The Bartonella henselae VirB/Bep system interferes with vascular endothelial growth factor (VEGF) signalling in human vascular endothelial cells

Authors


E-mail christoph.dehio@unibas.ch; Tel. (+41) 61 267 2140; Fax (+41) 61 267 2118.

Summary

The vasculotropic pathogen Bartonella henselae (Bh) intimately interacts with human endothelial cells (ECs) and subverts multiple cellular functions. Here we report that Bh specifically interferes with vascular endothelial growth factor (VEGF) signalling in ECs. Bh infection abrogated VEGF-induced proliferation and wound closure of EC monolayers as well as the capillary-like sprouting of EC spheroids. On the molecular level, Bh infection did not alter VEGF receptor 2 (VEGFR2) expression or cell surface localization, but impeded VEGF-stimulated phosphorylation of VEGFR2 at tyrosine1175. Consistently, we observed that Bh infection diminished downstream events of the tyrosine1175-dependent VEGFR2-signalling pathway leading to EC proliferation, i.e. phospholipase-Cγ activation, cytosolic calcium fluxes and mitogen-activated protein kinase ERK1/2 phosphorylation. Pervanadate treatment neutralized the inhibitory activity of Bh on VEGF signalling, suggesting that Bh infection may activate a phosphatase that alleviates VEGFR2 phosphorylation. Inhibition of VEGFR2 signalling by Bh infection was strictly dependent on a functional VirB type IV secretion system and thereby translocated Bep effector proteins. The data presented in this study underscore the role of the VirB/Bep system as important factor controlling EC proliferation in response to Bh infection; not only as previously reported by counter-acting an intrinsic bacterial mitogenic stimulus, but also by restricting the exogenous angiogenic stimulation by Bh-induced VEGF.

Introduction

Bartonella henselae (Bh) is a worldwide distributed, Gram-negative, zoonotic pathogen. In the feline reservoir host, this cat flea-borne pathogen rarely shows symptoms, while incidental transmission of Bh to humans by cat scratch or bite can result in a number of clinical outcomes (Dehio, 2008). Clinical manifestations range from long-lasting but self-limiting swelling of lymph nodes (cat scratch disease, CSD) in immunocompetent patients to tumour-like vasoproliferative lesions on the skin or in the inner organs known as bacillary angiomatosis (BA) or bacillary peliosis (BP) in patients with a compromised immune system (Florin et al., 2008). The close association of Bh with proliferating endothelial cells (ECs) in BA/BP (Chian et al., 2002) and the fact that antibiotic treatment leads to regression of those lesions (Koehler and Tappero, 1993) indicate an active role of Bh in triggering these vascular proliferations. Bh-triggered vasculoproliferative lesions resemble tumour angiogenesis – the pathological process of the formation of new capillaries out of pre-existing blood vessels. Foremost among the several different growth factors and their associated receptor tyrosine kinases (RTKs), the vascular endothelial growth factor (VEGF) family and VEGF receptors (VEGFRs) are essential regulators of angiogenesis under physiological as well as pathological conditions (Takahashi and Shibuya, 2005).

The secretion of VEGF from Bh-infected macrophages (Resto-Ruiz et al., 2002), recruited to the site of infection as a result of a pro-inflammatory response in ECs (Fuhrmann et al., 2001; Schmid et al., 2004), or from infected epithelial cells surrounding the blood vessel (Kempf et al., 2001), is proposed to contribute to BA/BP by promoting EC proliferation in a paracrine manner (Kempf et al., 2001; Resto-Ruiz et al., 2002). The secretion of VEGF from macrophages or epithelial cells was shown to be triggered by the trimeric autotransporter adhesin BadA (Bartonella adhesin A) via activation of hypoxia-inducible factor (HIF)-1 (Riess et al., 2004). Bh also demonstrates apotent direct mitogenic stimulation of human umbilical vein endothelial cells (HUVECs) (Conley et al., 1994). Furthermore, the inhibition of HUVEC apoptosis upon Bh infection might play indirectly a role in vasoproliferation by increasing cell survival (Kirby and Nekorchuk, 2002). Massive cytoskeletal actin rearrangements leading to a unique structure termed invasome (Dehio et al., 1997) is an additional striking change described for ECs following infection with Bh in vitro.

Invasome formation, anti-apoptosis and partly the NFκB-dependent pro-inflammatory activation of HUVECs depend on the VirB type IV secretion system (T4SS) of Bh and the thereby translocated seven Bartonella effector proteins (BepA to BepG) (Schmid et al., 2004; Schulein et al., 2005). Intriguingly, the potent VirB-independent direct stimulation of HUVEC proliferation is counterbalanced by the VirB/Bep system (Schmid et al., 2004). Moreover, a recently established three-dimensional spheroid sprouting assay for HUVECs (Korff and Augustin, 1999) allowed the assessment of the complex angiogenic properties of Bh in an elaborate in vitro model and revealed distinct but also adverse activities of the Beps in modulating capillary-like sprout formation (Scheidegger et al., 2009). Observations during the establishment of the spheroid-based in vitro model taking the proposed paracrine loop into account prompted us to study the effect of exogenous VEGF on Bh-infected ECs. Therefore we addressed in this study whether Bh and in particular its VirB/Bep system affects VEGF-stimulated angiogenic activity and intracellular signalling in HUVECs. Surprisingly, sprouting could not be induced by VEGF in spheroids formed from HUVECs pre-infected with Bh wild type. Moreover, VEGF-stimulated proliferation as well as migration was affected in HUVECs infected with Bh wild type. Immunoblot analyses demonstrate that the VirB/Bep system of Bh altered VEGF downstream signalling leading to the activation of the mitogen-activated protein kinase ERK1/2 – the major mediator of VEGF-triggered EC proliferation (Takahashi et al., 2001) – by interfering with phosphorylation of tyrosine1175 of VEGF receptor 2 (VEGFR2). The presented data emphasize the role of the VirB/Bep system as an important factor balancing the angiogenic potential of Bh– showing both pro- and anti-angiogenic characteristics – in the course of chronic infection of the human vasculature.

Results

Bartonella henselae infection interferes with VEGF-induced spheroid sprouting, proliferation and wound closure

Interference of VEGF activity on ECs by Bh infection was noticed first in the recently established three-dimensional spheroid-based in vitro model of Bartonella-triggered angiogenesis (Scheidegger et al., 2009). Spheroids from uninfected HUVECs or cells infected for 24 h with a multiplicity of infection (moi) of 300 with Bh wild type or the isogenic effector-less mutant ΔbepA–G were embedded in collagen and subsequently either stimulated with VEGF (25 ng ml−1) or left unstimulated. After 24 h the spheroids were imaged in bright field (Fig. 1A) and angiogenesis was quantified by calculating the cumulative sprout length (CSL) of individual spheroids (Fig. 1B). To characterize the effect of VEGF we calculated a VEGF sprout factor by dividing CSL of spheroids stimulated with VEGF by CSL of spheroids without VEGF (Fig. 1C). As previously reported (Scheidegger et al., 2009) HUVEC spheroids pre-infected with Bh wild type displayed an increase in CSL in comparison with uninfected control spheroids. Compared with wild type, the sprouting activity was reduced in spheroids of HUVECs pre-infected with the ΔbepA–G mutant (Fig. 1A and B). Surprisingly, spheroids made from HUVECs pre-infected with Bh wild type were not responsive to VEGF as CSL with and without VEGF were comparable resulting in a VEGF sprout factor of approximately one. In contrast, sprouting could be readily induced about 5.5-fold, respectively, 2.5-fold by VEGF for spheroids of uninfected HUVECs or cells pre-infected with the ΔbepA–G mutant (Fig. 1C).

Figure 1.

Infection with Bh wild type, but not the ΔbepA–G mutant, interferes with VEGF-induced spheroid sprouting, proliferation and wound closure.
A. Spheroids made from uninfected HUVECs or cells infected with the indicated Bh strains (moi = 300) for 24 h were embedded in collagen and either stimulated with VEGF or left untreated for an additional 24 h. Following fixation bright-field images of the embedded HUVEC spheroids were taken and representative images are shown.
B. Quantification of in vitro angiogenesis of collagen-embedded spheroids after 24 h by calculating the cumulative length (CSL) of sprouts originating from an individual spheroid. Shown are means ± SD of 10 spheroids per experimental group.
C. VEGF sprout factor calculated by dividing CSL stimulated by VEGF by unstimulated CSL. Shown are means ± SD of the indicated number of experiments.
D. Cell numbers of HUVECs grown as monolayers determined after 4 days of infection (moi = 50) in the presence or absence of VEGF. Differences between VEGF-stimulated and unstimulated samples were analysed by Student's t-test. P < 0.05 marked with an asterisk was considered statistically significant.
E. Representative bright-field images of HUVEC monolayers, uninfected or pre-infected (moi = 150) with the indicated strains for 24 h, followed by scratching with a pipette tip to wound the monolayers and stimulation with VEGF, and fixation after the indicated number of days before bright-field microscopy with phase contrast.

To further assess the effect of exogenous VEGF on ECs infected with Bh we used in addition two different classical in vitro assays. To determine the impact of Bh infection on VEGF-stimulated proliferation cell numbers were compared after 4 days in the presence or absence of VEGF (25 ng ml−1). HUVECs were seeded in 96-well plates and left uninfected or cells were infected with the indicated Bh strains at the moi of 50. Following fixation, cell numbers were determined by staining nuclei with DAPI, followed by automated image acquisition and cell counting using the CellProfiler software (Carpenter et al., 2006) (Fig. 1D). As reported previously the ΔbepA–G mutant strongly promoted cell proliferation and this VirB/Bep-independent stimulus was only slightly affected in Bh wild-type infection at the low moi used in this assay (Schmid et al., 2004). Addition of VEGF stimulated cell proliferation about fourfold for uninfected HUVECs. Proliferation of HUVECs infected with the ΔbepA–G mutant could be further increased 1.5-fold by exogenous VEGF. In sharp contrast, VEGF stimulation had no significant effect on HUVECs infected with wild-type bacteria.

A scratch wound assay was used to assess the effect of infection and exogenous VEGF on cell migration. To this end, HUVECs were seeded in six-well plates and infected with the indicated strains at an moi of 150 for 24 h. The confluent cell layers were wounded by scratching with a pipette tip, stimulated with VEGF (25 ng ml−1) and analysed by phase-contrast microscopy (Fig. 1E). Stimulation of uninfected HUVECs with VEGF led to complete closure of the wound within 2 days. Similarly, HUVECs pre-infected with the ΔbepA–G mutant and stimulated with VEGF readily moved to fill the damaged area. However, pre-infection with Bh wild type led to a prominent interference with migration, which could not be rescued by exogenous VEGF.

Taken together, pre-infection of HUVECs with Bh wild type abrogated sprouting, proliferation and wound closure induced by VEGF.

The VirB/Bep system blocks calcium flux upon stimulation with VEGF

Endothelial cell stimulation by VEGF leads to Ca2+ mobilization from intracellular stores (Brock et al., 1991). To further elucidate the nature of Bh interferences with VEGF signalling, HUVECs infected for 24 h with different moi of Bh wild type were loaded with the fluorescent Ca2+ indicator Fluo-4. Baseline fluorescence was measured before cells were stimulated with VEGF (25 ng ml−1). Application of VEGF to HUVECs induced a rapid increase in cytosolic calcium concentration ([Ca2+]i), followed by a sustained plateau phase slightly above baseline (Fig. 2A, control). With increasing moi the initial calcium peak was diminished to baseline level. At the moi of 150 the calcium peak was cut by half and the moi of 300 reduced the peak by approximately 70% (Fig. 2A). Treatment with the ionophore ionomycin (0.5 µM, Fig. 2A and B) or the intracellular calcium pump inhibitor thapsigargin (data not shown) showed that calcium stores are not emptied upon infection. ECs infected with a high moi (≥ 600) showed in general reduced [Ca2+]i levels, possibly due to cells suffering from bacterial overgrowth.

Figure 2.

Multiplicity of infection (moi)- and VirB/Bep-dependent interference with calcium flux upon stimulation with VEGF. Uninfected HUVECs or cells infected with the indicated Bh strains (moi = 300), if not indicated differently, for 24 h were loaded with the calcium indicator Fluo-4. Baseline fluorescence was measured for 2 min, then cells were stimulated with VEGF, thrombin or ATP as indicated and fluorescence was recorded for 7 min before Ionomycin was added if indicated.
A. Bh wild type interfered with calcium flux initiated by VEGF in an moi-dependent manner.
B. Interference by Bh infection depends on a functional VirB/Bep system.
C. GPCR (Thrombin or ATP)-triggered calcium flux was not influenced by Bh infection. Shown are Ca2+-flux plots of representative experiments (n ≥ 3).

Endothelial cells infected either with an moi of 300 of the ΔvirB4 mutant deficient for translocation via the VirB T4SS or the ΔbepA–G mutant behaved as uninfected ECs (Fig. 2B). Furthermore, heat-killed Bh wild type did not affect VEGF-induced increase in [Ca2+]i (data not shown). Thus, the moi-dependent inhibition of the VEGF-triggered calcium flux requires viable bacteria with a functional VirB T4SS and at least one if not more of the Bep effectors.

VEGF-induced calcium flux relies on the activation of phospholipase-Cγ (PLCγ). PLCγ hydrolyses the membrane phospholipid phophatidylinositol (4,5)-bisphosphate (PIP2) to generate the second messengers diacylglycerol (DAG) and inositol 1,4,5-triphosphate (IP3). IP3 generation further results in an increase of [Ca2+]i through opening of IP3-operated calcium-release channels (IP3R) of the endoplasmatic reticulum (ER), whereas DAG is the activator of protein kinase C (PKC) (Holmes et al., 2007). Yet, the accumulation of IP3 and the subsequent increase of [Ca2+]i is not only a consequence of RTK-activated PLCγ but also result from the activation of PLCβ by G protein-coupled receptors (GPCR) (Clapham, 2007). Bh wild type specifically blocked neither thrombin (1 U ml−1)-induced nor ATP (100 µM)-induced calcium flux (Fig. 2C). Experiments with basic fibroblast factor (bFGF) failed to induce any increase of [Ca2+]i in HUVECs (data not shown). Hence, these results suggest that Bh rather selectively affects the calcium flux in response to PLCγ activation coupled to RTK.

The VirB/Bep system affects VEGFR2 signalling upstream and downstream of calcium release

To determine whether Bh wild-type infection affects signalling events upstream and downstream of calcium release we assessed the phosphorylation and therefore activation of PLCγ and ERK1/2 respectively. To this end, HUVECs were infected with the indicated strains at an moi of 300 for 24 h. Subsequently, cells were stimulated with VEGF (25 ng ml−1) for 1–15 min. Total-cell lysates were harvested and separated by SDS-PAGE. VEGF treatment induced a transient phosphorylation of PLCγ in uninfected HUVECs, with a maximum response seen after 5 min. Despite showing similar PLCγ protein levels, PLCγ phosphorylation was strongly reduced in HUVECs infected with wild type, while infections with the mutants ΔvirB4 or ΔbepA–G behaved as uninfected cells (Fig. 3, panels 1 and 2; Fig. S1, panels 1 and 2). Consistently, phosphorylation of the ERK1/2 kinase in response to VEGF stimulation was specifically suppressed by infection with wild type compared with infections with the ΔvirB4 and ΔbepA–G mutants, as well as in uninfected cells (Fig. 3, panels 3 and 4; Fig. S1, panels 3 and 4). Under the same conditions, VEGF also induced the MAPK p38 after 15 min in uninfected HUVECs (Fig. 3, panels 5 and 6). The phosphorylation of p38 MAPK was generally increased with infection, but for wild type we did not observe any induction upon stimulation with VEGF. Hence, consistent with an abrogated calcium flux, infection of HUVECs with Bh wild type showed a pronounced interference with the activation of the VEGF-stimulated PLCγ-ERK1/2 signalling cascade.

Figure 3.

VEGFR2 downstream signalling is abrogated by infection with Bh wild type, but not the ΔbepA–G mutant. HUVECs were infected with the indicated strains (moi = 300) for 24 h or left uninfected before stimulation with VEGF for the indicated time. Total-cell extracts were separated by SDS-PAGE and immunoblots were prepared with the indicated antibodies.

The VirB/Bep system abrogates VEGFR2 signalling without altering receptor expression or cell surface localization

VEGF-A165, which was used in this study to stimulate HUVECs, binds to and stimulates autophosphorylation of the two related RTKs VEGFR1 (also FLT1) and VEGFR2 (also KDR). Inactivation of a single vegf-a allele in mice results in early lethality at embryonic day E11-12 due to deficient EC development and lack of vessels (Carmeliet et al., 1996). vegfr-2−/− animals die at E8.5 with a phenotype similar to that of vegf-a−/− animals (Shalaby et al., 1995). In contrast, vegfr-1−/− embryos die at E8.5-9 because of overgrowth and disorganization of blood vessels and not as a result of poor vascularization (Fong et al., 1999). Furthermore, disruption of the VEGFR1 tyrosine-kinase domain is compatible with normal vascular development (Hiratsuka et al., 1998). Even though binding VEGF with a 10-fold higher affinity than VEGR2, VEGFR1 signal transduction is extremely weak (Waltenberger et al., 1994). Whereas VEGFR1 may have a negative role, either acting as decoy receptor or by suppressing VEGFR2 signalling (Shibuya, 2006), VEGFR2 is regarded as a main transducer of VEGF-A165 in vasculogenesis and angiogenesis (Olsson et al., 2006). The aforementioned results showed that infection with Bh wild type interfered with VEGF-induced biological function and signalling. Hence, we examined whether VEGFR2 phosphorylation and therefore receptor activation was affected by infection. Total-cell lysates of uninfected HUVECs or cells infected for 24 h with Bh wild type or the ΔbepA–G mutant at moi of 300 and subsequently stimulated with VEGF (25 ng ml−1) for 1–15 min were separated by SDS-PAGE and analysed by Western blotting. Tyr1175 in the C-terminal tail of VEGFR2 was previously shown to represent the docking sites for downstream signalling via PLCγ and to have a critical role in the regulation of angiogenesis (Sakurai et al., 2005; Holmes et al., 2007). Consistent with the observed decrease in PLCγ activation, Bh interfered in a VirB/Bep-dependent manner with phosphorylation on this position (Fig. 4A, panel 2). Densitometric analysis of immunoblots after 5 min VEGF stimulation confirmed a significant reduction of Tyr1175 phosphorylation in HUVECs infected with Bh wild type in comparison with uninfected cells or cells infected with the ΔbepA–G mutant (Fig. 4B). Phosphorylation of another tyrosine, Tyr951, has been linked to migration (Zeng et al., 2001). Infection with Bh wild type also reduced phosphorylation of this site (Fig. 4A, panel 1). Thus, infection of HUVECs with Bh wild type interfered with phosphorylation of two tyrosine residues of VEGFR2 important for angiogenic signalling.

Figure 4.

VirB/Bep system-dependent diminution of VEGF-stimulated VEGFR2 phosphorylation.
A. HUVECs were infected (moi = 300) with the indicated strains or left uninfected for 24 h before stimulation with VEGF for the indicated time. Total-cell extracts were separated by SDS-PAGE and immunoblots were prepared with the indicated antibodies.
B. Densitometric analysis of immunoblots for phospho-Tyr1175 after 5 min VEGF stimulation. For each experimental condition the values were normalized to unstimulated control. Shown are means ± SD of nine independent experiments. The asterisk marks samples that differ statistically significantly (Student's t-test P < 0.05) from VEGF-stimulated samples infected with Bh wild type.

A decrease in VEGFR2 activation could be caused by a factor activated by Bh wild-type infection that prevents phosphorylation or promotes dephosphorylation of the receptor, or as well by changes in receptor expression or localization. To determine whether reduced Tyr1175 phosphorylation was due to downregulation of receptor expression we assessed VEGFR2 mRNA (kdr) expression in uninfected HUVECs or cells infected for 24 h with an moi of 300 with either Bh wild type or the ΔbepA–G mutant (Fig. 5A). As previously observed Bh wild type triggered the activation of the cAMP-dependent CREM/CREB pathway (crem) in a VirB/Bep-dependent manner (Schmid et al., 2006), but Bh wild-type infection had no influence on VEGFR2 mRNA levels in comparison with uninfected control ECs. The ΔbepA–G mutant though slightly increased expression of VEGFR2 mRNA. VEGFR2 protein expression was investigated by FACS analysis 24 h post infection and revealed no differences in intracellular receptor levels (Fig. 5B). Similar results were obtained with densitometric analysis of immunoblots for total VEGFR2 (Fig. S2). Furthermore, infection with Bh wild type did not specifically affect surface expression of VEGFR2 as shown by FACS analysis using an antibody directed against the extracellular domain of the receptor (Fig. 5C). Consistent with unchanged receptor levels, differences in the ratio of phospho-Tyr1175 to total VEGFR2 (Fig. 6A) indicate that infection with Bh wild type affected VEGF signalling not through decreased receptor expression or surface localization but via additional factors influencing the phosphorylation state of VEGFR2.

Figure 5.

The VirB/Bep system does not affect VEGFR2 expression nor surface localization. Expression of VEGFR2 on mRNA and protein level was assessed by real-time PCR and FACS, respectively, in uninfected HUVECs or HUVECs infected with the indicated strains (moi = 300) for 24 h.
A. Real-time PCR analysis of VEGFR2 (kdr) and Crem (crem) mRNA after infection for 24 h. Shown are means ± SD of three independent experiments.
B and C. FACS analysis of intracellular and surface VEGFR2 of HUVECS incubated with secondary antibody (dashed line) or stained with both primary and secondary antibodies (solid line). Shown are means ± SD of relative fluorescence intensities of four experiments and histograms of a corresponding representative experiment.
Samples marked with an asterisk (P < 0.05) differ statistically significantly from uninfected control samples using a Student's t-test.

Figure 6.

Inhibition of Tyr1175 phosphorylation by infection with Bh wild type is blocked by pervanadate treatment.
A. Densitometric analysis of phospho-Tyr1175 to total VEGFR2 ratios after 5 min VEGF stimulation of HUVECs pre-infected for 24 h (moi = 300) with the indicated Bh strains of left uninfected. For each experimental condition the values were normalized to unstimulated control. Shown are means ± SD of five independent experiments. The asterisk marks samples that differ statistically significantly (Student's t-test P < 0.05) from the corresponding unstimulated control sample.
B. HUVECs were infected with the indicated strains or left uninfected for 24 h. Total-cell extracts were separated by SDS-PAGE and immunoblots were probed with an anti-phosphotyrosine antibody. Infection with Bh wild type resulted in the loss of some prominent bands (see arrowheads) otherwise present in uninfected cell or cells infected with the ΔbepA–G mutant indicating the activation of a PTP.
C. HUVECs were left uninfected or infected for 24 h with Bh wild type or the ΔbepA–G mutant (moi = 300) before being pre-treated with pervanadate for 15 min prior to stimulation with VEGF for 5 min. Total-cell extracts were separated by SDS-PAGE and immunoblots were prepared with the indicated antibodies.

Inhibition of Tyr1175 phosphorylation by Bh infection is sensitive to pervanadate treatment

Reversible tyrosine phosphorylation is a central regulatory mechanism in cell signalling, which is mediated by numerous protein-tyrosine kinases (PTKs) and counteracting protein-tyrosine phosphatases (PTPs) (Hunter, 1995). During infection of HUVECs we observed that total-cell lysates of uninfected cells and cells infected with the ΔbepA–G mutant, respectively, displayed a comparable band pattern when probed with an anti-phosphotyrosine antibody. Yet, cell lysates of ECs infected with Bh wild type demonstrated a pronounced reduction in tyrosine phosphorylation of proteins of different molecular weights indicating the induction of a PTP with a pleiotropic effect on tyrosine phosphorylation (see arrowheads in Fig. 6B). Since infection with Bh wild type resulted in a reduced phosphorylation of at least two tyrosine residues of VEGFR2 the induction of a PTP could account for the impaired signalling by VEGFR2 in this condition.

To test the possible involvement of a PTP we used pervanadate, a tyrosine phosphatase inhibitor. HUVECs were left uninfected or infected for 24 h with the indicated strains, followed by pre-treatment with pervanadate (100 µM) for 15 min prior to stimulation with VEGF (25 ng ml−1) for 5 min. Total-cell lysates were separated by SDS-PAGE and analysed by immunoblotting. Pervanadate treatment increased the basal level of Tyr1175 phosphorylation in HUVECs and impaired the ability of Bh wild type to block VEGF-induced phosphorylation of this site (Fig. 6C). Hence, treatment of ECs with a tyrosine phosphatase inhibitor abrogated the capability of Bh wild type to interfere with VEGFR2 Tyr1175 phosphorylation, indicating the activity of a PTP induced upon infection.

Discussion

Both VirB/Bep-dependent and VirB/Bep-independent factors are thought to play a role in the remarkable capacity of Bh to trigger vasoproliferative processes resembling tumour angiogenesis, i.e. BA and BP (Schmid et al., 2004; Dehio, 2005; 2008; Pulliainen and Dehio, 2008). In addition to a direct stimulation of proliferation and inhibition of apoptosis in ECs the activation of a paracrine loop of pro-angiogenic factors such as VEGF, an essential regulator of physiological as well as pathological angiogenic processes (Takahashi and Shibuya, 2005), is thought to be involved in Bartonella-triggered vascular tumour formation (Resto-Ruiz et al., 2002; Riess et al., 2004; Dehio, 2005). In a previous study we used a spheroid-based in vitro angiogenesis model to address the angiogenic potential of Bh and its VirB/Bep system and demonstrated both promoting and also inhibiting actions of the Beps on capillary-like sprout formation (Scheidegger et al., 2009). To extend this model and especially to take into account the proposed paracrine loop we addressed in this study the effect of exogenous VEGF on HUVECs infected with Bh.

Surprisingly, upon Bh infection at elevated moi (≥ 150) the VirB/Bep system interfered with EC responsiveness to the potent EC-specific growth factor VEGF. Neither sprouting, nor proliferation, nor migration could be induced by VEGF if HUVECs were pre-infected with Bh wild type. Bh wild type inhibited phosphorylation of Tyr1175, one of the critical docking sites for proteins mediating VEGFR2 downstream signalling (Holmes et al., 2007), e.g. for PLCγ. PLCγ feeds into the signalling cascade leading to the activation of the MAPK ERK1/2, involved in VEGF-stimulated EC proliferation (Takahashi et al., 2001). Phosphorylation of PLCγ upon VEGF stimulation was inhibited by Bh in a VirB/Bep-dependent manner. Consequently, we observed also a lack of downstream calcium release as well as induction of ERK1/2 phosphorylation. The importance of PLCγ for VEGF function is highlighted by the fact that PLCγ1-deficient zebrafish embryos fail to respond to exogenous VEGF and that overall the mutant phenotype of these embryos resembles the ones of mutants lacking VEGF function (Lawson et al., 2003).

Interestingly, the extracellular adherence protein (Eap), a broad-spectrum adhesin secreted by the Gram-positive bacterium Staphylococcus aureus, was shown to cause a significant impairment of VEGF- and bFGF-induced ERK1/2 phosphorylation while not affecting AKT and p38 MAPK activation (Sobke et al., 2006). Furthermore, Eap also blocked the activation of the Raf/MEK/ERK cascade on exposure to the phorbol ester TPA which indicates interference downstream of PKC and actually independent of receptor activation. ERK1/2 is activated by the MAPK kinase MEK1/2 which itself is placed downstream of Raf-1. A key event in the activation of Raf-1 is its binding to RasGTP (Dhillon and Kolch, 2002). Sobke et al. observed a profound reduction in activated, GTP-bound Ras after VEGF, as well as bFGF stimulation in Eap-exposed HUVEC and propose that interference with Ras activation is the earliest target of Eap blockage of MAPK pathway induction. Next to the classical Ras/Raf/MEK/ERK cascade downstream of VEGFR2, VEGF also induces a Ras-independent pathway through PLCγ-activated PKC and Raf-1 (Takahashi et al., 1999). The relative contributions made by Ras-dependent versus Ras-independent pathways during VEGF-triggered MAPK signalling seem to be heterogeneous and vary depending on the origin of ECs. Concerning HUVECs, expression of a dominant negative Ras mutant had little effect on ERK1/2 activation, emphasizing the importance of the Ras-independent pathway in this type of EC (Yashima et al., 2001). In contrast to what has been described for Eap, the interference with VEGF signalling we have observed for Bh wild type seems to act on the level of receptor activation and to affect the Ras-independent pathway of ERK1/2 induction. Yet Bh and S. aureus share the ability to balance the EC angiogenic response preventing a possibly overshooting reaction of the host.

As mentioned above the earliest target of Bh interference with VEGF-signalling seems to be the VEGFR2 itself. Experiments performed with stimuli other than VEGF, such as the RTK bFGF or ATP or thrombin-triggering GPCR-coupled calcium flux, indicate that Bh interference is rather specific for VEGFR2. Furthermore, we found that infection did not alter VEGFR2 expression nor localization and therefore a decrease in receptor levels could not account for reduced VEGFR2 phosphorylation. However, probing total-cell extracts with an anti-phosphotyrosine antibody revealed a pleiotropic effect on the phosphorylation pattern of cells infected with Bh wild type, with loss of bands of high- as well as low-molecular-weight proteins. Furthermore, pervanadate treatment prior to VEGF stimulation abrogated the capacity of Bh wild type to block Tyr1175 phosphorylation indicating the activity of a PTP. Yet, the direct involvement of a PTP responsible for the observed interference with responsiveness of infected ECs to VEGF remains to be demonstrated.

The observation that in HUVECs infected with the Bh effector-free mutant ΔbepA–G VEGFR2 signalling was not affected indicates a role for Beps in the observed interference of Bh wild type with EC responsiveness towards VEGF. However, Ca2+ flux assays performed using infections with the ΔbepA–G mutant trans-complemented with single bep genes revealed that none of the Beps individually was able to interfere with the VEGF-stimulated increase of [Ca2+]i (F. Scheidegger and C. Dehio, unpubl. obs.), suggesting that at least two different Bep effector have to act in concert to inhibit VEGF signalling.

The data presented here on the VirB/Bep-dependent inhibition of VEGF signalling by Bh infection extend the previously reported VEGF-driven paracrine loop model of a Bh-triggered pro-angiogenic response (Kempf et al., 2001; Riess et al., 2004) in that Bh can control the triggered pro-angiogenic response of ECs in an moi-dependent manner. Interestingly, the Bh strain ‘Marseille’ (Drancourt et al., 1996) used to established the paracrine loop model expresses the non-fimbrial adhesin BadA as a crucial factor for the induction of VEGF secretion by infected epithelial cells (Riess et al., 2004) but does not express a functional VirB/Bep system (M. Truttmann, M. Faustmann and C. Dehio, unpubl. data), while the spontaneous streptomycin-resistant variant of the Houston-1 typing strain ATCC 49882T of Bh (Schmid et al., 2004) used in this study as wild type encodes a functional VirB/Bep system but due to a frameshift mutation in the parental clinical isolate (Regnery et al., 1992) is deficient for the BadA adhesion (Riess et al., 2004). Both the large and strongly expressed BadA adhesion (Riess et al., 2007) and the inducible VirB/Bep system composed of at least 19 proteins (Quebatte et al., 2010) are ‘costly’ virulence factors that are dispensable for bacterial growth in vitro and apparently are difficult to be both functionally maintained upon multiple passages in vitro (M. Truttmann, et al., unpubl. data). To further study the impact of the BadA-triggered and VEGF-driven paracrine pro-angiogenic loop and the control of the potentially overshooting angiogenic response of ECs by the here described VirB/Bep-mediated desensitization towards VEGF for Bh-triggered vasoproliferation will require further experimental work, ideally involving a Bh strain expressing both virulence factors in conjunction with appropriate cellular and animal infection models.

In summary, we demonstrated an intriguing effect of the VirB/Bep system, antagonizing a major signalling cascade of VEGFR2 following its activation by VEGF. Observations performed on the level of phosphorylation to assess activation of signalling proteins were reflected in assays such as proliferation, migration and sprout formation assessing the biological function of VEGF. Together with the fact that the VirB/Bep system masks a potent VirB-independent mitogenic activity (Schmid et al., 2004) and that bacterial effector BepG potently interferes with sprouting (Scheidegger et al., 2009), this strong interference of the VirB/Bep system with the signalling of one of the most important regulator of angiogenesis speaks against the VirB/Bep system as major pro-angiogenic factor in Bh pathogenesis. Along this line, the human-specific Bb strongly promotes vascular proliferation (Garcia et al., 1990) even though it has no VirB T4SS (Saenz et al., 2007). Thus, the VirB/Bep system more likely is a true host adaptation system, a modulator required for the regulation of the angiogenic activity of Bh during infection – antagonizing potent mitogenic stimuli but inducing minor perturbations in cell proliferation and cell death that facilitate chronic vascular infection.

Experimental procedures

Reagents

VEGF-A165 was purchased from ReliaTech (http://www.reliatech.de). Antibiotics, ATP and carboxymethylcellulose (methocel, 4000 centipoises) were from Sigma (http://www.sigmaaldrich.com). Collagen type I was isolated from rat tail tendons (Augustin, 2004). Thrombin and Ionomycin were purchased from Calbiochem (http://www.emdbiosciences.com). The mouse anti-phosphotyrosine antibody (clone 4G10) and mouse anti-actin antibody (clone C4) were purchased from Millipore (http://www.millipore.com). Rabbit anti-VEGFR2 (clone 55B11), rabbit anti-phospho VEGFR2 (Tyr1175, clone 19A10), mouse anti-p44/42 (ERK1/2) MAP Kinase, mouse anti-phospho p44/42 (Thr202/Tyr204, clone E10), rabbit anti-phospho p38 (Thr180/Tyr182, clone 12F8), rabbit anti-PLCγ1 and rabbit anti-phospho PLCγ1 (Tyr783) antibodies were purchased from Cell Signaling Technology (http://www.cellsignal.com). The antibody directed against the extracellular domain of VEGFR2 was from abcam (http://www.abcam.com). Secondary horseradish peroxidase (HRP)-conjugated antibodies were obtained from Amersham (http://www.amersham.com). Secondary Alexa633-conjugated antibody was from Molecular Probes (http://www.invitrogen.com).

Bacterial strains and growth conditions

The bacterial strains used in this study are listed in Table 1. Bartonella spp. were grown on Columbia agar (CBA) plates containing 5% defibrinated sheep blood in a humified atmosphere at 35°C and 5% CO2 for 2–3 days (Dehio et al., 1997). Strain RSE247, a spontaneous streptomycin-resistant strain of ATCC 49882T (Schmid et al., 2004), served as wild type in this study. Media were supplemented with 30 µg ml−1 kanamycin, 100 µg ml−1 streptomycin or 12.5 µg ml−1 gentamicin if needed.

Table 1.  Bacterial strains used in this study.
StrainGenotype or relevant characteristicsReference or source
B. henselae strains  
 RSE247Spontaneous Smr strain of ATCC 49882T, serving as wild typeSchmid et al. (2004)
 RSE242ΔvirB4 mutant of RSE247Schmid et al. (2004)
 MSE150ΔbepA–G mutant of RSE247Schulein et al. (2005)

Cell culture

HUVECs were isolated as described before (Dehio et al., 1997) and cultured in EGM medium (http://www.promocell.com) in a humified atmosphere at 37°C and 5% CO2.

Infection assay

HUVECs (passage 4–7) were plated in gelatine-coated well plates using EGM. The next day cells were washed twice with M199 with Earls salts (M199, Gibco, http://www.invitrogen.com) supplemented with 10% FCS and infected with the indicated moi of bacteria per cell in M199/10% FCS and incubated for 24 h.

Generation of EC spheroids

Endothelial cell spheroids of defined cell number were generated as described previously (Korff and Augustin, 1998; Scheidegger et al., 2009) with some modifications. In brief, uninfected, respectively, infected EC monolayers were trypsinized and suspended in EGM containing 20% methocel (Weber et al., 2008). Spheroids were generated by pipetting 400 ECs (in 25 µl) on quadratic Greiner Petri dishes (http://www.huberlaborworld.ch) and grown overnight as hanging drops in a humified atmosphere at 35°C and 5% CO2. Under these conditions, all suspended cells contribute to the formation of a single spheroid of defined cell number per drop.

Spheroid-based in vitro angiogenesis assay

The in vitro angiogenesis assay was performed as described (Weber et al., 2008; Scheidegger et al., 2009). The generated spheroids were harvested and suspended in the methocel solution containing 20% FCS. Subsequently, the ice-cold collagen solution (rat tail type I collagen in 0.1% acidic acid) was neutralized by adding 10% of 10-fold Medium 199 and approximately 10% 0.2 N NaOH to adjust the pH to 7.4. EC spheroid/methocel solution was mixed 1:1 with the neutralized collagen solution and 1 ml of the mixed solution containing approximately 50 EC spheroids was pipetted into individual wells of a pre-warmed 24-well plate to allow polymerization in the incubator in a humified atmosphere at 35°C, 5% CO2. After 30 min VEGF was added to an end-concentration of 25 ng ml−1 by pipetting 100 µl of a 10-fold concentrated working dilution on top of the polymerized gel. Plates were incubated in a humified atmosphere at 35°C, 5% CO2 for 24 h and fixed by adding 3.7% paraformaldehyde. Pictures were taken with a Leica DM IRBE inverted microscope using a MicroMAX camera (Princeton Instruments) using the 20× objective. In vitro angiogenesis was digitally quantified by measuring the cumulative length of the sprouts (CSL) that had grown out of each spheroid using the MetaMorph software analysing 10 spheroids per experimental group and experiment.

Proliferation assay

One thousand HUVECs were seeded in 96-well plates and infected with an moi of 50 in the presence or absence of VEGF (25 ng ml−1). After 4 days cells were fixed with 3.7% paraformaldehyde and exposed to DAPI (1:1000, 1 mg ml−1 stock). Plates were imaged with an automated fluorescence microscope (IXM, MDC Molecular Devices) and assessed quantitatively by using CellProfiler (Carpenter et al., 2006). Results are given as average cell number per image, analysing 10 images per well.

Scratch wound assay

A total of 1.2 × 105 HUVECs were seeded in six-well plates and infected with an moi of 150. Small areas of the confluent monolayers were disrupted by scratching three straight lines with a p200 pipette tip. Debris were removed by washing the cells once, stimulated with VEGF (25 ng ml−1) and followed by microscopy with a 10× objective. To obtain the same fields during the image acquisition we used a Leica DM-IRBE microscope with automated staging.

Ca2+ mobilization measurements

Five thousand HUVECs were seeded per well of a 96-well plate and unless stated differently infected with an moi of 300 in M199/10% FCS for 24 h. Medium was removed and cells were loaded with the fluorescent Ca2+ indicator Fluo-4 NW (http://www.invitrogen.com) for 45 min in the incubator in the dark. Following the manufacturer's protocol baseline fluorescence was measured for 2 min before cells were stimulated with VEGF (25 ng ml−1) or Thrombin (1 U ml−1), respectively, ATP (100 µM) and fluorescence was recorded for another 7 min before Ionomycin (0.5 µM) was added if indicated. Measurements were performed with a Synergy 2 Multi-Mode Microplate Reader (http://www.biotek.com).

Western blotting

A total of 6 × 105 HUVECs were seeded in 10 cm dishes and infected with an moi of 300 with the indicated strains for 24 h and if stated stimulated with 25 ng ml−1 VEGF for the indicated time. Proteins from total-cell lysates were separated by SDS-PAGE, transferred onto nitrocellulose membranes (Hybond-C, http://www.amersham.com) and examined using the indicated antibodies. The secondary HRP-conjugated antibody was visualized by enhanced chemiluminescence (http://www.perkinelmer.com). If indicated membranes were treated with strip buffer (0.2 M Tris-HCl pH 6.8, 2% w/v SDS, 0.7% v/v mercaptoethanol) and reprobed.

Real-time PCR

Total cellular RNA was isolated at 24 h after infection as described above. RNA manipulation and real-time PCR were performed as previously described (Dehio, 2005). Primers for gapdh were GAAGGTGAAGGTCGGAGTC (prMQ005) and GAAGATGGTGATGGGATTTC (prMQ006). Primers for crem were ATCGCCCGGAAGTTTGC (prMQ013) and CAGCTCTCGTTTGCGTGTTG (prMQ014). Primers for human kdr (VEGFR2) were CCAGACGGACAGTGGTATGGTT (prMQ082) and CACCATTCCACCAAAAGATGG (prMQ083).

FACS analysis

A total of 6 × 105 HUVECs seeded in 10 cm dishes and infected with an moi of 300 for 24 h were released from tissue culture plates by trypsination and collected by centrifugation. Intracellular staining of VEGFR2 was performed following the manufacturer's instructions (Cell Signaling Technology). For surface staining of VEGFR2 the cells were incubated with a 1:200 dilution of an antibody directed against the extracellular domain of the VEGFR2 (abcam) for 30 min at 4°C. Following a washing step with PBS containing 1% FCS the cells were incubated for another 30 min at 4°C with a 1:200 dilution of an anti-rabbit Alexa633-conjugated secondary antibody. The cells were washed, resuspended in PBS 1% FCS and analysed by FACS. FACS analysis was performed with a BD FACSCalibur flow cytometer. Relative fluorescence intensity was calculated by dividing the median fluorescence intensity of stained samples by the median fluorescence intensity of the secondary antibody signal.

Acknowledgements

We are grateful to Dr R. Jayachandran and P. Müller for technical assistance. We thank A. Pulliainen for critically reading of the manuscript and acknowledge the donation of human umbilical cords from the University Women's Hospital Basel. This work was supported by a PhD fellowship from the Misrock Foundation, Grant 31003A-109925 from the Swiss National Science Foundation, Grant 55005501 from the Howard Hughes Medical Institute and Grant 51RT-0-126008 (InfectX) from SystemsX.ch, the Swiss Initiative for Systems Biology.

Ancillary