The Psa Fimbriae of Yersinia pestis Interact with Phosphatidylcholine on Alveolar Epithelial Cells and Pulmonary Surfactant

Abstract

The pH 6 antigen (Psa) of Yersinia pestis consists of fimbriae with adhesive properties of potential importance for the pathogenesis of plague, including pneumonic plague. The Psa fimbriae mediate bacterial binding to human alveolar epithelial cells. The Psa fimbriae bound mostly to one component present in the total lipid extract from type II alveolar epithelial cells of the cell line A549 separated by thin-layer chromatography (TLC). The Psa receptor was identified as phosphatidylcholine (PC) by TLC using alkali treatment, molybdenum blue staining, and Psa overlays. The Psa fimbriae bound to PC in a dose-dependent manner, and binding was inhibited by phosphorylcholine (ChoP) and choline. Binding inhibition was dose dependent, although only high concentrations of ChoP completely blocked Psa binding to PC. In contrast, less than 1 μM of a ChoP-polylysine polymer inhibited specifically the adhesion of Psa-fimbriated Escherichia coli to PC, and type I (WI-26 VA4) and type II alveolar epithelial cells. These results indicated that the homopolymeric Psa fimbriae are multimeric adhesins. Psa also bound to pulmonary surfactant, which covers the alveolar surface as a product of type II alveolar epithelial cells and includes PC as the major component. The observed dose-dependent interaction of Psa with pulmonary surfactant was blocked by ChoP. Interestingly, surfactant did not inhibit Psa-mediated bacterial binding to alveolar cells, suggesting that both surfactant and cell membrane PC retain Psa-fimbriated bacteria on the alveolar surface. Altogether, the results indicate that Psa uses the ChoP moiety of PC as a receptor to mediate bacterial binding to pulmonary surfactant and alveolar epithelial cells.

Yersinia pestis is transmitted by fleas or aerosols, causing bubonic or pneumonic plague by infecting regional lymph nodes or the lungs and septicemic plague when local containment is bypassed (25). Several virulence factors have been identified. A plasmid-encoded type III secretion system functions to export Yops and LcrV proteins that are delivered to the extracellular milieu, the plasma membrane, or the cytosol of a host target cell. The Yops and LcrV act in concert to inhibit phagocytosis and downregulate inflammation (4, 6). The extracellular bacteria deliver effector proteins into the host cell in a contact-dependent process (27). Two potential adhesins expressed by Y. pestis are the plasminogen activator protein (Pla) and the pH 6 antigen (Psa). Pla is an outer membrane protease that cleaves and activates plasminogen, a property that is important for the ability of Y. pestis to infect via the peripheral routes (32). Pla also enables Y. pestis to adhere to several noncollagenous matrix proteins and enhances HeLa cell invasion (7, 13, 14). Y. pestis infections induce anti-Psa antibodies (2) and Y. pestis lacking Psa exhibits reduced virulence in mice, the 50% lethal dose after intravenous application of Y. pestis KIM5 being reduced at least 100-fold (17). Thus, Psa is both expressed in vivo and implicated in bacterial virulence. It has also been speculated that Psa is involved in the binding of the organism to target cells to allow effective intracellular delivery of Yops (33), although later findings with the mouse macrophage cell line RAW264.7 did not support this possibility (11). Nevertheless, the latter study showed that Y. pestis Psa promotes resistance to phagocytosis by RAW264.7 cells, independently of Yops and capsule antigen fraction 1 (11). Psa expression was shown to occur in the macrophage cell line RAW264.7, depending on the acidification of intracellular compartments (18). The operon encoding Psa has been cloned and sequenced (18). The latter study revealed that this antigen belongs to a class of adhesins that are exported and assembled as 4-nm-thick fibrils via a chaperone-usher pathway (34) and consists of the homopolymeric structure of PsaA, a subunit protein of 15 kDa. Expression of Psa on the surface of the bacteria is induced when the bacteria are grown between pH 5 and 6.7 and between 35 and 41°C (2). Psa mediates agglutination of erythrocytes of many species (3) and binds to some β1-linked galactosyl residues in glycosphingolipids (24). In Yersinia pseudotuberculosis, it was reported that Psa mediates bacterial binding to the epithelial cell line HEp-2 (37). In addition, purified Psa binds to apolipoprotein B-containing lipoproteins in human plasma, mainly low-density lipoprotein, and it was suggested that the lipid moiety of the lipoprotein is responsible for the interaction (20). Is has also been proposed that Psa acts as an Fc receptor for human immunoglobulin G1 (IgG1), IgG2, and IgG3 (38).

To determine the role of Psa in pneumonic plague, we have recently analyzed its interaction with three human respiratory tract epithelial cells (19). All three studied cell lines, and particularly the type II alveolar epithelial cell line A549, showed a significantly higher interaction when the bacteria expressed Psa. In the present study, we searched for the A549 cell receptor for Psa. We found that Psa bound specifically to the phosphatidylcholine (PC) of cell lipid extracts, as well as to commercial PC preparations. Since PC is the principal component of pulmonary surfactant, which is secreted by type II alveolar epithelial cells, we also studied the Psa interaction with pulmonary surfactant. Psa bound to the PC present on type I and type II alveolar epithelial cells and in pulmonary surfactant. This adhesion was blocked by phosphorylcholine (ChoP), indicating that the interaction of Psa with the polar moiety of PC is essential for binding.

MATERIALS AND METHODS

Reagents.

Phospholipids and glycolipids were purchased from Matreya (Pleasant Gap, PA). Survanta (beractant [25-mg phospholipids/ml and less than 1.0 mg/ml protein]; Ross Products Division, Abbott Laboratories, Columbus, OH) is a natural bovine lung extract containing phospholipids (80% of the total lipids being PC), neutral lipids, fatty acids, and surfactant proteins B and C. Rat surfactant from bronchoalveolar lavage and Survanta were generously supplied by S. R. Bates (Institute for Environmental Medicine, University of Pennsylvania). Rabbit polyclonal antibodies against Psa were prepared in our laboratory using recombinant Psa isolated from E. coli SE5000 (psa⁺). Unless specified, reagents were purchased from Sigma (St. Louis, MO).

Bacterial strains, human cells, and growth conditions.

The nonfimbriated Escherichia coli host strain SE5000 (MC4100 recA56 Fim⁻) was used (30). Plasmid pCS267 is a pBR322 derivative which carries the psaA, psaB, and psaC genes, encoding the pH 6 antigen subunit and chaperone and usher proteins, respectively (19). Bacteria were routinely grown at 37°C in Luria-Bertani (LB) medium (22), supplemented with ampicillin (200 μg/ml) when appropriate. Cells of the human type II alveolar epithelial line A549 (ATCC CCL185) and type I alveolar epithelial line WI-26 VA4 (ATCC CCL95.1; American Type Culture Collection, Rockville, MD) were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Rockville, MD) high-glucose formulation with GlutaMAX supplemented with 10% inactivated fetal bovine serum (FBS) at 37°C in 5% CO₂.

Isolation of fimbriae.

It was observed that recombinant pH 6 antigen (Psa) was present in significant amounts in the culture supernatant of E. coli SE5000/pCS267 grown overnight in LB medium at 37°C. Thus, Psa was isolated from spent culture medium after removal of the bacteria by centrifugation at 8,000 × g for 30 min. A saturated solution of ammonium sulfate was added to the supernatant to a final concentration of 30%. After overnight incubation on ice, the supernatant was centrifuged at 10,000 × g for 30 min, and the pellet was resuspended in Tris-buffered saline (TBS; 10 mM Tris-HCl [pH 7.4], 154 mM NaCl). Excess ammonium sulfate was removed by extensive dialysis against TBS, and the protein concentration was determined to be 0.28 mg/ml culture supernatant (21). The purity of the fimbriae was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining, showing essentially only the 15-kDa PsaA subunit band (>95% purity).

Preparation of cellular lipids.

A549 cells grown to confluence were washed with phosphate-buffered saline (PBS; 10 mM NaHPO₄, 1.8 mM KH₂PO₄, 2.7 mM KCl, 137 mM NaCl [pH 7.5]) and harvested mechanically. Lipids from cells (equivalent to 1.5 mg total cellular protein content) were successively extracted with 1.0 ml chloroform-methanol-water (4:8:3 [vol/vol/vol]) and chloroform-methanol (1:1 [vol/vol]) for 30 min. Lipid extracts were pooled, dried under N₂, and resuspended in chloroform-methanol-water (60:30:4.5 [vol/vol/vol]). In order to remove glycerophospholipids from the cellular lipid extract, a mild alkaline methanolysis was done (5). After drying, a lipid aliquot was incubated in 1.0 ml of 0.25 M NaOH (in methanol) at 50°C for 10 min. Acetic acid (0.05 ml) was added followed by chloroform (2.0 ml) and water (0.67 ml). After two phases appeared, the upper phase was removed and the lower phase was washed with methanol-water (1:1 [vol/vol]) and applied to thin-layer chromatography (TLC) plates.

Analysis of cellular lipids separated by TLC.

Cellular lipids were separated on TLC aluminum-backed silica gel G plates (Merck, Darmstadt, Germany) with chloroform-methanol-water (65:25:4 [vol/vol/vol]) used as a solvent system (8, 28, 36). Glycolipids were stained with orcinol-sulfuric acid reagent (29) and phospholipids were detected with molybdenum blue reagent (Sigma). For the overlay assay, the plates were sequentially treated with polyisobutylmethacrylate (0.4% in n-hexane), blocked in 0.5% bovine serum albumin (BSA) in PBS for 1 h at room temperature, overlaid with purified Psa fimbriae (50 μg/ml in 1% BSA-PBS) for 1 h, and washed three times with PBS-0.02% Tween 20. Bound fimbriae were visualized with rabbit anti-Psa antiserum (19) in combination with a horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (Cappel, MP Biomedicals, Aurora, Ohio), with 4-chloro-1-naphthol and H₂O₂ as the substrate.

Determination of binding by ELISA.

Binding was determined by enzyme-linked immunosorbent assay (ELISA) as follows. The wells of polyvinyl chloride microtiter plates (96 wells; Falcon, Becton Dickinson, Franklin Lakes, NJ) were coated with 1 μg (each) of phospholipids (PC, phosphatidylserine [PS], phosphatidylethanolamine [PE], phosphatidylinositol [PI], or sphingomyelin [SM]) or galactosylceramide (GalCer) in methanol (100%) or chloroform-methanol (1:1 [vol/vol]), as described previously (12). Both coating methods gave comparable binding results. The lipids were dried at 37°C, the wells were blocked with 0.5% BSA-PBS for 1 h, and dilution rows of Psa fimbriae in 0.5% BSA-PBS were prepared. After incubation at room temperature for 1 h and washing with PBS, bound fimbriae were detected with rabbit anti-Psa antiserum as described above, except that o-phenylenediamine was used as the chromophore. For the binding inhibition assays, Psa fimbriae (0.25 to 1.0 μg/ml in 0.5% BSA-TBS) were preincubated with twofold serially diluted ChoP, phosphorylserine (SerP), galactose, lactose, or glucose (starting at a concentration of 500 mM in TBS) for 30 min at room temperature. The mixtures were transferred to the lipid-coated plates, and fimbrial binding was determined with rabbit anti-Psa antiserum in combination with HRP-conjugated anti-rabbit IgG and o-phenylendiamine (A₄₅₀). For the pulmonary surfactant studies, polyvinyl chloride plates were coated with Survanta (1.0 μg of phospholipids/ml) or rat surfactant (includes approximately 80 μg/ml PC and 0.16 mg/ml protein; diluted 1/10 in PBS) overnight at 37°C. Psa binding and ChoP inhibitory effect were assayed as previously described for lipids. Determination of the 50% inhibitory concentration (IC₅₀) was done using the Hill equation (KaleidaGraph 4.0; Synergy Software, Reading, PA).

Preparation of polymeric inhibitors.

ChoP, SerP, or α-lactose 1-phosphate (LacP) were cross-linked to polylysine (molecular weight [MW], 15,000 to 30,000) by using EDC [1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride] to form stable phosphoramidate bonds. Briefly, ChoP (165.0 mg/ml, or SerP [45 mg/ml] or LacP [275 mg/ml]), EDC (143.8 mg/ml), and polylysine (1.0 mg/ml) were dissolved and mixed in 0.1 M imidazole (pH 6.0) and incubated for 5 h at 37°C, followed by extensive dialysis against TBS. The presence of cross-linked molecules was confirmed by nondenaturing PAGE (40), mass spectrometry (DNA & Proteomics Core Facility, School of Veterinary Medicine, University of Pennsylvania), and ELISA with specific anti-ChoP (monoclonal antibody [MAb] TEPC15; Sigma) or SerP (MAb clone 1H6; Upstate) antibodies. The efficiency of cross-linking was quantified by phosphorus analysis (39). Polymerized ChoP-polylysine (200 μg/ml) was digested with trypsin from Sigma (2 μg/ml; 9,800 BAEE [N-α-benzoyl-l-arginine ethyl ester] units per mg of protein) overnight at 37°C in 25.0 mM ammonium bicarbonate (pH 8.0).

Liposome-mediated bacterial agglutination.

Bacterial recognition of phospholipids was also determined by assessment of the aggregation of bacteria with various phospholipid vesicles. The following liposome vesicles were prepared: PC-containing liposomes with PC-dicetylphosphate (DP)-cholesterol (CHO) at a molar ratio of 5:1:4, PE-containing liposomes with PE-DP-CHO at a molar ratio of 5:1:4, PC- and GalCer-containing liposomes with PC-DP-CHO-GalCer at a molar ratio of 5:1:4:0.5, and PE- and GalCer-containing liposomes with PE-DP-CHO-GalCer at a molar ratio of 5:1:4:0.5. For liposome preparations, lipids dissolved in methanol-chloroform (1:1 [vol/vol]) were mixed and dried under N₂ gas and then vacuum dried for 30 min. The dried pellets were resuspended in isopropanol (50 μl), and PBS (450 μl) was added while mixing. For bacterial agglutination tests, slide agglutinations were performed by mixing 2.5 μl of liposomes (0.2 mg/ml) with 10 μl of E. coli psa⁺ (SE5000/pCS267) or lacking psa (SE5000/pBR322) grown overnight at 37°C and washed three times in PBS.

Psa fimbria binding to A549 cells.

Binding of purified Psa fimbriae was performed on monolayer of A549 cells cultured in 96-well tissue culture plates. Cells were washed with PBS and fixed with 1% formaldehyde in PBS for 20 min at room temperature. All of the experimental conditions used to assay Psa binding were the same as those described for the ELISA experiments. Negative controls without addition of fimbriae to cells gave A₄₅₀ values of approximately 0.05 (similar to those of controls without cells).

Recombinant E. coli psa⁺ cell binding to alveolar epithelial cells and pulmonary surfactant.

Bacterial binding assays were done essentially as described previously (11, 19). Briefly, A549 and WI-26 VA4 cells cultured to confluence in 24-well tissue culture plates were washed three times with PBS and DMEM without FBS was added. E. coli psa⁺ (SE5000/pCS267) or E. coli lacking psa (SE5000/pBR322) was grown as described earlier, washed with PBS, retaken in DMEM without FBS, and added to the cells at a multiplicity of infection of 10. After incubation at 37°C for 1 h, duplicate wells of infected cell monolayer were washed five times with PBS and 0.1% Triton X-100 in water was added. After 15 min, cell lysates were collected and serially diluted 10-fold in PBS, and aliquots were inoculated onto LB agar plates with suitable antibiotics to assess viable bacterial CFU. For the binding inhibition assays with water-soluble inhibitors (ChoP, SerP, galactose, choline, and the polylysine polymers), the inhibitors were added to the cells at the same time as the bacteria. For the binding inhibition assays with pulmonary surfactant (Survanta), three approaches were used. Survanta was added to the epithelial cells, the plates were centrifuged (600 × g, 10 min) and incubated for 10 min at 37°C, and the supernatant was removed or not removed before the bacteria were added. For the third alternative, Survanta was incubated for 30 min at 37°C with the bacteria before addition of the mixture to the epithelial cells. The range of the concentrations of Survanta in the wells was from 0.05 to 6.25 mg phospholipids per ml. The statistical analyses were conducted using Student's t test.

RESULTS

Interaction of Psa fimbriae with lipids from A549 cells.

In order to search for the presence of Psa receptors on A549 cells, cellular proteins and lipids were analyzed by standard ligand blotting assays. Total membrane proteins separated by SDS-PAGE and transferred to nitrocellulose membranes failed to bind to Psa fimbriae (not shown). In contrast, the fimbriae bound to one major band of the total lipid extract separated by TLC (Fig. 1A, lane 1). This lipid band could also be detected with the molybdenum blue reagent that is specific for phosphorus (Fig. 1B, lane 1), but not with the orcinol-sulfuric acid reagent that is specific for carbohydrates (Fig. 1C, lane 1). This result indicated that the compound recognized by Psa corresponded to a phospholipid. To investigate whether the Psa receptor could be a minor glycolipid component that comigrated with the detected phospholipid band, an aliquot of the total lipid extract was treated by mild alkali hydrolysis to which only sphingolipids are resistant. Essentially no phospholipids could be stained with molybdenum blue after the alkali treatment, indicating that most if not all of the phospholipids had been hydrolyzed (Fig. 1B, lane 2). Psa fimbriae did not bind to any component in the alkali-resistant fraction (Fig. 1A, lane 2). In addition to the cellular lipids, standard phospholipids were separated by TLC and phosphorus detection was carried out to identify the phospholipid with receptor activity on A549 cells. The lipid that interacted with Psa comigrated only with PC (Fig. 1B, lane 3) and not with any orcinol-stained glycosphingolipid used as a control (Fig. 1C, lane 3). Psa was also able to bind to purified PC from egg yolk and weakly to sphingomyelin (Fig. 1A, lane 3). These results identified PC as a putative receptor for the Psa fimbriae.

FIG. 1.

Psa binding to lipids from A549 cells on TLC plates. (A) Psa fimbriae were overlaid on TLC-separated lipids, and binding was detected with rabbit polyclonal anti-Psa antibody followed by incubation with HRP-conjugated anti-rabbit IgG. (B) Detection of phospholipids by molybdenum blue staining. (C) Detection of glycolipids with orcinol. Lane 1, A549 cell lipid extract corresponding to 300 μg of cellular protein; lane 2, A549 cell lipid extract after alkali methanolysis. Lane 3 contained commercial PE, PC, PS, PI, and SM (5.0 μg each) for panels A and B and neutral glycosphingolipid Qualimix (4.0 μg) containing GalCer, lactosylceramide (LacCer), globotriaosylceramide (Gb₃Cer), and globotetraosylceramide (Gb₄Cer) for panel C.

Binding of Psa fimbriae to lipids and specific binding inhibition by the polar moieties of lipids.

Psa binding to standard phospholipids was quantified on microtiter plates (Fig. 2). Psa bound with the highest affinity to PC, showing a typical saturation curve, contrasting with Psa binding to equivalent amounts of sphingomyelin. No interactions between Psa and phosphatidylserine, phosphatidylethanolamine, or phosphatidylinositol could be detected. To identify the lipid domain involved in the interaction between PC and Psa, ChoP was tested as a blocking agent. ChoP inhibited the binding of Psa to PC in a dose-dependent manner (data not shown), with an IC₅₀ of 24 mM (Table 1), maximum inhibition (96%) being obtained at a concentration of 125 mM ChoP. In contrast, phosphorylserine did not demonstrate a significant effect on the Psa-PC interaction at the highest concentration tested (Table 1). Interestingly, choline itself was able to inhibit Psa binding to PC, indicating that the phosphate of ChoP on PC was not required for Psa recognition. A previous study reported that Psa binds to GalCer and some other β1-linked galactosyl residues in glycosphingolipids (24). This was confirmed by finding that Psa bound to GalCer in a dose-dependent manner (data not shown). The interaction was inhibited by lactose, which contains a β1-linked galactosyl residue, or galactose, but not by ChoP (Table 1) or choline. Conversely, the highest concentration of lactose or galactose used did not inhibit Psa binding to PC (Table 1). These results indicated that the ChoP or choline moiety of PC plays an essential role for the specific recognition of PC by Psa. In addition, the data also showed that Psa has different binding domains for its interactions with ChoP and β1-linked galactosyl residues, respectively.

FIG. 2.

Psa binding to solid-phase-coated phospholipids. PC, SM, PS, PE, or PI was adsorbed onto polyvinyl chloride microtiter wells (1.0 μg per well) and incubated with increasing concentrations of Psa fimbria. Bound fimbria was detected with an antifimbrial antibody and an HRP-conjugated secondary antibody. The data are means ± standard error of three values and represent one of three reproducible experiments.

TABLE 1.

Inhibition of Psa binding to PC or GalCer by monomeric or polymeric inhibitors

Inhibitor	IC50 (μM)
	PC	GalCer
ChoP (monomer)	24,235	>500,000
Choline	5,443	>500,000
SerP (monomer)	>250,000	NDb
Polylysine-ChoPa	0.031	>50.00
Polylysine-SerPa	>50.00	>50.00
Galactose	>250,000	41,420
Lactose	>250,000	71,540
Glucose	ND	>500,000
Polylysine-LacPa	>50.00	2.29

^aAverages of 18 linked ChoP, 15 linked SerP, or 14 linked LacP per polylysine (mol/mol) and average of 150 lysines per polylysine molecule (MW of 15,000 to 30,000).

^bND, not determined.

Inhibition of Psa binding to PC or GalCer by synthetic polymers.

Although the interactions of Psa with PC or GalCer were inhibited by specific competitors, the IC₅₀ values of the competitors were high (Table 1). These results could be due to the fact that ChoP is a soluble monomer, whereas Psa is a homopolymeric organelle that potentially exposes hundreds of binding domains for PC molecules. To investigate this possibility, ChoP was multimerized by cross-linking it to polylysine. This ChoP polymer had a binding IC₅₀ for PC that was 7.5 × 10⁵ times lower than the IC₅₀ for the ChoP monomer (Table 1). Since each polylysine-ChoP molecule had an average of 18 covalently linked ChoP moieties (as determined by quantitative phosphorus analysis), 4.2 × 10⁴ times less ChoP was needed when polymerized to obtain the same inhibitory effect as for the monomer. In contrast, no binding inhibition was detected with the highest concentration of a similarly synthesized polylysine-SerP polymer (Table 1), confirming the specificity of the ChoP-Psa interaction. To demonstrate that the described binding inhibition was only due to receptor polymerization, the polylysine-ChoP polymer was trypsinized and assayed for binding to PC. As shown in Fig. 3, trypsinization increased the IC₅₀ approximately 10³-fold. That trypsinized polylysine-ChoP had a lower IC₅₀ than ChoP suggested that the trypsinization was not complete, consistent with the mass spectrometry profile (not shown). Trypsinized and nontrypsinized polylysine-SerP were not inhibitory, as expected.

FIG. 3.

Inhibition of Psa binding to PC. PC was adsorbed onto polyvinyl chloride microtiter wells and incubated with Psa fimbriae (0.25 μg/well) and increasing concentrations of the following inhibitors: trypsinized polylysine-ChoP (empty circles), untreated polylysine-ChoP (filled circles), trypsinized polylysine-SerP (empty diamonds), or untreated polylysine-SerP (filled diamonds). PC adsorption onto the wells and Psa detection were as described in the legend to Fig. 2. The data are means ± standard error of three values and are representative of two reproducible experiments.

The need to use polymeric molecules to inhibit efficiently Psa binding to its receptors was further investigated with a synthesized polylysine-LacP polymer. Similar to the inhibition of Psa binding to PC by polymerized ChoP, polymerizing LacP increased drastically its inhibitory power for the Psa-GalCer interaction (Table 1). Since each polylysine-LacP molecule had an average of 14 covalently linked LacP moieties, 2.2 × 10³ times less LacP was needed when polymerized to obtain the same inhibitory effect as for the monomer. The polylysine-phosphorylserine polymer did not inhibit the Psa binding to GalCer, confirming the specificity of the polylysine-LacP inhibitor for the Psa-GalCer interaction.

Interaction of Psa-expressing bacteria with phosphatidylcholine-containing liposomes.

The accessibility of lipid receptors to ligands is more restricted when the receptors are membrane associated than when they a solid phase is coated with them. To determine whether Psa recognizes PC as a membrane constituent, PC-containing liposomes were prepared and their capacity to agglutinate Psa fimbriated E. coli was investigated. PC-containing liposomes, but not PE-containing liposome controls, mixed with E. coli expressing Psa fimbriae promoted strong bacterial agglutination (Table 2). The incorporation of GalCer into PE-liposomes resulted in the agglutination of E. coli psa⁺ cells. In contrast, no agglutination was observed when E. coli psa mutant cells were mixed with PC-, PE-, or PE-GalCer-containing liposomes. Incubation of the fimbriated E. coli strain with PBS did not result in the formation of autoaggregates. Thus, the Psa fimbriae allow bacteria to bind to PC-containing membranes.

TABLE 2.

Liposome-mediated bacterial agglutination

Liposome	Agglutination in E. coli cellsa
	psa+	psa mutant
PC	++	−
PC-GalCer	+++	−
PE	−	−
PE-GalCer	++	−

^aAgglutination was quantified as follows: +++, immediate very strong reaction; ++, strong reaction after 30 s; +, weak reaction after 1 min; −, no reaction for 2 min.

ChoP inhibits the binding of Psa and Psa-fimbriated bacteria to alveolar epithelial cells.

We studied the binding of purified Psa fimbriae to confluent monolayers of type I (WI-26 VA4) and type II (A549) alveolar epithelial cells in 24-well plates. The purified fimbriae bound to cells in a concentration-dependent manner, reaching 50% and 100% binding at fimbrial concentrations of 0.1 μg/ml and 1.0 μg/ml, respectively. This binding was inhibited in a dose-dependent manner by preincubation of Psa with different concentrations of ChoP prior to the assay (IC₅₀ = 300 mM ChoP). Further studies examined the ability to inhibit Psa-mediated bacterial binding to A549 and WI-26 VA4 cells. Consistent with our previous results (19), a recombinant E. coli strain that expressed Psa fimbriae adhered to A549 cell monolayers 11 times better than the corresponding E. coli strain that did not express the fimbriae (Table 3). The presence of 500 mM ChoP or 50 mM choline efficiently inhibited E. coli psa⁺ cell binding to the A549 cells (80% and 77% inhibition, respectively) and the WI-26 VA4 cells (86% and 83%, respectively). In contrast, 500 mM phosphorylethanolamine did not effectively interfere with E. coli psa⁺ cell binding to the two alveolar epithelial cells (≤11% inhibitions). Similarly, 50 mM serine had no inhibitory effects. Although 500 mM galactose did not efficiently inhibit binding to the A549 cells (11% inhibition), it weakly inhibited binding to the WI-26 VA4 cells (34% inhibition).

TABLE 3.

Inhibition of adhesion of E. coli psa⁺ and psa mutant cells to alveolar epithelial cells^a

Inhibitor	% E. coli adherencea
	A549 cells		WI-26 VA4 cells
	Psa+	Psa−	Psa+	Psa−
None	12.28 ± 1.34b	1.10 ± 0.34	8.15 ± 0.61b	2.04 ± 0.12
ChoP (500 mM)	2.44 ± 0.20c	0.96 ± 0.20	1.14 ± 0.12c	1.81 ± 0.12
Choline (50 mM)	2.85 ± 0.07c	1.08 ± 0.27	1.39 ± 0.17c	1.86 ± 0.13
EtnPd (500 mM)	10.93 ± 0.63	1.05 ± 0.32	8.26 ± 0.38	1.98 ± 0.28
Galactose (500 mM)	10.93 ± 0.35	1.09 ± 0.34	5.41 ± 0.83	2.28 ± 0.30
Polylysine-ChoP (100 nM)	5.86 ± 0.22c	NDe	3.39 ± 0.29c	ND
Polylysine-SerP (100 nM)	10.46 ± 0.72	ND	7.41 ± 0.22	ND
Polylysine-LacP (100 nM)	10.19 ± 0.54	ND	6.89 ± 0.04	ND

^aAdherence is expressed as (no. of cells bound/no. inoculated) × 100. The data are means ± standard error of three independent experiments.

^bStatistically significant difference between E. coli Psa⁺ and E. coli Psa⁻ cells (P < 0.01).

^cStatistically significant difference with no inhibitor (P < 0.01).

^dEtnP, phosphorylethanolamine.

^eND, not determined.

In addition to studying monomeric inhibitors, multimerized inhibitory molecules cross-linked to polylysine were shown to be effective in the nanomolar range. As shown in Table 3, 100 nM polylysine-ChoP was an effective inhibitor of the E. coli psa⁺ cell binding to A549 and WI-26 VA4 cells (52% and 58% inhibition, respectively), whereas the same concentrations of polymeric SerP or LacP did not significantly affect this binding (<15% or <20% inhibition, respectively). Although higher concentrations of polylysine-ChoP (200 nM) were even more inhibitory (73% for both cell types), polylysine-SerP (200 nM) presented some weak inhibitory properties (27% and 25% inhibition, respectively, for the two cell types) that were probably due to nonspecific interactions at these concentrations of polylysine polymers. Polylysine-LacP at 200 nM also acted as a weak inhibitor, which was only significant for the WI-26 VA4 cells (36% inhibition; P ≤ 0.05). Taken together, all of the results indicated that bacterial adhesion to the A549 cells conferred by the Psa fimbriae involved essentially PC. Although Psa-mediated bacterial adhesion to the WI-26 VA4 cells involved mainly PC, galactose-containing receptors were likely also involved.

Interaction of pulmonary surfactant with Psa fimbria.

Pulmonary surfactant is a lipoprotein complex that is synthesized and secreted by alveolar type II epithelial cells into the thin liquid layer that lines the lung epithelium. It is well established that lipids comprise over 90% of pulmonary surfactant, the remainder being made up of the surfactant proteins SP-A, -B, -C, and -D. At least 80% of the surfactant lipids are phospholipids, while the remainder are composed largely of neutral lipids such as cholesterol. Phosphatidylcholine lipids account for 70 to 80% of surfactant phospholipids (35). Considering the PC-enriched environment present in pulmonary surfactant, Psa binding to the commercial surfactant preparation Survanta and to rat surfactant from bronchoalveolar lavage was assayed. A concentration-dependent binding of purified Psa fimbriae to both surfactant preparations was observed by ELISA (Fig. 4A). Binding to Survanta reached saturation at fimbrial concentrations of 10 to 20 μg/ml, whereas binding to rat surfactant was essentially completed at 5.0 to 10 μg/ml. ChoP inhibited Psa binding to Survanta and rat surfactant in a concentration-dependent manner (Fig. 4B). The maximum blocking effect (90% inhibition) was observed for a ChoP concentration of 250 mM, when 0.5 μg/ml of Psa was added. These results indicated that the phosphorylcholine moiety of pulmonary surfactant PC interacts directly with the Psa fimbriae.

FIG. 4.

Psa binding to solid-phase-coated pulmonary surfactant. Polyvinyl chloride microtiter wells were used to adsorb rat surfactant (filled circles) or Survanta (empty squares). Binding of increasing concentrations of Psa fimbriae (A) and the inhibitory effect of ChoP on Psa binding (B) were determined as described in the legend to Fig. 2. The data are means ± standard error of three values and are representative of two reproducible experiments.

Since pulmonary surfactant covers the cell surfaces in the alveoli, surfactant PC might inhibit Psa-mediated bacterial binding to alveolar epithelial cells. To investigate this possibility, pulmonary surfactant, in the form of Survanta, was added to either A549 or WI-26 VA4 cells. Preincubation of the epithelial cells or Y. pestis with surfactant containing up to 6.25 mg phospholipids/ml did not significantly affect the interaction between each type of epithelial cell and the bacteria. Since surfactant remains associated with alveolar cell surfaces as mono- or multilamellar layers of lipids, it is likely that the added surfactant layered itself on the cell surface or on local surfactant without changing the nature of a presented PC-containing surface.

DISCUSSION

We recently showed that Psa mediated bacterial binding to a number of human respiratory tract epithelial cells (19). Psa binding and the adhesion of Psa-fimbriated recombinant E. coli or Y. pestis to type II alveolar epithelial cells (A549) were the most striking findings. These results pointed to the presence of a specific Psa receptor or receptors on the surface of these cells. In contrast, the Psa fimbriae are not needed for bacterial binding to macrophages such as the RAW164.7 cells (11). Macrophages, being professional phagocytes, express a large mosaic of scavenging receptors to optimize bacterial binding and engulfment. Thus, bacterial recognition by macrophages is anticipated to be less dependent on one specific adhesin, such as Psa. Payne et al. reported that the Y. pestis Psa fimbriae bind to β1-linked galactosyl residues of some glycosphingolipids, which are found on a range of host cells (24). Here, when total lipid extracts from A459 cells were analyzed for the interaction with Psa fimbriae using the TLC-overlay technique, only one predominant band was detected. This lipid could not be stained with a carbohydrate-specific reagent, and after alkali treatment of the cellular lipid extract, which results in the saponification of the phospholipids but leaves the glycosphingolipids intact, no TLC-separated lipid of the A549 cells was recognized by the Psa fimbriae. These results suggested that a phospholipid, but not a glycosphingolipid, acted as a Psa receptor on A549 cells. This phospholipid was further identified as PC, using TLC and the most common mammalian phospholipids as comparative standards, together with a phosphorus-specific stain. Psa also bound to commercial PC from egg yolk on TLC, confirming that PC is the A549 receptor for the Psa fimbriae. Interestingly, Makoveichuk et al. observed that Psa interacts with the lipid moiety of plasma lipoproteins and with cell membranes (20). These results are consistent with our findings, since all the lipoproteins and liposomes that were recognized by Psa in the latter study have large surfaces exposing polar lipid heads, including the choline moiety of PC (26).

It is noteworthy that in the A549 lipid extract, carbohydrate staining allowed us to visualize some glycosphingolipids whose relative migrations corresponded to short-carbohydrate-chain glycosylceramides (including hydroxylated GalCer [not shown]). Although lactosylceramide and hydroxylated GalCer were previously reported to interact with Psa fimbriae, we did not detect this interaction under the experimental conditions used for TLC. However, weak binding of Psa to TLC-separated commercial hydroxylated GalCer could be distinguished by adding an excess of this lipid to the TLC plate. For this, it was necessary to assay at least 10 times more pure hydroxylated GalCer than the amount of cerebroside found in the A549 cellular lipid extract used to detect PC as a Psa receptor (data not shown). Similarly, we did not detect Psa binding to any other A549 glycolipid under the conditions used to identify PC as a receptor, despite a previous report which showed that pure asialo-GM1, an A549 receptor for Pseudomonas aeruginosa (10), interacts with the Psa fimbriae. Thus, it is likely that the amounts of cellular lipid extract applied to the TLC that were adequate for recognition of PC by Psa, were not sufficient for the detection of potential weak interactions of Psa with asialo-GM1 or any other glycosphingolipid.

The Psa fimbriae bound to PC and to sphingomyelin in a concentration-dependent manner, although the binding affinity to PC was significantly stronger. The Psa-PC interaction could be completely inhibited by the presence of ChoP or choline, indicating the importance of the chemical structure of choline for the binding domain of the receptor. Psa did not bind to phospholipids that lacked a ChoP or choline moiety, such as PE, PS, or PI. Since some of the fatty acids of phospholipids that are adsorbed on a solid phase are expected to be exposed and accessible to ligands, our results strongly suggest that only the ChoP or choline moiety of PC, and not its fatty acid chains, is involved in the ligand-receptor interaction. Moreover, unlike ChoP monomers that inhibited in the millimolar range, polymerized ChoP, prepared by cross-linking ChoP to polylysine molecules, was inhibitory in the nanomolar range. In contrast the SerP-polylysine was not inhibitory. These results confirmed that the ChoP domain of PC interacts specifically with Psa. They also indicated that the Psa ligand has multiple binding domains for ChoP, in agreement with the homopolymeric structure of the Psa fimbria. Although the affinity of ChoP of a Psa binding moiety might be relatively weak, it is suggested that PC headgroup proximity on host cell membranes (or arrayed on a solid phase) permits multiple simultaneous interactions with the polymeric adhesins of the Psa fimbria. This is in contrast with other well-studied fimbriae, such as the type 1 and P fimbriae of Escherichia coli, which are heteropolymers that carry only a few minor adhesive subunits with higher affinities for their monomeric soluble receptor analogues. The specificity of the PC-Psa interaction was further supported by results on binding experiments of Psa-fimbriated bacteria with PC- and PE-containing liposomes. In contrast to TLC overlay or ELISA binding assays, ligand binding to receptors that are embedded into liposomes more closely models the adhesion to cell membrane receptors. Only Psa-fimbriated E. coli cells, and not nonfimbriated E. coli cells, were aggregated by PC-containing liposomes. The absence of interactions of Psa-fimbriated E. coli cells with the PE-containing liposomes attested again to the specificity of the Psa-PC interaction.

That Psa specifically recognizes certain glycosphingolipids carrying β1-linked galactosyl residues, such as GalCer, as previously described (24), was further confirmed with LacP- and SerP-polylysine polymers, only the former one being inhibitory at micromolar concentrations. The Psa-PC and Psa-GalCer interactions were independent of each other. Galactose and lactose, the latter one used as a monomer or polymer, interfered only with the Psa-GalCer, but not with the Psa-PC interaction. Conversely, monomeric or polymeric ChoP disturbed only Psa binding to PC, but not to GalCer. These results indicate that Psa fimbriae utilize separate domains for binding to PC and GalCer.

Choline and ChoP monomers or polymers inhibited significantly the binding of Psa-fimbriated E. coli cells to A549 and WI-26 VA4 cells. Although galactose and LacP polymers also inhibited bacterial adhesion to the WI-26 VA4 cells, the inhibition was significantly less efficient, suggesting that the bacteria bound to the alveolar epithelial cells mainly by using receptors with the polar choline moiety. Because PC contains choline and is a major component of the outer leaflet of mammalian cell membranes, and because of the strong avidity of the Psa-fimbriated bacteria for membrane-embedded PC, PC is likely the most relevant cellular receptor for Psa. Psa-mediated binding to PC on cell membranes is consistent with our previous findings suggesting that Psa inhibits the intracellular uptake of Y. pestis by binding to a receptor that does not activate any internalization process (19). Although galactose-containing secondary receptors might modulate Psa-mediated bacterial binding, no protein receptor could be detected when total cellular membranes were analyzed by standard ligand blotting assays (40; data not shown). The relative importance of β1-linked galactosyl- and PC-exposing receptors on the surface of the respiratory tract remains to be determined in mammals.

Type II alveolar epithelial cells secrete surfactant, which is a lipoprotein complex found in the fluid lining the alveolar surface of the lungs. Pulmonary surfactant is composed of approximately 90% lipid and 10% protein. The majority of pulmonary surfactant lipids are phospholipids; among them, the most abundant is PC (80%) (35). Thus, it was highly relevant to study potential interactions between Psa and surfactant. Two different types of pulmonary surfactants were investigated: the surfactant contained in rat broncheoalveolar lavage and Survanta, a commercial surfactant extracted from bovine lungs. As expected, considering the high percentage of PC present in pulmonary surfactant, Psa binding and inhibition of binding by ChoP were both concentration dependent for the two surfactant preparations. Since surfactant covers the pulmonary surface, surfactant might act as a protective layer that inhibits the binding of Psa-fimbriated bacteria to all of the pulmonary cells. Although the latter possibility was unlikely, since Psa-fimbriated bacteria bound efficiently to the A549 type II alveolar epithelial cells, which secrete surfactant, addition of various amounts of surfactant to the type I (WI-26 VA4) or type II (A549) cells did not interfere with the adhesive process. In the lungs, surfactant forms a surface monolayer of lipids (air-water interface) and a subphase (aqueous hypophase) consisting of lamellar bodies that unravel into tubular myelin, the source of the lipids composing the surface film (15). During the tidal breathing process, compression and expansion cycles extract or insert most lipids from or into the surface layer of the surfactant film (9). Thus there is a constant cyclical movement that thickens and thins the surfactant film. It is anticipated that bacteria bound to PC in surfactant lipids will participate in these kinetic events, bringing many bacteria into contact with alveolar cell membranes. In addition to this movement of surfactant components during breathing, the alveolar epithelial cells that produce surfactant also recycle surfactant by endocytosis, an additional process that should permit Y. pestis to come into direct contact with the cell membrane. Interestingly, the potential biological relevance of surfactant PC and recycling has also been described for adenoviral infections of alveolar epithelial cells (1).

Since surfactant covers the whole respiratory surface in the lungs, PC might play an important role for the early interactions of Psa-fimbriated Y. pestis with mammalian pulmonary cells after aerosol transmission. Thus, it is tempting to speculate that aerosol sprays containing soluble ChoP polymers might exhibit prophylactic properties by inhibiting Y. pestis contact with respiratory tract epithelial cells. However, whether aerosols and expectorates created by humans or mammals with pneumonic plague carry Psa-fimbriated Y. pestis is not known.

We have previously shown that Y. pestis binds better to three types of human respiratory tract epithelial cells if the bacteria express Psa (19). Thus, in contrast to macrophages, which don't need Psa to interact with Y. pestis (11), Psa-mediated Y. pestis binding to surfactant and respiratory tract epithelial cells might optimize the needed cell contact to activate the T3SS for the intracellular delivery of effector molecules to epithelial cells. The indirect detection of new adhesins in the absence of the pH 6 and F1 antigen also suggested the existence of additional Y. pestis ligands for respiratory tract epithelial cells (19). The relative function of each of these different pathogen-host interactions might be modulated by temporal and spatial variables during the infectious process. Moreover, the role of the alveolar epithelial cells in the early anti-inflammatory process of primary pneumonic plague (4, 16, 23, 31) remains to be determined.