|Home | About | Journals | Submit | Contact Us | Français|
Trimeric autotransporter adhesins (TAAs) are important virulence factors of Gram-negative bacteria responsible for adherence to extracellular matrix (ECM) and host cells. Here, we analyzed three different TAAs (Bartonella adhesin A [BadA] of Bartonella henselae, variably expressed outer membrane proteins [Vomps] of Bartonella quintana, and Yersinia adhesin A [YadA] of Yersinia enterocolitica) for mediating bacterial adherence to ECM and endothelial cells. Using static (cell culture vials) and dynamic (capillary flow chambers) experimental settings, adherence of wild-type bacteria and of the respective TAA-negative strains was analyzed. Under static conditions, ECM adherence of B. henselae, B. quintana, and Y. enterocolitica was strongly dependent on the expression of their particular TAAs. YadA of Y. enterocolitica did not mediate bacterial binding to plasma or cellular fibronectin under either static or dynamic conditions. TAA-dependent host cell adherence appeared more significant under dynamic conditions although the total number of bound bacteria was diminished compared to the number under static conditions. Dynamic models expand the methodology to perform bacterial adherence experiments under more realistic, bloodstream-like conditions and allow dissection of the biological role of TAAs in ECM and host cell adherence under static and dynamic conditions.
The first and most decisive step in bacterial infections is the adherence to the host. Trimeric autotransporter adhesins (TAAs) are widely represented in alpha-, beta-, and gammaproteobacteria and play important roles in bacterial pathogenicity (23). TAAs build a characteristic trimeric, lollipop-like surface structure and share a modular organization composed of different domains. These domains, called membrane anchor, stalk, neck, and head, are conserved modules which are present in nearly all TAAs (18, 23) (Fig. 1). The autotransport domain (membrane anchor) defines this TAA family. It secretes a complex structure, the passenger domain, consisting of stalk, neck, and N-terminal head domains. The head, which is a globular structure, represents the most adhesive part of the TAA.
The prototypical TAA is represented by Yersinia adhesin A (YadA) from enteropathogenic Yersinia species (Yersinia enterocolitica and Yersinia pseudotuberculosis), which cause a variety of diseases in humans, ranging from diarrhea to septicemia, mesenteric lymphadenitis, and reactive arthritis (6). YadA (originally designated P1 ) is critical for the colonization of the intestinal mucosa (14). The structure and function of this TAA have been investigated extensively (reviewed in reference 11). Expression of YadA is essential for establishing infections (34, 39). The virulence of Y. enterocolitica is correlated with the ability of YadA to adhere to extracellular matrix (ECM) components (16, 39) and epithelial and polymorphonuclear cells (14, 31) and to mediate autoagglutination (35).
Bartonella adhesin A (BadA) represents a major virulence factor of Bartonella henselae and consists of 3,082 amino acids (aa) with a mass of 328-kDa per monomer (28). The repetitive 22 neck-stalk elements define the enormous length of BadA (~240 nm), whereas in YadA (422 aa; 43 kDa) the presence of a nonrepetitive stalk results in a length of ~23 nm (17). BadA is of major importance for adhesion of B. henselae to endothelial cells (ECs) and to extracellular matrix components. In immunocompetent patients, B. henselae causes cat scratch disease and endocarditis, whereas immunosuppressed individuals can suffer from vasculoproliferative disorders such as bacillary angiomatosis (2). BadA has been identified as the key factor involved in the induction of an angiogenic host cell response (28). Bartonella quintana, which causes, e.g., trench fever (also called 5-day fever, a chronic and relapsing intraerythrocytic bacteremia), bacillary angiomatosis, and endocarditis, contains four different adhesins (variably expressed outer membrane proteins VompA to VompD) which are similar to BadA in their domain composition but of minor length (950 to 1,014 aa). Apparently, only VompA, -B, and -C, but not VompD, are expressed on conventionally cultured B. quintana. Vomps are responsible for autoagglutination and binding to ECM components (41) but are not involved in bacterial cell adhesion to epithelial HeLa-229 and phagocytic THP-1 cells (32). Expression of Vomps by B. quintana is essential to cause chronic bacteremia in macaques (24).
The aim of the study was to elucidate whether the three TAAs, BadA of B. henselae, Vomps of B. quintana, and YadA of Y. enterocolitica, share similar biological functions in the process of bacterial adherence to ECM components and to endothelial cells compared under identical experimental conditions. Moreover, we wanted to know whether dynamic flow infection models (more similar to in vivo situations) deliver comparable results to static infection models, which are widely used in pathogenicity research. Using these two approaches, we observed that BadA, Vomps, and YadA all act as promiscuous binding partners to ECM components and endothelial cells, with differences occurring under static and dynamic conditions. Under static experimental conditions, specific YadA-mediated bacterial binding to endothelial cells was not detectable but was uncovered in dynamic flow models.
B. henselae and B. quintana were grown for 4 days in Schneider's medium supplemented with 10% fetal calf serum (FCS) and 5% sucrose at 37°C in a humidified atmosphere with 5% CO2 (29). Stock suspensions were prepared by freezing the bacteria in Luria-Bertani (LB) broth containing 20% glycerol at −80°C. The number of viable bacteria per aliquot was determined by counting the CFU in serial dilutions from the frozen stocks, which were cultivated on Columbia blood agar (CBA) plates for 14 days. Y. enterocolitica strains were grown in LB broth overnight at 27°C, diluted to an optical density at 600 nm (OD600) of 0.2, and grown at 37°C for a further 3 h to induce YadA expression. Antibiotics were used at the following concentrations: for B. henselae BadA−, 30 μg/ml kanamycin; for Y. enterocolitica WA-314(pYV+), 10 μg/ml nalidixic acid; for Y. enterocolitica WA(pYVO8-A-0), 10 μg/ml nalidixic acid and 50 μg/ml kanamycin. Strains that were used in this study are summarized in Table 1.
Expression of TAAs (B. henselae BadA, B. quintana Vomps, and Y. enterocolitica YadA) was assessed by immunofluorescence using specific antibodies. For this purpose, bacteria were resuspended in phosphate-buffered saline (PBS) and air dried on glass slides. After fixation with 3.75% PBS-buffered paraformaldehyde (PFA) for 10 min at 4°C, bacteria were incubated with the respective primary antibodies (rabbit anti-BadA IgG , rabbit anti-B. henselae IgG targeting Vomps of B. quintana , and rabbit anti-YadA IgG ) and with Cy2-conjugated secondary antibodies (anti-rabbit IgG; Dianova, Hamburg, Germany) for 1 h each at room temperature. Finally, bacterial DNA was stained with 4′,6′-diamidino-2-phenylindole (DAPI) for 10 min. After each staining procedure, bacteria were washed extensively three times with PBS. Finally, slides were mounted with Fluoprep medium (bioMérieux, Nürtingen, Germany) and analyzed with a Leica DMRE fluorescent microscope equipped with a Spot RT monochrome digital camera and the related Spot advanced software (Visitron, Puchheim, Germany).
For ultrastructural studies, bacterial colonies of B. henselae (BadA+ and BadA−) and B. quintana (Vomp+ and Vomp−) were grown for 4 days on CBA, fixed with 2.5% glutaraldehyde in PBS for 20 min at room temperature, and kept for 20 h at 4°C. Cells were covered with 2% agarose, and blocks containing single colonies were cut out. After postfixation with 1% osmium tetroxide in 100 mM phosphate buffer, pH 7.2, for 1 h on ice, these blocks were rinsed with aqua bidest, treated with 1% aqueous uranyl acetate for 1 h at 4°C, dehydrated through a graded series of ethanol, and embedded in Epon. Sections were stained with 1% aqueous uranyl acetate and lead citrate and analyzed in a Philips CM10 electron microscope at 60 kV using a 30-μm objective aperture.
For static adherence assays, coverslips were coated with 1 ml of ECM components diluted in PBS at 10 μg/ml for 24 h at room temperature (Table 2 lists the ECM components). Homogenous coating of coverslips was proven by plasma fibronectin-, laminin-, and collagen III-specific immunofluorescent labeling (primary antibodies were mouse-anti plasma fibronectin [BD-Transduction Laboratories, Heidelberg, Germany],rabbit-anti-laminin, and rabbit anti-collagen III [both Abcam, Cambridge, United Kingdom]; secondary antibodies were Cy2-labeled anti-mouse IgG and DyLight 488-labeled anti-rabbit IgG [Dianova]) and confocal laser scanning microscopy (CLSM). For the respective negative controls, specific primary antibodies were omitted from the staining procedure (see Fig. S1 in the supplemental material). The coating solution was removed, and 1.0 × 107 bacteria (resuspended in 1 ml of PBS) were added to the coated coverslips by centrifugation (at 300 × g for 5 min). After an incubation period of 30 min (humidified atmosphere, 37°C, and 5% CO2), coverslips were washed three times with PBS. Bound bacteria were fixed with 3.75% PFA, stained with DAPI, and analyzed as described above.
For adherence assays under dynamic flow conditions, multichannel slides (μ-Slide VI0.4 flow kit; Ibidi, Martinsried, Germany) were coated with 10 μg/ml ECM components for 24 h at room temperature. Bacteria were resuspended in M199 medium (Biochrom AG, Berlin, Germany) at a concentration of 1.0 × 108 bacteria per ml. Bacterial suspensions were pumped for 15 min at room temperature through the channels with a shear stress (τ) of 1.0 dyne/cm2 using syringes (Braun, Melsungen, Germany) and perfusor pumps (MTS, Schweinfurt, Germany). After a washing step with PBS, bound bacteria were fixed with 3.75% PFA, stained with DAPI, and analyzed as described above. Per experiment, bacterial adherence of all six strains used in this study was tested in parallel (Fig. 2).
For live imaging of BadA-mediated adherence of B. henselae to collagen III under flow conditions, time-lapse microscopy was performed. First, B. henselae BadA+ and BadA− cells were fluorescently stained by carboxyfluorescein N-hydroxysuccinimidyl ester (diluted 1:200 in 100 mM NaHCO3) for 30 min on ice as described previously (36). For quenching, bacteria were resuspended in 50 mM Tris-HCl for 5 min on ice. After each incubation step, bacteria were centrifuged for 5 min at 1,300 × g and washed three times in PBS. Time-lapse microscopy was performed at a shear stress of 0.25 dyne/cm2 using a Zeiss Axiovert 200 microscope equipped with an Axiocam MRm digital camera (Carl Zeiss AG, Jena, Germany) and a Ludl filter wheel set controlled by the MAC 5000 system (Ludl Electronic Products Ltd., Hawthorne, NY). Movies were processed with the Axiovision, version 4, software (Carl Zeiss).
Human umbilical vein endothelial cells (HUVECs) were cultured in endothelial cell growth medium (PromoCell, Mannheim, Germany), as described previously (20). For static adherence assays, 1.0 × 105 cells were seeded onto collagen G-coated coverslips, grown overnight without antibiotics, and infected with 1.0 × 107 of the respective bacterial strains by centrifugation (multiplicity of infection [MOI] of 100; 300 × g for 5 min). Cells were incubated for 30 min in a humidified atmosphere at 37°C and 5% CO2. After samples were washed gently with prewarmed cell culture medium, the infection was stopped by the addition of 3.75% PFA.
For dynamic adherence assays, HUVECs (50 μl of a suspension of 1.0 × 105 cells) were seeded overnight in growth medium without antibiotics on collagen G-coated multichannel slides (Ibidi). Bacteria (MOI of 1,000) were resuspended in 1 ml of EC growth medium and pumped with a shear stress (τ) of 0.125 dyne/cm2 for 15 min at 37°C through the HUVEC-coated channels. After samples were washed gently with prewarmed cell culture medium for 5 min, the infection was stopped by the addition of 3.75% PFA.
Bacterial adherence was quantified using CLSM as described previously (20). Cells were permeabilized by incubation with 0.1% Triton X-100 (Sigma, Steinheim, Germany) in PBS for 15 min. The actin cytoskeleton was stained with tetramethyl rhodamine isothiocyanate (TRITC)-labeled phalloidin (Sigma) for 1 h, and DNA was stained with DAPI for 10 min. Finally, slides were mounted with Fluoprep medium. Cellular fluorescence was evaluated using a Leica DM IRE 2 confocal laser scanning microscope. Two different fluorochromes were detected simultaneously representing red (TRITC) and blue (DAPI) channels. Images were digitally processed with Photoshop, version 7.0 (Adobe Systems, Mountain View, CA). Bacterial adherence was quantified by counting cell-adherent bacteria on 100 endothelial cells per bacterial strain.
All experiments were performed at least three times and revealed comparable results. Differences between mean values of experimental and control groups were analyzed by a Student's t test. A value of P <0.01 was considered to be statistically significant.
First, we wanted to prove whether the strains used in this study express the particular TAAs (BadA, Vomps, and YadA). For this purpose, immunofluorescence analysis was performed for B. henselae BadA+ and BadA− cells, B. quintana Vomp+ and Vomp− cells, and Y. enterocolitica YadA+ and YadA− cells. Results revealed bright green fluorescence signals located on the surfaces of the respective wild-type strains, but these signals were absent on the respective mutant strains (Fig. 3). These results were consistent with those from TEM analysis for B. henselae BadA+ and BadA− strains and B. quintana Vomp+ and Vomp− strains (Fig. 4). Here, BadA appeared as a long and hairy structure of enormous length (~250 nm) originating from the outer membrane. As suspected earlier, electron-dense globular structures at the top of the adhesin layer potentially represent the head domain. Vomps, which were formerly not detectable at a satisfying resolution, form a dense fringe on the surface of B. quintana cells and are significantly shorter (~40 nm) than BadA proteins from B. henselae. Again, the Vomps span from the outer membrane to an electron-dense globular structure at the top, probably representing the head domain (Fig. 4, arrows). Even with different electron microscopy protocols, the expression of YadA of Y. enterocolitica could not be demonstrated by TEM. However, Western blotting confirmed YadA expression in wild-type Y. enterocolitica strain whereas expression of this protein was absent in the respective Y. enterocolitica YadA− strain (data not shown).
So far, all experiments elucidating the binding of B. henselae, B. quintana, or Y. enterocolitica to ECM components have been performed under static conditions (1, 16, 19, 28, 41). To compare TAA-dependent binding affinities of the bacteria and to set an internal standard for the following experiments, the adhesion of the bacterial strains used here to plastic (polystyrene), to collagen I, III, and IV, to laminin, to plasma fibronectin, to cellular fibronectin, to hyaluronic acid, and to vitronectin (Table 2) was analyzed statically using immobilized ECM compounds. Bacteria were centrifuged on ECM-coated coverslips, incubated for 30 min, stained with DAPI, and analyzed by fluorescence microscopy. All TAA-expressing bacteria (including autoagglutinated bacteria) adhered considerably more to all ECM components tested than their corresponding TAA-deficient mutant strains (Fig. 5). Remarkably, plasma and cellular fibronectin were the only substrates which did not support adhesion of YadA+ and YadA− Y. enterocolitica strains. This indicates that YadA is not involved in fibronectin binding, confirming earlier findings (16). In contrast, BadA and Vomps mediate fibronectin binding of B. henselae and B. quintana. Moreover, all analyzed TAAs seem to support a promiscuous rather than a specific attachment of bacteria to ECM components under static conditions.
Static conditions are the in vitro default setting in pathogenesis research and cellular microbiology although, in vivo, the interaction of bacteria with ECM components occurs only rarely under such settings. However, studying the attachment of bacteria under dynamic flow conditions more closely mimics the infection process as bacterial binding and therefore the infection process are influenced by continuous shear stress (25, 26). Dynamic infection assays have not previously been employed for either Bartonella spp. or Yersinia spp. We established a flow chamber system which allowed the analysis of bacterial binding under dynamic conditions (Fig. 2 and Materials and Methods). Immobilized ECM components and plastic (cyclic olefin copolymer, similar in its features to polyethylene) were exposed to a constant flow of a solution containing a defined amount of the bacterial strains used here at a shear stress of 1.0 dyne/cm2. After 15 min of flow, bound bacteria were fixed, stained with DAPI, and analyzed by fluorescence microscopy. Generally, the number of attached bacteria was lower in flow chamber settings than in static experiments. The number of autoagglutinated bacteria (visible as clumps by immunofluorescence) was also significantly diminished under dynamic flow conditions. Again, YadA did not play an evident role in fibronectin binding of Y. enterocolitica. Binding of B. henselae to ECM components was BadA dependent even under dynamic conditions; however, a preferential binding to collagen III, laminin, cellular and plasma fibronectin, hyaluronate, and vitronectin became evident (Fig. 6; see also the time-lapse microscopy of B. henselae BadA+ and BadA− strains on collagen III in the supplemental video material). Compared to binding to hyaluronate and vitronectin, binding of B. quintana to collagen I, III, and IV, to laminin, and to cellular and plasma fibronectin was strongly diminished. These differences suggest a flow-dependent preference of Vomp-mediated binding of B. quintana to hyaluronate and vitronectin. For Y. enterocolitica, YadA mediated preferentially collagen I binding.
Endothelial cells represent a further major docking site for systemic infections, and Bartonella spp. are endotheliotropic pathogens (for a review, see reference 7). A crucial role of B. henselae BadA in the infection process of endothelial cells was demonstrated (28), whereas this was less obvious for the Vomps of B. quintana (32). Interaction of Y. enterocolitica with endothelial cells has also been described previously (3, 8, 13); however, the particular role of YadA in this process has not been elucidated.
We assayed the TAA-dependent adhesion of B. henselae, B. quintana, and Y. enterocolitica to endothelial cells (HUVECs) in static and dynamic infection models. The MOI of 100 used in static infection assays was increased in the dynamic flow assays to 1,000, which takes into account the decreased bacterial attachment under shear stress. Furthermore, the dynamic infection model was adapted to flow conditions and shear rates of 0.125 dyne/cm2, which prevail in small capillary vessels. Samples were fixed, endothelial cells were stained by TRITC-phalloidin, and bacterial DNA and cellular nuclei were stained by DAPI. Quantification of adherent bacteria was performed by using CLSM and counting cell-adherent bacteria for 100 cells. The static infection assays revealed significant differences between B. henselae BadA+ and BadA− strains or B. quintana Vomp+ and Vomp− strains. In contrast, YadA expression failed to influence significantly endothelial cell adherence of Y. enterocolitica under static conditions (Fig. 7).
Using the more physiological flow chamber system, results confirmed the static data obtained for B. henselae and B. quintana. However, the number of adherent bacteria was lower under dynamic conditions. Particularly, the numbers of nonspecifically adhering TAA-negative bacteria were reduced by the fluid flow. Therefore, shear stress increased the stringency and, hence, specificity of the TAA-expressing pathogens in endothelial cell attachment. Despite the reduced numbers of bacteria, the ratio of adherent BadA+ to BadA− B. henselae cells increased from 8.4- to 13.4-fold, and the ratios of adherent Vomp+ to Vomp− B. quintana cells increased from 4.3- to 46.1-fold from static to dynamic conditions, respectively. YadA-mediated endothelial cell adherence became evident only under dynamic flow conditions. Here, the attachment of YadA+ Y. enterocolitica to endothelial cells was 10.5-fold greater than that of YadA− bacteria (Fig. 8), whereas under static conditions, this ratio remained almost unchanged (1.3-fold). Thus, only the dynamic infection model revealed YadA-mediated attachment of Y. enterocolitica to endothelial cells.
The biological role of TAAs in specific binding to ECM components and host cells is of continual interest in pathogenicity research of many Gram-negative bacteria (23). This is mainly due to the fact that adherence to matrix and host cells is the first and crucial step in the infection process of the host. Unfortunately, results obtained with different bacteria (e.g., Bartonella spp. and Yersinia spp.) are not directly comparable as different authors use different experimental approaches and settings. This leads to difficulties in interpreting data of, e.g., TAA-specific adhesion. Experiments measuring the binding of Bartonella spp. and Yersinia spp. to ECM components have been performed previously under static conditions (1, 16, 19, 28, 41). Our studies employed both classical static and newly established dynamic flow infection models. When we compared these two approaches, we uncovered previously unrecognized differences in preferential binding specificities of the three TAAs BadA, Vomps, and YadA to ECM components and endothelial cells, which are essential docking sites in local or systemic infections.
Most infection processes occur under mass transport conditions, which influence the rate of microbial adhesion to surfaces significantly (5). In vivo, these conditions are most prevalent in blood vessels, heart valves, or the gastrointestinal tract; here, bacterial adherence to the ECM or host cells is mandatory for causing infections, and this process strongly depends on the particular rheological situation. It is clear that the process of bacterial binding, e.g., via TAAs or other adhesins, must be crucially influenced by shear stress. This idea is starting to attract more attention in infectious disease research (e.g., in infections with Neisseria meningitidis ). Dynamic models should enable the analysis of bacterial adherence under more realistic conditions than provided by static experiments. The range of physiological in vivo shear stress rates spans from 2 dyne/cm2 (venules) to ~6 dyne/cm2 (arterioles) but can decrease even down to 0.004 dyne/cm2 in capillary networks (27). We used shear stress rates of 1.0 dyne/cm2 for ECM and 0.125 dyne/cm2 for endothelial cell infection models, which are within this range. Shear stress higher than 0.125 dyne/cm2 resulted in the detachment of endothelial cells from the channel slides.
In fact, under flow conditions it turned out that bacterial aggregates do not appear to bind to ECM-coated surfaces (in contrast to binding in static settings) (Fig. 5 and and6).6). The most reasonable explanation for this phenomenon could be that bacterial aggregates are dragged physiologically by flow velocity. Furthermore, dynamic flow conditions revealed differences in ECM binding which were not evident in static experiments; in particular, Vomp-dependent adherence of B. quintana to collagen I, III, and IV, to laminin, and to fibronectins was highly diminished, and this result is in line with earlier observations detecting no fibronectin binding by B. quintana (32). Similarly, a significant YadA-mediated adherence of Y. enterocolitica was observed only for collagen I under flow conditions (Fig. 6), as was suggested earlier (16). Only BadA of B. henselae seems to mediate adherence to all ECM components tested here under both static and dynamic conditions. These results suggest that an increasing number of the neck-stalk repeats present in BadA (22 elements), Vomps (5 elements), and YadA (1 element) might be a reason for the higher promiscuity in ECM binding of longer adhesins (BadA, 250 nm; Vomps, 40 nm [Fig. 4]; YadA, 23 nm ). Such a hypothesis fits an earlier report that expression of a truncated B. henselae BadA lacking 21 neck-stalk repeats resulted in a fibronectin-binding-deficient phenotype (19). It needs to be mentioned that the head domain of BadA has a more complex architecture (consisting of a YadA-like head and additional GIN and Trp-ring subdomains ) than VompA, -B, and -C of B. quintana or YadA of Y. enterocolitica (Fig. 1). These subdomains might have an additional and enhanced effect on the ECM-binding capability of BadA.
In our experiments, the missing YadA-mediated fibronectin binding of Y. enterocolitica was evident both under static and dynamic conditions. These findings are in accordance with data published by Heise and Dersch demonstrating that YadA of Y. enterocolitica does not bind fibronectin. The molecular reason for this phenomenon could be explained by the presence of a 30-aa fibronectin binding element (position 53 to 83) in YadA of Yersinia pseudotuberculosis that is missing in YadA of Y. enterocolitica (16). However, it should be mentioned that controversial data on YadA-dependent fibronectin binding exist (33, 40) and that a specific collagen binding site in the YadA head was not identified (22). It has even been shown that Y. enterocolitica YadA binds promiscuously to collagen II and III fragments, and these experiments did not reveal specific binding motifs within collagens (21). From the ligand spectrum of the bacterial adhesion assays, no common structural motifs in the ECM molecules can be deduced as they represent different structural groups, such as collagens and laminins. Moreover, they belong to proteins and to sugar-derived glycosaminoglycan molecules, such as hyaluronate. Some of the ligands are found in the interstitial connective tissue, such as the fibrillar collagens, whereas laminins are typically found in basement membranes, which are borderline structures of the connective tissue and separate different tissue compartments (10). All these observations suggest that domains of TAAs might mediate promiscuous and unspecific binding, resulting in a broad variety of potential binding partners.
YadA-dependent binding of Y. enterocolitica to endothelial cells was detectable only when flow conditions were present (Fig. 7 and and8).8). This is a surprising result and indicates that specific bacterial binding phenotypes which are masked under static infection conditions are uncovered in dynamic flow settings. A potential explanation for this observation might be that static adherence assays are prone to suffer from unspecific and low-affinity binding mechanisms in contrast to dynamic flow assays, where specific and stronger host cell binding becomes obvious. Thus, shear rates and mechanical forces increase the stringency of the binding interactions between the bacterial adhesion and their ligands of the ECM or on endothelial cells. For B. henselae it has been demonstrated that the head domain of BadA is crucially involved in endothelial cell adherence (19), whereas the role of the Vomp head domains in cell adherence of B. quintana is less clear. When B. quintana was exposed to HeLa-229 cells and THP-1 macrophages, cell adherence did not differ between Vomp+ and Vomp− strains (32). However, in the current setting, a clear effect of Vomp expression on endothelial cell adherence was detected both under static and dynamic conditions. It can be assumed that the binding of B. quintana to endothelial cells is also mediated by the head domains of VompA, -B, and -C although direct experimental evidence for this hypothesis is still lacking. Additionally, we observed the Vomps of B. quintana for the first time at a satisfying resolution by electron microscopy, revealing the presence of a globular electron-dense structure at the top of the Vomps, most likely representing the head domain (Fig. 4). The reason why B. quintana exhibits different adhesion phenotypes to endothelial and epithelial cells or macrophages remains unclear and might be explained by a specific binding partner expressed on endothelial cells. However, fibronectin bridging of bacteria to α5β1 integrins, which has been demonstrated to contribute to BadA-dependent endothelial cell binding of B. henselae (28), has not been demonstrated for B. quintana yet.
Taking these results together, shear stress is a key determinant in defining whether bacteria bind to matrix and vasculature. Our data demonstrate the use of dynamic flow infection models in infections with B. henselae, B. quintana, and Y. enterocolitica. Our experiments deciphered TAA-dependent mechanisms underlying bacterial adherence to ECM components and endothelial cells. Apparently, static infection experiments tend to mask certain specific adherence phenotypes, which can be uncovered by employing dynamic infection models. Therefore, dynamic models seem to represent a valuable and necessary extension to the widely used static models in infection research.
We thank Albert Haas (Bonn, Germany) for providing fluorescent labeling protocols of bacteria and Kerstin Laib-Sampaio (Tübingen, Germany) for expert help in time-lapse microscopy.
V.A.J.K. was partially supported by a grant from the Deutsche Forschungsgemeinschaft (DFG-SFB 766). J.A.E. is supported by the DFG through the Excellence Cluster, Cardio-Pulmonary System (EXC147/1).
†Supplemental material for this article may be found at http://iai.asm.org/.
Published ahead of print on 2 May 2011.