Factors influencing the cell adhesion and invasion capacity of Mycoplasma gallisepticum

Background The cell invasiveness of Mycoplasma gallisepticum, the causative agent of respiratory disease in chickens and infectious sinusitis in turkeys, may be a substantial factor in the well-known chronicity of these diseases and in the systemic spread of infection. To date, not much is known about the host factors and mechanisms involved in promotion or obstruction of M. gallisepticum adherence and/or cell invasion. In the current study, the influence of extracellular matrix (ECM) proteins such as fibronectin, collagen type IV and heparin, as well as plasminogen/plasmin, on the adhesion and cell invasion levels of M. gallisepticum to chicken erythrocytes and HeLa cells was investigated in vitro. Two strains, Rhigh and Rlow, which differ in their adhesion and invasion capacity, were analyzed by applying a modified gentamicin invasion assay. Binding of selected ECM molecules to M. gallisepticum was proven by Western blot analysis. Results Collagen type IV, fibronectin, and plasminogen exerted positive effects on adhesion and cell invasion of M. gallisepticum, with varying degrees, depending on the strain used. Especially strain Rhigh, with its highly reduced cell adhesion and invasion capabilities seemed to profit from the addition of plasminogen. Western and dot blot analyses showed that Rhigh as well as Rlow are able to adsorb horse fibronectin and plasminogen present in the growth medium. Depletion of HeLa cell membranes from cholesterol resulted in increased adhesion, but decreased cell invasion. Conclusion ECM molecules seem to play a supportive role in the adhesion/cell invasion process of M. gallisepticum. Cholesterol depletion known to affect lipid rafts on the host cell surface had contrary effects on cell adherence and cell invasion of M. gallisepticum.


Background
For a long period of time, Mycoplasma spp., very small, wall-less prokaryotes, were considered obligate extracellular bacteria until in 1989 Lo et al. described intracellular organisms in an AIDS-patient, which were later identified as M. fermentans [1]. To date, other Mycoplasma spp., such as M. penetrans [2], M. genitalium [3], M. pneumoniae [4], M. suis [5], M. synoviae [6], M. bovis [7], M. hyorhinis [8], and M. hominis [9] have likewise been shown to be capable of invading non-phagocytic eukaryotic cells. Mycoplasma gallisepticum, the causative agent of chronic respiratory disease in chickens and infectious sinusitis in turkeys has been shown to be able to invade HeLa cells and chicken embryo fibroblasts (CEF) in vitro [10] and chicken red blood cells (RBC) in vitro and in vivo [11]. Furthermore, M. gallisepticum has been proven to cross the mucosal barrier and to spread systemically in vivo [12]. However, even closely related M. gallisepticum strains can differ markedly in their pathogenicity for chickens [13,14]. The strains R low and R high derive from different passages of strain R grown in artificial medium [14,15]. The lowpassage, hemadsorption-positive and virulent strain R low (10th passage) was shown to be cell-invasive in vitro and in vivo, whereas the high-passage, hemadsorptionnegative strain R high (164th passage) displays a highly reduced virulence and exhibits only marginal cell invasiveness [10][11][12]16].
Many publications over the past years addressed the bacterial proteins involved in mycoplasma cell adhesion and invasion ( [17][18][19] for reviews). However, host factors involved in cell invasion have been investigated to a much lesser extent. The extracellular matrix (ECM) occupies the space between animal cells and is composed of secreted proteoglycans and non-proteoglycan polysaccharides or proteins like collagens, fibronectin, and laminin [20,21]. These macromolecules influence adhesion, migration, proliferation, and differentiation of eukaryotic cells [22], and they serve also as substrate for the attachment and internalization of pathogenic microorganisms [23,24]. ECM-and plasminogen-binding properties have also been reported for a limited number of Mycoplasma species like M. penetrans [25], M. hyopneumoniae [26,27], M. bovis [28] and M. gallisepticum [29][30][31]. So far, only for M. fermentans and M. pneumoniae an influence of ECM molecules and plasminogen on adhesion and invasion capabilities has been documented [32][33][34]. For M. gallisepticum strains R low and R high , published data is limited to differential binding properties of the organism to fibronectin [29], heparin [30] and plasminogen [31].
Another host factor that plays a role in bacterial invasion processes is cholesterol, the major component of lipid rafts [35]. Cholesterol seems to play a major role in the invasion process of M. fermentans, as invasion rates were 70% lower in cholesterol-depleted HeLa cells, whereas adhesion rates were not influenced [32].
In the current study, our aim was to investigate hemadsorption-positive and -negative strains of M. gallisepticum for their capability to adhere to and invade HeLa cells and chicken red blood cells in the presence of selected ECM molecules and plasminogen. The role of cholesterol availability on the host cell membrane for the adhesion and invasion of M. gallisepticum was also examined.
Adherence and invasion rates of M. gallisepticum were analyzed together in a modified GIA as described elsewhere [10,11]. Briefly, for the GIA using RBC, an overnight culture of M. gallisepticum was centrifuged at 10,000 × g for 10 min, resuspended in DMEM and diluted to reach a final concentration of 4 to 8 × 10 4 CFU per ml. One ml of this suspension was mixed with 20 μl of a stock solution of the respective ECM molecules. A 250-μl aliquot of this mixture was incubated with 250 μl DMEM containing gentamicin (final concentration 400 μg/ml) for 3 h at 37°C to verify the lethal concentration of gentamicin. The residual 750 μl of the mixture was co-incubated for 2 h at 37°C with 750 μl RBC in DMEM containing 1.5 × 10 8 RBC. A 500-μl aliquot of this mixture was then diluted 1:10 in PBS and 100-μl samples were plated on HFLX agar plates to count the number of input bacteria. The remaining ECM-mycoplasma-RBC mixture was divided into 2 samples, and centrifuged at 300 × g for 3 min. The resulting pellets (two per ECM molecule) were washed once with PBS to remove non-adherent bacteria, and resuspended in 500 μl DMEM each. One resuspended pellet was further diluted 1:10 in PBS and plated on HFLX agar plates to determine the total number of adherent and intracellular bacteria. The other resuspended pellet was treated with 500 μl DMEM containing gentamicin (final concentration 400 μg/ml) and incubated for 3 h at 37°C. After the gentamicin treatment, 100 μl-samples were plated to count the intracellular bacteria.
The GIA using HeLa cells was carried out in 24-well cell culture plates. If HeLa cell membranes had to be depleted from cholesterol, the cells were preincubated in the presence of 5 mM MβCD for 30 min at 37°C and 5% CO 2 as described elsewhere [37]. After thoroughly washing to remove any MβCD, HeLa cells were co-incubated with 500 μl-samples of ECM-treated bacteria (4 to 8 × 10 4 CFU per ml DMEM) for 2 h at 37°C in a 5% CO 2 atmosphere (2 wells per ECM molecule). After 2 h of incubation, the HeLa cells were washed with PBS to remove non-adherent mycoplasma, trypsinized for 10 min (0.05% trypsin-EDTA; Gibco BRL) and the cells of one duplicate well were treated with 1 ml DMEM containing gentamicin (400 μg/ml) to kill all extracellular mycoplasma. After 3 h of incubation, the number of intracellular mycoplasma was determined by plating 100 μl of serial dilutions in PBS onto HFLX agar plates. The other duplicate well was used to determine the number of adherent mycoplasma and treated the same way except that no gentamicin was added. The number of input bacteria was determined from serial dilutions of mycoplasma samples before ECM treatment. All experiments were arranged in duplicate assays and repeated two (R high ), or three (R low ) times, respectively.

Western blots
Lysates of whole cell protein were produced from 1-ml samples of M. gallisepticum grown in HFLX or E. coli grown in Luria-Bertani (LB) medium according standard methods. For S. pneumoniae, biomass was collected directly from Columbia III agar plates with 5% sheep blood (Becton Dickinson, Heidelberg, Germany). Denatured whole cell proteins were separated by SDS-gel electrophoresis and transferred to nitrocellulose membranes (Whatman, Dassel, Germany). The blots were blocked with 2% blocking reagent (Blotting Grade Blocker, Bio-Rad) and incubated with either anti-fibronectin or antiplasminogen antibodies (ABIN125555 or ABIN285631; antibodies-online GmbH, Aachen, Germany), diluted 1:1,000 in TBS. Horseradish peroxidase-labeled goat anti-rabbit IgG (P0217; DAKO Hamburg, Germany), diluted 1:2,000, was used as the secondary antibody, and the blots were developed using 4-chloro-1-naphthol (4CN) according to the manufacturer's instructions (Bio-Rad).

Dot blots
One-ml samples of overnight cultures of R high , R low , E. coli, S. pneumoniae and S. canis were washed in equal amounts of PBS and five μl were dropped onto nitrocellulose and allowed to dry. Blocking and further processing was performed as described for Western blots.

Statistical analysis
Invasion and adhesion frequencies are expressed as the mean ± standard deviation of n independent values. Statistical analysis of the data was performed running the SPSS 18.0 software package (SPSS Inc. Chicago, USA). Group means were compared by Single-factor ANOVA and post hoc Tukey test, and P-values of < 0.05 were considered statistically significant.

Results
The influence of ECM molecules and other factors on adhesion and invasion rates of two M. gallisepticum strains, R low and R high , was investigated by a modified version of the Gentamicin Invasion Assay (GIA). The adhesion rate of strain R low to chicken RBC was about 77%. A slight increase was seen with EACA, alone or in combination with plasminogen or plasminogen activated to plasmin by the addition of the human urokinase plasminogen activator ( Figure 1A). A more remarkable increase was observed when the hemadsorption-negative and avirulent strain R high was incubated together with plasmin. EACA in combination with plasminogen resulted in adhesion rates three times higher than those of R high alone, and fibronectin apparently decreased the adhesion capacity of R high for RBC.
However, standard deviations were high and thus differences were not statistically significant. The previously reported low cell invasion rates of R low for RBC [11] could be confirmed and were not notably influenced by any ECM molecule ( Figure 1B). A slight decrease in the mean invasion rate was observed in the presence of plasminogen. R high showed almost no cell invasion capabilities in this set-up; only with plasminogen a significantly increased cell invasion (p < 0.05) was observed (0.014 ± 0.003%).
A positive influence on the adhesion of strain R low to HeLa cells was observed upon the addition of plasminogen, type IV collagen, plasminogen with EACA, and plasmin ( Figure 2A). However, none of these elevations was statistically significant. R high exhibited eight-fold lower adhesion rates to HeLa cells than R low , however, there was a statistically significant enhancement in adhesion of R high when adding fibronectin (p ≤ 0.001), plasminogen (p = 0.022) or plasmin (p ≤ 0.001) in comparison to R high w/o ( Figure 2A). As all results obtained with cellular-or plasma-derived fibronectin were comparable, only the results for cellular fibronectin are shown in the figures.
Cell invasion rates of R low into HeLa cells were increased by the addition of plasminogen, plasmin and collagen type IV ( Figure 2B). The combination of EACA with either plasminogen or plasmin decreased invasion rates. Surprisingly, we also recorded enhanced cell invasion rates of R low with EACA alone, compared to the mean invasion rates of R low alone ( Figure 2B).
To get an insight into the role of cholesterol/lipid rafts concerning the cell invasion and/or adhesion process, HeLa cell membranes were depleted from cholesterol using MβCD prior to the addition of the bacteria. Whereas cholesterol depletion had a positive effect on the adhesion of R low (Figure 2A), cell invasion was slightly decreased ( Figure 2B). However, the observed differences in cell invasion were not statistically significant.
As the addition of fibronectin led to an increased adhesion of R high to HeLa cells, while R high has been reported to lack the fibronectin-binding proteins found in R low [29], Western and dot blot analyses of cell lysates or intact cells were performed. Fibronectin was detected on whole cells of R low as well as R high without the necessity of preincubating the mycoplasma cultures with supplemented fibronectin ( Figure 3B). Streptococcus pneumoniae, a species known to possess fibronectin-binding proteins, also appeared positive on dot blots while E. coli proved negative. Western blot analysis of R low and R high cell lysates with fibronectin-specific antibodies resulted in the detection of a band at the same size as the fibronectin used as positive control or in the cell lysate of S. pneumoniae ( Figure 3A). No difference regarding the occurrence of fibronectin in samples of R low and R high could be observed. When tested for the presence of plasminogen on the bacterial cell surface, both R low and R high proved positive in dot blot assays ( Figure 4A). While Streptococcus canis strains G1 and G2, which were used as positive and negative controls for plasminogen binding according to a recently published report by Fulde et al. [38], exhibited either intense or marginal reaction with the Plg-specific antibodies, R low and R high showed an equally strong reaction. On Western blots, protein bands of 158, 89, 64 and 53 kDa were labeled in the cell lysates of R low and R high ( Figure 4B).

Discussion
ECM molecules, like fibronectin or vitronectin, have been described to play a strategic role for pathogenic bacteria. They serve as a bridge component between the pathogen that binds to the ECM molecule and to host cells, which naturally have receptors for these ECM molecules [39,40], thereby bringing the bacterium in close contact with the host cell. A more sophisticated exploit of interactions between ECM molecules has been described for Streptococcus pyogenes which is able to adhere to collagen via fibronectin bound to the bacterial surface [41]. Plasminogen might act similarly as a bridging component in rare events [42]. However, for cell invasive bacteria plasminogen might be of better use in its activated form, the serine protease plasmin. Besides exhibiting a crucial role in fibrinolysis and ECM remodeling, the proteolytic activity of plasmin could empower the plasmin-equipped bacterium more directly to penetrate host cells or transmigrate through monolayers, thereby assisting to cross the epithelial barrier [42][43][44]. M. fermentans has been shown to profit from binding to plasminogen or ECM molecules. When co-incubated with HeLa cells, M. fermentans exhibited an increased adherence to HeLa cells when pretreated with collagen type III or IV or plasminogen [32,34]. In this study, we report the likewise positive effect of plasminogen, as well as a slight enhancement due to fibronectin on the adherence capacity of M. gallisepticum. Fibronectin and plasminogen can of course be found in the serum used in the culture medium, which was proven by the ability to detect these molecules on the surface of R low and R high without further adding them (Figures 3 and 4). Nevertheless, to study their distinct effects on the invasion and adhesion capabilities of R low and R high , fibronectin and plasminogen were added in abundance. The hemadsorption-positive strain R low showed an increased adherence to HeLa cells when preincubated with plasminogen. Strain R high, lacking the major cytadherence proteins GapA and CrmA, exhibits drastically decreased cytadherence capacities that have been well documented [45]. However, it seems that the  Detection of plasminogen in cultures of R low and R high . R low and R high were also equally well stained on dot blot assays using a plasminogen-specific antibody (A). As positive controls Streptococcus canis strain G1 (+) [38] and plasminogen (Plg) were used, while S. canis strain G2 (−) [38] or HFLX medium served as negative controls. On Western blots, prominent protein bands of 158, 89, 64 and 53 kDa were stained equally well in cell lysates of R low and R high (B). MWM, molecular weight marker. rather marginal adhesion rates of R high to HeLa cells can be enhanced by plasminogen, plasmin and fibronectin (Figure 2A). The recruitment of plasmin or fibronectin as a bridging molecule could represent a secondary anchoring system of M. gallisepticum besides the primary and accessory cytadhesion proteins like GapA and CrmA. As, on the other hand, adhesion rates of R high with plasmin or fibronectin were still lower than adhesion rates of R low alone, it seems that the loss of GapA and CrmA on bacterial site cannot be overcome by the ECM-mediated adhesion.
The positive effect of fibronectin on the adhesion capacity of R high to HeLa cells was surprising, as in 2006 R high was reported to lack fibronectin-binding proteins [29]. When we tested whole cells and whole cell-lysates of M. gallisepticum strains R low and R high for fibronectinbinding properties, it turned out that fibronectin could be detected on the surface of mycoplasma whole cells as well as in cell lysates, irrespective of the strain tested. In contrast to May et al. [29], who investigated the TX-114 insoluble proteins of M. gallisepticum R high , we did not exclude the cytoplasmic proteins from our analyses. In M. pneumoniae, also cytoplasmic proteins like EF-Tu and PDHB were shown to appear at the surface of the mycoplasma cell and to bind fibronectin [46]. Also in other Mycoplasma species like M. genitalium, EF-Tu was found to be present at the cell surface [47]. Interestingly, although the EF-Tu proteins of M. genitalium and M. pneumoniae share a 96% identity, the M. genitalium protein does not bind fibronectin. Recently, Balasubramanian et al. showed that 3 amino acid residues (S343, P345, and T357) of the M. pneumoniae EF-Tu were essential for its binding activity to fibronectin, and that these critical amino acids are changed to alanine residues in the EF-Tu of M. genitalium [47]. The EF-Tu of M. gallisepticum, on the other hand, shares a lower overall identity with the EF-Tu of M. pneumoniae, but is higher conserved for these three Fn-binding residues (Q343,K345, T357), leaving it open to speculate about an involvement of M. gallisepticum R high EF-Tu in the adhesion process by fibronectin binding.
Detection of fibronectin on the surface of M. gallisepticum, routinely cultured in HFLX, indicates that the fibronectin concentration in the horse serum is high enough to allow the pathogen to adsorb substantial amounts. Absorbance of protein components from the growth medium by M. gallisepticum has already been reported for transferrin or immunoglobulins [48] and for the latter IgG-binding proteins have been recognized [49]. Whether binding of fibronectin plays a role for M. gallisepticum in vivo, remains to be elucidated.
The influence of another family of ECM molecules, collagen type IV and V, onto M. gallisepticum adhesion and invasion was also investigated in the present study. Whereas collagen type V revealed no marked effects (data not shown), type IV collagen had a slightly positive effect on the adherence and invasion capacity of R low to HeLa cells. For M. fermentans a supporting effect of collagen type III and V on the adherence capacity to HeLa cells had been reported [34]. In contrast, collagen type IV had no effect on M. fermentans adherence. This difference in the mode of action concerning the same family of compounds (collagens) could perhaps be explained by the different nature of the three collagens investigated, as collagen type III and V exhibit fibrillar structures, and are mostly found in extensible connective tissue, whereas collagen type IV is rather linked to the basal lamina. Concomitantly, this difference might also reflect the different natural habitats of M. fermentans and M. gallisepticum.
Plasminogen is not an ECM molecule per se but after its full activation to plasmin, it exerts a serine protease function on the fibrinolytic system, as well as on many ECM molecules. Binding of plasmin or plasminogen to fibrin and other ECM molecules predominantly occurs via its high-and low affinity lysine-binding sites (LBS) [50], and therefore can be easily blocked by EACA, a lysine analogue. If plasminogen is used as a bridging molecule by M. gallisepticum, then binding of EACA to the LBS of plasminogen should compensate for any positive effect of plasminogen on adherence and/or invasion. The results from the invasion assays support this hypothesis, as R low, pretreated with plasmin or plasminogen ( Figure 2B), showed higher invasion rates to HeLa cells than the untreated control and addition of EACA indeed had a negative impact on the plasminogen-elevated invasion rates. For M. fermentans, plasminogen was shown to enhance cell invasion as well as cell adhesion [32,51], whereas no effect was seen with M. pneumoniae [33]. When we analyzed the cell adherence of M. gallisepticum, an increase after treatment with plasminogen or plasmin was observed ( Figure 2A). To our surprise, however, EACA had no negative effect on the plasminogen-elevated cell adhesion, but rather boosted the cell adhesion. Until recently, EACA has been described as a potent inhibitor of plasminogenbinding to bacteria, [51,52], no one has yet investigated the effect of EACA alone on bacterial cell invasion or adhesion capacity. The increase in cell adhesion of R low and R high in the presence of EACA alone (Figure 2A) may indicate that another M. gallisepticum surface protein is participating in the multifactorial adhesion process. It is well established that M. gallisepticum contains several cell surface proteins with hemagglutinating (pMGA/VlhA [53]), or adhesive properties like Mgc2 [54], α-enolase [31] or the uncharacterized proteins P30, P48, P50, P80 [55].
Cholesterol depletion from HeLa cell membranes resulted in a decreased invasion capacity of M. gallisepticum, suggesting that cholesterol-rich regions in the cell membrane like lipid rafts are needed for the invasion process. Many bacteria have been described to enter host cells via receptors being present in lipid rafts and/or caveolae [reviewed by [56,57]. An alternative strategy for mycoplasma entry could be the engulfment via the lipid rafts. So far, entry into eukaryotic cells via engulfment without the use of receptors has only been described for very small objects like viruses and bacterial toxins [35]. Mycoplasmas belong to the smallest bacteria, they do not contain a cell wall, and their bodies seem easily deformable. M. gallisepticum has been observed to squeeze itself through 220-nm [58] and apparently even 100-nm pore filters (own observation). As lipid rafts in vivo have diameters up to 700 nm [57], M. gallisepticum are flexible and small enough to maybe enter the eukaryotic cell by squeezing through lipid rafts. As envisioned by others before [17,59] the fusogenic properties of the mycoplasma lipidic cell wall might additionally enable the pathogen to interact with the lipid-rich raft domains. While cholesterol depletion lowered the cell invasion rates of M. gallisepticum, adherence capabilities to HeLa cells were improved. This could be explained by the fact, that cholesterol removal from membranes results in dispersion of the raft-associated lipids and proteins [56], thus destroying the possible entry site, but at the same time generating more sites for the mycoplasma to adhere to. With M. fermentans, plasminogen-preincubated bacteria adhered to cholesteroldepleted HeLa cells as well as to cholesterol-containing cells, but internalization of M. fermentans was almost completely inhibited after cholesterol depletion [32].

Conclusion
Taking these results together, it seems that M. gallisepticum is utilizing a second anchoring system besides its own primary and accessory cytadhesion proteins like GapA and CrmA that relies on the recruitment of ECM-or ECM-like molecules like collagen type IV, fibronectin or plasminogen/plasmin. Cholesterol depletion of HeLa cells had contrary effects on the adherence and cell invasion of M. gallisepticum, calling us to investigate the role of lipid rafts on the host cell surface in these mechanisms.