Neisseria meningitidis producing the Opc adhesin binds epithelial cell proteoglycan receptors

Abstract

Neisseria meningitidis possesses a repertoire of surface adhesins that promote bacterial adherence to and entry into mammalian cells. Here, we have identified heparan sulphate proteoglycans as epithelial cell receptors for the meningococcal Opc invasin. Binding studies with radiolabelled heparin and heparin affinity chromatography demonstrated that Opc is a heparin binding protein. Subsequent binding experiments with purified ³⁵SO₄-labelled epithelial cell proteoglycan receptors and infection assays with epithelial cells that had been treated with heparitinase to remove glycosaminoglycans confirmed that Opc-expressing meningococci exploit host cell-surface proteoglycans to gain access to the epithelial cell interior. Unexpectedly, Opa28-producing meningococci lacking Opc also bound proteoglycans. These bacteria also bound CEA receptors in contrast to the Opc-expressing phenotype, suggesting that Opa28 may possess domains with specificity for different receptors. Opa/Opc-negative meningococci did not bind either proteoglycan or CEA receptors. Using a set of genetically defined mutants with different lipopolysaccharide (LPS) and capsular phenotype, we were able to demonstrate that surface sialic acids interfere with the Opc–proteoglycan receptor interaction. This effect may provide the molecular basis for the reported modulatory effect of capsule and LPS on meningococcal adherence to and entry into various cell types.

Introduction

Many bacterial pathogens colonize the human host through interactions between specific bacterial surface adhesins and complementary binding sites at the mammalian cell surface. The identification of host receptors exploited by bacteria is important as their distribution may clarify features, such as host range and tissue tropism of infections. Furthermore, detailed knowledge about the receptor structure may help define domains on the bacterial adhesin that are critical for establishing disease and may provide a basis for the development of novel therapeutic or preventive strategies. Many pathogens, including Neisseria meningitidis, have evolved a diverse array of surface structures to interact with host cells. N. meningitidis possesses two classes of filamentous appendages, called class I and II pili (Potts and Saunders, 1988; Virji et al., 1989; Aho et al., 1997; Nassif et al., 1997), which emanate from the cell surface and facilitate bacterial adherence to human cells in concert with the membrane- and pilus-associated protein, PilC1. This protein facilitates pilus assembly and may be directly involved in the anchoring of the bacteria to the epithelial cell surface (Nassif et al., 1994; Virji et al., 1995; Rahman et al., 1997; Ryll et al., 1997). Pili are considered to be the prime attachment-promoting factor for encapsulated meningococci (Virji et al., 1992a; Nassif et al., 1997). Capsule production, however, is subject to phenotypical variation (Masson and Holbein, 1985; Hammerschmidt, 1996a,b), and non-encapsulated meningococci are frequently isolated from their natural port of entry, the human nasopharynx (Craven et al., 1980; Cartwright et al., 1987; Jones et al., 1992). In the laboratory, this unencapsulated phenotype often does not require pili for bacterial adherence, suggesting that, in the absence of the negatively charged capsule, other surface-located adhesins can interact with the host cells. This event may ultimately result in bacterial-directed uptake by the mammalian cells, cytotoxicity and penetration of the nasopharyngeal epithelial barrier (Stephens et al., 1983; 1991; Rayner et al., 1995; de Vries et al., 1996). Several meningococcal surface adhesins have been identified, including the Opc outer membrane protein and members of the opacity (Opa) outer membrane protein family (Achtman et al., 1988; Aho et al., 1991; Olyhoek et al., 1991). Opc and Opa are 24–35 kDa proteins that share many features, although their corresponding genes and predicted two-dimensional structures are unrelated (Aho et al., 1991; Olyhoek et al., 1991; Merker et al., 1997). Both types of proteins may exist as multimers, confer colonial opacity and have been implicated in facilitating meningococcal adherence and entry into various types of host cells (Achtman et al., 1988; Virji et al., 1992b; 1993; de Vries et al., 1996). Furthermore, both adhesins are variably expressed at the meningococcal surface because of reversible mutations in a repetitive sequence in the promoter (Opc) or coding region (Opa) of the respective genes (Kawula et al., 1988; Sarkari et al., 1994). This surface variation occurs in the human host (Tinsley and Heckels, 1986; Woods and Cannon, 1990; Achtman et al., 1991; Patrick et al., 1993) and may serve as an adaptive mechanism that enables the bacteria to spread to different anatomical niches by creating bacterial phenotypes with different receptor specificities (Virji et al., 1993; de Vries et al., 1996).

Recently, several types of receptors for the meningococcal Opa/Opc adhesins have been identified. Certain Opa proteins, but not Opc, have been demonstrated to bind different members of the carcinoembryonic antigen (CEA or CD66) receptor family (Virji et al., 1996a,b), which are present on endothelial cells, polymorphonuclear cells (PMNCs) and certain epithelia. Opc-producing meningococci, on the other hand, interact with the serum glycoprotein vitronectin and appear to use this molecule to attach to integrins that are present on the apical surface of endothelial cells (Virji et al., 1994). Opc-producing meningococci also promote bacterial entry into certain types of epithelial cells (Virji et al., 1993; de Vries et al., 1996). This event occurs in the absence of vitronectin, and its molecular basis is unknown. In the present study, we set out to identify the epithelial cell receptor for Opc. Following a similar approach that led to the identification of syndecan-like receptors for gonococcal Opa proteins, we present here evidence that cell-surface proteoglycans are a prime receptor for Opc on cultured Chang epithelial cells.

Results

Inhibition of Opc-mediated meningococcal infection of Chang cells by heparin

Infection of Chang conjunctiva epithelial cells with non-encapsulated variants of meningococcus strain B1940siaA producing the Opc protein resulted in a mean number of about 80 adherent and nine intracellular meningococci per epithelial cell at 3 h after infection. Meningococcal variants of the same phenotype except for the Opc protein interacted only poorly with the epithelial cells and were not internalized (Fig. 1A). These data are consistent with the proposed function of Opc as a bacterial adhesin/invasin (Virji et al., 1992b; 1993; de Vries et al., 1996).

Figure 1. . A. Adherence and entry into Chang epithelial cells of N. meningitidis B1940siaA that either lacks Opc (Opc⁻) or produces Opc (Opc⁺) (3 h of infection). The behaviour of the Opc-producing variant was also tested in the presence of heparin (Opc⁺ + Hp, 100 μg ml⁻¹). B. Effect of preincubation of the meningococci (Mn) or Chang cells (Chang) with 100 μg ml⁻¹ heparin for 10 min on B1940siaA/Opc⁺ adherence and entry into Chang cells. Data are the mean ± SEM of four experiments.

In search of a binding site for Opc at the epithelial cell surface, we first added heparin to the infection assay with the idea in mind that heparin may inhibit any interaction involving heparin binding proteins, such as the previously implicated serum glycoprotein vitronectin (Virji et al., 1994) and proteoglycans, which act as receptors for certain gonococcal Opa adhesins (Chen et al., 1995; van Putten and Paul, 1995). Inclusion of heparin in the infection system caused a dose-dependent reduction in bacterial adherence and completely abolished the internalization of strain B1940siaA by the epithelial cells. Maximal inhibition of cellular infection was observed at 10 μg ml⁻¹ heparin (Fig. 1A). Preincubation of either meningococci or the epithelial cells with heparin for 10 min followed by removal of unbound compound before the start of the infection demonstrated that the inhibitory effect was caused primarily by an effect of heparin on the meningococci (Fig. 1B).

Binding of radiolabelled heparin to meningococci expressing Opc protein

Radiolabelled heparin was used to evaluate the negative effect of heparin on the Opc-mediated interaction with Chang epithelial cells. Strain B1940siaA was incubated with tritiated heparin, and the association of label was followed through time. As shown in 2Fig. 2A, [³H]-heparin bound to strain B1940siaA expressing Opc, while binding was virtually abolished in the presence of an excess of unlabelled heparin. Binding was observed at both 37°C and 4°C (data not shown) and reached equilibrium within 10 min of incubation (Fig. 2A). Similar results were obtained for strain H44/76 producing Opc (variant H44/763) (data not shown). No binding was observed for the Opc-negative strain (Fig. 2B). Measurement of [³H]-heparin binding in the presence of increasing amounts of unlabelled heparin resulted in a dose-dependent reduction in bound label. Saturation of binding sites was achieved at a concentration of about 10 μg ml⁻¹ heparin (Fig. 2B). Scatchard analysis suggested non-co-operative binding of heparin to a single class of receptors (Fig. 2C). Calculation of the number of binding sites indicated a binding capacity of approximately 520 fmol of heparin per 5 × 10⁷ bacteria with a rather high receptor affinity (K_d) of 1.4 × 10¹⁰ M⁻¹. Assuming that each heparin binding protein bound one molecule of heparin, this implies the presence of about 6200 heparin binding proteins at the surface of the Opc-producing strain.

Figure 2. . A. Kinetics of [³H]-heparin binding to B1940siaA/Opc⁺ (closed circles). Binding in the presence an 100 μg ml⁻¹ heparin was considered as non-specific (open triangles). B. Displacement curve demonstrating the binding of [³H]-heparin to B1940siaA/Opc⁺ in the presence of increasing concentrations of unlabelled compound (closed circles). No binding was observed for B1940siaA lacking Opc (open circle). C. Scatchard analysis of data shown in B. Bound heparin is expressed in fmol per 5 × 10⁷ bacteria. Data are the mean ± SEM of at least four experiments.

Opc is a heparin-binding protein

In order to ascertain that the Opc protein binds heparin, we isolated outer membranes from the Opa-negative variants H44/763 and 4, which differed only in Opc expression. Membrane proteins were solubilized using 5% Zwittergent Z3,14, and the mixture was passed over a heparin–Affigel matrix. After extensive washing of the column with buffer containing 0.1 M NaCl, proteins that were bound to the heparin matrix were eluted with a stepwise salt gradient (0.2–2.0 M NaCl). Eluted material was collected, precipitated and analysed by SDS–PAGE and Western blotting. Silver staining of gels in conjunction with immunoblotting with the Opc-specific antibody B306 demonstrated single elution of the Opc protein at 0.2 M NaCl, indicating that Opc is a heparin binding protein (Fig. 3). At higher salt concentrations (0.4 M NaCl), several other outer membrane proteins were eluted (Fig. 3). These proteins were also isolated from the Opc-negative strain, which was unable to bind heparin, suggesting that they did not contribute significantly to heparin binding in the native bacteria.

Figure 3. . Heparin affinity chromatography of solubilized outer membranes of strains H44/763 (Opc⁺) and 4 (Opc⁻). Membranes were passed over a heparin matrix, and eluted fractions were analysed by SDS–PAGE and silver staining (A) and for the presence of Opc by Western blotting with the Opc-specific antibody B306 (B). The membranes showed no detectable levels of Opa proteins as estimated from their immunoreactivity with the antibody 4B12/CII (data not shown; see Fig. 5). OM, outer membrane fraction; 0.2 M NaCl, elution with buffer plus 0.2 M NaCl; 0.4 M NaCl, elution with buffer plus 0.4 M NaCl; 2.0 M NaCl, elution with buffer plus 2 M NaCl. Molecular weight markers are in kilodaltons (kDa).

Host cell-surface proteoglycans are receptors for the Opc⁺ meningococcal phenotype

The structural relatedness of heparin and heparin sulphate led us to hypothesize that epithelial cell heparan sulphate proteoglycans may also be recognized by Opc and that these molecules may serve as host receptors for Opc-expressing meningococci, anchoring the bacteria to the cell surface. We tested this concept by preincubating Chang epithelial cells with the enzyme heparitinase, which specifically cleaves heparan sulphate moieties before infection. This treatment has previously been shown to remove glycosaminoglycans effectively, keeping the phagocytic properties of the epithelial cells intact (van Putten and Paul, 1995). Heparitinase treatment (75 mU ml⁻¹ for 2 h) greatly reduced meningococcal adherence and decreased (strain B1940) or even abolished (strain H44/76) bacterial uptake by the mammalian cells (Fig. 4A). To substantiate that cell-surface heparan sulphate proteoglycans served as receptors for Opc-expressing meningococci further, we labelled Chang cells with ³⁵SO₄ for 24 h and isolated the soluble extracellular domain of heparan sulphate proteoglycans (ectodomain; van Putten and Paul, 1995). Binding assays with meningococcal variants that differed in their expression of Opc showed specific binding of the purified soluble receptor to Opc-positive bacteria only (Fig. 4B). Binding was completely abolished by the addition of an excess (10 μg ml⁻¹) of heparan sulphate and was only observed for variants that expressed Opc. Together, these data strongly suggest that meningococci producing Opc exploit the heparan sulphate moiety on proteoglycan receptors to adhere to and enter Chang epithelial cells.

Figure 4. . A. Effect of heparitinase treatment of Chang cells on Opc-mediated adherence and entry into Chang cells of strains B1940siaA and H44/763. B. Binding of ³⁵SO₄-labelled extracellular proteoglycan receptor domain (HSPG) derived from Chang cells to Opc-producing meningococci was measured in the absence (hatched bars) and in the presence (filled bars) of 10 μg ml⁻¹ heparan sulphate. Data are the means ± SEM of five experiments.

Relationship between proteoglycan receptor binding and CEA receptor recognition

We demonstrated previously for meningococcus strain H44/76 that the expression of Opc promoted entry into Chang cells, while the presence of Opa28 was primarily associated with meningococcal entry of primary nasopharyngeal cells (de Vries et al., 1996). To assess whether this cell tropism was based on different receptor specificities of the adhesins involved, we tested both bacterial phenotypes for binding of proteoglycan receptors and of CEA receptors previously implicated as receptors for meningococccal Opa proteins (Virji et al., 1996a). Binding of CEA was measured by incubating meningococci with soluble CEA (CD66e) for 10 min, followed by removal of unbound receptor and detection of the bound molecules by SDS–PAGE and Western blotting of whole-cell lysates with horseradish peroxidase (HRP)-conjugated anti-CEA antibody. Soluble CEA bound only to the Opa-producing variant and not to Opc-producing meningococci (Fig. 5, left), in agreement with the data of Virji et al. (1996a). Proteoglycan binding assays, however, showed that both the Opc-producing and the Opa28-positive meningococci were able to bind this class of receptors (Fig. 5, right). These data indicate that, while Opc specifically recognizes proteoglycan receptors, Opa28-producing meningococci may exploit both CEA and proteoglycans, dependent on receptor availability. The binding of heparan sulphate proteoglycan receptors by Opa28-producing meningococci may explain their reported (de Vries et al., 1996) adherence and low level of bacterial entry into Chang cells, which lack CEA cellular receptors (J. P. M. van Putten, unpublished observation).

Figure 5. . Receptor specificities of meningo- coccal Opa28 and Opc. Meningococcal strains H44/761 producing Opa28 (lane 1), H44/763 producing Opc (lane 2) and H44/764, which lacks both adhesins (lane 3), were incubated with pure CEA, lysed and subjected to SDS–PAGE and Western blotting. Bound CEA, Opa28 and Opc were detected with either CEA receptor (αCEA), Opa- (αOpa) or Opc- (αOpc) specific antibodies and HRP-conjugated secondary antibodies in the ECL protocol (left). The same strains were also incubated with purified ³⁵SO₄-labelled extracellular fragment of proteoglycans (HSPG), and receptor binding activity was determined by measurement of bound radioactivity(right). Note that H441 binds both proteoglycans and CEA, in contrast to H443, which recognizes only HSPG. H444, lacking both Opa and Opc, bound neither of the receptors.

Modulation of proteoglycan receptor binding by capsule and lipopolysaccharide (LPS) sialylation

Capsule expression and incorporation of sialic acids into lipopolysaccharides (LPS) by meningococci have previously been reported to modulate Opc-associated adherence to epithelial cells (Virji et al., 1992b; Hammerschmidt et al., 1994). In the light of the current findings, we sought the molecular basis for this event by testing the effects of capsule production and LPS sialylation on Opc-mediated proteoglycan receptor binding. In these experiments, we used a series of recombinant serogroup B meningococcal strains with defined defects in capsule and/or LPS biosynthesis genes. The construction and phenotypical characteristics of these strains have been described (Table 1). Infection experiments with Chang epithelial cells using Opc-producing variants of the different recombinants (Fig. 6A) showed that strains B1940 wt (capsulation⁺, LPS sialylation⁺), B1940galE (capsulation⁺, LPS sialylation⁻) and B1940siaD (capsulation⁻, LPS sialylation⁺) were poorly internalized by the epithelial cells compared with strains lacking surface sialic acids (B1940siaA and B1940cps) (Fig. 6B), consistent with previous observations (Hammerschmidt et al., 1996b). Proteoglycan receptor-binding assays with the same recombinants showed that the presence of sialylated LPS (wt, siaD ) dramatically reduced the ability of the bacteria to bind soluble proteoglycan receptor (Fig. 6C). This is consistent with the notion that surface sialic acids block Opc-mediated interactions with epithelial cells by preventing binding of this adhesin to cellular proteoglycan receptor. The encapsulated galE mutant, however, did bind soluble receptor, although it interacted sparsely with the epithelial cells (Fig. 6C). Thus, capsule may exert its anti-phagocytic effect by preventing contact with receptors present at the negatively charged epithelial cell surface rather than by interference with the receptor binding properties of Opc.

Table 1.  . Characteristics of bacterial strains used in this study. Mutant B1940siaA was derived from the wild-type strain B1940 and has an insertional activation (IS1301 ) of the siaA gene; B1940siaD was constructed by inactivation of the siaD gene with Tn1725, which confers resistance to chloramphenicol; B1940cps was obtained by complete deletion of the cps locus, including the galE gene, and insertion of an erythromycin resistance gene; B1940galE was established by replacing the galE and rfb genes in the region D with a chloramphenicol resistance cassette; H44/761, 3 and 4 are natural variants of the strain H44/76.

Figure 6. . Effect of capsule and LPS phenotype on Opc function. Genetically defined capsule and LPS variants (see Table 1) producing equal amounts of Opc as determined by Western blotting (A) were tested for their ability to adhere and enter Chang epithelial cells (B) and to bind proteoglycan (HSPG) receptors (C). Results in B and C are the mean ± SEM of 8–12 experiments.

Discussion

A typical feature of the pathogenic Neisseria species is their extensive surface variation that creates a vast repertoire of different bacterial phenotypes. This plasticity probably reflects optimal bacterial adaptation to the exclusively human host and may provide the potential to colonize the human nasopharynx and, in the absence of appropriate immune defence, to disseminate to other tissues. The relationship between surface variation and spread to different infection niches is illustrated by the dramatic changes in pili, LPS, capsule, Opa and Opc production that have been observed during the various stages of meningococcal disease (Craven et al., 1980; Tinsley and Heckels, 1986; Cartwright et al., 1987; Woods and Cannon, 1990; Achtman et al., 1991; Jones et al., 1992; Patrick et al., 1993). A major challenge is to relate these changes in surface composition to the molecular events that lead to the establishment of infection. The present work serves this aim by unravelling the nature of an epithelial cell receptor for the meningococcal Opc adhesin and by providing a molecular basis for the reported negative modulatory effect of capsule and LPS on certain Opc functions.

Our data indicate that heparan sulphate proteoglycans are prime epithelial receptors for the meningococcal Opc adhesin. Experimental evidence that is provided includes (i) Opc-positive meningococci bind heparin in a specific and saturable fashion; (ii) Opc is retained on a heparin–affinity matrix; (iii) Opc-producing strains bind radiolabelled purified epithelial proteoglycan receptors; and (iv) enzymatic digestion of the heparan sulphate moieties of glycosaminoglycans destroys the Opc-mediated interaction with Chang epithelial cells. Opc has previously been reported to facilitate meningococcal adherence to umbilical vein endothelial cells through binding of the serum glycoprotein vitronectin, which in turn anchors the bacteria to integrins present at the apical endothelial cell surface (Virji et al., 1994). Vitronectin appears not to be required for Opc-mediated entry into Chang epithelial cells, as our experiments were performed in the absence of serum or vitronectin. Thus, multiple receptors may exist for Opc. At this point, however, caution is needed, inasmuch as, in Chinese hamster ovary cells, gonococci producing OpaA use cellular proteoglycans for bacterial adherence but require vitronectin to complete the bacterial entry process (Duensing and van Putten, 1997). Thus, in certain cell types, proteoglycan and vitronectin-dependent bacterial uptake pathways may co-operate. To our knowledge, a possible involvement of proteoglycan receptors in the reported vitronectin-mediated interaction of Opc-producing meningococci with endothelial cells has not been investigated.

The identification of Opc as a protein that binds heparin and heparan sulphate proteoglycan receptors classifies this molecule in a group of bacterial adhesins (for review, see Rostand and Esko, 1997) that includes distinct members of the gonococcal Opa protein family (Chen et al., 1995; van Putten and Paul, 1995), Bordetella pertussis fimbriae (Geuijen et al., 1996) and filamentous haemagglutinin (FHA; Hannah et al., 1994), the ActA protein of Listeria monocytogenes (Alvarez-Dominguez et al., 1997) and the major outer membrane protein (MOMP) of Chlamydia species (Su et al., 1996). The exact nature of the proteoglycan receptors recognized by these adhesins/invasins and the signalling pathways that lead to uptake of these bacterial pathogens are unknown, although gonococcal MS11-OpaA have been demonstrated to bind to syndecan-like molecules present on various types of epithelial cells (van Putten and Paul, 1995). This bacterium enters this cell type via a bacteria-directed phagocytosis-like process that requires tyrosine kinase activity (Meyer et al., 1994) and is accompanied by a transient recruitment of F-actin at the sites of bacterial entry (Grassméet al., 1996). Considering the functional similarities between gonococcal OpaA and meningococcal Opc, including the ability to facilitate bacterial entry into Chang epithelial cells, it can be seen that a similar mechanism is exploited by Opc-producing meningococci.

The domain on Opc that facilitates interaction with proteoglycan receptors remains to be defined. Typically, binding of glycosaminoglycans involves an array of basic amino acids interspaced by hydropathic residues (Cardin and Weintraub, 1989). Recently, a structural model for the Opc outer membrane protein has been proposed, and the protein has been predicted to consist of 10 transmembrane segments and five surface-exposed loops. Helical wheel diagrams of the various predicted loops indicate regions of high-positive charge density in the surface-exposed loops 1, 2 and 3 (data not shown). Loop 2 has been proposed previously to be involved in the binding of vitronectin, as monoclonal antibodies that map to this loop block vitronectin-mediated interactions of Opc-producing meningococci with endothelial cells (Virji et al., 1992b; 1994; Merker et al., 1997). As both vitronectin and Opc are heparin binding proteins, it is tempting to speculate that the binding of vitronectin involves the formation of a Opc–heparin–vitronectin complex. This would locate the heparin binding domain on Opc in close approximation to the binding site for vitronectin. Alternatively, Opc may carry independent binding domains for heparin and vitronectin. The recently described procedure to manipulate the amino acid sequence of individual surface-exposed loops of Opc genetically (Merker et al., 1997) may provide a valuable tool to discriminate between these possibilities and to define the heparin binding domain.

The ability of meningococci to bind proteoglycan receptors was not restricted to the Opc-producing bacterial phenotype. Similar binding properties were observed for meningococci of strain H44/76 producing Opa28 (and no Opc). This finding is of particular interest, because Opa28 has been associated with meningococcal entry into human nasopharyngeal cells, the target tissue at the port of entry (de Vries et al., 1996). Meningococcal Opa proteins have previously been reported to exploit members of the CEA receptor family that are present on endothelial cells, PMNCs and certain epithelia (Virji et al., 1996a,b). Our data confirm that Opa-producing meningococci recognize CEA cellular receptors but, in addition, demonstrate that Opa can bind proteoglycan receptors. This finding may explain the observed interactions between Opa-producing bacteria and Chang cells (Virji et al., 1993; de Vries et al., 1996), which do not carry CEA receptors (unpublished observations). Theoretically, the recognition of different classes of receptors may provide the Opa⁺ bacterial phenotype with more flexibility with respect to cell tropism by enabling interactions with cells that express proteoglycan receptors, CEA receptors or both. The possible presence of two binding domains with different receptor specificity on a single bacterial adhesin is not unique, as gonococcal MS11-OpaC has similar characteristics (Bos et al., 1997; Chen et al., 1997). Thus, while meningococcal Opc acts functionally like OpaA in gonococcus strain MS11, Opa28 of strain H44/76 may be the meningococcal homologue of MS11-OpaC.

Several studies suggest that Opc-mediated interactions with epithelial cells are negatively modulated by capsule- and/or LPS-associated sialic acids at the bacterial surface (Virji et al., 1992b; Hammerschmidt et al., 1994; de Vries et al., 1996). The availability of purified soluble proteoglycan receptor derived from Chang epithelial cells gave the opportunity to explore the molecular bases for these effects. By using a set of genetically defined serogroup B meningococcal mutants defective in capsule, LPS or sialic acid biosynthesis, we clearly demonstrated that the incorporation of sialic acid into the glycose moiety of the LPS prevented appropriate binding of the Opc adhesin to the proteoglycan receptor. This lack of receptor binding activity correlated well with the inability of these sialylated bacterial phenotypes to interact with mucosal cells. The finding that the encapsulated galE mutant lacking the acceptor site for sialic acid on its LPS still bound soluble proteoglycan receptor but was unable to interact effectively with the same receptors at the epithelial cell surface may indicate that capsule prevents Opc-mediated adherence by steric hindrance rather than by direct interference with Opc function. Thus, capsule- and LPS-associated sialic acids may modulate Opc-mediated bacterial invasiveness through different mechanisms. These observations together with the in vitro and in vivo data that capsule rather than sialylated LPS contributes to the resistance of meningococci to killing by antibodies and complement (Hammerschmidt et al., 1994; Vogel et al., 1996) fit the scenario that a fine tuning of the sialylation status of serogroup B meningococci plays a key role in the pathogenesis of meningococcal disease by creating bacterial phenotypes optimally adapted to survive in the various infection niches.

Experimental procedures

Bacteria and cell lines

The origin, genotype and phenotypical characteristics of the N. meningitidis strains B1940 and H44/76 and their derivatives have been described previously (Frosch et al., 1990; Hammerschmidt et al., 1994; de Vries et al., 1996; Hammerschmidt et al., 1996a) and are summarized in Table 1. Bacteria were grown routinely on GC agar base [composition per litre: 3.75 g of trypticase peptone (BBL), 7.5 g of thiotone E (BBL), 4 g of K₂HPO₄, 1 g of KH₂PO₄, 5 g of NaCl, 1 g of soluble starch (BBL), 1% bacto-agar (Difco), 1% IsoVitaleX (Difco), pH 7.20] at 37°C in 5% CO₂. Opc and Opa expression were monitored by SDS–PAGE and Western blotting (Weel et al., 1991) with the specific monoclonal antibodies B306 and 4B12 respectively (generously provided by M. Achtman). Opc variants produced large amounts of Opc, unless indicated otherwise. Capsule and LPS phenotypes were determined as described previously (van Putten, 1993; Hammerschmidt et al., 1994; de Vries et al., 1996). For use in receptor binding assays and infection experiments, bacteria were grown either on agarose-based GC medium or in 10 ml of HEPES medium [composition per litre: 10 mM HEPES, 145 mM NaCl, 5 mM KCl, 5 mM glucose, 1 mM CaCl₂, 1 mM MgCl₂, 15 g l⁻¹ proteose peptone no. 3 (Difco) and 1% IsoVitaleX, pH 7.4] in 50 ml polypropylene tubes in a gyratory waterbath shaker (250 r.p.m. at 37°C for 2 h). Chang conjunctiva epithelial cells (ATCC CCL20.2) and Chinese hamster ovary (CHO-K1) cells (ATCC CCL61) were grown in RPMI-1640 plus 5% fetal bovine serum (FBS) (tissue culture medium; Life Technologies) in 25 cm² flasks (Corning), unless indicated otherwise.

Infection experiments

For use in infection experiments, epithelial cells were grown onto 12 mm circular glass coverslips in a 24-well plate (1 ml of tissue culture medium per well) for 48 h at 37°C in 5% CO₂. Before the addition of bacteria, the medium was replaced by 1 ml of either serum-free Dulbecco's modified Eagle medium (DMEM; Life Technologies — CHO cells) or HEPES buffer (composition as HEPES medium but without proteose peptone no. 3 and with 0.2% IsoVitaleX — Chang cells). Bacteria were added at a bacteria to host cell ratio of 50:1, and cells were incubated at 37°C for 3 h, unless indicated otherwise. In some experiments, bacteria or host cells were preincubated in HEPES buffer without (control) or with heparin (100 μg ml⁻¹ heparin 171 U mg⁻¹ sodium; ICN) for 10 min, followed by three washes with HEPES buffer. When heparin or heparan sulphate (heparan sulphate IV, a generous gift from Celsus Laboratories) were included in the infection assay, these compounds were added just before the addition of the bacteria. Heparitinase (heparinase III; Sigma) treatment was essentially carried out as described previously (van Putten and Paul, 1995). Infection was stopped by washing the monolayers three times with Dulbecco's phosphate-buffered saline (DPBS; composition per litre: 140 mM NaCl, 2.5 mM KCl, 8.1 mM Na₂HPO₄, 1.5 mM KH₂PO₄, 1 mM CaCl₂, 1 mM MgCl₂, pH 7.4) and fixed in DPBS plus 0.1% glutaraldehyde/1% paraformaldehyde. Bacterial adherence and entry were scored by microscopy after immunogold–silver staining and/or crystal violet staining of the specimen (van Putten et al., 1994; van Putten and Paul, 1995; de Vries et al., 1996). The validity of this procedure and its advantage over the gentamicin assay have been described previously (van Putten et al., 1990; 1994; van Putten, 1991; van Putten and Paul, 1995). Results are expressed as the mean number of adherent and intracellular bacteria per epithelial cell.

Heparin affinity chromatography

Bacterial heparin-binding proteins were isolated as described previously (van Putten and Paul, 1995) with minor modifications. In brief, meningococcal outer membranes were prepared by the lithium acetate method (Pannekoek et al., 1992), treated with DNase (2 U for 1 h at 37°C) and solubilized in 5% Zwittergent Z3,14 in 1 ml of buffer (50 mM Tris-HCl, 10 mM EDTA, 0.1 M NaCl, pH 8.0) for 1 h at 37°C. After clearance by centrifugation (12 000 g for 10 min), this solution (containing 100 μg of protein) was mixed (end-over-end rotation) with 1 ml of heparin–Affigel (Bio-Rad) pre-equilibrated in the same buffer containing 0.5% Z3,14 for 1 h at 20°C. The slurry was then packed into a column (10 × 1 cm), and the column was washed with 10 ml of buffer containing 0.5% Z3,14. Bound material was eluted with the same buffer in a stepwise salt gradient (0.1–2.0 M NaCl). Fractions (1 ml) were collected, precipitated in 80% ethanol (16 h at −20°C) and resuspended in 40 μl of distilled water. Each fraction (1–5 μl) was analysed by SDS–PAGE and Western blotting. Proteins eluted in the second or third fraction of each elution step. These fractions are shown in Fig. 3. Opc was detected with the monoclonal antibody B306 and HRP-conjugated goat anti-mouse anti-IgG in the enhanced chemiluminescence (ECL) protocol (Amersham).

Receptor binding assays

Binding of [³H]-heparin (NEN-Dupont) was measured by incubating 5 × 10⁷ bacteria in 150 μl of HEPES buffer in the presence of 45 nCi of [³H]-heparin (0.7 mCi ml⁻¹). Approximately 10% of the labelled heparin was found to be biologically active at the time of the assays, resulting in a final concentration of radiolabel of 0.042 μg ml⁻¹. Binding assays were performed for 10 min at 4°C, unless indicated otherwise. The reaction was stopped by centrifugation (12 000 × g for 2 min at 20°C), and the pellets were washed twice with 150 μl of HEPES buffer. Scintillation fluid (Biosafe II; RPI) was added, and bacteria-associated radioactivity was measured in a Beckman scintillation counter model LS 3800. The same procedure was followed for binding of the radiolabelled extracellular fragment of heparan sulphate proteoglycans. Metabolic labelling of Chang epithelial cells and isolation of the ³⁵SO₄-labelled extracellular proteoglycan fragment (ectodomain) was carried out as described previously (van Putten and Paul, 1995). Non-specific binding was defined as radioactivity bound in the presence of 100 μg ml⁻¹ unlabelled heparin. For Scatchard analysis, non-specific binding was subtracted from specific binding, and heparin was assumed to have a mean molecular weight of 6000 (measured by SDS–PAGE). Linear regression analysis was used for curve fitting, with the x-intercept representing the total number of binding sites and the −1/slope giving the K_d. Data were analysed with the program GRAPHPAD PRISM 2.0. CEA receptor binding was determined by incubating 5 × 10⁷ bacteria with 2 μg of purified CEA (Zymed Laboratories) in 250 μl of HEPES buffer for 10 min at 20°C, followed by the removal of unbound receptor by centrifugation (12 000 × g for 2 min at 20°C). The bacteria were washed twice with 150 μl of HEPES buffer, transferred to fresh tubes and solubilized by boiling (10 min at 96°C) in 20 μl of SDS–PAGE sample solution. Whole-cell lysates were separated by SDS–PAGE (8–15% gradient gels) and transferred onto nitrocellulose. Bacteria-associated CEA was detected with a HRP-conjugated polyclonal rabbit anti-CEA antibody (1:100 dilution; Dako) at a in the ECL protocol.

Acknowledgements

We thank Dr John Swanson for critical reading of the manuscript.