Neisseria meningitidis NadA is a new invasin which promotes bacterial adhesion to and penetration into human epithelial cells


E-mail; Tel. (+39) 0577 243414; Fax (+39) 0577 278508.


Neisseria  meningitidis  is  a  human  pathogen,  which is a major cause of sepsis and meningitis. The bacterium colonizes the upper respiratory tract of approximately 10% of humans where it lives as a commensal. On rare occasions, it crosses the epithelium and reaches the bloodstream causing sepsis. From the bloodstream it translocates the blood–brain barrier, causing meningitis. Although all strains have the potential to cause disease, a subset of them, which belongs to hypervirulent lineages, causes disease more frequently than others. Recently, we described NadA, a novel antigen of N. meningitidis, present in three of the four known hypervirulent lineages. Here we show that NadA is a novel bacterial invasin which, when expressed on the surface of Escherichia coli, promotes adhesion to and invasion into Chang epithelial cells. Deletion of the N-terminal globular domain of recombinant NadA or pronase treatment of human cells abrogated the adhesive phenotype. A hypervirulent strain of N. meningitidis where the nad A gene was inactivated had a reduced ability to adhere to and invade into epithelial cells in vitro. NadA is likely to improve the fitness of N. meningitidis contributing to the increased virulence of strains that belong to the hypervirulent lineages.


Neisseria meningitidis is a capsulate Gram-negative bacterium responsible for meningitis and sepsis, two devastating diseases that can kill children and young adults. N. meningitidis colonizes the nasopharynx of 3–30% of healthy individuals and for reasons that are not yet understood, in people lacking bactericidal antibodies, the colonization can develop into invasive diseases (Goldschneider et al., 1969a,b).

To cause disease, the bacterium needs to traverse the mucosal barrier and enter into the bloodstream, causing septicaemia; here it can interact with endothelial cells of brain microvessels and cross the blood–brain barrier, resulting in fulminating meningitis. Colonization of the epithelium is the first step in the multistage adhesion cascade, followed by invasion of the cell, intracellular persistence and transcytosis. These events are modulated by the interaction of virulence factors with their host cell receptors, and signals are sent from pathogen to host and host to pathogen in the adhesion cascade (Merz and So, 2000).

The factors known to play a role in adhesion of meningococcus are type IV pili, PilC, a pilus-associated protein that binds to CD46 human receptor (Virji et al., 1991; Nassif et al., 1994; Kallstrom et al., 1997; Taha et al., 1998; Scheuerpflug et al., 1999) and App (Adhesion and penetration protein), an autotransporter protein highly homologous to Hap protein of Haemophilus influenzae (Hadi et al., 2001; Serruto et al., 2003). The other major class of adhesins are represented by the Opa and the Opc proteins (Simon and Rest, 1992; Virji et al., 1993), which mediate the adhesion and invasion of non-capsulate meningococci to both endothelial and epithelial cells by interacting with CD66 and heparan sulphate proteoglycan receptors (Virji et al., 1996; de Vries et al., 1998). To describe the adhesion of meningococcus to epithelial cells, Nassif et al. (1999) have proposed a two-step model. Initially, bacteria attach to host cells via pili, forming bacterial clumps at the apical side of cell monolayer. The second step corresponds to more diffuse adherence, with loss of piliation and subsequent tight adhesion (Pujol et al., 1999; Deghmane et al., 2000), which can be followed by internalization of bacteria into host cells. Entry into epithelial cells may protect bacteria against host clearance mechanisms and/or may play an important role in traversal of the epithelial barrier for subsequent spreading. However, although many bacterial and host factors have been identified, several aspects of meningococcal pathogenesis remain unknown.

Recently, we described NadA (Neisseria adhesin A), a novel antigen proposed as a vaccine candidate against N. meningitidis serogroup B (Comanducci et al., 2002). The nadA gene is present in approximately 50% of N. meningitidis strains comprising hypervirulent lineages ET37, ET5 and cluster A4 and is absent in the hypervirulent lineage III, in Neisseria gonorrhoeae and in commensal Neisseriae. The gene is well conserved and it clusters into three well-defined alleles. The NadA protein forms surface-exposed high-molecular-weight oligomers that are stable in SDS-PAGE. Furthermore, the recombinant purified protein induces a bactericidal antibody response and binds to epithelial cells in vitro. On the basis of the analysis of the amino acid sequence, NadA presents a tripartite structural organization with an N-terminal globular domain (‘head’ domain), an intermediate α-helix region with high propensity to form coiled-coil structures (coiled-coil ‘stalk’) and a conserved C-terminal membrane anchor domain (Comanducci et al., 2002). The predicted molecular structure, the ability of forming oligomers and interacting with host cells in vitro suggest that NadA belongs to the novel class of non-fimbrial adhesins that includes the pathogenicity factors YadA and UspAs, recently named ‘Oca’ family (Oligomeric coiled-coil adhesin) and characterized by oligomerization associated with the conserved C-terminal membrane anchor domain. Furthermore, for these proteins the binding domains mediating adhesion are probably located within the N-terminal moiety (Cornelis et al., 1998; Hoiczyk et al., 2000; Lafontaine et al., 2000; Roggenkamp et al., 2003; Desvaux et al., 2004).

In this study we expressed NadA in Escherichia coli to investigate its biological function. We found that NadA is exported to the surface, assembled in oligomers anchored to the outer membrane, and confers to E. coli the ability to adhere and invade the epithelial cells. The ‘head’ domain of the protein is the region responsible of the binding activity. A role of NadA in adhesion and invasion is also shown in N. meningitidis.


Expression and localization of NadA in E. coli

To study the biological role of NadA, different forms of the protein (Fig. 1A) were expressed in E. coli using the pET expression vector and the BL21(DE3) strain as expression host. The recombinant strains E. coli-NadA, E. coli-NadAΔ1-23 and E. coli-NadAΔ351-405 were generated and the proteins were expressed with or without addition of isopropyl-1-thio-β- d-galactopyranoside (IPTG) (induced and uninduced condition respectively). Western blot analysis of whole-cell lysate from E. coli-NadA, grown under induced conditions, showed two major bands of about 40 and 42.5 kDa, corresponding to NadA mature protein, after the cleavage of the signal peptide, and to the unprocessed form respectively. In addition, two high-molecular-weight bands were observed (Fig. 1B, lane 2). These latter bands were the only bands present when bacteria were grown under uninduced conditions (Fig. 1B, lane 3). The higher-molecular-weight band migrates to the same position as oligomeric NadA detected in meningococcus (Fig. 1B, lane 1). Western blot analysis of E. coli-NadAΔ1-23 and E. coli-NadAΔ351-405 total-cell lysates showed bands of 40 and 35 kDa, respectively (Fig. 1B, lanes 4 and 5), corresponding to the expected molecular weight of each monomeric protein. No reactive bands were observed in E. coli-pET, which contains the vector alone, used as negative control (Fig. 1B, lane 6).

Figure 1.

A. Schematic representation of the various NadA forms used in this study. The leader peptide (LP), the head domain, the coiled-coil region and the membrane anchor domain are shown.
B. Western blot analysis of the expression of NadA in E. coli. Total-cell lysate of: meningococcus strain 2996 (positive control) (lane 1); E. coli-NadA (induced condition) (lane 2); E. coli-NadA (uninduced condition) (lane 3); E. coli-NadAΔ1-23 (induced condition) (lane 4); E. coli-NadAΔ351-405 (induced condition) (lane 5); E. coli-pET (negative control) (lane 6); E. coli-NadAΔ30-87 (lane 7); E. coli-NadAΔ24-42 (lane 8); E. coli-NadAΔ43-70 (lane 9); E. coli-NadAΔ71-87 (lane 10). The assay was performed using a rabbit polyclonal serum against NadAΔ351-405 and a secondary peroxidase-conjugate antibody.
C. FACS analysis on whole-cell bacteria performed using an anti-NadAΔ351-405 antibody. Grey profile indicates the control strain E. coli-pET.

These results show that only the construct expressing full-length NadA produces the high-molecular-weight forms of the protein that are heat-stable in SDS gel. Furthermore, only in bacteria grown under uninduced condition, NadA was totally processed and assembled in oligomers.

To investigate whether these proteins were surface-exposed, we performed FACS analysis on whole-cell bacteria. E. coli-NadA, grown in the absence of IPTG, was stained by an anti-NadA antibody and a peak of fluorescence, which is absent in E. coli-pET, was observed (Fig. 1C). In contrast, FACS analysis on E. coli-NadAΔ1-23 and E. coli-NadAΔ351-405 showed that NadA lacking the leader peptide or the predicted anchor domain is not surface-exposed (Fig. 1C).

Surface localization and oligomerization of full-length NadA were also confirmed by Western blot analysis of outer membrane proteins. As shown in Fig. 2A, in the outer membrane fraction of E. coli-NadA two high-molecular-weight bands corresponding to NadA oligomers and a weak band corresponding to NadA monomer are present. These bands are absent in the outer membrane from the control strain E. coli-pET.  We conclude that in E. coli, as in meningococcus, full-length NadA is efficiently expressed, transported to the outer membrane and assembled in very stable oligomers.

Figure 2.

Localization of NadA in E. coli.
A. Western blot analysis of outer membrane proteins from E. coli-NadA (lane 1) and E. coli-pET (lane 2).
B. SDS-PAGE analysis of periplasmic and supernatant fractions from E. coli-NadAΔ351-405 (lanes 1 and 3) and E. coli-pET (lanes 2 and 4 respectively). The 35 kDa band indicated by an arrow was confirmed by Western blotting (not shown). Western blots were performed using a serum against NadAΔ351-405.

Finally, we examined the cellular localization of NadAΔ351-405. Periplasmic and supernatant fractions from E. coli-NadAΔ351-405 were prepared and analysed by SDS-PAGE. As shown in Fig. 2B a band of 35 kDa, corresponding to the monomeric protein, was found in both the periplasm and the supernatant (lanes 1 and 3 respectively). This band was absent in the control strain (lanes 2 and 4). However, biochemical analysis of purified NadAΔ351-405 showed that the protein is present in solution as a trimer, which is not stable to denaturating conditions (S. Savino, personal communication). We conclude that in E. coli, NadAΔ351-405 is translocated into the periplasmic space and released into the culture supernatant as a trimeric protein, suggesting that the C-terminal 351–405 region is responsible for anchoring to the outer membrane and stable oligomerization.

NadA mediates adhesion of E. coli to human epithelial cells

We have previously shown that the purified NadAΔ351-405 binds to Chang conjunctival epithelial cells in vitro in a dose-dependent manner (Comanducci et al., 2002). We examined the ability of NadA to promote in vitro adherence of E. coli to human epithelial cells. Chang cell monolayers were infected with uninduced E. coli-NadA or E. coli-pET and the number of cell-associated colony-forming units (cfu) was calculated by plating bacteria. The results of one representative experiment (Fig. 3) showed that E. coli-NadA had 30-fold higher adhesion to Chang cells than E. coli-pET strain. Adhesion of E. coli-NadA to Chang cells was also confirmed using immunofluorescence confocal microscopy. After incubation with E. coli-NadA or with E. coli-pET, Chang cells were fixed and bacteria were detected using an anti-E. coli serum. E. coli expressing NadA were found to adhere to cultured Chang epithelial cells (Fig. 4A), while no adhesion was observed with the control strain (Fig. 4B). The same experiments performed with cultured HUVECs (human umbilical vein endothelial cells) gave negative results (data not shown). These data indicate that NadA is able to promote adhesion of E. coli to human epithelial cells but not to HUVECs.

Figure 3.

Adhesion to epithelial cells of E. coli expressing different forms of NadA. Chang monolayers were infected with approximately 3 × 107 bacteria (moi 100) for 3 h at 37°C. Adherence was expressed as cfu per well. Values represent the mean ± standard deviation of a representative experiment performed in triplicate.

Figure 4.

Immunofluorescence microscopy showing adhesion to and invasion into epithelial cells of E. coli expressing NadA.
A and B. Chang monolayer was infected with E. coli-NadA (A) or with E. coli-pET (B). Bacteria were detected using a primary anti-E. coli serum and a secondary antibody, Alexa Fluor 488, and viewed using confocal laser scanner microscopy.
C–E. The same area of Chang monolayer infected with E. coli-NadA. Extracellular bacteria were stained with an anti-E. coli serum and a secondary antibody, Alexa Fluor 488 (green, C). After cell's permeabilization, total bacteria were labelled using an anti-E. coli serum and a secondary antibody, Alexa Fluor 586 (red, D). The overlay (E) of images (C) and (D) showed extracellular bacteria stained in yellow (arrows) and intracellular bacteria stained in red (thin arrows).

NadA mediates invasion of E. coli to human epithelial cells

To investigate whether NadA mediates the internalization of E. coli into epithelial cells, a Chang monolayer was incubated with E. coli-NadA and invasion analysed using the double immunofluorescence technique (Heesemann and Laufs, 1985). The results are illustrated in Fig. 4. The extracellular bacteria were visualized with green fluorescence (Fig. 4C), while the total bacteria were visualized with red fluorescence (Fig. 4D). The image generated by superimposition of pictures C and D (Fig. 4E) showed the presence of intracellular bacteria (stained in red), while the extracellular bacteria were stained in yellow. No red fluorescent bacteria were detected when the experiment was performed without membrane permeabilization (data not shown).

We quantified the invasion of E. coli-NadA using a gentamicin protection assay on Chang monolayers infected with E. coli-NadA or E. coli-pET. The results from one representative experiment, reported in Fig. 5, show that the presence of NadA induces more than 1 log increase in the number of intracellular bacteria. Furthermore, when the gentamicin protection assay was performed in the presence of the microfilament inhibitor cytochalasin D, the invasion ability of E. coli-NadA was drastically reduced and was comparable to that of the negative E. coli-pET strain (Fig. 5).

Figure 5.

Invasion into epithelial cells of E. coli expressing NadA. Chang monolayers were infected with approximately 3 × 107 bacteria (moi 100) for 3 h at 37°C. Invasion was calculated using the gentamicin protection assay with and without pre-incubation of the cells with cytochalasin D (1 µg ml−1), and expressed as cfu per well. Values represent the mean ± standard deviation of a representative experiment performed in triplicate.

These data show that NadA is able to mediate invasion of E. coli into epithelial cells by a microfilament-dependent process.

Electron microscopy analysis

The adhesion and invasion process mediated by NadA was further analysed by scanning (SEM) and transmission (TEM) electron microscopy on Chang monolayers infected with E. coli-NadA strain (Fig. 6). SEM analysis revealed many bacteria associated to the cell membrane (Fig. 6A and B). In addition, TEM analysis demonstrated a more intimate association of E. coli-NadA with the epithelial cell surface (Fig. 6C). This close interaction was often associated with the formation of cellular membrane protrusions (Fig. 6D) and followed by the internalization of the bacteria into cytoplasmic vacuoles (Fig. 6E).

Figure 6.

Scanning and transmission electron microscopy of epithelial cells infected with E. coli-NadA.
A and B. SEM showing many bacteria associated with Chang cell surface.
C and D. TEM showing the close interaction of E. coli-NadA with the cell membrane and the formation of membrane protrusions at the cell–bacterium contact site.
E. TEM showing an internalized bacterium inside a cytoplasmic vacuole indicated by an arrow.

The N-terminal region of NadA contains the binding domain

We investigated the role of the N-terminal globular domain of NadA (amino acids 24–87) in host-cell interaction. On the basis of the secondary structure analysis, this domain was additionally divided into three putative domains: a predicted α-helix-rich region with hydrophobic residues (amino acids 24–42), an internal part without a defined secondary structure (amino acids 43–70), and an additional domain with a predicted α-helix propensity (amino acids 71–87). Truncated NadA proteins devoid of the entire N-terminal region from residues 30–87 (the amino acids 24–29 were left to allow the processing of the leader peptide and therefore the right maturation of the protein), or lacking one of the three N-terminal domains (Fig. 1A) were expressed in E. coli BL21(DE3) strain. Western blot and FACS analysis showed that all four mutated proteins were expressed as surface-exposed high-molecular-weight proteins (Fig. 1B, lane 7–10, Fig. 1C), indicating that they maintained the proper oligomeric structure. These strains were tested in the adhesion assay using immunofluorescence analysis, and the results showed that all strains expressing each NadA N-terminal variants had completely lost the capacity to adhere to Chang epithelial cells (data not shown).

To confirm these results the strains E. coli-NadAΔ24-42, E. coli-NadAΔ43-70 and E. coli-NadAΔ71-87 were used to infect Chang epithelial cells and the number of cfu was then determined. The data, reported in Fig. 3, show that the level of adhesion of the three strains to human cells was comparable to that of the negative control. These results confirm the hypothesis that the N-terminal region of NadA contains the binding domain and suggest that either the removal of each of the three subdomains alters the binding site or that the entire region is important in the interaction with human cells.

Characterization of the binding specificity

We analysed the ability of NadAΔ351-405 to bind to different human cell lines. Various human cell lines were incubated with three concentrations of the purified protein, and the binding was measured by FACS analysis. As shown in Fig. 7A, NadAΔ351-405 was able to bind to three out of four human epithelial cell lines tested (Chang, Hep-2 and HeLa) although to different extents. In addition, NadAΔ351-405 failed to bind to HUVECs, confirming the results obtained by adhesion assays. Furthermore, we tested the ability of NadAΔ351-405 to bind to a panel of extracellular matrix components (fibronectin, laminin, collagen I, III and IV, and heparan sulphate) by ELISA. The results showed that NadA was not able to bind any of the proteins tested (data not shown). Finally, to explore the biochemical nature of the cellular receptor, Chang cells were treated with pronase or phospholipase A2, to modify the protein or lipid content of the cell membrane respectively (Okuma et al., 2001). Subsequently, the binding of NadAΔ351-405 was determined by FACS analysis. Only the pronase treatment was able to reduce the capability of NadAΔ351-405 to bind to Chang cells (Fig. 7B). These results confirm that NadA binds specifically to epithelial cells and suggest that a protein receptor is likely to mediate this interaction.

Figure 7.

Characterization of the binding activity of NadAΔ351-405 to human cells using FACS analysis.
A. Comparison of binding specificity of NadAΔ351-405 to different human cell lines.
B. Biochemical properties of binding. Chang cells were treated with medium alone (black bars), pronase (1000, 500 and 250 µg ml−1) or phospholipase A2 (800, 200 and 50 µg ml−1) and subsequently incubated with recombinant NadAΔ351-405 (200 µg ml−1).
In both cases, the binding was analysed by FACS using an anti-NadAΔ351-405 antibody. The values are reported as net mean fluorescence intensity (MFI).

Role of NadA to adhesion and invasion of N. meningitidis

To investigate whether the contribution of NadA to bacterial adhesion and invasion described in E. coli is also relevant to N. meningitidis, a nadA isogenic mutant was constructed in the capsulate-piliated MC58 strain (MC58 NadA strain). The ability of MC58 NadA strain to adhere to Chang epithelial cells was investigated after 3 h of infection and compared with that of the wild-type MC58. The results, reported in Fig. 8A, show approximately a threefold reduction in the associated MC58 NadA bacteria as compared with the wild-type strain, indicating that NadA has a role in the adhesion of meningococcus to epithelial cells.

Figure 8.

Role of NadA in the adhesion and invasion of meningococcus.
A. Association of the wild-type capsulate MC58 and the isogenic mutant MC58 NadA strain to Chang cells after 3 h of incubation (moi 100).
B. Invasion of the acapsulate MC58 strain (MC58 SiaD) and the strains MC58 SiaD/NadA, MC58 SiaD/Opc, MC58 SiaD/NadA/Opc into Chang cells (moi 100). Invasion was calculated using the gentamicin protection assay after 5 h of infection.
The number of associated and internalized bacteria is expressed as cfu per well. Values represent the mean ± standard deviation of one out of four experiments performed in triplicate.

In accordance with previous observations (Virji et al., 1992; Unkmeir et al., 2002), we found that the number of internalized capsulate bacteria was very low (data not shown). Therefore, to evaluate the role of NadA in the invasion of meningococcus into epithelial cells, we constructed a knockout mutant of the nadA gene in the acapsulate MC58 SiaD strain (MC58 SiaD/NadA strain). In addition, in order to compare the contribution of NadA with that of Opc, a well-characterized invasin of meningococcus (Virji et al., 1992; de Vries et al., 1996), a single Opc and a double NadA/Opc mutants were also generated in MC58 SiaD strain (MC58 SiaD/Opc and MC58 SiaD/NadA/Opc strain respectively). The invasion of these mutants was compared with that of the parent strain (MC58 SiaD) by incubating bacteria with a monolayer of Chang cells. After 5 h of infection, the monolayers were subjected to gentamicin treatment to determine the number of internalized bacteria. The results of one representative experiment represented in Fig. 8B show that MC58 SiaD/NadA and MC58 SiaD/Opc strains are, respectively, 2.8 and 5.9 times less invasive as compared with the parent strain. Furthermore, the mutant MC58 SiaD/NadA/Opc was 10.6 times less invasive than the parent strain. The NadA negative mutant also showed a trend towards a lower adhesion compared with the parent strain; however, the difference was not statistically significant, even in the absence of Opc (data not shown).

In conclusion, these results indicate that NadA contributes to the invasion of meningococcus into epithelial cells. We also showed that the contribution of NadA is similar to that of Opc and that the two proteins seem to play an independent role in invasion.


In this work we used E. coli to study the biological role of NadA, a novel meningococcal adhesin included into the recently defined ‘Oca’ family (Desvaux et al., 2004). We showed that in E. coli, as in meningococcus, NadA is transported to the outer membrane where it forms high-molecular-weight oligomers, which are very stable to boiling and denaturating conditions (SDS-PAGE). It has been supposed that the highly conserved C-terminal region of the members of the Oca family is responsible for outer membrane insertion and oligomerization of these proteins (Roggenkamp et al., 2003). Here we demonstrate that for NadA the anchor is also responsible for its surface localization and assembly in SDS-PAGE-stable oligomeric structures. In fact, the protein that is devoid of the predicted anchor domain is no more membrane-anchored but is released into the E. coli culture supernatant through the periplasmic space. The mechanism that mediates this phenomenon is not understood, however, we cannot exclude the possibility that secretion in the extracellular medium is a consequence of the high level of expression and accumulation of the protein in the periplasm. Interestingly, we also found that the secreted protein is assembled in oligomers that are not SDS-PAGE resistant. This aspect is under investigation (S. Savino, personal communication).

We have previously reported that recombinant-purified NadA binds to human epithelial cells (Comanducci et al., 2002). In this study we further investigated the adhesive properties of NadA using in vitro adhesion assays and we showed that oligomeric NadA expressed on the E. coli surface mediates adherence of bacteria to Chang epithelial cells. On the other hand, we were unable to detect any adhesion to HUVECs. These results indicate that homology of NadA with Oca proteins is conserved at functional level. Meningococcus is an extracellular pathogen but it is also present inside the cells. Among the factors involved in the human cell interaction identified so far, Opc and Opa proteins have been widely described as able to mediate invasion of acapsulate meningococcus to epithelial and endothelial cells (Virji et al., 1992; 1993; Unkmeir et al., 2002); however, experimental evidence suggests that other bacterial components could be involved (Pujol et al., 1997). We investigated the ability of NadA to promote internalization of E. coli into Chang epithelial cells. Using the in vitro gentamicin protection assay, we found that approximately 10% of adherent bacteria are internalized, showing that NadA mediates invasion of E. coli into epithelial cells.

We demonstrated that this uptake is associated with cytoskeletal rearrangement, as pre-incubation with the actin microfilament-disrupting agent cytochalasin D totally abolished the E. coli NadA-mediated invasion. Moreover, TEM analysis revealed the presence of intimately associated bacteria to the cell surface and the formation of membrane protrusions at the bacterial contact site, likely to be involved in bacterial internalization.

To investigate the specific contribution of NadA in the adhesion and invasion process of N. meningitidis, we inactivated the nadA gene in the capsulate MC58 strain and in its acapsulate variant (MC58 SiaD). We demonstrated that the capsulate isogenic nadA mutant shows a reduced ability to adhere to Chang epithelial cells as compared to the wild-type MC58, while only a trend of lower adhesion was observed in the absence of the capsule. It is known that the polysaccharide capsule can functionally mask membrane proteins. However, we have previously shown that NadA is accessible to antibody binding even in the presence of capsule (Comanducci et al., 2002). Therefore, NadA is surface-exposed and available to contribute to meningococcal adhesion even in a capsulate strain. In contrast, in an acapsulate background the contribution of NadA in adhesion could be masked by the simultaneous action of many other factors that are also surface-exposed.

As the number of invasive capsulate bacteria is usually low, we analysed the NadA-mediated invasion only in a capsule-deficient strain. We demonstrated that the acapsulate NadA mutant showed a statistically significant reduced ability of about threefold to invade Chang cells. In this study we also compared the contribution in invasion of NadA with that of Opc, a well-characterized invasin. In our system, the acapsulate Opc strain was about six times less invasive as compared with the acapsulate MC58, accordingly with published data (de Vries et al., 1996). It is important to note that NadA and Opc have a similar contribution in the invasion of meningococcus. Of particular interest is the fact that the strain lacking both NadA and Opc showed a 10-fold reduction as compared with the parent strain, suggesting that NadA and Opc act independently during the invasion process of meningococcus.

We therefore conclude that NadA is a new meningococcal factor involved in the colonization of epithelium and cell invasion.

Most of the various adhesive functions of YadA have been assigned to specific domains present in the N-terminal region of the protein (Tamm et al., 1993; Roggenkamp et al., 1995; 1996). As recently shown, a properly folded globular head domain, as well as a trimeric form of YadA, is required for collagen binding (Nummelin et al., 2004). We decided to investigate whether the N-terminal globular domain of NadA (‘head’ domain) has any role in the protein's function. This domain has been additionally subdivided into three different structural domains and various NadA deletion mutants lacking the entire ‘head’ and each single domain were expressed in E. coli. All the mutated proteins form surface-exposed stable oligomers, suggesting that the oligomeric structure was maintained and that the ‘head’ region is not required for oligomerization and surface localization of NadA. Furthermore, these strains, tested in an in vitro adhesion assay, failed to bind to Chang epithelial cells, indicating that the entire N-terminal domain of NadA seems to be necessary to mediate binding to these cells and that this process could require a region properly folded.

Finally, we explored the nature of NadA-mediated interaction to host cells using the purified recombinant protein. Our results show that after pronase treatment, the capability of NadA to bind Chang epithelial cells is drastically reduced, suggesting that a protein receptor molecule could mediate this binding. Furthermore, this hypothetical NadA receptor seems to be differentially expressed by different human epithelial cell lines and absent in the HUVECs. Furthermore, no binding was detected to any of the extracellular matrix components tested.

Based on the results presented in this study, we propose that NadA is organized in functional domains. Our model is illustrated in Fig. 9.

Figure 9.

Model of structure–function relationship of NadA in E. coli. As in meningococcus, full-length NadA is exported to the E. coli surface and forms stable oligomers having functional binding capacity to human epithelial cells (a). NadAΔ30-87, lacking the N-terminal globular domain, is still able to form stable surface-exposed oligomers, but has lost the adhesive properties (b). NadAΔ351-405, devoid of the predicted anchor domain, is assembled in a trimer released into the culture supernatant, through the periplasmic space, and maintains its binding capacity (c). OM, outer membrane; P, periplasm; IM, inner membrane.

It is tempting to speculate that NadA, which is present in three of the four hypervirulent lineages, is a virulence factor that increases the fitness of hypervirulent bacteria to colonize and/or cause disease by increasing their adhesion and invasion capabilities. It is, however, interesting that other hypervirulent lineages are also successful pathogens even if they do not have NadA. This suggests that meningococcus may use different factors to achieve the same goal. An understanding of the interrelationship of NadA with other bacterial determinants of pathogenicity should provide important insights into mechanisms by which N. meningitidis causes disease.

Experimental procedures

Plasmid construction

In order to express the full-length protein, nadA gene was amplified by polymerase chain reaction (PCR) from N. meningitidis 2996 strain using as primers the oligonucleotides nadA-FOR (cgcggatcccatatg-AAACACTTTCCATCC, NdeI site) and nadA-REV (cccgctcgag-TTACCACTCGTAAT TGACGCC, XhoI  site).  The  mutated  genes  coding  for NadAΔ1-23 and NadAΔ351-405 were generated using the oligonucleotides nadAnl-FOR (cgcggatcccatatg-GCCACAAACGAC GAC, NdeI site) and nadA-REV, nadA-FOR and nadAc-REV (cccgctcgag-TTAACCCACGTTGTAAGGTTG, XhoI site) respectively. The digested DNA fragments were cloned into pET21b vector (Novagen). The plasmids containing the mutated genes coding for NadAΔ30-87, NadAΔ24-42, NadAΔ43-70 and NadAΔ71-87 were generated by a two-step cloning. PCRa (nadA-FOR and nadA23-REV cgcggatccgctagcTGCCAGT GCGCCG, additional NheI site; amplifying the region corresponding to amino acids 1–23) and PCRb (nadAh-FOR cgcggatccgctagcaacgacgacgatAAAGTCGTGACTAACCTG, additional NheI site and nadA-REV; amino acids 88–405) were used to express NadAΔ30-87. PCRa and PCRc (nadA43-FOR cgcggatccgctagcAACAATGGCCAAGAAATCAA, additional NheI site and nadA-REV; amino acids 43–405) were used to express NadAΔ24-42. PCRd (nadA-FOR; nadA42-REV cgcggatccGTAGGCAGCAGCAATGG, additional BamHI site; amino acids 1–42) and PCRe (nadA71-FOR cgcggatc cGTAGGCAGCAGCAATGG, additional BamHI site and nadA-REV; amino acids 71–405) were used to express NadAΔ43-70. Finally, PCRf (nadA-FOR; nadA70-REV cgcggatc cGTCTTTTTTGGTAATTGTGCCGT, additional BamHI site; amino acids 1–70) and PCRg (nadA88-FOR cgcggatc cAAAGTCGTGACTAACCTGAC, additional BamHI site; nadA-REV; amino acids 88–405) were used to express NadAΔ71-87. The insertion of NheI or BamHI site results in Alanine-Serine or Glycine-Serine additional amino acids respectively. The ligation products were transformed into E. coli DH5α (Invitrogen) and E. coli BL21(DE3; Novagen) was used as expression host.

DNA cloning and E. coli transformation were performed according to the standard protocols (Sambrook et al., 1989). E. coli strains were cultured at 37°C in Luria–Bertani (LB) broth supplemented with 100 µg ml−1 ampicillin.

Gene expression and cell fractionation

Escherichia coli strains were grown at 37°C for 14–16 h, then bacteria were recovered by centrifugation (uninduced condition) or diluted in fresh medium and grown at 37°C until OD600 = 0.4–0.8. Protein expression was induced by addition of 1 mM IPTG (Sigma) for 3 h (induced condition).

Whole-cell lysates were obtained by resuspending bacteria in SDS-sample buffer 1× and boiling for 5–10 min. Total-cell extract from N. meningitidis strain 2996 was prepared as previously described (Comanducci et al., 2002). Outer membrane proteins were prepared following the rapid procedure described by Carlone et al. (1986). Briefly, bacteria were harvested, suspended in 1 ml of 10 mM Hepes buffer (pH 7.4) and sonicated on ice. The cell membranes were recovered by successive centrifugations at 15600 g at 4°C in a microcentrifuge. Cytoplasmic membranes were solubilized by addition of an equal volume of 2% Sarkosyl in 10 mM Hepes (pH 7.4). The outer membranes were then recovered by centrifugation and suspended in 10 mM Hepes buffer.

The periplasmic fraction was prepared by suspending bacteria in TRIS 50 mM (pH 8)-Sucrose 25%, adding polymixin B (final concentration 100 µg ml−1) and incubating the suspension for 30 min at room temperature. After centrifugation for 30 min at 13000 r.p.m. (4°C), the supernatant was recovered and mixed with SDS-loading buffer 4×. To prepare culture supernatant, bacteria were harvested at 13 000 r.p.m. for 10 min at 4°C. The supernatant was filtered through a 0.22 µm filter, and precipitated with 10% final concentration of trichloroacetic acid (TCA) for 2 h in ice. After centrifugation for 30 min at 13 000 r.p.m. (4°C), the pellet was washed twice with 70% ethanol and resuspended in SDS-loading buffer 1×. All the samples were diluted in SDS-sample buffer 4× and boiled for 5–10 min.

Equal amounts of proteins were separated using NuPAGE Gel System, according to the manufacturer's instructions (Invitrogen). Proteins were revealed by Coomassie-blue staining or transferred onto nitrocellulose membranes for Western blot analysis. Western blot was performed using a rabbit polyclonal anti-serum against purified NadAΔ351-405 (dilution 1:3000) and a secondary peroxidase-conjugate antibody (DAKO).

Surface detection of NadA in E. coli using FACS analysis

Surface-detection of NadA in E. coli was performed using FACS analysis. Approximately 2 × 106 bacteria were incubated for 1 h with an anti-NadAΔ351-405 serum (1:1000) and subsequently for 30 min with R-phycoerythrin (PE)-conjugated goat F(ab)2 antibody to rabbit IgG (diluted 1:100, Jackson ImmunoResearch Laboratories). All antibodies were diluted in PBS with 1% fetal bovine serum (FBS). Samples were analysed with a FACS-Scan flow cytometer (Beckton-Dickinson).

Purification of NadAΔ351-405

Escherichia coli-NadAΔ351-405 was grown under induced condition as described. All subsequent procedures were performed at 4°C. The culture supernatant, containing NadAΔ351-405, was filtered using 0.22 µm filter, concentrated seven times by ultrafiltration and then dialysed against 20 mM Tris HCl pH 7.6 at 4°C. The material was cleared by centrifugation at 39 000 g for 20 min and loaded on a Q Sepharose XL column (Pharmacia), washed, and eluted with 400 mM NaCl in the same buffer. The elution product was then diluted to 1.8 M NaCl and loaded on a Phenyl Sepharose 6 Fast Flow (High Sub, Pharmacia). The protein was eluted with 50 mM NaCl. To reduce the high endotoxin content, a Hydroxylapatite ceramic column Chromatography (HA Macro. Prep) was subsequently performed. Anti-NadAΔ351-405 serum was prepared as described (Comanducci et al., 2002).

Cell cultures

Chang (Wong-Kilbourne derivative, clone 1-5c-4, human conjunctiva), HeLa (human cervical epitheloid carcinoma), HEC-1B (human endometrium) and Hep-2 (human laryngeal epidermoid carcinoma) were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 15 mM l-glutamine, antibiotics and 10% heat-inactivated FBS. HUVECs (Promocell) were maintained in endothelial cell growth medium according to the manufacturer's instructions. All cells were grown at 37°C in 5% CO2.

Infection assays using E. coli strains

For adhesion and invasion assays, Chang cells were seeded on 24-well tissue culture plates (1 × 105 cells per well), on Laboratory-Tek chamber slides (Nunc) (immunofluorescence adhesion assay) or on PET membrane (Falcon) (electron microscopy studies). After 24 h incubation in an antibiotic-free medium, approximately 3 × 107 bacteria were added per well in DMEM supplemented with 1% FBS and incubated for 3 h at 37°C in 5% CO2. After removal of non-adherent bacteria by washing, cells were lysed with 1% saponin (Sigma) and serial dilutions of the suspension were plated onto LB agar to calculate the cfu.

To determine the number of intracellular bacteria, the infected Chang monolayer was treated with gentamicin (200 µg ml−1) for 2 h at 37°. After washing, cells were lysed and bacteria recovered and plated. To assess the role of actin microfilament, the cells were pre-incubated for 30 min at 37°C before infection with the microfilament inhibitor cytochalasin D (1 µg ml−1, SIGMA) (Rosenshine et al., 1994). The cytochalasin D was present during the 3 h of infection. Adhesion and invasion were analysed as described.

For adhesion immunofluorescence assay cells and bacteria were incubated as described. After washing of non-bound bacteria, cells were fixed in 3.7% paraformaldhehyde for 30 min, washed, incubated for 1 h with a rabbit polyclonal anti-E. coli serum (1:100, DAKO) and for 30 min with Alexa Fluor 488 goat anti-rabbit IgG (1:1000, Molecular Probes). For differential staining of extracellular and intracellular bacteria, fixed cells were incubated with: (i) rabbit polyclonal anti-E. coli serum, (ii) secondary antibody Alexa Fluor 488 goat anti-rabbit IgG, (iii) 10 min with triton 1% in PBS to permeabilize the cell membrane, (iv) rabbit polyclonal anti-E. coli serum and (v) secondary antibody Alexa Floor 586 goat anti-rabbit IgG (modified by Kwok et al., 2002). All washing steps and antibodies dilutions were performed in PBS with 1% FBS. Labelled preparations were mounted with SlowFade Light antifade kit (Molecular Probes) and analysed with a confocal microscope (Bio-Rad Laboratories) or with a Zeiss Axiophot immunofluorescence microscope. Extracellular and total bacteria were examined using the green and red filter sets respectively.

For electron microscopy studies samples were processed as described by Pujol et al. (1997). Briefly, cells were fixed with 2.5% paraformaldehyde + 2.5% glutaraldehyde in cacodylate sucrose buffer overnight at 4°C. After washing in the same buffer, the cells were postfixed with 1% osmium tetroxide + 0.15% ruthenium red in 0.1 M cacodylate-buffer for 1 h at 4°C. After washing in purified water the stain was blocked with 1% uranyl acetate. Samples were then dehydrated in a graded ethanol series. For SEM, samples were then dried by the critical point method using CO2 in a Balzers Union CPD 020, sputter-coated with gold in a Balzers MED 010 unit, and observed with a JEOL JSM 5200 electron microscope. For TEM, samples were fixed and dehydrated as described above and embedded in Epon-based resin. Thin sections were cut with Reichert Ultracut ultramicrotome using a diamond knife, collected on copper grids, stained with uranyl acetate and lead citrate and observed with a JEOL 1200 EX II electron microscope.

Binding assay

Binding assay was performed as previously described (Comanducci et al., 2002). Briefly, human cells were non-enzymatically detached using cell dissociation solution (CDS, Sigma), harvested and suspended in RPMI medium supplemented with 1% FBS. Approximately 1 × 105 cells were mixed with different concentrations of NadAΔ351-405 or medium alone for 30 min at 4°C. Cells were then incubated with mouse polyclonal serum against NadA (diluted 1:500) for 1 h at 4°C, and with R-phycoerythrin (PE)-conjugated goat F(ab)2 antibody to mouse Ig (diluted 1:100; Jackson ImmunoResearch Laboratories) for 30 min at 4°C. Cells were analysed with a FACS-Scan flow cytometre (Beckton-Dickinson). The mean fluorescence intensity (MFI) for each population was calculated.

To enzymatically modify cell surface proteins, 1 × 105 Chang cells were incubated with three concentrations of pronase (1000, 500, 250 µg ml−1) (Sigma) or phospholipase A2 (800, 200, 50 µg ml−1) (Sigma) in FBS-free RPMI medium for 30 min at 37°C in 5% CO2, and subsequently incubated with 200 µg ml−1 NadAΔ351-405 or medium alone for 30 min at 4°C. The binding was detected by FACS as described.

Construction of knockout mutants in meningococcus

The meningococcus knockout mutants used in this study were generated by allelic exchange in capsulate MC58 and acapsulate MC58 SiaD strains (Masignani et al., 2003) using the plasmid pBSnadAERM and pBSopcKAN. The plasmid pBSnadAERM was obtained amplifying nadA gene from +3 to +1089 from MC58 using nadA-FOR3 (gctctagag AACACTTTCCATCCAAAG, XbaI site) and nadA-REV3 (ccgctcgag-TTACCACTCGTAATTGACG, XhoI site). The digested fragment was cloned into pBluescript and the plasmid transformed into E. coli DH5. The plasmid was then digested with HincII in order to interrupt and replaced nadA gene with the ermC gene. To generate the plasmid pBSopcKAN, the upstream (260 bp) and downstream (510 bp) flanking regions of opc gene were amplified from MC58 using UPopc-FOR (gctctaga-CCGATAAATACACACAAAC, XbaI site) and UPopc-REV (tcccccggg-CAGGGCAATCATGGCAC, SmaI site); DOWNopc-FOR (tcccccggg-CGGCATAAAAT TCTGATGA, SmaI site) and DOWNopc-REV (cccgctcgag CGAGTATTCACGTCGG, XhoI site) respectively. The fragments were subsequently cloned into pBluescript and the plasmid was transformed into E. coli DH5. The plasmid was then digested with SmaI in order to insert the kanamycin cassette. Naturally competent N. meningitidis strains MC58 and MC58 SiaD were transformed with pBSnadAERM or pBSopcKAN as described (Masignani et al., 2003). Finally, the plasmid pBSopcKAN was subsequently used to transform the MC58 SiaD/NadA to generate the MC58 SiaD/NadA/Opc strain.

The presence of truncated nadA gene or opc gene was confirmed by PCR and the lack of expression of the proteins was confirmed by Western blot analysis (data not shown). All these strains were characterized for the expression of pilin, PilC, OpA and Opc or NadA by Western blot analysis. The results showed comparable levels of expression in all strains (data not shown).

Association of N. meningitidis to epithelial cells

The interaction of N. meningitidis with cultured Chang epithelial cells was studied using a method similar to that described by Serruto et al. (2003). Briefly, the epithelial cells were seeded in 96-well tissue culture plates (104 cells per well) and grown to confluency. The culture medium was removed and the cells were washed three times with Hank's balance salt solution (HBSS) and fresh medium added. Bacteria grown on GC agar were washed in PBS once, resuspended in culture medium in the presence of 2% FBS, and added (~2 × 106) to monolayers [multiplicity of infection (moi) 1:100] in triplicate. Cells and bacteria were incubated at 37°C for 3 or 5 h in 5% CO2. The inoculating dose of bacteria was confirmed by serial dilution and plating. To determine association, monolayers were washed four times with HBSS to remove non-adherent bacteria. The remaining bacteria were released by the addition of 1% saponin and incubation at 37°C for 15 min and the number of associated bacteria was determined by serial dilution and plating. To determine invasion, monolayers were subjected to gentamicin treatment for 90 min. The intracellular bacteria were released by the addition of 1% saponin (15 min at 37°C) and the number was determined by serial dilution and plating. Statistical significance of the number of invasive bacteria of the parent and knockout strains was determined using the Student's t-test on four independent experiments performed in triplicates. P-values of <0.05 were judged to be of statistical significance.


We thank Emanuele Papini for assistance with confocal microscopy, Pasquale Zizza for technical help, Catherine Mallia for editing and Giorgi Corsi for artwork. This work was supported by a grant from the European Commission within the fifth Framework Programme: Mucosal Immunization and Vaccine Development (MUCIMM), Contract No. QLK2CT 1999 00228.