containing toxic concentrations of hemin, and the ability to infect the feline mammalian reservoir host. Our data provide new insight into the function of the B. henselae Hbp and suggest that Hbp found in the outer membrane contribute to acquisition of scarce heme in the mammalian bloodstream, whereas upregulated production of Hbp-containing outer membrane vesicles (OMV) permits B. henselae survival within the gut of the arthropod vector by mitigating the toxicity of high heme concentrations.


Bacterial strains and growth conditions.

Low-passage (2 to 3 passages from frozen stock) B. henselae strains were used for all experiments. B. henselae wild-type strain JK33R is a low-passage strain that was isolated directly from the bacillary angiomatosis lesion of an AIDS patient who also was bacteremic with B. henselae. JK33S is a derivative of JK33R that was generated after several passages of JK33R on chocolate agar, resulting in a phenotypic change in colony morphology from rough to smooth. This transition in colony morphology usually occurs early during in vitro passage of all B. henselae strains and has been attributed to a disruption in the badA gene in B. henselae Marseille (41). The smooth phenotype also is present in the ATCC strain Houston-1, commonly used for in vitro experiments, and maintenance of infectivity of smooth derivatives in the feline host has been well documented (58).

All B. henselae strains were grown on fresh chocolate agar plates (23) for 5 to 7 days in candle extinction jars. For growth at temperatures corresponding to the arthropod vector (28°C) or the mammalian host (37°C), B. henselae was initially grown from frozen stocks on chocolate agar for 7 days at 35°C. Bacteria were then passaged to fresh chocolate agar plates and grown at either 37°C for 5 days or 37°C for 1 day, followed by 28°C for 4 days, and then harvested. The liquid medium used in B. henselae experiments was M199 medium supplemented with glutamine, sodium pyruvate, and 20% fetal bovine serum (M199S) (23).

Escherichia coli strains were grown in Luria-Bertani liquid medium. When required, kanamycin (Kan) was used at a final concentration of 50 μg/ml, cefazolin (Cef) at 20 μg/ml, nalidixic acid (Nal) at 2 μg/ml, and gentamicin (Gen) at 10 μg/ml. For sacB negative selection, a sterile-filtered sucrose solution was added to chocolate agar to give a final sucrose concentration of 5%. Strains and plasmids used in this paper are described in Table 1.

Table 1
Bacterial strains and plasmids used in this study

Hemin preparation.

For hemin supplementation of chocolate agar plates, hemin stock was prepared by dissolving 1 g hemin chloride (Sigma, St. Louis, MO) in 15 ml 1.5 M NaOH. For hemin added to M199S liquid medium or 100 mM Tris buffer, pH 8, 1 g hemin chloride was combined with 15 ml 1.5 M NH4OH. After resuspension, hemin solutions were centrifuged at 1,000 × g for 10 min to remove the precipitated free iron. The supernatant was filtered through a 0.2-μm filter to remove residual free iron and contaminants (E. Wyckoff and S. Payne, personal communication). The hemin concentration was calculated by measuring the absorbance of each solution at 572 nm, where an absorbance of 5.5 is equivalent to 1 mM (57). Solutions were stored in the dark at 4°C and used within 1 week. The filter-sterilized hemin solution was added to autoclaved chocolate agar prior to pouring. Hemin solutions or pH-identical 1.5 M NH4OH control solutions were added to the liquid media prior to hemin binding and toxicity experiments.

Comparison of transposon efficiencies in B. henselae.

To optimize transposon mutagenesis for B. henselae strains JK33R and JK33S, we compared the efficiency of pBT20, a plasmid containing the mariner transposon and transposase (26), with that of pJKTn5-GentR, which contains the Tn5 transposon (22). To generate transposon insertions in B. henselae strains by biparental conjugation, the donor strain E. coli S17-1 (Table 1) carrying pBT20 or pJKTn5-GentR was grown overnight and then diluted 1:10 in fresh media and grown to an optical density at 600 nm (OD600) of approximately 1.0. E. coli from 1 ml of culture was washed twice in 1 ml M199S, and the cell pellet was resuspended in 1 ml M199S. Then, 10 μl of the washed E. coli donor strain was combined with 500 μl M199S and pipetted onto a confluent plate of B. henselae JK33S grown on chocolate agar for 5 days at 35°C. Bartonella and donor E. coli were mixed on the plate surface, dried for 5 min, and then incubated for 5 h at 35°C in a candle extinction jar to allow conjugation. Dilutions of conjugation cell suspensions were plated on chocolate agar containing Gen, Cef, and Nal to select for recipients with transposon insertions or Cef and Nal to enumerate total Bartonella CFU. Plates were incubated for 7 to 10 days, and CFU were counted to calculate transposon efficiency (Genr Cefr Nalr colonies/Cefr Nalr colonies).

Transposon mutagenesis screen for colonies with altered hemin pigmentation.

After comparing efficiencies of the two transposon constructs, we subsequently utilized the mariner transposon conjugated into the strain B. henselae JK33S because of the greater efficiency of mutagenesis and large numbers of transformants generated (see Results). Serial dilutions of the conjugation suspension were plated onto chocolate agar plates containing 1 mM hemin and Gen, Cef, and Nal. Plates were incubated for 7 to 10 days and then screened visually for bacterial colonies that had an altered pigmentation compared with that of the wild-type parent strain. These mutants were selected and further characterized.

After identification of transposon-mutagenized B. henselae strains that had altered pigmentation, a two-step arbitrary PCR protocol was used to identify the location of transposon insertions (9). Genomic DNA (gDNA) was isolated from each insertion-mutagenized strain, and ~50 ng gDNA was used as a template in four initial PCRs, each with primer Rnd1-TnM and 1 of 4 semirandom primers (Arb1A to -D) optimized for use in Bartonella (Table 2). Products from this arbitrary PCR were used as a template for a second PCR with Rnd2-TnM and Arb2 primers. Each mariner insertion site was then confirmed using sequence-specific primers to amplify and sequence the targeted gene from gDNA.

Table 2
Oligonucleotide primers used in this study

To confirm that each mariner insertion strain contained only a single transposon insertion, 3 μg of gDNA from each strain was digested overnight with BclI (New England BioLabs, Beverly, MA), which does not cut within the mariner transposon. Digests were separated, transferred, and probed using standard protocols (43). Primers TnM1 and TnM2 (Table 2), which flank the mariner transposon, were used to amplify a 1,502-bp DNA fragment that was used as a probe. The probe was labeled and samples were visualized using an alkaline phosphatase direct labeling system and CDP-Star detection reagent (GE Health Sciences, Piscataway, NJ).

RNA preparation and RT-PCR analysis of hbp gene expression.

After growth of B. henselae strains at 37°C and 28°C, bacteria were resuspended in stop solution (2.5% phenol, 47.5% ethanol, 50% M199 medium [17]). Cells were pelleted, and each bacterial pellet was resuspended in 150 μl lysis solution (0.4 mg/ml lysozyme, 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA) and incubated at room temperature for 5 min. Total RNA was extracted in TRIzol reagent (Invitrogen, Carlsbad, CA). The air-dried RNA pellet was resuspended in 100 μl of DNase reaction mix (9 μl Turbo DNase, 2 μl RNase inhibitor, 10 μl buffer, 79 μl double-distilled water [ddH2O]; Applied Biosystems/Ambion, Austin, TX) and incubated at 37°C for 30 min. RNA was column purified using an RNeasy minikit (Qiagen, Inc., Valencia, CA). Purified RNA was DNase treated again, with RQ1 RNase-free DNase (Promega, Inc., Madison, WI), and repurified. The quantity and quality of RNA were measured using a NanoDrop spectrophotometer (Thermo Fisher Scientific). cDNA was synthesized from 0.5 μg total RNA using a Superscript III first-strand synthesis kit (Invitrogen). For nonquantitative reverse transcriptase PCR (RT-PCR), 2 μl of cDNA was used as a template for PCR, and products were amplified using hbp-specific primers (Table 2) and then separated on a 3% agarose gel.

To measure relative expression of each hbp gene by reverse transcriptase quantitative PCR (RT-qPCR), hbp-specific primers (Table 2) were designed and tested against a dilution series of gDNA to confirm a primer efficiency of 95 to 105%. All primer sets yielded only one product, as measured by a dissociation curve. glyA and purA, which encode serine hydroxymethyltransferase and adenylosuccinate synthetase, respectively, were amplified as reference genes. RT-qPCR was performed using an MX3000P qPCR machine (Stratagene/Agilent Technologies, Santa Clara, CA). cDNA was synthesized from 0.2 μg total RNA and diluted with 200 μl nuclease-free water. Five μl diluted cDNA from each B. henselae strain grown at 37°C and 28°C was combined with 1 μM each specific primer and SYBR master mix (Applied Biosystems, Carlsbad, CA) in a total reaction volume of 25 μl. Reaction conditions were 95°C for 15 min and 40 cycles of 95°C for 30 s, 50°C for 1 min, and 72°C for 45 s, followed by a product dissociation curve from 55°C to 95°C. The quantification cycle (Cq) was measured for each reaction. Assays were performed in triplicate, and Cq was measured for each. The fold change was calculated for each hbp gene from the same cDNA sample with the 2−ΔΔCq method (28), using the average Cq of glyA as a normalizing control, under the same growth conditions. The fold change values from three independent experiments were then averaged, and standard errors of the means were calculated.

Generation of an isogenic hbpC null mutant in B. henselae JK33S.

Targeted allelic replacement was used to generate a complete, in-frame deletion of the hbpC gene. Specific primers were used to amplify genomic regions upstream (839 bp) and downstream (791 bp) of B. henselae hbpC (Table 2). These two products were used in an overlap PCR to generate a genomic fragment with a complete deletion of the coding sequence of hbpC flanked by BamHI sites. This was then cloned into pCR2.1-TOPO (Invitrogen) and verified by sequencing, yielding pJR158 (Table 1). pJR158 was inserted into pJM05 (29), yielding pJR159. E. coli S17-1 harboring pJR159 was used as the donor strain for biparental conjugation in a two-step, sacB-mediated mutagenesis (29) of hbpC in B. henselae JK33S. After selection on 5% sucrose plates, Kans Sucr colonies were screened by PCR and sequencing to identify allelic replacement resulting in hbpC deletion. For Southern blot confirmation of hbpC deletion, gDNA was isolated from JK33S and JK33S ΔhbpC and probed with the BamHI fragment of pJR158, containing the regions flanking hbpC.

Subcellular fractionation of B. henselae cell lysates and immunoblot analysis.

Four confluent plates of each B. henselae strain were used for each fractionation. The Sarkosyl-based fractionation protocol was performed as described previously (7), but inner membrane protein (IMP) and cytosolic (CYT) fractions were used without further precipitation or concentration. The final OMP pellet was solubilized in 10 mM HEPES, pH 8.0, with 0.1% sodium dodecyl sulfate (SDS). Total protein concentrations of whole-cell lysate (WCL), CYT, IMP, and OMP fractions were determined (MicroBCA protein assay kit; Pierce, Rockford, IL), and 2.5 μg total protein per sample was separated by 12% SDS-PAGE.

Proteins separated by SDS-PAGE were transferred to nitrocellulose membranes and incubated with a 1:5,000 dilution of mouse anti-HbpA (anti-Pap31) antibody (60) or a 1:2,000 dilution of rabbit anti-His antibody (Cell Signaling Technology, Beverly, MA), followed by a horseradish peroxidase (HRP)-conjugated secondary antibody, and visualized after addition of ECL reagent (Pierce).

In-gel trypsin digestion and mass spectrometry.

Individual protein spots of interest were excised from Coomassie-stained SDS-polyacrylamide gels and cut into 1-mm2 pieces. In-gel trypsin digestion, high-performance liquid chromatography–tandem mass spectrometry (HPLC–MS-MS), and protein identification by comparison to eubacterial protein sequences within the MSDB database ( were performed as described previously (7).

Hemin binding by B. henselae and E. coli.

To measure the amount of hemin bound by each strain, a protocol for Bartonella hemin binding was used as described previously (32), with modifications. Two confluent plates of each B. henselae strain were grown at 37°C, and two at 28°C. Bacteria were pelleted and washed three times, and bacterial aggregates were disrupted by passage through a 27-gauge needle 4 times. Each bacterial suspension was diluted to a final OD600 of 1. Thirty μg hemin was added to each tube, and the suspension was incubated for 1 h in a 5% CO2 atmosphere and mixed by inversion every 15 min. Three identical tubes without bacteria were prepared and incubated in parallel. After incubation, cells were pelleted for 2 min at 16,000 × g, and the supernatant was removed and centrifuged again to completely remove any bacteria. The amount of hemin present after centrifugation was quantified by measuring absorbance of the supernatant at 400 nm. The amount of hemin bound was calculated and compared to that in a control reaction mixture without bacteria, which was similarly processed. Binding assays for each strain were performed in triplicate. Ninety-five percent confidence intervals were calculated, and a two-tailed t test was performed to measure statistical significance.

To test the ability of HbpC to bind hemin in a heterologous expression system, we generated a construct that contained the hbpC coding sequence under the control of a T7 promoter and a lac operator for expression in E. coli. The hbpC coding sequence was amplified without the predicted native signal sequence and subcloned into pET22b+, downstream of the pelB periplasmic signal sequence (pelBss), yielding pJR215. For hemin binding experiments, overnight cultures of E. coli BL21(DE3) cells containing pJR215 were diluted 1:100 and grown for 2 h at 37°C. To induce expression of the PelBSS::HbpC::His6 construct, isopropyl β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 1 mM and cells were induced for 2 h at 30°C. Induced and uninduced pET22b+ and uninduced pET22b+ empty vector controls were grown under identical conditions. Induced and uninduced cells were washed 3 times, diluted to a final OD600 of 1, incubated with 30 μg hemin, and assayed as described above for B. henselae strains. Ninety-five percent confidence intervals were calculated, and a two-tailed t test was performed to measure statistical significance.

Comparative resistance of B. henselae strains to hemin toxicity.

B. henselae strains were grown at 37°C or 28°C. For each strain, cells from 2 confluent plates were resuspended, washed, and then resuspended in 1 ml M199S. To disrupt bacterial aggregates, cells were passed through a 27-gauge needle 5 times and diluted to an OD600 of 1. Bacteria were incubated with hemin at a final concentration of 5 mM or with an equal volume of control solution without hemin, adjusted with HCl to match the pH of the hemin stock solution. Incubation was carried out in 5-ml round-bottomed polystyrene tubes at 37°C or 28°C in 5% CO2 for 24 h. After incubation, cells were washed twice and resuspended in M199S and then serially diluted and plated in triplicate onto chocolate agar plates for CFU quantification after 7 to 10 days of incubation at 35°C. To calculate percent survival, CFU after incubation with hemin were divided by total CFU. Ninety-five percent confidence intervals for replicates were calculated, and two-tailed t tests were used to measure statistical significance.

OMV purification and analysis.

OMV were isolated from B. henselae strains grown on chocolate agar plates, using a modification of the protocol previously described for isolation of OMV from Pseudomonas aeruginosa plate-grown biofilms (38, 47). B. henselae was scraped from the surfaces of 4 plates per strain and resuspended in 1 ml sterile 0.9% (wt/vol) NaCl. A homogeneous solution was obtained after pipetting the bacterial suspension, and cell density was calculated by measuring the OD600. Approximately the same number of cells was used for each strain in further preparations. Cell suspensions were centrifuged for 10 min at 2,000 × g at 4°C, the supernatant was retained, and the pellet was resuspended in 1 ml fresh 0.9% NaCl and centrifuged. This was repeated twice, and all supernatants were pooled and filtered through 0.22-μm syringe-top filters (Millipore) to remove any remaining whole cells. OMV were recovered from the resulting filtrates by ultracentrifugation at 150,000 × g for 1.5 h at 4°C in a Ti45 rotor (Beckman Instruments, Fullerton, CA). The vesicle pellet was resuspended with 50 mM HEPES buffer, the total protein present in each OMV fraction was quantified, and equal amounts of total protein were separated by SDS-PAGE and immunoblotted.

OMV were visualized by transmission electron microscopy (TEM) using a negative staining method as described previously (20). Five μl of the OMV sample was placed on Formvar- and carbon-coated grids for 5 min. Excess fluid was removed, and grids were placed sample side down on a drop of 2% uranyl acetate for 1 to 2 s. Excess fluid was removed, and the sample was air dried for at least 1 h before examination by TEM (Philips Tecnai 10; FEI, Hillsboro, OR).

For semiquantitative analysis of vesicles generated by each strain grown at 37°C or 28°C, the OMV fraction was diluted 1:10 with phosphate-buffered saline (PBS), mixed gently, and incubated with a final concentration of 3.3 μg/ml FM 4-64 (Invitrogen), a membrane stain, for 10 min at 37°C (30). FM 4-64, PBS, and 2.5 μM hemin samples were used as controls. Assays were performed in triplicate. To calculate relative lipid quantity, samples were excited at 530 nm and emission was measured at 645 nm. Fluorescence units were normalized to the average value for the B. henselae JK33S wild-type strain grown at 37°C by dividing arbitrary fluorescence units measured for each sample by those of B. henselae JK33S grown at 37°C to determine the relative quantity of OMV produced. To estimate hemin quantity in each fraction, the OD400 was measured in triplicate for each OMV sample and compared to a standard curve (18). For both membrane and hemin measurements, 95% confidence intervals were calculated, and two-tailed t tests were used to measure statistical significance.

Experimental infection of cats with isogenic B. henselae HbpC strains.

To assess the role of HbpC in infection, two specific-pathogen-free (SPF) cats per strain were inoculated with B. henselae JK33S, JK33S hbpC*, and B. henselae JK33S ΔhbpC, as described previously (55). The six SPF cats were obtained from the Feline Research Colony at the School of Veterinary Medicine, University of California, Davis. Animals were raised in a flea- and tick-free environment, and all procedures involving animals followed NIH protocols and were approved by and performed according to guidelines of the Institutional Animal Care and Use Committee of University of California, Davis (protocol number 15537). Prior to experimental infection, cats were confirmed to be antibody and culture negative for Bartonella spp. by indirect immunofluorescence assay and blood culture, respectively. Bacteria used for the inoculum were grown on chocolate agar for 7 days at 35°C and then passaged to fresh chocolate agar and grown at 35°C for 4 days before resuspension in M199 medium supplemented with glutamine and pyruvate. Each of the three different strain inocula had a final concentration of 3 × 108 to 4 × 108 CFU per ml.

Each cat was inoculated intradermally with 0.5 to 0.6 ml of inoculum, distributed between 2 sites behind the shoulder blade. Cats were examined every day for the first 2 weeks of infection and every week thereafter. For culture and serology, blood was drawn and collected from each animal in plastic 2-ml EDTA tubes (Becton Dickinson, Franklin Lakes, NJ). After centrifugation, the red blood cell pellet was plated onto fresh chocolate agar plates and incubated at 35°C in candle extinction jars for 3 weeks, and bacterial colonies were enumerated weekly as CFU/ml of blood.


B. henselae is efficiently mutagenized by the mariner-based transposon construct pBT20.

To identify B. henselae genes involved in hemin transport or uptake, it was first necessary to identify an efficient system of transposon mutagenesis. We compared a Tn5-based system (plasmid pJKTn5-GentR [22]) with a high-efficiency, mariner-based system (plasmid pBT20 [26]). The efficiency of mutagenesis with the mariner transposon was ~10,000-fold greater than that with Tn5 (Table 3). The mariner transposon construct also was capable of mutagenizing B. quintana with similar efficiency (Table 3). All subsequent mutagenesis experiments were performed using the mariner transposon construct pBT20 with a B. henselae JK33S recipient strain, because of the high efficiency of both conjugative transfer and transposon mutagenesis.

Table 3
Conjugation efficiencies of mariner and Tn5 transposons in Bartonella

Visual screening of B. henselae mariner insertion strains identifies mutants with an altered hemin accumulation phenotype.

We hypothesized that B. henselae strains with alterations in hemin accumulation would appear lighter (less hemin uptake) or darker (more uptake) than the parent strain when grown on plates supplemented with 1 mM hemin. Visual screening of B. henselae colonies with transposon insertions for an altered pigmentation phenotype compared to that of the wild type (Fig. 1A) identified 12 darkly pigmented strains, each of which contained a single transposon insertion (Fig. 1B). Approximately 20,000 colonies were screened, which is calculated to represent a genome coverage of greater than 99.9% (i.e., greater than 99.9% chance that each gene in the B. henselae genome was hit at least once) (43). Of the 12 hyperpigmented strains, BH15 displayed the darkest pigmentation phenotype (Fig. 1A) and was selected for further characterization.

Fig 1
mariner transposon mutagenesis of B. henselae JK33S yields strains with darker pigmentation than that of the wild-type parent. (A) Dark colonies, such as BH15, are easily observed with a visual screen of mutagenized Bartonella colonies grown on chocolate ...

The mutant strain BH15 contains a transposon insertion upstream of the B. henselae hemin binding protein C (hbpC) gene.

The BH15 mutant contains a mariner transposon insertion 57 bp upstream of the coding region of hbpC, at the 5′ end of the hbpCAB locus, with the mariner transposon Ptac promoter oriented in the direction of transcription of the downstream hbpC gene (Fig. 2A). We designated the mutant strain BH15 B. henselae JK33S hbpC*.

Fig 2
Insertion of the mariner transposon construct pBT20 upstream of a locus containing three hemin binding protein (hbp) genes in mutant strain BH15 (hbpC*) yields constitutive expression of hbpC. (A) Sequencing of BH15 shows that the mariner insertion site ...

To determine if hbpC mRNA is generated from the Ptac promoter on the inserted transposon, we performed RT-PCR on RNA isolated from B. henselae JK33S and JK33S hbpC*. By use of internal hbpC primers, a 242-bp product was observed from cDNA isolated from both B. henselae JK33S and the hbpC* transposon mutant, which demonstrated that hbpC is transcribed in both strains (Fig. 2B). After amplification with primers that span the junction of the transposon and the hbpC promoter region, a 399-bp product was observed only in cDNA from the B. henselae JK33S hbpC* strain (Fig. 2B). Thus, transcription of hbpC in the hbpC* strain is attributable to constitutive expression from the Ptac promoter present within the inserted mariner transposon.

The isogenic B. henselae JK33S ΔhbpC deletion mutant is nonpolar.

To further understand the function of HbpC, we generated a B. henselae ΔhbpC strain in JK33S. DNA hybridization and blotting of gDNA from B. henselae JK33S and the ΔhbpC mutant, using the genomic region flanking hbpC as a probe, demonstrated that hbpC is deleted in JK33S ΔhbpC (Fig. 3A).

Fig 3
Generation of a B. henselae JK33S hbpC null mutant and comparison of pigmentation of the three isogenic strains JK33S wild type, JK33S ΔhbpC, and JK33S hbpC*. (A) A Southern blot of genomic DNA extracted from individual B. henselae JK33S (wt) ...

We then compared in vitro phenotypic characteristics of B. henselae JK33S ΔhbpC to those of B. henselae JK33S hbpC* and the JK33S parental strain. These 3 isogenic strains grew at similar rates on chocolate agar under standard conditions for Bartonella (37°C, 5% CO2, humidified atmosphere). The pigmentation of the ΔhbpC strain was lighter than that of B. henselae JK33S hbpC*, and pigmentation of the ΔhbpC strain was similar to that of the B. henselae JK33S wild-type strain at both 37°C and 28°C (Fig. 3B and C, respectively). The pigmentation observed in these strains suggests that hemin binding levels are similar in both the B. henselae JK33S ΔhbpC strain and the JK33S parent strain in vitro on chocolate agar but that hemin binding by these two strains is much less than that in the JK33S hbpC* strain.

We next tested whether disruption of hbpC in the B. henselae ΔhbpC mutant has a polar effect on, or perturbs, transcription of the downstream genes hbpA and hbpB. These 3 hbp genes are tandemly arranged on the chromosome, in a locus spanning approximately 5 kb (Fig. 2A). To determine if the B. henselae hbpC* and ΔhbpC mutant strains transcribe hbpA and hbpB, we first assayed for the presence of hbp mRNA in the B. henselae JK33S wild-type strain and the hbpC* and ΔhbpC mutant strains by using RT-PCR. hbpA and hbpB mRNA was present in all 3 strains at both 28°C and 37°C (Fig. 4A), and the hbpC transcript was absent in the ΔhbpC strain at both temperatures. Therefore, the hbpA and hbpB genes were transcribed in B. henselae strain JK33S ΔhbpC.

Fig 4
Constitutive transcription or deletion of hbpC alters transcription of other hbp genes. (A) hbpA and hbpB are transcribed in the B. henselae JK33S ΔhbpC mutant grown at 28°C and 37°C. No hbpC mRNA is detected in B. henselae JK33S ...

Changes in hbpC transcription alter mRNA levels of other hbp genes.

We next tested whether deletion or constitutive transcription of hbpC results in quantitative changes in mRNA levels of other hbp genes, including hbpD and hbpE, which are located elsewhere in the genome. We quantified levels of each of the 5 hbp transcripts in all 3 strains (B. henselae JK33S, JK33S ΔhbpC, and JK33S hbpC*) at both 37°C and 28°C using RT-qPCR. Changes in hbpC mRNA levels were not calculated for the ΔhbpC mutant, in which the hbpC coding sequence is deleted. hbpC mRNA levels were increased in the JK33S hbpC* strain at both 37°C and 28°C, compared with those in the JK33S parent strain (Fig. 4B). Interestingly, hbpA transcript levels in B. henselae JK33S hbpC* were decreased at both temperatures (Fig. 4B). In the B. henselae JK33S wild-type strain, we observed that hbpC mRNA levels were increased at 28°C compared with those at 37°C (Fig. 4C). In contrast, we found that hbpA mRNA levels were increased at 28°C in the B. henselae JK33S wild-type strain, which is not observed in B. quintana (2). mRNA levels of hbpB, hbpD, and hbpE were decreased in all 3 strains grown at 28°C compared with levels at 37°C (Fig. 4C). Therefore, mRNA levels of the hbp genes in B. henselae were regulated by temperature. Changes in hbpC expression due to constitutive expression or deletion of hbpC also altered the abundance of mRNA for the other hbp genes.

Constitutive transcription of the hbpC gene in B. henselae JK33S hbpC* results in increased HbpC protein expression.

We next evaluated whether the constitutive expression of hbpC from the Ptac promoter in B. henselae JK33S hbpC* results in increased HbpC protein expression. Visualization of OMP using SDS-PAGE demonstrated the presence of a unique band (Fig. 5A, JR1) in the OMP fractions of B. henselae JK33S hbpC* grown at both 28°C and 37°C. This band, at approximately 30 kDa, had the predicted molecular mass of HbpC and was not observed in the OMP fraction from strains JK33S wild type and JK33S ΔhbpC grown at either 28°C or 37°C (Fig. 5A). Expression of the protein JR1 also was not observed in the CYT and IMP fractions from the JK33S hbpC* strain but was faintly visible in the WCL fraction of the JK33S hbpC* strain grown at 28°C (see Fig. S1 in the supplemental material). We excised the JR1 band from B. henselae JK33S hbpC* grown at 28°C; mass spectrometry confirmed that JR1 is HbpC, with a high level of significance and sequence coverage (Table 4). These data confirmed that constitutive expression of hbpC leads to an increase in HbpC levels in B. henselae JK33S hbpC* grown at both 28°C and 37°C and that HbpC localizes to the OMP fraction.

Fig 5
Subcellular fractionation of B. henselae JK33S, JK33S ΔhbpC, and JK33S hbpC* protein lysates shows that constitutive expression of hbpC at 28°C and 37°C results in increased HbpC in the outer membrane. (A) After Sarkosyl-based ...
Table 4
Identification of B. henselae proteins by high-performance liquid chromatography–electrospray tandem mass spectrometry (HPLC–ESI MS-MS)

B. henselae HbpA protein is differentially expressed at arthropod (28°C) and mammalian (37°C) temperatures and is downregulated by constitutive HbpC expression.

Subcellular fractionation of cell lysates was performed with B. henselae JK33S wild-type, ΔhbpC, and hbpC* strains grown at 28°C. Equal amounts of total protein for all three strains were separated by SDS-PAGE, revealing three additional bands in the OMP fraction in strains grown at 28°C compared to results for strains grown at 37°C, at molecular masses of approximately 30 to 35 kDa (Fig. 5A, JR2, JR3, and JR4). Mass spectrometry identified JR2 as HbpA, and JR3 appeared to be a mixture of the competence lipoprotein ComL precursor and HbpA (Table 4). ComL is an OMP with a molecular weight similar to that of HbpA. JR4, the ~30-kDa protein that was significantly induced at 28°C, also was found to be HbpA by mass spectrometry, with a high level of significance.

We immunoblotted the OMP fractions of B. henselae JK33S wild-type, ΔhbpC, and hbpC* strains with an anti-HbpA (anti-Pap31) monoclonal antibody (60). This antibody detected multiple bands, including proteins at the predicted mass of HbpA (30 to 35 kDa), in OMP fractions from B. henselae strains grown at 28°C and 37°C. Proteins at higher molecular masses (presumed to be HbpA oligomers) also were present in OMP fractions from B. henselae JK33S wild type and JK33S ΔhbpC grown at 28°C, but not 37°C (Fig. 5B). This correlates with the significant upregulation of hbpA mRNA levels observed for B. henselae JK33S wild type and JK33S ΔhbpC at 28°C compared with 37°C (Fig. 4C). Although HbpA protein expression was increased at 28°C in JK33S wild type and JK33S ΔhbpC, HbpA protein expression (Fig. 5B) and hbpA mRNA levels (Fig. 4B) were decreased in JK33S hbpC* grown at both temperatures. These data demonstrate that both changes in environmental conditions (e.g., temperature) and altered expression of hbpC affect hbpA mRNA and protein expression.

Upregulated expression of HbpC in B. henselae and heterologous expression of HbpC in E. coli confer the ability to bind an increased amount of hemin.

Although the specific function of HbpC is unknown, we hypothesized that it binds hemin, because the related protein HbpA is known to bind hemin in vitro (32, 60). We found that hemin binding by the B. henselae JK33S hbpC* strain grown at both temperatures increases significantly compared to hemin binding by the JK33S wild-type and ΔhbpC mutant strains (Fig. 6A). No significant difference was observed in the amounts of hemin bound by the JK33S wild-type strain and the ΔhbpC mutant. These data demonstrated that although HbpC binds hemin, proteins other than HbpC, such as HbpA, contribute to the majority of hemin binding activity in the JK33S wild-type strain. The increased hemin binding capacity of the hbpC* strain at both 37°C and 28°C is likely due to the overexpression of HbpC.

Fig 6
Overexpression of HbpC in B. henselae JK33S hbpC* or E. coli BL21(DE3) significantly increases hemin binding capacity compared with that for control strains. (A) B. henselae cells were incubated with 30 μg/ml (~0.05 mM) hemin for 1 h. ...

To further examine the ability of HbpC to mediate hemin binding in the absence of other Hbp, we heterologously expressed B. henselae hbpC in E. coli BL21(DE3), which does not encode any Hbp homologs. The expression of PelBSS::HbpC::His6 in this strain was confirmed by immunoblotting with anti-His antibodies (see Fig. S2 in the supplemental material). We found that expression of PelBSS::HbpC::His6 in E. coli confers the ability to bind exogenous hemin (Fig. 6B). These data demonstrate that HbpC is sufficient to mediate hemin binding in the absence of other B. henselae Hbp.

Constitutive expression of HbpC in B. henselae confers resistance to toxic concentrations of hemin.

We hypothesized that HbpC binds and sequesters excess hemin, contributing to the survival of Bartonella in the presence of high concentrations of heme in the arthropod gut at 28°C. We tested whether increased levels of HbpC (JK33S hbpC*) or absence of HbpC (JK33S ΔhbpC) affects the ability of B. henselae to resist the toxic effects of high hemin concentrations (5 mM) in vitro. Incubation of B. henselae JK33S, JK33S ΔhbpC, and JK33S hbpC* for 24 h at 37°C in medium supplemented with 5 mM hemin revealed that a significantly higher percentage of B. henselae JK33S hbpC* bacteria than wild-type or JK33S ΔhbpC bacteria survive exposure to high concentrations of hemin (Fig. 7). This protective effect also was observed for JK33S hbpC* bacteria grown and incubated with hemin at 28°C. These data suggest that constitutive expression of hbpC in B. henselae JK33S hbpC* attenuates the toxic effects of high concentrations of hemin.

Fig 7
Constitutive expression of hbpC protects B. henselae against hemin toxicity. B. henselae JK33S, JK33S ΔhbpC, and JK33S hbpC* were incubated with 5 mM hemin or a pH-identical control solution in M199S medium for 24 h, and surviving bacterial cells ...

B. henselae produces OMV, and HbpC and HbpA are found in these OMV.

We sought to determine whether B. henselae generates OMV, as observed in the cell-free supernatant of many Gram-negative bacteria, and to investigate whether Hbp are associated with these vesicles. We isolated OMV from B. henselae strains, demonstrating for the first time that Bartonella generates OMV. TEM of negatively stained OMV fractions from JK33S, JK33S ΔhbpC, or JK33S hbpC* grown at 28°C demonstrated that all 3 B. henselae JK33S strains produce vesicles of sizes between 20 nm and 100 nm (Fig. 8A).

Fig 8
B. henselae produces outer membrane vesicles (OMV) that contain HbpA and HbpC, and constitutive expression of HbpC increases the amount of hemin associated with OMV. (A) OMV isolated from B. henselae JK33S, JK33S ΔhbpC, and JK33S hbpC* grown on ...

When protein constituents of the OMV isolated from B. henselae hbpC* grown at 28°C and 37°C were resolved by SDS-PAGE, prominent bands were observed (Fig. 8B, left) at the molecular masses predicted for HbpC and HbpA. The constituent proteins in the bands marked by an asterisk appear to be a mix of HbpC and HbpA, as determined by mass spectrometry (Table 4, JR5) and by immunoblotting with anti-HbpA antibody (Fig. 8B, right). These data demonstrate that B. henselae generates OMV and that HbpC and HbpA are present in these vesicles.

OMV expression is increased at 28°C, and constitutive expression of HbpC significantly increases the hemin binding capacity of OMV.

To determine whether the growth temperature of B. henselae affects the quantity of OMV generated and the amount of hemin present, we prepared OMV from equivalent amounts of B. henselae JK33S, JK33S ΔhbpC, or JK33S hbpC* grown at 37°C and 28°C. We used a semiquantitative method to measure the amount of membrane present from each strain by incubating OMV fractions with FM 4-64, a membrane dye (30). All three strains grown at 28°C produced significantly greater quantities of membrane than the corresponding strain grown at 37°C (Fig. 8C).

To quantify differences in hemin bound by OMV when HbpC is overexpressed, we used a spectrophotometric method (18) to estimate the amount of hemin present in these vesicles. Comparing the 3 strains, each grown at 37°C and 28°C, we found that significantly more hemin is associated with OMV isolated from the B. henselae JK33S hbpC* strain (Fig. 8D). In addition, OMV pellets from the JK33S hbpC* strain were substantially darker than the pellets from the same mass of JK33S and JK33S ΔhbpC, especially at 28°C (Fig. 8E). This corroborates the increased hemin binding by the JK33S hbpC* strain OMV shown in Fig. 8D. These data show that more OMV are produced when B. henselae is grown at the arthropod vector temperature of 28°C and that OMV from the JK33S hbpC* strain sequester significantly more hemin.

The JK33S ΔhbpC and hbpC* strains have a decreased ability to infect cats.

We sought to identify the effect of perturbed hbpC gene expression on the ability of B. henselae to colonize the bloodstream of the natural feline host reservoir. We inoculated two SPF cats with B. henselae JK33S, two with B. henselae JK33S ΔhbpC, and two others with B. henselae JK33S hbpC*. We quantified the level of bacteremia over 140 days by culturing B. henselae from the bloodstream of each inoculated animal (Fig. 9). The maximal levels of bacteremia in the two animals infected with JK33S were 1.7 × 104 and 4.7 × 104 CFU/ml, similar to the maximal bacteremia levels observed in experimental feline infections with other B. henselae type I strains (55, 58). Bacteremia levels after infection with JK33S ΔhbpC appeared attenuated compared with those after infection with JK33S, with a maximum of 3.9 × 102 and 5.7 × 102 CFU/ml. No colonies were isolated from the blood of either animal infected with B. henselae JK33S hbpC* at any time during the experiment. Perturbation of HbpC expression therefore appears to affect the ability of B. henselae to colonize the bloodstream of the feline host.

Fig 9
B. henselae JK33S, JK33S ΔhbpC, and JK33S hbpC* demonstrate differing levels of virulence in vivo in the feline natural host. Two cats each were inoculated intradermally with 2 × 108 CFU of one of the 3 bacterial strains, and blood was ...


Bartonella species require heme for growth, can access heme under conditions of stringent heme sequestration, and yet withstand environmental exposure to much higher concentrations of heme than other bacteria. To understand how Bartonella tolerates the widely disparate concentrations of heme encountered in the mammalian bloodstream and the gut of the arthropod vector, we sought to identify B. henselae mutants with perturbed hemin interactions. To accomplish this, it was first essential to optimize a method for high-efficiency transposon mutagenesis in low-passage human isolates of B. henselae (e.g., JK33) and B. quintana. We directly compared 2 transposon systems used previously in Bartonella species mutagenesis (mariner based [42, 52] and Tn5 based [40]) and found the pBT20 mariner system to be ~10,000-fold more efficient than the pJKTn5-GentR plasmid at mutagenizing B. henselae (Table 3).

Using a visual screen, we identified 12 transposon mutants that appear hyperpigmented compared with wild-type B. henselae JK33S. This phenotype suggested an increase in hemin uptake, binding, or both. Interestingly, we did not identify any mutants with decreased pigmentation. This could occur because inactivation of heme-related genes is lethal or because we were unable to visually discern a difference in in vitro phenotype between the wild-type strain and strains with decreased hemin uptake.

We studied the most darkly pigmented mutant strain, BH15, and found that the mariner transposon inserted directly upstream of hbpC, a member of a family of conserved OMP implicated in Bartonella virulence. Increased uptake of hemin by this strain could have resulted from constitutive hbpC expression from the tac promoter (Ptac) encoded within the transposon, which is known to function in B. henselae (27), or from derepression of hbpC expression due to disruption of any cis-acting element upstream of hbpC. We found that hbpC expression in BH15 occurs due to constitutive expression from the mariner pBT20 Ptac promoter (Fig. 2). The Hbp were named based on the in vitro hemin binding activity of HbpA from B. quintana (10, 32) and B. henselae (60). Within B. henselae and B. quintana, the five Hbp share 45 to 56% amino acid identity. We hypothesized, and were able to demonstrate, that HbpC also is involved in hemin binding. A hemin-binding homolog of Hbp, Omp31, has been identified in Brucella species (15), the closest phylogenetic relative of Bartonella (35). However, the mechanism by which Hbp or Omp31 contributes to pathogenesis has not been defined.

We sought to better understand the role of HbpC in B. henselae, focusing initially on the effect of alteration in hbpC transcription on expression of other hbp genes. mRNA expression data from the ΔhbpC strain suggested that hbpC is not cotranscribed with hbpA and hbpB (Fig. 4A). This corroborates data for the related species B. quintana, demonstrating that only hbpC transcription is increased at arthropod-like temperatures (28 to 30°C) compared to that at 37°C (2). B. henselae hbpC and hbpA also were found to have differing transcriptional profiles during in vitro infection of human endothelial cells (36). We found that changes in expression of hbpC affect mRNA levels of the other hbp genes, particularly hbpA, which is decreased in the hbpC* strain (Fig. 4C). A study on B. quintana demonstrated that alteration of expression of a different hbp gene, hbpA, also altered transcript levels of other hbp genes (32). Of note, we were unable to construct isogenic hbpA and hbpCAB deletion mutants using JK33S despite repeated attempts (data not shown), suggesting that hbpA is an essential gene in B. henselae.

We also examined the effect of temperature on hbp mRNA levels. The hbpC transcript level in the B. henselae JK33S wild-type strain was regulated by temperature, with increased hbpC mRNA levels at 28°C compared to those at 37°C (Fig. 4C), as was observed for the B. quintana hbpC ortholog (2). However, in contrast to B. quintana, in which hbpA transcription levels are quantitatively similar at 37°C and 30°C (2), we found that hbpA transcript expression is significantly increased in both B. henselae wild-type and ΔhbpC strains at 28°C (Fig. 4C) and that hbpB, hbpD, and hbpE expression is decreased. The differences in B. henselae and B. quintana hbp gene transcription in response to decreased temperature could reflect the differences encountered in the environments of the respective arthropod vectors of B. henselae (cat flea) and B. quintana (human body louse). For instance, the cat flea experiences a much greater temperature range than does the human body louse (21).

Quantification of hbp transcript levels in the isogenic B. henselae strains at different temperatures and heme concentrations suggested that transcriptional regulation is coordinated according to environmental conditions. Positive and negative regulation of hbp expression could change the profile of the Hbp present in the bacterium, to take advantage of the different functional role for each Hbp in the response to ambient heme concentration and temperature in different niches. However, the mechanism(s) by which differential expression of the B. henselae hbp genes is regulated is not known.

Alignment of the region upstream of the transcriptional start site of each hbp gene in B. quintana, B. henselae, and B. bacilliformis has identified a putative 14-bp cis-acting element called the H box (hbp family box) (3). In B. quintana, expression of hbp may be regulated in part by the iron response regulator (Irr), whose transcription is increased under low-hemin conditions (3). Irr was shown to bind to the promoter region of hbpC near the putative H box by electrophoretic mobility shift assays, although direct binding to the H box was not tested (3). In addition to a cis-acting element, expression of different hbp genes in a given environmental context could be linked with a two-component system (TCS). In Corynebacterium diphtheriae, the ChrAS TCS is involved in heme-dependent gene activation (4), including upregulation of an ABC transporter involved in resistance to heme toxicity (5). B. henselae contains several TCSs that could respond to altered heme levels in the hbpC* mutant, affecting expression of other hbp genes. It has been reported that expression of hbpA, hbpB, and hbpD is altered in a B. henselae mutant that contains an in-frame deletion in the response regulator batR, a member of the BatR/BatS TCS, which is hypothesized to sense and respond to environmental changes in pH (36).

Although we observed that overexpression of HbpC increases hemin binding, the Hbp do not have homology to classical hemin receptors, and the specific mechanisms involved in the binding of hemin and the subsequent uptake of hemin into Bartonella are unknown. Secondary and tertiary sequence analysis of HbpC, using the Conserved Domain Database and HHPred algorithms, revealed that HbpC has some structural homology with porins that contain a membrane-spanning β-barrel, e.g., OmpA of E. coli and Omp25/Omp31 of Brucella. Although we did not identify any previously described heme-binding ligands, we hypothesize that HbpC binds heme directly via a novel mechanism and/or makes the outer membrane more permeable to heme by allowing it to pass through its pore.

We observed that HbpA resolves into multiple species of slightly differing molecular masses at both temperatures (immunoblots in Fig. 5B and and8B,8B, right). Others have observed this, and two-dimensional (2D) SDS-PAGE with matrix-assisted laser desorption ionization–time-of-flight MS (MALDI-TOF MS) documented the simultaneous presence of HbpA proteins with slightly differing molecular masses and pI (39), of unknown significance. In addition, immunoreactive species were identified at molecular masses consistent with oligomers of HbpA (immunoblots in Fig. 5B and and8B,8B, right), which also have been observed in B. henselae (11, 60) and B. quintana (10). Porins from other Gram-negative organisms, such as Rhodobacter capsulatus and Neisseria meningitidis, are observed as stable trimers in crystal structures, where each β-barrel is a single monomer (50, 54). The trimeric composition of many porins suggests the possibility that monomers of different Hbp could form stable heterotrimeric and/or heterooligomeric structures, because of the amino acid conservation between HbpA and HbpC. Such a direct interaction of HbpA with HbpC in the OMV could be one explanation for the increased presence of higher-molecular-mass forms of HbpA observed in the OMV of the hbpC* strain at 28°C (Fig. 8B). Interestingly, higher-molecular-mass HbpA species are decreased in the OMP fraction of the hbpC* strain at 28°C (Fig. 5B). Interaction of HbpA with HbpC in a heterooligomeric structure could shift the localization of HbpA to the OMV. It will be interesting to examine the direct interactions between HbpA and HbpC in the OMP and OMV to better understand the interplay among these proteins.

Acquisition of heme as a source of iron and prosthetic groups for proteins is critical for survival of Bartonella, but heme also is a biologically reactive molecule. The interaction of hemin with reactive oxygen species, such as hydrogen peroxide, leads to oxidative damage of amino acids and lipids (16). Bartonella species can tolerate extremely high concentrations of heme (up to 1 to 2 mM) compared to other bacteria (33, 34, 44). In comparison, Staphylococcus aureus growth is severely limited in 10 to 20 μM hemin (49, 51), and Neisseria meningitidis growth is inhibited by ~0.2 mM hemin (37). Surprisingly, this tolerance of B. henselae for high levels of hemin occurs even though the genome does not encode a heme oxygenase homolog to degrade and detoxify heme (1).

We established that B. henselae and E. coli strains expressing high levels of HbpC remove more hemin from media than the respective isogenic wild-type strains (Fig. 6), consistent with binding of hemin by HbpC. Interestingly, although HbpA levels are greater at 28°C (Fig. 5B), all three B. henselae strains bind more hemin at 37°C than at 28°C (Fig. 6). This could be because of an increase in the fraction of total cellular HbpA and HbpC localized to OMV at 28°C and the significantly greater production of OMV at 28°C. Hbp in the OMV appear to be associated with hemin, especially at 28°C (Fig. 8D and E). Note that the contribution of OMV-localized Hbp to hemin binding in the three isogenic strains cannot be evaluated using the hemin binding assays with results shown in Fig. 6, because OMV are pelleted only by ultracentrifugation and not by the slow-speed centrifugation used in the hemin binding assay.

Our data implicate Hbp-containing OMV in the mitigation of heme toxicity. We have demonstrated for the first time that Bartonella produces OMV and that Hbp are present in OMV fractions (Fig. 8B; Table 4, JR5). It has been hypothesized that binding of toxic agents by OMV, such as binding of exogenous gentamicin by P. aeruginosa, is a general mechanism utilized to protect bacterial cells, e.g., within a biofilm (47). Interestingly, E. coli releases OMV when under cell envelope stress due to toxic, misfolded proteins and possibly as a general stress mechanism (31). In addition, electron micrographs from supernatants of Brucella suis cultures undergoing acidic shock show vesicles that contain the Hbp homologs Omp31 and Omp25, although these vesicles have not been studied further (6).

We found that more OMV are produced at 28°C than at 37°C (Fig. 8C) and that the B. henselae JK33S hbpC* OMV bind greater amounts of hemin (Fig. 8D). The increased amount of hemin bound in these OMV correlates with the increased number of HbpC protein molecules present in the JK33S hbpC* vesicles. We hypothesize that the decreased toxicity observed in the hbpC* strain is due to the overexpression of HbpC, which results in binding and sequestration of additional hemin through the generation of OMV containing HbpC. OMV released from the bacterial surface would thus decrease the local, cytotoxic concentration of hemin. This would represent a new mechanism to enable a pathogenic bacterium to persist within arthropod vectors containing high levels of toxic molecules, like heme, until transmission to their alternate niche, the mammalian host bloodstream.

Infection of the natural feline host reservoir revealed distinct phenotypes for the three isogenic strains. Infection with wild-type JK33S produced maximal titers of 2 × 104 to 5 × 104 CFU/ml, comparable to titers reported after feline inoculation with other wild-type B. henselae strains (58). Strikingly, the hbpC* strain was unable to colonize the feline host bloodstream, perhaps directly due to the increased hemin sequestration resulting from constitutive expression of HbpC at 37°C, although there could be some indirect contribution from the concomitant decrease in HbpA in the hbpC* strain (Fig. 4).

The B. henselae JK33S ΔhbpC mutant did not have a substantial defect in hemin binding or in protection from toxicity in vitro, compared to the JK33S wild-type strain (Fig. 6 and and7).7). This suggests that the hemin binding functions of the Hbp are overlapping or redundant. It is known that B. henselae HbpA also binds hemin in vitro (10, 11), that HbpA is more abundant than HbpC in the wild-type strain (Fig. 5A and B), and that the hbpA null mutation appears to be lethal (data not shown). Thus, we predict that HbpA makes an important contribution to hemin binding in both the wild type and the ΔhbpC mutant, at both temperatures, attenuating the in vitro phenotype. Inoculation of the feline host with the ΔhbpC mutant did reveal a defect in the ability to colonize the feline bloodstream in vivo (Fig. 9), although the number of animals inoculated was small. This difference could be due to environmental conditions present in vivo in the bloodstream that are not fully recapitulated in vitro. Future experiments will focus on identifying other potential environmental factors that affect hbp expression in vivo, e.g., pH and partial O2 pressure (pO2), and on quantifying transcription of the hbp genes in vivo to further understand the role of HbpC and HbpA while B. henselae is in the host bloodstream.

Our data support a model in which B. henselae wild-type bacteria integrate environmental signals, including temperature and hemin concentration, to coordinate and modulate both the expression of the Hbp and the production of OMV. At 37°C, B. henselae increases expression of HbpA in the outer membrane to bind scarce heme in the mammalian bloodstream and decreases production and shedding of OMV. At 28°C, B. henselae increases expression of both HbpA and HbpC and increases the production and release of Hbp-containing OMV; the OMV sequester hemin, protecting B. henselae from the toxic levels of ambient heme in the arthropod gut.

Supplementary Material

Supplemental material:


We thank Eric Johansen and Steven Hall at the UCSF Mass Spectrometry Core Facility for assistance with mass spectrometry and protein identification, David Rehkopf for assistance with statistical analysis, Ivy Hsieh at the San Francisco Veterans Affairs Medical Center for assistance with OMV TEM micrographs, Volkhard Kempf for providing the anti-Pap31 (anti-HbpA) antibody, Joanne Engel for providing pJKTn5-GentR, and Stephen Lory for providing the mariner transposon construct.

This research was supported by funds from the California HIV/AIDS Research Program of the University of California (J.A.R.) and funds from the National Institutes of Health under Ruth L. Kirschstein National Research Service awards F32AI078627 (J.A.R.) and F32AI062004 (D.H.W.). J. E. Koehler received funding support from a California HIV/AIDS Research Program award, a Burroughs Wellcome Fund Clinical Scientist Award in Translational Research, and NIH grants U54AI065359 and R01AI52813. Feline infection was supported by a grant from the Center for Companion Animal Health (George and Phyllis Miller Feline Research Fund), School of Veterinary Medicine, University of California, Davis.


Published ahead of print 9 January 2012

Supplemental material for this article may be found at


1. Alsmark CM, 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. 101:9716–9721. [PubMed]
2. Battisti JM, Sappington KN, Smitherman LS, Parrow NL, Minnick MF. 2006. Environmental signals generate a differential and coordinated expression of the heme receptor gene family of Bartonella quintana. Infect. Immun. 74:3251–3261. [PMC free article] [PubMed]
3. Battisti JM, et al. 2007. Transcriptional regulation of the heme binding protein gene family of Bartonella quintana is accomplished by a novel promoter element and iron response regulator. Infect. Immun. 75:4373–4385. [PMC free article] [PubMed]
4. Bibb LA, Kunkle CA, Schmitt MP. 2007. The ChrA-ChrS and HrrA-HrrS signal transduction systems are required for activation of the hmuO promoter and repression of the hemA promoter in Corynebacterium diphtheriae. Infect. Immun. 75:2421–2431. [PMC free article] [PubMed]
5. Bibb LA, Schmitt MP. 2010. The ABC transporter HrtAB confers resistance to hemin toxicity and is regulated in a hemin-dependent manner by the ChrAS two-component system in Corynebacterium diphtheriae. J. Bacteriol. 192:4606–4617. [PMC free article] [PubMed]
6. Boigegrain RA, et al. 2004. Release of periplasmic proteins of Brucella suis upon acidic shock involves the outer membrane protein Omp25. Infect. Immun. 72:5693–5703. [PMC free article] [PubMed]
7. Boonjakuakul JK, et al. 2007. Proteomic and immunoblot analyses of Bartonella quintana total membrane proteins identify antigens recognized by sera from infected patients. Infect. Immun. 75:2548–2561. [PMC free article] [PubMed]
8. Bowers TJ, Sweger D, Jue D, Anderson B. 1998. Isolation, sequencing and expression of the gene encoding a major protein from the bacteriophage associated with Bartonella henselae. Gene 206:49–52. [PubMed]
9. Caetano-Anolles G. 1993. Amplifying DNA with arbitrary oligonucleotide primers. PCR Methods Appl. 3:85–94. [PubMed]
10. Carroll JA, Coleman SA, Smitherman LS, Minnick MF. 2000. Hemin-binding surface protein from Bartonella quintana. Infect. Immun. 68:6750–6757. [PMC free article] [PubMed]
11. Chenoweth MR, Greene CE, Krause DC, Gherardini FC. 2004. Predominant outer membrane antigens of Bartonella henselae. Infect. Immun. 72:3097–3105. [PMC free article] [PubMed]
12. Chomel BB, et al. 1995. Bartonella henselae prevalence in domestic cats in California: risk factors and association between bacteremia and antibody titers. J. Clin. Microbiol. 33:2445–2450. [PMC free article] [PubMed]
13. Chomel BB, et al. 1996. Experimental transmission of Bartonella henselae by the cat flea. J. Clin. Microbiol. 34:1952–1956. [PMC free article] [PubMed]
14. Crichton R. 2001. Inorganic biochemistry of iron metabolism: from molecular mechanisms to clinical consequences, 2nd ed John Wiley and Sons, Inc., London, United Kingdom.
15. Delpino MV, Cassataro J, Fossati CA, Goldbaum FA, Baldi PC. 2006. Brucella outer membrane protein Omp31 is a haemin-binding protein. Microbes Infect. 8:1203–1208. [PubMed]
16. Everse J, Hsia N. 1997. The toxicities of native and modified hemoglobins. Free Radic. Biol. Med. 22:1075–1099. [PubMed]
17. Gaynor EC, et al. 2004. The genome-sequenced variant of Campylobacter jejuni NCTC 11168 and the original clonal clinical isolate differ markedly in colonization, gene expression, and virulence-associated phenotypes. J. Bacteriol. 186:503–517. [PMC free article] [PubMed]
18. Genco CA, Odusanya BM, Brown G. 1994. Binding and accumulation of hemin in Porphyromonas gingivalis are induced by hemin. Infect. Immun. 62:2885–2892. [PMC free article] [PubMed]
19. Gutteridge JM, Smith A. 1988. Antioxidant protection by haemopexin of haem-stimulated lipid peroxidation. Biochem. J. 256:861–865. [PubMed]
20. Harris JR. 1999. Negative staining of thinly spread biological particulates. Methods Mol. Biol. 117:13–30. [PubMed]
21. Harwood RF, James MT. 1979. Entomology in human and animal health, 7th ed Macmillan Publishing Co., Inc., New York, NY.
22. Kang PJ, et al. 1997. Identification of Pseudomonas aeruginosa genes required for epithelial cell injury. Mol. Microbiol. 24:1249–1262. [PubMed]
23. Koehler JE, Quinn FD, Berger TG, LeBoit PE, Tappero JW. 1992. Isolation of Rochalimaea species from cutaneous and osseous lesions of bacillary angiomatosis. N. Engl. J. Med. 327:1625–1631. [PubMed]
24. Koehler JE, et al. 1997. Molecular epidemiology of bartonella infections in patients with bacillary angiomatosis-peliosis. N. Engl. J. Med. 337:1876–1883. [PubMed]
25. Koehler JE, Tappero JW. 1993. Bacillary angiomatosis and bacillary peliosis in patients infected with human immunodeficiency virus. Clin. Infect. Dis. 17:612–624. [PubMed]
26. Kulasekara HD, et al. 2005. A novel two-component system controls the expression of Pseudomonas aeruginosa fimbrial cup genes. Mol. Microbiol. 55:368–380. [PubMed]
27. Lee AK, Falkow S. 1998. Constitutive and inducible green fluorescent protein expression in Bartonella henselae. Infect. Immun. 66:3964–3967. [PMC free article] [PubMed]
28. Livak KJ, Schmittgen TD. 2001. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 25:402–408. [PubMed]
29. MacKichan JK, Gerns HL, Chen YT, 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. 76:788–795. [PMC free article] [PubMed]
30. McBroom AJ, Johnson AP, Vemulapalli S, Kuehn MJ. 2006. Outer membrane vesicle production by Escherichia coli is independent of membrane instability. J. Bacteriol. 188:5385–5392. [PMC free article] [PubMed]
31. McBroom AJ, Kuehn MJ. 2007. Release of outer membrane vesicles by Gram-negative bacteria is a novel envelope stress response. Mol. Microbiol. 63:545–558. [PMC free article] [PubMed]
32. Minnick MF, et al. 2003. Five-member gene family of Bartonella quintana. Infect. Immun. 71:814–821. [PMC free article] [PubMed]
33. Myers WF, Cutler LD, Wisseman CL., Jr 1969. Role of erythrocytes and serum in the nutrition of Rickettsia quintana. J. Bacteriol. 97:663–666. [PMC free article] [PubMed]
34. Myers WF, Osterman JV, Wisseman CL., Jr 1972. Nutritional studies of Rickettsia guintana: nature of the hematin requirement. J. Bacteriol. 109:89–95. [PMC free article] [PubMed]
35. O'Connor SP, Dorsch M, Steigerwalt AG, Brenner DJ, Stackebrandt E. 1991. 16S rRNA sequences of Bartonella bacilliformis and cat scratch disease bacillus reveal phylogenetic relationships with the alpha-2 subgroup of the class Proteobacteria. J. Clin. Microbiol. 29:2144–2150. [PMC free article] [PubMed]
36. Quebatte M, et al. 2010. The BatR/BatS two-component regulatory system controls the adaptive response of Bartonella henselae during human endothelial cell infection. J. Bacteriol. 192:3352–3367. [PMC free article] [PubMed]
37. Rasmussen AW, Alexander HL, Perkins-Balding D, Shafer WM, Stojiljkovic I. 2005. Resistance of Neisseria meningitidis to the toxic effects of heme iron and other hydrophobic agents requires expression of ght. J. Bacteriol. 187:5214–5223. [PMC free article] [PubMed]
38. Renelli M, Matias V, Lo RY, Beveridge TJ. 2004. DNA-containing membrane vesicles of Pseudomonas aeruginosa PAO1 and their genetic transformation potential. Microbiology 150:2161–2169. [PubMed]
39. Rhomberg TA, et al. 2004. Proteomic analysis of the sarcosine-insoluble outer membrane fraction of the bacterial pathogen Bartonella henselae. Proteomics 4:3021–3033. [PubMed]
40. Riess T, Anderson B, Fackelmayer A, Autenrieth IB, Kempf VA. 2003. Rapid and efficient transposon mutagenesis of Bartonella henselae by transposome technology. Gene 313:103–109. [PubMed]
41. Riess T, et al. 2004. Bartonella adhesin A mediates a proangiogenic host cell response. J. Exp. Med. 200:1267–1278. [PMC free article] [PubMed]
42. Saenz HL, et al. 2007. Genomic analysis of Bartonella identifies type IV secretion systems as host adaptability factors. Nat. Genet. 39:1469–1476. [PubMed]
43. Sambrook J, Russell DW. 2001. Molecular cloning: a laboratory manual, 3rd ed Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
44. Sander A, Kretzer S, Bredt W, Oberle K, Bereswill S. 2000. Hemin-dependent growth and hemin binding of Bartonella henselae. FEMS Microbiol. Lett. 189:55–59. [PubMed]
45. Schaible UE, Kaufmann SH. 2004. Iron and microbial infection. Nat. Rev. Microbiol. 2:946–953. [PubMed]
46. Schmitt TH, Frezzatti WA, Jr, Schreier S. 1993. Hemin-induced lipid membrane disorder and increased permeability: a molecular model for the mechanism of cell lysis. Arch. Biochem. Biophys. 307:96–103. [PubMed]
47. Schooling SR, Beveridge TJ. 2006. Membrane vesicles: an overlooked component of the matrices of biofilms. J. Bacteriol. 188:5945–5957. [PMC free article] [PubMed]
48. Simon R, Priefer U, Puhler A. 1983. A broad host range mobilization system for in vivo genetic engineering: transposon mutagenesis in Gram negative bacteria. Nat. Biotechnol. 1:784–791.
49. Stauff DL, et al. 2008. Staphylococcus aureus HrtA is an ATPase required for protection against heme toxicity and prevention of a transcriptional heme stress response. J. Bacteriol. 190:3588–3596. [PMC free article] [PubMed]
50. Tanabe M, Nimigean CM, Iverson TM. 2010. Structural basis for solute transport, nucleotide regulation, and immunological recognition of Neisseria meningitidis PorB. Proc. Natl. Acad. Sci. U. S. A. 107:6811–6816. [PubMed]
51. Torres VJ, et al. 2007. A Staphylococcus aureus regulatory system that responds to host heme and modulates virulence. Cell Host Microbe 1:109–119. [PMC free article] [PubMed]
52. Vayssier-Taussat M, et al. 2010. The Trw type IV secretion system of Bartonella mediates host-specific adhesion to erythrocytes. PLoS Pathog. 6:e1000946. [PMC free article] [PubMed]
53. Wandersman C, Delepelaire P. 2004. Bacterial iron sources: from siderophores to hemophores. Annu. Rev. Microbiol. 58:611–647. [PubMed]
54. Weiss MS, et al. 1991. Molecular architecture and electrostatic properties of a bacterial porin. Science 254:1627–1630. [PubMed]
55. Werner JA, et al. 2007. Experimental infection of domestic cats with passaged genotype I Bartonella henselae. Vet. Microbiol. 122:290–297. [PubMed]
56. Wilks A, Burkhard KA. 2007. Heme and virulence: how bacterial pathogens regulate, transport and utilize heme. Nat. Prod. Rep. 24:511–522. [PubMed]
57. Windholz M. 1976. The Merck index: an encyclopedia of chemicals, drugs, and biologicals, 9th ed John Wiley and Sons, Inc., London, United Kingdom.
58. Yamamoto K, et al. 2003. Infection and re-infection of domestic cats with various Bartonella species or types: B. henselae type I is protective against heterologous challenge with B. henselae type II. Vet. Microbiol. 92:73–86. [PubMed]
59. Zangwill KM, et al. 1993. Cat scratch disease in Connecticut. Epidemiology, risk factors, and evaluation of a new diagnostic test. N. Engl. J. Med. 329:8–13. [PubMed]
60. Zimmermann R, Kempf VA, Schiltz E, Oberle K, Sander A. 2003. Hemin binding, functional expression, and complementation analysis of Pap 31 from Bartonella henselae. J. Bacteriol. 185:1739–1744. [PMC free article] [PubMed]

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