The VirB type IV secretion system of Bartonella henselae mediates invasion, proinflammatory activation and antiapoptotic protection of endothelial cells


  • Michael C. Schmid,

    1. Division of Molecular Microbiology, Biozentrum, University of Basel, Klingelbergstrasse 70, 4056 Basel, Switzerland.
    Search for more papers by this author
  • Ralf Schulein,

    1. Division of Molecular Microbiology, Biozentrum, University of Basel, Klingelbergstrasse 70, 4056 Basel, Switzerland.
    Search for more papers by this author
  • Michaela Dehio,

    1. Division of Molecular Microbiology, Biozentrum, University of Basel, Klingelbergstrasse 70, 4056 Basel, Switzerland.
    Search for more papers by this author
  • Geertrui Denecker,

    1. Division of Molecular Microbiology, Biozentrum, University of Basel, Klingelbergstrasse 70, 4056 Basel, Switzerland.
    Search for more papers by this author
    • Present address: Department for Molecular Biomedical Research, Ghent University, Technologiepark 927, 9052 Ghent, Belgium.

  • Ilaria Carena,

    1. Division of Molecular Microbiology, Biozentrum, University of Basel, Klingelbergstrasse 70, 4056 Basel, Switzerland.
    Search for more papers by this author
  • Christoph Dehio

    Corresponding author
    1. Division of Molecular Microbiology, Biozentrum, University of Basel, Klingelbergstrasse 70, 4056 Basel, Switzerland.
    Search for more papers by this author


Bartonella henselae is an arthropod-borne zoonotic pathogen causing intraerythrocytic bacteraemia in the feline reservoir host and a broad range of clinical manifestations in incidentally infected humans. Remarkably, B. henselae can specifically colonize the human vascular endothelium, resulting in inflammation and the formation of vasoproliferative lesions known as bacillary angiomatosis and bacillary peliosis. Cultured human endothelial cells provide an in vitro system to study this intimate interaction of B. henselae with the vascular endothelium. However, little is known about the bacterial virulence factors required for this pathogenic process. Recently, we identified the type IV secretion system (T4SS) VirB as an essential pathogenicity factor in Bartonella, required  to  establish  intraerythrocytic  infection  in the mammalian reservoir. Here, we demonstrate that the VirB T4SS also mediates most of the virulence attributes associated  with  the  interaction  of B. henselae during the interaction with human endothelial cells. These include: (i) massive rearrangements of the actin cytoskeleton, resulting in the formation of bacterial aggregates and their internalization by the invasome structure; (ii) nuclear factor κB-dependent proinflammatory activation, leading to cell adhesion molecule expression and chemokine secretion, and (iii) inhibition of apoptotic cell death, resulting in enhanced endothelial cell survival. Moreover, we show that the VirB system mediates cytostatic and cytotoxic effects at high bacterial titres, which interfere with a potent VirB-independent mitogenic activity. We conclude that the VirB T4SS is a major virulence determinant of B. henselae, required for targeting multiple endothelial cell functions exploited by this vasculotropic pathogen.


Bartonella henselae is a zoonotic pathogen of growing medical importance. In the feline reservoir host, this arthropod-borne bacterium causes intraerythrocytic bacteraemia. Transmission to humans occurs by cat scratch or bite or the bite of an infected cat flea. Depending on the immune status of the infected individual, B. henselae can cause a variety of clinical manifestations. Immunocompetent patients develop cat scratch disease (a necrotizing lymphadenopathy with fever), endocarditis or neuroretinitis. In immunocompromised patients, B. henselae causes vasoproliferative lesions, which result in the formation of tumours of the skin or inner organs (bacillary angiomatosis) or blood-filled cysts in the liver and spleen (bacillary peliosis) (Karem et al., 2000). Among all human pathogenic bacteria, this remarkable capacity to trigger vasoproliferative tumour growth is limited to Bartonella spp. (Dehio, 2003).

Within vasoproliferative lesions B. henselae is found in close association with proliferating endothelial cells. Clearance of infection by antibiotic treatment results in complete regression of vascular lesions. These findings suggest that B. henselae specifically colonizes the vascular endothelium and produces a mitogenic factor that acts locally and temporarily. Primary human umbilical vein endothelial cells (HUVEC) are used as in vitro system to study this interaction with the vascular endothelium (Dehio, 2001; 2003). Bartonella henselae invades and colonizes HUVEC by two distinct routes, either as individual bacteria through a classical endocytotic pathway, or as bacterial aggregates which are formed on the cell surface, followed by their engulfment and internalization via the invasome structure. Invasome-formation and internalization is an actin-dependent process resulting in massive cytoskeletal rearrangements (Dehio et al., 1997). Remarkably, B. henselae can stimulate the proliferation and migration of HUVEC even without direct contact, suggesting that bacteria secrete a vascular mitogen (Maeno et al., 1999). Endogenous growth factors (e.g. vascular endothelial  cell  growth  factor,  VEGF)  released  by B. henselae-infected macrophages in vitro may also contribute to endothelial cell proliferation in vivo (Kempf et al., 2001; Resto-Ruiz et al., 2002; Dehio, 2003). The proinflammatory activation described for B. henselae-infected endothelial cells (Fuhrmann et al., 2001) could contribute to the recruitment of macrophages and other inflammatory cells competent for secretion of VEGF, which typically infiltrate vasoproliferative lesions. The nuclear factor κB(NF-κB)-dependent  proinflammatory  response  of B. henselae-infected HUVEC is characterized by cell surface expression of cell adhesion molecules (i.e. ICAM-1 and E-selectin), which result in increased neutrophil rolling and adherence (Fuhrmann et al., 2001). Finally, B. henselae protects HUVEC from apoptotic cell death as shown by the suppression of caspase activation and DNA fragmentation. The resulting increased endothelial cell survival is considered to contribute to the formation of vasoproliferative lesions in vivo (Kirby and Nekorchuk, 2002).

Recently, we have identified the VirB type IV secretion system (T4SS) as an essential pathogenesis factor in the related species B. tribocorum (Schulein and Dehio, 2002). T4SS are multicomponent transporters of Gram-negative bacteria with functions as diverse as the delivery of effector proteins into eukaryotic target cells in pathogenesis or DNA transfer in bacterial conjugation (Christie, 2001). The highly conserved VirB T4SS of B. tribocorum and B. henselae are encoded by operons of 10 genes (virB2-10) (Padmalayam et al., 2000; Schulein and Dehio, 2002). Unlike wild-type B. tribocorum, mutants deleted for virB4 are unable to cause intraerythrocytic bacteraemia (Schulein and Dehio, 2002), which is the hallmark of infection in the mammalian reservoir host (Schulein et al., 2001). Further analysis revealed that the VirB T4SS is required at an early infection stage before the onset of intraerythrocytic bacteraemia (Schulein and Dehio, 2002). Based on the observation that the virB operon of B. henselae is induced during infection of endothelial cells in vitro (Schmiederer et al., 2001), we reasoned that the VirB system may be involved in mediating endothelial interaction in both the mammalian reservoir and the incidental human host. Here, we used genetic analysis of the VirB system of B. henselae in combination with cell assays to study the role of this T4SS during HUVEC infection. We demonstrate that most of the known physiological changes associated with B. henselae infection of HUVEC, i.e. actin remodelling, and the induction of a proinflammatory and antiapoptotic response are dependent on the VirB system. Moreover, we show that VirB mediates cytostatic or even cytotoxic effects at high bacterial titres, which interfere with the VirB-independent mitogenic activity of B. henselae. Together, our data demonstrate that the VirB T4SS represents a major virulence determinant of Bartonella mediating subversion of multiple vascular endothelial cell functions.


Mutagenesis of the VirB T4SS of B. henselae

As a genetic basis for analysing the role of VirB in mediating B. henselae–HUVEC interaction, we constructed an apolar in frame deletion in the virB4 gene and complemented the resulting ΔvirB4 mutant with full-length virB4 in trans (ΔvirB4/pvirB4). The isogenic wild-type, ΔvirB4 and ΔvirB4/pvirB4 strains were used for parallel infection in cell assays described below.

Course of bacterial growth in endothelial cell co-culture, cell invasion, and intracellular survival of wild-type and ΔvirB4 mutant bacteria

We first tested whether the ΔvirB4 mutant is attenuated in the capacity to grow in co-culture with, invade into, or survive within endothelial cells. Therefore HUVEC were infected with a multiplicity of infection (MOI) of 100 bacteria per cell (MOI = 100) for 0 h, 6 h, 30 h, 54 h and 78 h. To quantify the total number of cultivatable bacteria in HUVEC co-culture, the infected cell monolayers were lysed, and colony forming units (cfu) were determined by plating of serial dilutions. Figure 1A shows a parallel course of growth of wild-type and ΔvirB4 mutant bacteria over the entire time-course. The number of bacteria changed little during the first 30 h of infection, but a strong increase followed at later time points (54 h and 78 h). A parallel experiment, in which gentamicin was used to kill extracellular bacteria (gentamicin-protection assay), determined the course of cell invasion (for time points 0 h, 6 h and 30 h) and intracellular survival (for time-points 30 h, 54 h and 78 h). Figure 1B, lower panel, shows a similar number of intracellular wild-type and ΔvirB4 mutant bacteria at 6 h of infection. The number of intracellular bacteria strongly increased at 30 h of infection, however, with significantly more ΔvirB4 mutant than wild-type bacteria. To follow the course of intracellular survival of bacteria after 30 h of infection, extracellular bacteria were killed during a 2 h incubation period with gentamicin (30 to 32 h), followed by washing to remove gentamicin and incubation with normal medium for the remaining period of infection (total of 54 h or 78 h). The effectiveness of gentamicin treatment for killing extracellular bacteria in these samples is shown in the upper panel of Fig. 1B (total number of bacteria at 54 h and 78 h). The lower panel illustrates that wild-type and ΔvirB4 mutant bacteria show an equal drop in the number of cultivable intracellular organisms during one day (1.6-fold reduction at 54 h versus 30 h), and a further drop till 78 h of infection.

Figure 1.

Course of bacterial growth in endothelial cell co-culture, cell invasion and intracellular survival of wild-type and ΔvirB4 bacteria. Human umbilical vein endothelial cells were infected with wild-type or ΔvirB4 mutant bacteria. (A, B, upper panel) Total cfu or (B, lower panel) gentamicin-protected cfu (only intracellular bacteria) were determined after the indicated period of infection. (B) To follow the intracellular survival after 30 h of infection, extracellular bacteria were killed by gentamicin treatment for 2 h, followed by washing steps and further incubation in normal medium for the remaining time of the indicated infection period. Mean values ± SD of triplicate samples are presented for one out of two similar experiments.

Taken together, quantification of cfu during HUVEC infection did not provide any indication of attenuation of the ΔvirB4 mutant as compared to wild-type bacteria. The course of growth in co-culture with HUVEC and of intracellular survival was similar for both genotypes. Notably, the ΔvirB4 mutant displayed even an enhanced capacity for cell invasion compared to wild-type bacteria.

VirB is required for invasome-mediated cell entry and the associated massive actin rearrangements

We next investigated whether VirB is required for bacterial internalization into HUVEC by either of two described pathways, via (i) endocytosis of individual bacteria or via (ii) the invasome-mediated engulfment of large bacterial aggregates formed on the cell surface (Dehio et al., 1997). Human umbilical vein endothelial cells infected for 48 h were differentially stained for intra- and extracellular bacteria and for F-actin and examined by confocal microscopy (Fig. 2). Bacteria of all isogenic strains were internalized by conventional endocytosis as shown by intracellular bacteria localizing to a perinuclear position (Fig. 2B–D, bottom plane, red-coloured bacteria in the overlay). In contrast, invasome structures containing a bacterial aggregate (Fig. 2B and D, top plane) and the associated ring-shaped actin rearrangements (Fig. 2B and D, bottom plane) were exclusively triggered by wild-type and ΔvirB4/pvirB4 bacteria. Infection with ΔvirB4 did not result in any obvious cytoskeletal rearrangement compared to the actin cytoskeleton of uninfected cells (Fig. 2A and C). Quantification of invasome-positive cells, infected with three different infection doses (MOI = 30, 100, and 300) of the three isogenic strains revealed a strict dependency of invasome formation on a functional VirB system (Table 1). Invasome formation by the complemented mutant strain ΔvirB4/pvirB4 was indistinguishable from wild-type at higher infection doses, whereas phenotypic restoration was reduced for the lower bacterial titres (i.e. MOI = 30, Table 1).

Figure 2.

VirB mediates massive actin rearrangements resulting in invasome-mediated uptake of bacterial aggregates. Human umbilical vein endothelial cells were infected for 48 h (A) without bacteria (control) or with either (B) wild-type (C) ΔvirB4, or (D) ΔvirB4/pvirB4 bacteria. Cells were immunocytochemically stained for F-actin (upper left quadrant, green), extracellular bacteria (upper right quadrant, blue), and total bacteria (lower left quadrant, red). The overlay of all three channels is presented in the lower right quadrant, distinguishing extracellular bacteria (appearing in pink as a mixture of red and blue) from intracellular bacteria (appearing in red). Confocal pictures were taken (A–D) in a plane just above the cell substratum where actin stress fibres localize (bottom) or (B, D) in a parallel plane 3 µm above (top). The scale bar represents 10 µm.

Table 1. Quantification of invasome formation.
Strain% invasome-positive cells ± SD
MOI = 30MOI = 100 MOI = 300
Wild-type66 ± 396 ± 298 ± 2
Δ virB4  0 ± 0 0 ± 0 0 ± 0
Δ virB4/ pvirB426 ± 480 ± 395 ± 3

Notably, the impairment of ΔvirB4 mutant bacteria to accumulate in an invasome structure resulted in increased endocytic uptake into perinuclear localizing phagosomes (Fig. 2B–D, bottom plane, red-coloured bacteria in the overlay). Given the different time-frames of endocytosis (minutes) and invasome-mediated invasion (up to one day) (Dehio et al., 1997), this finding may explain the increased number of gentamicin-protected (intracellular) ΔvirB4 mutant versus wild-type bacteria at the 30 h time-point of infection (Fig. 1C).

Taken together, we observed an absolute requirement of VirB for invasome formation and the associated massive actin rearrangements.

VirB mediates an NF-κB-dependent proinflammatory response

We next analysed whether VirB mediates the proinflammatory response activated in B. henselae-infected HUVEC (Fuhrmann et al., 2001). As typical markers of an NF-κB-dependent proinflammatory response, the cell surface expression of ICAM-1 and the secretion of the chemokine IL-8 into the culture medium were quantified. Compared to uninfected cells, IL-8 secretion was elevated by ΔvirB4 infection in a time-dependent (30 h and 54 h) and infection-dose dependent manner (MOI = 30, 100 or 300) (Fig. 3A). Infection with wild-type displayed the same time- and infection dose-dependency, whereas IL-8 levels were always significantly increased relative to ΔvirB4. The complemented mutant displayed partial phenotypic restoration (Fig. 3A), as in results reported above for invasome formation. Surface expression of ICAM-1 measured at 54 h of infection with a MOI = 100 (Fig. 3B) was in full accordance with the data obtained for IL-8 secretion. Consistent results were also obtained for the transcriptional activity of NF-κB in HEK293T cells measured by a transfected luciferase reporter plasmid after 30 h of infection with a MOI = 100 (Fig. 3C).

Figure 3.

VirB mediates an NF-κB-dependent proinflammatory phenotype. (A, B) HUVEC or (C) HEK293T cells were infected for (A) 30 h or 54 h (B) 54 h or (C) 30 h without bacteria (control) or with either wild-type, ΔvirB4, or ΔvirB4/pvirB4 bacteria with (A) a MOI as specified (B) MOI = 300 or (C) MOI = 100.
A. IL-8 in the culture supernatants was quantified by ELISA.
B. Cell surface expression of ICAM-1 was determined by flow cytometry.
C. NF-κB activity was determined by luciferase activity expressed from a transfected NF-κB reporter plasmid. All assays were repeated at least three times. Mean values ± SD of triplicate samples are presented for one representative experiment.

Together, these data show that the strong NF-κB-dependent  proinflammatory  response  triggered  by B. henselae infection is mediated primarily by VirB, whereas also VirB-independent processes contribute to this response.

VirB mediates suppression of apoptosis

We next tested whether the VirB system mediates suppression of apoptosis in B. henselae-infected HUVEC (Kirby and Nekorchuk, 2002). We measured the activity of executioner caspases-3 and -7 as early apoptotic markers by using a specific fluorogenic substrate. The basal activity of these caspases in uninfected cells, which likely accounts for a low level of spontaneous apoptosis, was suppressed by infection with all three bacterial strains at a MOI = 100 (Fig. 4A). Treatment with the apoptotic inducer actinomycin D (Kirby and Nekorchuk, 2002) resulted in an equal increase in caspase-3 and  -7 activity in cells which were either uninfected or infected with ΔvirB4. In contrast, infection with wild-type or ΔvirB4/pvirB4 resulted in a complete suppression of actinomycin D-induced caspase activation (Fig. 4B). Suppression of caspase activation by wild type, but not by the ΔvirB4 mutant, was consistently seen for the tested range of infection doses (MOI = 30, 100 and 300, see Fig. 4C). Consistent with the partial phenotypic restoration of the complemented mutant ΔvirB4/pvirB4 reported above, this strain suppressed caspase activation only at higher infection doses (MOI = 100 and 300, see Fig. 4C). The analysis of loss of lipid membrane asymmetry (Fig. 4D) and DNA-fragmentation (Fig. 4E) as late apoptotic markers also demonstrated that wild-type and ΔvirB4/pvirB4, but not ΔvirB4, efficiently suppress actinomycin D-induced apoptosis.

Figure 4.

VirB mediates protection of endothelial cells from apoptosis. (A–E) HUVEC were infected for 24 h without bacteria (control) or with wild-type, ΔvirB4, or ΔvirB4/pvirB4 bacteria with an MOI = 100 if not specified differently. Cells were then incubated (B, C, D, right panel of E) with actinomycin D to trigger apoptosis or (A, left panel of E) without actinomycin D as control. (A, B) After the time-points indicated or (C–E) after 12 h the assay was terminated and various markers of apoptosis were assayed. (A–C) Caspase-3 and -7 activities were measured with a specific fluorogenic peptide substrate. (D) The loss of membrane asymmetry was quantified by flow cytometry analysis of FITC-annexin V-stained cells. Counterstaining with PI allowed differentiation of necrotic cells (upper right quadrant of the dot plot) and apoptotic cells (lower right quadrant). The percentages of cells localizing to these quadrants are indicated. (E) DNA fragmentation was measured by ELISA. (A–C, E) Mean values ± SD of triplicate samples are presented for one representative out of three replica experiments. (E) Actinomycin d-treated samples differing statistically significantly from untreated control samples are marked by an asterisk (P < 0.05) or two asterisks (P < 0.01).

We conclude that VirB is strictly required for the antiapoptotic activity of B. henselae on HUVEC.

VirB mediates cytostatic and cytotoxic effects at high bacterial titres, which interfere with HUVEC proliferation stimulated in response to a potent bacterial mitogen

Finally, we investigated whether VirB is involved in B. henselae-stimulated HUVEC proliferation (Conley et al., 1994; Maeno et al., 1999). During a 5-day proliferation assay in serum-containing culture medium deprived for specific growth factors, the number of uninfected cells dropped slightly (Fig. 5A and B), likely due to spontaneous apoptosis. Unexpectedly, we observed that for all infection doses tested (MOI = 10, 30 or 100) ΔvirB4 strongly stimulated HUVEC proliferation, resulting in a ∼threefold increase in cell number already on day 3. On day 5 the ∼eightfold increase in cell number even surpassed the effect caused by the potent mitogen VEGF (Fig. 5A and B). In contrast, infection with wild type or ΔvirB4/pvirB4 at the lowest dose tested (MOI = 10) resulted in only a ∼twofold increase in cell number on day 3 without further increase till day 5. This indicates that VirB mediates a cytostatic effect at the elevated bacterial titre reached in the extended course of this co-cultivation experiment (compare to Fig. 1A). The higher infection doses (MOI = 30 or 100) also resulted in a ∼twofold increase in cell number on day 3, whereas cell numbers sharply dropped at day 5, indicating that at very high bacterial titres VirB can even mediate cytotoxicity (Fig. 5B). These data clearly show: (i) that the potent mitogenic activity of B. henselae is VirB-independent, reaching maximal activity already at a low infection dose (MOI = 10); (ii) that the cytostatic and cytotoxic effects caused by B. henselae at high titre are entirely dependent on VirB, and (iii) that the VirB-mediated cytostatic/cytotoxic effects interfere with the activity of the VirB-independent mitogen in an infection dose-dependent manner.

Figure 5.

VirB mediates cytostatic/cytotoxic effects at high bacterial titres, which interfere with HUVEC proliferation stimulated by a potent VirB-independent bacterial mitogen. HUVEC in serum-containing medium deprived for specific growth factors were infected for 5 days without bacteria (control) or with either wild-type, ΔvirB4 or ΔvirB4/pvirB4 at the indicated MOI. VEGF (50 ng ml−1) was used as a positive control.
A. Digital pictures of phase-contrast microscopic fields are illustrated for day 1, 3 and 5 (MOI = 30 for bacterial infections).
B. Graph of the proliferation indices (= cell number on given day relative to day 0, mean ± SD of triplicate samples) for day 1, 3 and 5 is illustrated for one representative out of three replica experiments. Proliferation indices which differ statistically significantly from control (day 1) are marked by asterisks (one for P < 0.05, two for P < 0.01).


Vasoproliferative tumour lesions developed by patients infected with B. henselae are the result of a remarkable interaction between the pathogen and vascular endothelial cells. The use of HUVEC as in vitro model allowed to differentiate between four prominent changes of endothelial cell function in response to B. henselae infection (Dehio, 2003), which we have shown here to be all modulated by the T4SS VirB – the data are summarized schematically in Fig. 6. (i) The massive cytoskeletal rearrangements resulting in invasome-mediated uptake of bacterial aggregates (Dehio et al., 1997) are entirely dependent on VirB. This unique invasion process competes with bacterial internalization by conventional endocytosis (Dehio et al., 1997), therefore the ΔvirB4 mutant impaired in invasome-mediated invasion displayed increased internalization by the classical endocytic pathway. Internalization via the VirB-dependent invasome represents a novel paradigm for the invasion of bacteria into host cells and may serve as a cellular colonization mechanism for endothelial cells associated with the formation of vasoproliferative lesions (Dehio, 2003). (ii) The NF-κB-dependent proinflammatory response (Fuhrmann et al., 2001), considered to contribute indirectly to vasoproliferative growth (Kempf et al., 2001; Resto-Ruiz et al., 2002; Dehio, 2003), is activated to a low level by VirB-independent factors (e.g. LPS), whereas the majority of this phenotype is clearly contributed by VirB. (iii) The inhibition of apoptosis, which results in increased endothelial cell survival and thereby directly contributes to vasoproliferative tumour formation (Kirby and Nekorchuk, 2002), is strictly dependent on VirB. (iv) Mitogenic signalling resulting in endothelial cell proliferation (Dehio, 2003) was the only known physiological change triggered by B. henselae infection that we could not assign to VirB. However, VirB turned out to interfere with this proliferative response in an unexpected manner. Remarkably, ΔvirB4 elicited a proliferative response at least as strong as that by the potent mitogen VEGF. HUVEC proliferation was maximal already at the lowest bacterial infection dose tested (MOI = 10), whereas even the highest dose (MOI = 100) did not cause any sign of cytotoxicity. In sharp contrast, both wild type and ΔvirB4/pvirB4 caused cytostatic or even cytotoxic effects in an infection dose-dependent manner, which masked most of the potent mitogenic activity of B. henselae. VirB-mediated cytotoxicity was only ob-served for elevated bacterial titres reached rather late in the course of a 5-day proliferation assay, which was performed as a co-cultivation experiment as described before (Conley et al., 1994; Maeno et al., 1999). Modification of the assay conditions by either daily replacement of the culture medium to remove an excess of bacteria or antibiotic killing of bacteria on day 2 abolished VirB-mediated cytotoxicity and resulted in a similar proliferation response elicited by all three isogenic strains (data not shown). Whether VirB-mediated cytostatic and cytotoxic effects represent an in vitro artifact because of elevated bacterial titres or whether they have a physiological role in vivo (e.g. to control an excessive proliferation response) needs to be addressed further.

Figure 6.

Scheme of the role of the VirB T4SS in mediating B. henselae– vascular endothelial cell interaction.

Collectively, our data demonstrate that the VirB T4SS of B. henselae modulates several fundamentally important endothelial cell functions in vitro, indicating that it represents a central virulence factor for establishing chronic infection of the human vascular endothelium. Our previous work in the B. tribocorum/rat model showed that VirB is also required for establishing intraerythrocytic bacteraemia – the hallmark of Bartonella-infection in the mammalian reservoir (e.g. in the cat for B. henselae) (Schulein and Dehio, 2002). In this case, VirB is essential for colonizing a yet undefined primary niche in which the pathogen gains competence for the subsequent erythrocyte infection process (Schulein and Dehio, 2002). Circumstantial evidence suggests that endothelial cells represent a major constituent of this primary niche (Dehio, 2003). Thus, it appears likely that the VirB system represents an essential virulence determinant for vascular endothelial cell infection both in mammalian reservoir hosts (i.e. for establishing intraerythrocytic infection) and in incidentally infected humans (i.e. for establishing the prominent vascular pathologies). Together with surface expression of VirB components, such as the immunogenic 17 kDa antigen (Anderson et al., 1995; Padmalayam et al., 2000), this T4SS system represents an essential surface structure for pathogenicity and therefore a promising target for vaccine development.

T4SS of other human pathogens have been recognized to mediate important virulence features: the Cag system of Helicobacter pylori stimulates proinflammatory activation and a motile phenotype of gastric epithelial cells, the Dot/Icm system of Legionella pneumophila and the VirB system of Brucella spp. are required for the establishment of an intracellular replication niche in macrophages (Nagai and Roy, 2003). However, none of these T4SS has been associated with an equally wide range of physiological changes in target cells as we have reported here for the B. henselae VirB system in vascular endothelial cells. Crucial for the molecular understanding of VirB function will be the characterization of effector molecules translocated by this T4SS into infected endothelial cells. Whereas the VirB system shares striking sequence conservation with plasmid conjugal-transfer systems (Schulein and Dehio, 2002), we have recently identified several effector proteins translocated by this T4SS into infected endothelial cells (unpublished data). The knowledge of VirB-translocated effectors and their molecular function in modifying endothelial cell processes related to the cytoskeleton, inflammation, apoptosis, and proliferation would improve our understanding of the infection biology of the vasculotropic pathogen B. henselae. Moreover, these bacterial effectors could provide valuable tools for the specific manipulation of endothelial cells, and may allow to uncover novel aspects in vascular biology.

Experimental procedures

Bacterial strains and growth conditions

The bacterial strains and plasmids used in this study are listed in Table 2. Escherichia coli NovaBlue was used for cloning steps and the dapE. coli strain β2150 for plasmid mobilization to B. henselae (Dehio and Meyer, 1997). Escherichia coli strains were grown at 37°C overnight in Luria–Bertani broth, supplemented with 50 mg l−1 kanamycin, 25 mg l−1 gentamicin, or 1 mM diaminopimelic acid. Bartonella henselae ATCC 49882T (Regnery et al., 1992) and its derivatives were grown for 3–7 days on Columbia agar containing 5% defibrinated sheep blood (CBA agar) in a humidified atmosphere with 5% CO2 at 35°C. The spontaneous streptomycin-resistant mutant RSE247 obtained by selection of B. henselae ATCC 49882T on 100 mg l−1 streptomycin served as wild-type strain. Kanamycin (50 mg l−1) or gentamicin (10 mg l−1) were added to select for transconjugants after mating (Dehio and Meyer, 1997), and 0.5 mM IPTG and 100 mg l−1 streptomycin to select for loss of cointegrates of the mutagenesis vector (Schulein and Dehio, 2002).

Table 2. Bacterial strains and plasmids used in this study.
Strain or plasmidGenotype or relevant characteristicsReference or source
B. henselae strains
ATCC 49882T‘Houston-1’, isolated from a bacteraemic HIV-patient Regnery et al. (1992)
RSE242ΔvirB4 mutant of RSE247This work
RSE247Spontaneous SmR strain of ATCC 49882TThis work
RSE364RSE242 containing pvirB4This work
E. coli strains
β2150 F′ lacZDM15 lacI q traD36 proA + B + thrB1004 pro thi strA hsdS lacZΔM15ΔdapA::erm (ErmR) pir(Dehio and Meyer, 1997)
NovaBlue endA1 hsdR17(r K12–m K12 +) supE44 thi-1 recA1 gyrA96 relA1 lac [F proA + B + lacIqM15::Tn10 (TcR)]Novagen, Madison
pRS14 oriT, oriColE1, gfpmut2, lacIq, rpsL, KmR, mutagenesis vector for generating a ΔvirB4 in-frame deletion in B. tribocorum Schulein and Dehio (2002)
pRS20tra mob+ GmR, expression vector for virB4 of B. tribocorum Schulein and Dehio (2002)
pRS25Derivative of pRS14 used to generate a ΔvirB4 in frame deletion in B. henselaeThis work
pRS100pRS20 containing a 0.4 kb fragment carrying the putative virB promoter of B. henselaeThis work
pvirB4Derivative of pRS100 containing full length virB4 of B. henselaeThis work

Generation of a VirB in frame deletion mutant and complementation in trans

DNA manipulations were carried out by standard procedures (Sambrook et al., 1989). Plasmid pRS25 used for generating a ΔvirB4 mutant was constructed as follows: the BamHI insert of pRS14 (Schulein and Dehio, 2002) was replaced by a 1060 bp BamHI fragment of the virB locus containing a 2103 bp in frame deletion in virB4. This fragment was constructed by megaprime PCR from two PCR products. Product 1 of 0.56 kb was amplified with primers prRS07 and prRS08 and contained the first 123 bp of the virB4 gene and upstream sequences. Product 2 of 0.50 kb was amplified with primers prRS09 and prRS10 and contained the last 129 bp of the virB4 gene and downstream sequences. Megapriming and PCR amplification with primers prRS07 and prRS10 were performed as described (Schulein and Dehio, 2002).

pRS25 was used to generate the ΔvirB4 mutant RSE242 in the RSE247 wild-type background by a two-step gene replacement procedure as recently described for B. tribocorum (Schulein and Dehio, 2002). Polymerase chain reaction with primers prRS285 and prRS286 priming in virB2 and virB5, respectively, was used to confirm the chromosomal deletion of virB4 in RSE242.

The virB4 complementation plasmid pvirB4 was constructed as follows: a 0.40 kb fragment containing the putative virB promoter upstream of virB2 was amplified with primers prRS198 and pRS199. This fragment was digested with ClaI, filled-in, and ligated with the 4.8 kb vector backbone of plasmid pRS20 (Schulein and Dehio, 2002) after digestion with XhoI and ClaI and fill-in, giving pRS100. A 2.39 kb PCR fragment containing the complete virB4 gene was amplified by primers prRS200 and pRS201, digested with XhoI, and ligated with XhoI-digested pRS100, resulting in pvirB4. pvirB4 was conjugated into RSE242, resulting in the trans-complemented mutant strain ΔvirB4/pvirB4 (RSE364). For details of the oligonucleotides used see Table 3.

Table 3. Oligonucleotides used in cloning steps and verification of chromosomal deletion in B. henselae.
NameSequenceaRestriction site
  • a

    . Restriction endonuclease cleavage sites are underlined.


Cell lines and cell culture

HUVEC were isolated as described (Dehio et al., 1997) and cultured in EGM medium (PromoCell, Heidelberg, Germany). The human embryonic kidney cell line HEK293T was cultured in DMEM medium with Glutamax (Gibco, Carlsbad, CA) containing 10% fetal calf serum (FCS, Life Technologies, Rockville, MD). All cells were maintained at 37°C in a humidified atmosphere containing 5% CO2.

Infection assay

Unless otherwise indicated, HUVEC (passage 3–7) were plated at a density of 2–3 × 104 cells/cm2 in EBM medium complemented with the ‘Supplement Pack’ except for gentamicin and ampho B (PromoCell). The next day, cells were washed twice with medium M199 with Earls salts (M199, Gibco) supplemented with 10% FCS. Unless stated differently, cells were infected with a MOI = 100 bacteria per cell (Kirby and Nekorchuk, 2002) in M199/10% FCS and incubated for the indicated time. If specified, 100 nM actinomycin D (Sigma-Aldrich, St Louis, MO) was added to trigger apoptosis as described before (Kirby and Nekorchuk, 2002).

Bacterial growth in HUVEC co-culture and gentamicin protection assay

Human umbilical vein endothelial cells seeded in 24-well plates were infected with a MOI = 100 for the time indicated and immediately centrifuged at 1200 g for 5 min to associate bacteria with the cellular surface. To determine the total number of cultivatable bacteria, samples were directly lysed with a final concentration of 1% saponin (Roth, Karlsruhe, Germany) in PBS, followed by incubation for 15 min at 37°C. Serial dilutions of the cell lysates were plated on CSB-agar and colony forming units (cfu) were determined. To determine intracellular bacteria, an incubation with gentamicin sulphate (125 µg ml−1 in culture medium) for 2 h at 37°C and two washing steps were added before lysis with 1% saponin. To monitor intracellular survival, extracellular bacteria were killed after 30 h of infection by incubation for 2 h with gentamicin sulphate (125 µg ml−1 in M199/10% FCS), followed by two washing steps and further incubation in fresh culture medium.

Immunocytochemistry and quantification of invasome-positive cells

Human umbilical vein endothelial cells were infected for 48 h, fixed, and stained for F-actin, extracellular, and total bacteria as described (Dehio et al., 1997), except that TRITC-phalloidine was used to label F-actin, and Cy2- and Cy5-conjugated goat anti-rabbit antibodies (Dianova, Hamburg, Germany) to label extracellular and total bacteria respectively. A Leica TCS NT confocal laser scanning microscope equipped with an argon/krypton/HeNe mixed gas laser (Leica Lasertechnik, Heidelberg, Germany) was used for confocal microscopy.

To quantify invasome-positive cells, stained specimen were examined under a Leica DM-IRBE inverted fluorescence microscope using a 10 × objective. Fifty cells for each of three random microscopic fields were examined for the presence of invasome structures based on the massive actin-rearrangments and the associated bacterial aggregates and the mean percentage of invasome-positive cells ± SD was calculated.

Determination of IL-8 secretion and ICAM-1 expression

Human umbilical vein endothelial cells seeded in 6-well plates were infected with a MOI = 300 for the indicated time. IL-8 in the culture supernatant was quantified by the human IL-8 DuoSet ELISA kit (R and D Systems, Minneapolis, MN). Adherent cells were harvested by mild trypsinization and resuspended in PBS containing 3% FCS and stained for 1 h at 4°C with FITC-conjugated mouse anti-human ICAM-1 polyclonal antibodies (R and D Systems). Stained cells were analysed with a FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ).

Caspase activity assay

After infection of HUVEC in 24-well plates for 24 h apoptosis was induced for the indicated time. Subsequently, a fluorogenic assay with the caspase-3 and -7 substrate Ac-DEVD-amc (Peptide Institute, Osaka, Japan) was carried out as described (Vercammen et al., 1998). The release of fluorescent 7-amino-4-methylcoumarin (amc) was measured by fluorometry (excitation at 355 nm, and emission at 460 nm).

Annexin V assay

After infection of HUVEC for 24 h in 6-well plates apoptosis was induced for 12 h. Cells were then collected by mild trypsinization and briefly centrifuged together with the culture supernatant. The cell pellet was washed, resuspended, and stained with Annexin V Alexa Fluor 488 (Molecular Probes, Eugene, OR). Propidium iodide (PI, 1 µg ml−1) was added to counterstain necrotic cells and samples were then analysed with a FACSCalibur flow cytometer (Becton Dickinson).

Cell death detection ELISA

After infection of HUVEC for 24 h in 6-well plates apoptosis was induced for 12 h. Then, cells were lysed and the cytoplasmic histone-associated DNA oligonucleosome fragments were quantified by a cell death detection ELISA (Roche Diagnostics, Indianapolis, IN).

Measurement of nuclear factor κB activity

HEK293T cells were seeded at 1 × 104 cells/cm2 and on the next day co-transfected via the calcium phosphate precipitation method with the NF-κB reporter plasmid pconaLuc (LMBP3248 from the LMBP collection,, containing a luciferase gene under control of a consensus binding site of NF-κB, and the control plasmid pEFlacZ (Invitrogen). After 16 h of incubation, the medium was replaced and cells were infected with a MOI = 100. Thirty hours later cells were lysed with NP-40 buffer (Denecker et al., 2001) and NF-κB activity was determined by measuring the luciferase activity present in the cell extracts by a luminescence reader in counts per second (Janssens et al., 2002). Luciferase values were normalized for differences in transfection efficiency on the basis of β-galactosidase activity in the same extracts.

Proliferation assay

Human umbilical vein endothelial cells were plated at 4 × 103 cells/cm2 in 6-well plates and infected with different MOI as indicated. For the following 5 days, the cell number in each well was determined by capturing digital images (MicroMAX camera from Princeton Instruments, Trenton, NJ, with MetaMorph2 software) of three random microscopic fields per well using the 10 × phase-contrast objective of a Leica DM-IRBE inverted microscope. Cell numbers were counted for each image and the mean and standard deviation were calculated for the three microscopic fields taken per well. The proliferation index was calculated by dividing the cell number by the mean of the cell number counted in the same well at the beginning of the experiment (day 0).

Statistical analysis

All experiments were repeated at least three times with triplicate samples. Statistical significance was determined using Student's t-test.


We thank H. L. Saenz and C. J. Thompson for critical reading of this manuscript. We are grateful to N. Devaux for excellent technical assistance. The Bruderholzspital Basel is acknowledged for providing human umbilical cords. This work was supported by a grant from the Swiss National Science Foundation (3100–061777.00/1) to C.D.