Neisseria meningitidis is a common and usually harmless inhabitant of the mucosa of the human nasopharynx, which, in rare cases, can cross the epithelial barrier and cause meningitis and sepsis. Biofilm formation favours the colonization of the host and the subsequent carrier state. Two different strategies of biofilm formation, either dependent or independent on extracellular DNA (eDNA), have been described for meningococcal strains. Here, we demonstrate that the autotransporter protease NalP, the expression of which is phase variable, affects eDNA-dependent biofilm formation in N. meningitidis. The effect of NalP was found in biofilm formation under static and flow conditions and was dependent on its protease activity. Cleavage of the heparin-binding antigen NhbA and the α-peptide of IgA protease, resulting in the release of positively charged polypeptides from the cell surface, was responsible for the reduction in biofilm formation when NalP is expressed. Both NhbA and the α-peptide of IgA protease were shown to bind DNA. We conclude that NhbA and the α-peptide of IgA protease are implicated in biofilm formation by binding eDNA and that NalP is an important regulator of this process through the proteolysis of these surface-exposed proteins.
Neisseria meningitidis is the aetiological agent of meningococcal disease, a systemic infection that is a major cause of morbidity and mortality in children and adolescents in developing countries. The bacterium usually is a harmless inhabitant of the human nasopharynx, but, in rare cases, it is able to penetrate the mucosal barriers and reach the bloodstream, where it multiplies and causes sepsis. One of the virulence factors is the capsule. Invasive N. meningitidis strains express either one of six different capsular polysaccharides classified as A, B, C, W135, X and Y. The capsule confers resistance to complement-mediated killing (Kahler et al., 1998) and phagocytosis (Kolb-Mäurer et al., 2001), and it is required for intracellular survival during infection of tissue culture cells (Spinosa et al., 2007). Non-capsulated bacteria are often found in healthy individuals (Cartwright et al., 1987). During colonization of the nasopharynx, adhesion of the bacteria to epithelial cells is hindered by the capsule (Virji et al., 1993), which may physically impair the binding of major neisserial adhesins to host cell receptors (Spinosa et al., 2007). In the nasopharynx, the meningococci appear as microcolonies (Sim et al., 2000) resembling biofilm structures (Costerton et al., 1995).
Biofilms are surface-attached microbial communities embedded in a self-produced extracellular matrix of polymeric substances (Costerton et al., 1995). This structure provides a protective niche in which the cells become more resistant to antibacterial agents, such as antibodies or antibiotics. Thus, N. meningitidis biofilm structures could be formed for protection against the defence mechanisms of the host in order to guarantee persistence during the carrier stage. Encapsulated bacteria cannot generate biofilms on abiotic surfaces (Yi et al., 2004; Lappann et al., 2006), but they were reported to form biofilms on epithelial cells (Neil et al., 2009).
Several factors, including type IV pili (Yi et al., 2004), the two-partner secretion system protein A (Neil and Apicella, 2009) and extracellular DNA (eDNA), are involved in biofilm formation. eDNA is an essential component for initial cell-to-cell or cell-to-substratum attachment in many bacterial pathogens (Moscoso et al., 2006; Berne et al., 2010; Harmsen et al., 2010). Recently, it was reported that eDNA can also be involved in the initiation of biofilm formation by meningococci (Lappann et al., 2010).
In this study, we investigated the role of several meningococcal autotransporters in biofilm formation. Autotransporters form a very versatile group of secreted proteins. They consist of a secreted passenger domain and a translocator domain, which is needed for the transport of the passenger across the outer membrane. The translocator domains of different autotransporters are structurally very similar, but the passenger domains are highly diverse in function. In several microorganisms, autotransporters have been identified that are involved in auto-aggregation (Zhang et al., 2004; Klemm et al., 2006; Alamuri and Mobley, 2008), a process related to biofilm formation. Eight different autotransporters have been identified in N. meningitidis, but only six of them are expressed in strain MC58 (van Ulsen and Tommassen, 2006), i.e. AusI, App, IgA protease (IgAp), NalP, NhhA and NadA. The former four are proteases. A potential role for some of them in bacterial pathogenesis has been demonstrated (Lin et al., 1997; Serruto et al., 2003; Vidarsson et al., 2005), but no studies have addressed their possible implication in biofilm formation or auto-aggregation. NalP is a secreted subtilisin-like serine protease, whose expression is phase variable by slipped-strand mispairing (Saunders et al., 2000). It is involved in the proteolytic cleavage of several other autotransporters (van Ulsen et al., 2003; van Ulsen et al., 2006) and surface-exposed lipoproteins (Roussel-Jazédé et al., 2010; Serruto et al., 2010) suggesting that NalP is a regulator of the outer membrane and secretome composition. The objective of this study was to investigate the possible role of NalP and other autotransporter proteases in biofilm formation. We newly discovered the involvement of several surface-exposed proteins in eDNA-mediated biofilm formation and demonstrate that NalP is involved in the regulation of biofilm formation by proteolytically targeting these proteins.
NalP expression alters biofilm formation under static conditions
Neisseria meningitidis has evolved two different strategies for biofilm formation, an eDNA-dependent one, which is used, among others, by cc32 strains, and an eDNA-independent one used by clonal complex (cc) 11 and cc8 strains (Lappann et al., 2010). The first type forms biofilms within 1 h, while longer periods are required for the latter. Since it was reported that biofilm formation is inhibited by capsular polysaccharide (Yi et al., 2004), we used unencapsulated mutants of the N. meningitidis reference strains H44/76 (cc32) and B16B6 (cc11), designated HB-1 and BB-1, respectively, in our experiments. First, biofilm formation was studied under static growth conditions. In agreement with previous results (Lappann et al., 2010), biofilms of cc32 strain HB-1 were detected after 1 h of growth, while at least 8 h were required before we detected biofilms of cc11 strain BB-1. After 12 h or 36 h, we observed a significant reduction in biofilm mass formed by HB-1 and BB-1, respectively, compared with early biofilms (data not shown).
In both genetic backgrounds, biofilm formation was significantly increased when the nalP gene was inactivated, while no significant effect on biofilm formation was observed when the genes for the other autotransporter proteases, App, IgAp or AusI, were knocked out (Fig. 1A). The wild-type strains and corresponding nalP mutants showed similar growth characteristics (Fig. S1A and B), which excludes an enhanced growth rate of the nalP mutants as the cause of their increased biofilm formation. The nalP mutant phenotype was complemented when NalP was expressed in trans from plasmid pEN300 but not when a mutant derivative of NalP with a substitution of the active-site Ser427 by Ala was expressed (Fig. 1B), indicating that the protease activity of NalP is required for the observed phenotype.
The effect of NalP was also examined in cc8 reference strain 2996 and in a panel of seven patient isolates collected in the Netherlands between 1960 and 2007 and belonging to cc32 or cc11 (Table 1). First, capsule-deficient derivatives of these strains were isolated and expression of NalP was evaluated by Western blotting (data not shown). The biofilm-forming capacity of the capsule-deficient strains varied widely and did not correlate to their NalP-expression status (Fig. S2A) or their growth rates (Fig. S1C–E). Like cc32 reference strain HB-1, all isolates of cc32 formed biofilms within 1 h, while biofilm formation in isolates of cc11 and cc8 was generally slower and more variable (Fig. S2A). Subsequently, nalP was knocked out in the isolates that did express NalP (strains 2070077, 69 and 2996); in all cases, the mutants showed enhanced biofilm formation as compared with their parent (Fig. S2B). In all isolates that did not express NalP (strains 2070755, 2071066, 2071749, 348 and 2011872), the chromosomal nalP gene was deleted, and plasmid pEN300, carrying nalP under lac-promoter control, was introduced. In all these cases, induction of the expression of NalP with isopropyl-β-d-1-thiogalactopyranoside (IPTG) reduced biofilm formation (Fig. S2B). In conclusion, although the invasive strains analysed displayed variable biofilm-forming abilities, the negative effect of NalP expression was evident in all of them, confirming our observations in HB-1 and BB-1 and suggesting a general mechanism.
aery, erythromycin-resistance cassette; kan, kanamycin-resistance cassette; cat, chloramphenicol-resistance cassette. For disease isolates, serogroup, sequence type (ST), clonal complex (cc), isolation year and expression status of NalP, as determined by Western blotting, are provided.
Derivative of H44/76 (B:15:P1.7,16) of cc32 with the capsule locus replaced by ery
We noticed that strains HB-1 and BB-1 formed a lattice after 12 or 24 h of growth in static conditions and that the appearance of these structures was different in the nalP mutants of these strains (Fig. 2). No lattice structures were detected immediately after adding bacteria from cultures grown for 6 h in shaking conditions to the wells of the plates, which suggests that static conditions are required for lattice formation. In both strains disperse bacterial aggregates of different sizes were observed after 12 h of incubation, while large aggregates with intervening spaces, following a branch-like organization, were observed after 24 h (Fig. 2). The lattice formed more slowly in BB-1 than in HB-1 and in both strains, the lattice developed more slowly in the nalP mutant than in the corresponding wild-type strain as evidenced by a less defined structure at 12 h (Fig. 2). In strain BB-1 and its derivative, this difference was also evident at 24 h. The wild-type phenotype was restored in the nalP mutants when NalP was expressed from plasmid pEN300 after addition of IPTG (Fig. 2). Thus, these observations show macroscopically the effect of NalP expression on the bacterial population structure. Interestingly, these structures were not formed when DNase I was added to the culture medium, indicative of the involvement of eDNA in their formation.
NalP expression alters biofilm formation under flow conditions
Meningococci have been shown to form biofilms also in continuous flow systems on abiotic and biotic surfaces (Lappann et al., 2006; Neil et al., 2009). Such systems assure the continuous supply of nutrients and maintain shear forces that may mimic natural dynamic conditions. Therefore, bacterial culture in flow-chamber systems represents the golden tool to study biofilm physiology.
Strains HB-1 and BB-1 and the corresponding nalP mutants were grown for 14 h in a continuous flow cell system and biofilm development was monitored by real-time fluorescence microscopy. The dynamic formation of the biofilms is shown in Movies S1–S3, and the characteristics of the architecture of 14-h-old biofilms are shown in Fig. 3. Strain HB-1 initiated biofilm formation through the interaction of microcolonies with the substratum (Movie S1). After 14 h of growth, an increase in biofilm biomass was visible which then covered 40–60% of the substratum. The biofilm was organized in large macrocolonies with large intervening spaces within the biomass (Fig. 3A) resembling the lattice structures observed after growth under static conditions (Fig. 2A). In contrast, attachment of strain BB-1 to the substratum was initiated through smaller sized microcolonies and single cells (Movie S2). After 14 h of growth, the biomass covered almost the entire substratum, and a flat biofilm with small microcolonies and, as compared with HB-1, smaller intervening spaces was evidenced. Similar structures as observed for HB-1 and BB-1 were previously reported for unencapsulated derivatives of strains MC58 (cc32) and 2120 (cc11) respectively (Lappann et al., 2006; Lappann and Vogel, 2010). Statistical analysis of image stacks acquired by confocal microscopy of live/dead-stained biofilm biomass using the comstat software revealed a significantly lower roughness coefficient in BB-1 (10-fold reduction) as compared with HB-1 (Fig. 3B), indicating two different biofilm structures.
The nalP mutant derivatives of both strains showed considerable alteration in the biofilm architecture compared with the corresponding wild-type strains (Fig. 3; biofilm development for HB-1ΔnalP is shown in Movie S3). In general, the nalP mutants produced more confluent layers than the wild-type strains as evidenced by an increase in substratum coverage, which is clearly revealed by a reduction of the size of the intervening spaces within the biomass (approximately three- to fivefold reduction). Thus, NalP expression appears to decrease the binding of the meningococci to the substratum. Statistical comparison using comstat software revealed a significant reduction of the roughness coefficient in the biomass of the nalP mutants as compared with the corresponding wild-type strains (Fig. 3B), further confirming the alteration of the biofilm structure. In conclusion, these analyses confirm the involvement of NalP in biofilm formation also under flow conditions.
Composition and role of eDNA in biofilm formation
To evaluate whether eDNA has a role in the increased biofilm formation observed in the nalP mutants, the bacteria were incubated with DNase I. This treatment dramatically reduced initial biofilm formation in both strain HB-1 and its nalP mutant derivative under static conditions (Fig. 4A), while heat-inactivated DNase I had no effect (results not shown). Thus, the enhanced biofilm formation in the nalP mutant is dependent on eDNA. In accordance with previous results (Lappann et al., 2010), DNase treatment did not significantly affect biofilm formation in cc11 strain BB-1. However, it did reduce biofilm formation of the nalP mutant of this strain to similar levels as observed in the parent strain (Fig. 4A). Thus, although eDNA is not involved in the initial biofilm formation in BB-1, it enhances biofilm formation when NalP is not expressed in this strain.
Lappann et al. (2010) showed that biofilm formation in DNase I-treated bacteria of an unencapsulated derivative of cc32 strain MC58 was largely restored by adding crude DNA, but weakly with purified chromosomal DNA. Note that crude DNA is chromosomal DNA that is released from the bacteria by autolysis and such preparations additionally contain a large variety of proteins (vide infra). Using this approach, we evaluated whether different amounts or composition of the eDNA might explain the enhanced biofilm formation in the nalP mutant. Quantification of the amounts of released DNA did not reveal significant differences between the parental strains and the corresponding nalP mutants (Fig. S3A). Next, we examined crude DNA preparations from HB-1 and the nalP mutant on agarose gels. In both cases, a single band with a high molecular size was detected that migrated more slowly in the gel in the preparations of the nalP mutant (Fig. 4B). A similar difference was observed when crude DNA of strain BB-1 and its nalP-mutant derivative were compared (data not shown). The band was degraded after treatment with DNase I, while RNase A had no effect (Fig. 4B and results not shown). Interestingly, treatment of the crude DNA preparations with proteinase K abrogated the difference in electrophoretic mobility of the crude DNA preparations (Fig. 4B), suggesting that the DNA from the nalP mutants contains substantially more bound proteins. Comparison of the protein profiles of the crude DNA samples by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) revealed a large variety of proteins but no obvious differences between preparations from the parent and the nalP mutant (Fig. S3B). Interestingly, treatment of the crude DNA preparations with the non-ionic detergent Elugent to solubilize membrane vesicles also abrogated the difference in their electrophoretic mobility (Fig. 4B). Together, we postulate that the altered composition of outer-membrane-associated proteins in the nalP mutant increases the binding of outer-membrane vesicles to the eDNA, resulting in the lower electrophoretic mobility of the eDNA of the mutant.
We next asked whether the different crude DNA preparations of wild-type and mutant bacteria could restore biofilm formation of DNase-treated bacteria to different extents. However, we found that biofilm formation of DNase I-treated HB-1 cells was restored to the same extent with crude DNA from either the nalP mutant or the wild-type strain (Fig. 4C). Similarly, both DNA preparations were equally effective in restoring biofilm formation of DNase-treated cells of the nalP mutant of HB-1 (Fig. 4C). Thus, in spite of the differences in associated materials (Fig. 4B), there are no qualitative biofilm-stimulating differences between both DNA preparations. Of interest, crude DNA extracted from cultures of strain BB-1 was equally effective in restoring biofilm formation of DNase-treated HB-1 cells (data not shown). Thus, crude DNA from a strain that is not dependent on eDNA for biofilm formation is apparently able restore the eDNA-dependent biofilm formation of strain HB-1.
Role of NhbA and the α-peptide of IgAp in biofilm formation
Since eDNA from the wild-type and nalP-mutant bacteria are apparently equally effective in mediating biofilm formation, we next considered the possibility that NalP exerts its effect on biofilm formation by its capacity to release other proteins from the cell surface. NalP is known to cleave other autotransporters (van Ulsen et al., 2003; van Ulsen et al., 2006), the LbpB component of the lactoferrin receptor (Roussel-Jazédé et al., 2010) and the heparin-binding protein NhbA (Serruto et al., 2010). Since heparin, like DNA, is highly negatively charged, we decided to investigate the possible role of NhbA in biofilm formation. NhbA binds heparin in vitro through an Arg-rich region located at amino acid residues 296–305. This region, which might also be involved in binding eDNA, is removed by NalP-mediated cleavage (Serruto et al., 2010). Hence, it is conceivable that NalP exerts its negative effect on biofilm formation by cleaving NhbA. It should be noted, however, that NalP cleaves a substantial proportion (> 50%), but not all NhbA molecules expressed by the bacteria (Serruto et al., 2010). Therefore, even when NalP is expressed, still some intact NhbA molecules remain present at the bacterial cell surface, which could be responsible for the lower level of eDNA-dependent biofilm formation in the wild-type strain.
To investigate the involvement of NhbA in biofilm formation, we constructed an nhbA deletion mutant. Indeed we found that the nhbA mutation in strain HB-1 dramatically impaired biofilm formation under static conditions (Fig. 5A). This phenotype was complemented when an ectopic copy of the nhbA gene was introduced on the chromosome in the hrtA locus (Fig. 5A). In contrast, the nhbA mutation did not affect the eDNA-independent biofilm formation in strain BB-1 (Fig. 5A) consistent with our hypothesis that NhbA participates in biofilm formation by binding eDNA.
The effect of the nhbA mutation on biofilm formation was also investigated in the continuous flow system. The development of the biofilm is shown in Movie S4, and its architecture is compared with that of the wild type and the nalP mutant in Fig. 3. After 14 h of growth, the biofilm biomass of the nhbA mutant of strain HB-1 covered ∼ 20% of the substratum (cf. 40–60% in the case of strain HB-1) with very large intervening spaces within the biomass (Fig. 3A), in agreement with the reduction of biofilm formation observed under static conditions. When the flow was increased 50-fold for 10 min, the biofilm of nhbA mutant was completely removed from the flow cell, while biofilms of the wild-type strain and the nalP mutant remained unaltered (data not shown). These observations indicate that NhbA has, directly or indirectly, an important role in the binding of the bacteria to the substratum. Because of the poor association of the biofilm with the substratum, comstat analysis revealed a drastic reduction of the roughness coefficient of the biofilm as compared with the wild type (Fig. 3B). The substratum coverage of biofilms of the nhbA mutant in BB-1 was similar to that of the wild-type but minor differences were observed (Fig. 3A). Visual inspection of the biofilm architecture suggested a slight increase of intervening spaces within the biomass as compared with the wild type (Fig. 3A). The roughness coefficient was slightly reduced (Fig. 3B), but the difference was not statistically significant. In contrast to the biofilms of the nhbA mutant of HB-1, this biofilm was not affected by increase of the flow rate, confirming that NhbA has no essential role in the binding of this eDNA-independent biofilm-forming strain to the substratum. All together, these observations show that NhbA plays an important role in biofilm formation via the eDNA-dependent route.
As shown in Fig. 1A, inactivation of nalP in strain HB-1 drastically enhances biofilm formation. If NalP exerts its negative effect on biofilm formation via cleavage of NhbA, one would expect that inactivation of nalP in the nhbA mutant does not stimulate biofilm formation. However, we found that biofilm formation in an nhbA nalP double mutant of HB-1 was higher than that in the ΔnhbA single mutant but not significantly higher than that in the HB-1 parental strain (Fig. 5A). Therefore, NalP-mediated cleavage of NhbA explains partially, but not totally, the negative effect of NalP on biofilm formation.
To identify additional NalP targets involved in biofilm formation, we considered the role of NalP in the processing of other autotransporters. Both IgAp and App are synthesized with a so-called α-peptide attached. These α-peptides contain nuclear targeting signals (Pohlner et al., 1987), which are highly positively charged. NalP releases IgAp and App from the cell surface with the α-peptide attached. In the absence of NalP, shorter versions of these proteins lacking the α-peptide are released through autocatalytic processing (van Ulsen et al., 2003). The α-peptides, which remain at the cell surface in the absence of NalP (V. Roussel-Jazédé, unpubl. obs.), may bind DNA and explain the increased biofilm formation of the ΔnalP ΔnhbA double mutant relative to the ΔnhbA single mutant. To test this possibility, we switched to strain HB-2, which, like HB-1, is an unencapsulated derivative of strain H44/76, but the capsule-locus deletion is markerless allowing for the introduction of additional mutations. HB-2 is as efficient in biofilm formation as HB-1 (data not shown) and single mutations in nalP, app, iga and nhbA, and an nhbA nalP double mutation all showed similar effects on biofilm formation as observed in HB-1 (Fig. 5B; cf. Figs 1A and 5A). Next, iga and app were disrupted in the ΔnalP single mutant as well as in the ΔnalP ΔnhbA double mutant. Deletion of app did not significantly affect biofilm formation in either of the strains (Fig. 5B). However, inactivation of iga resulted in a drastic reduction in biofilm formation already in the ΔnalP single mutant (Fig. 5B), consistent with the hypothesis that the α-peptide of IgAp that remains exposed at the cell surface in a nalP mutant contributes to biofilm formation. Disruption of the iga gene in the ΔnalP ΔnhbA double mutant reduced biofilm formation to similar very low levels as observed in HB-2 ΔnhbA (Fig. 5B). The various mutations introduced in strain HB-2 did not alter the growth characteristics (Fig. S1F). Together, our results demonstrate a role for NhbA and the α-peptide of IgAp in biofilm formation, presumably by binding eDNA. The presence of either one of these proteins on the bacterial cell surface will stimulate eDNA-dependent biofilm formation (Table 2). When both proteins are present, as in the case of the ΔnalP mutant, biofilm formation is increased. In contrast, when neither of the proteins is present, as is the case in the ΔnhbA single mutant or in the ΔnalP ΔnhbA Δiga triple mutant, eDNA-dependent biofilm formation is drastically reduced (Table 2). Thus, while NhbA and the α-peptide of IgAp stimulate eDNA-dependent biofilm formation, NalP regulates biofilm formation by cleaving the positively charged polypeptides from the cell surface.
Table 2. Relation between the amount of surface-exposed NhbA and α-peptide of IgAp and the efficiency of biofilm formation in various mutant derivatives of strain HB-2
ΔnalP ΔnhbA Δiga
NhbA and the α-peptide of IgAp are DNA-binding proteins
We next asked if indeed NhbA and the α-peptide of IgAp bind DNA. To test this possibility, the polypeptides were expressed as recombinant proteins in Escherichia coli and purified. The DNA-binding ability of both proteins was tested using electrophoretic mobility shift assays. Both recombinant proteins retarded the migration of the target DNA on agarose gels, while no gel shift was observed when the DNA was incubated with bovine serum albumin (BSA) as a control (Fig. 6). These results confirm the DNA-binding capacities of NhbA and the α-peptide of IgAp.
Biofilms are highly structured communities of bacteria on solid surfaces, which can confer protection against antibiotics or cellular defence systems of the host (Leid et al., 2005; Jensen et al., 2010). The capacity of N. meningitidis to form biofilms on biotic and abiotic surfaces has previously been demonstrated (for a review, see Lappann and Vogel, 2010), but the mechanisms required for their formation have only been started to become elucidated. It has been described that eDNA, released by autolysis, can serve as an important matrix component of biofilms in N. meningitidis (Lappann et al., 2010; Lappann and Vogel, 2010) and in other microorganisms (Whitchurch et al., 2002). However, how eDNA stimulates biofilm formation was not known. Here, we identified three cell-surface-exposed meningococcal proteins implicated in eDNA-dependent biofilm formation, i.e. NhbA, the α-peptide of IgAp and NalP.
NhbA was previously identified as a protein that binds heparin and heparan sulphate proteoglycans (Serruto et al., 2010). Since both structures, like DNA, are negatively charged macromolecules, we considered the possibility that NhbA binds eDNA and in this way contributes to biofilm formation. In fact, heparin columns are classically used in biochemistry to isolate DNA-binding proteins, already indicating that NhbA might be a DNA-binding protein. Indeed, we could demonstrate in gel shift experiments that NhbA binds DNA, and we also demonstrated that inactivation of the nhbA gene drastically reduces eDNA-dependent biofilm formation but not the eDNA-independent initiation of biofilm formation that occurs in some meningococci, like strain BB-1. Our demonstration that NhbA is involved in eDNA-dependent biofilm formation may be relevant for other pathogens as well. Many pathogens are known to produce proteins that bind heparin or heparan sulphate, both linear, highly sulphated glycosaminoglycans (Rostand and Esko, 1997). By binding glycosaminoglycans, these proteins are thought to mediate adherence to host tissues and also to increase resistance to the bactericidal activity of serum. Our observation that NhbA of N. meningitidis mediates eDNA-dependent biofilm formation suggests that also the heparin-binding proteins of other bacteria might have an additional or perhaps even primary role in biofilm formation.
IgAp of Neisseria gonorrhoeae was the first autotransporter ever described (Pohlner et al., 1987). It is synthesized as a large precursor with an N-terminal signal sequence that directs transport across the inner membrane via the Sec machinery and a C-terminal β-barrel domain that inserts into the outer membrane and mediates the transport of the passenger domain across the outer membrane. The passenger consists of three moieties: the protease domain of ∼ 106 kDa, a small γ-peptide of 3.1 kDa and an α-peptide of ∼ 45 kDa (Pohlner et al., 1987). In N. meningitidis, either the entire passenger is release from the bacterial cell surface by NalP-mediated processing or only the protease domain is released by autocatalytic processing (van Ulsen et al., 2003). In the latter case, the α-peptide remains attached via the β-barrel domain at the bacterial cell surface (V. Roussel-Jazédé, unpubl. results). The function of the large α-peptide is not really known. It has been shown that it contains nuclear targeting signals that indeed can direct an attached reporter protein to the nucleus of eukaryotic cells (Pohlner et al., 1995). However, it is not known whether IgAp is indeed transported to the nucleus at any stage of the commensal or pathogenic lifestyle of the bacteria and what the function of nuclear localized IgAp could be. Even more enigmatic is the possible role of the nuclear targeting signals that remain on the bacterial cell surface when NalP is not expressed. Since nuclear targeting signals consists of stretches of positively charged amino acids, we considered the possibility that the α-peptide of IgA protease, like NhbA, binds DNA and is involved in this way in eDNA-dependent biofilm formation. We could indeed demonstrate in gel shift assays that the α-peptide binds DNA. Furthermore, our results demonstrate that iga expression stimulates eDNA-dependent biofilm formation if NalP-mediated release of the α-peptide from the cell surface is prevented. Also another autotransporter, App, is synthesized with an α-peptide containing nuclear targeting signals, which is released from the cell surface by NalP-mediated processing (van Ulsen et al., 2003). However, inactivation of app did not significantly affect biofilm formation in a ΔnalP mutant or a ΔnalP ΔnhbA double mutant, suggesting that the α-peptide of App, in contrast to that of IgAp, has no role in biofilm formation. Probably, this is due to the much lower expression level of App as compared with that of IgAp (van Ulsen et al., 2003).
NalP is a secreted subtilisin-like serine protease of N. meningitidis. Disruption of the nalP gene revealed earlier its implication in the processing of other secreted autotransporters (van Ulsen et al., 2003) and in the cleavage of cell-surface-exposed lipoproteins (Roussel-Jazédé et al., 2010; Serruto et al., 2010). However, the physiological role of these proteolytic events has not been fully elucidated. Here, we report that NalP expression impacts on biofilm formation through its cleavage of IgAp and NhbA. In a wild-type strain, NalP releases IgAp from the cell surface with the α-peptide attached (van Ulsen et al., 2003), and a considerable proportion of the NhbA molecules is cleaved resulting in the release of the positively charged heparin-binding domain into the medium (Serruto et al., 2010). The eDNA-dependent biofilm formation in such strain is largely due to the remaining intact NhbA molecules on the cell surface, since inactivation of the nhbA gene abrogates biofilm formation. If NalP is not expressed, the number of positively charged polypeptides on the cell surface, i.e. uncleaved NhbA molecules and α-peptides of IgAp (and App), drastically increases, resulting in an increased capacity to bind eDNA. Indeed, eDNA-dependent biofilm formation is also drastically increased in such a strain, and NhbA and the α-peptide of IgAp synergistically contribute to the process as appears from the reduction in biofilm formation in double and triple mutants (Fig. 5B). Now, also the decreased electrophoretic mobility of eDNA from the ΔnalP mutant can be explained. This is not due to the binding of positively charged polypeptides; these are released into the medium of the wild-type strain. However, N. meningitidis is well known for its high capacity to release outer-membrane vesicles (OMVs) into the medium. If the outer membrane contains large quantities of intact NhbA and α-peptides, as is the case in the nalP mutant, these OMVs will bind eDNA with high affinity resulting in the observed protease- and detergent-sensitive decrease in the electrophoretic mobility of the eDNA.
It is important to note that the expression of nalP is phase variable due to slipped-strand mispairing, implying that also the presence of intact NhbA and α-peptides at the bacterial cell surface is indirectly controlled by phase variation. Thus, NalP acts as a regulator of biofilm formation. Probably, NalP-mediated variation of surface-associated NhbA and α-peptides affects the distribution of cells between the sessile and planktonic community, which may contribute to dispersal and colonization of new niches and hosts. Also, it is interesting to note that NalP expression affected biofilm formation not only in strain HB-1, but also in strain BB-1 (Fig. 1A) and other strains of cc8 and cc11 (Fig. S2B). It has been proposed that meningococci use two different strategies for biofilm formation, one dependent and the other independent of eDNA (Lappann et al., 2010). We confirmed that HB-1 and BB-1 are representatives of both pathways. However, although initiation of biofilm formation in strain BB-1 was insensitive to DNase, we noticed that inactivation of nalP in this strain enhanced biofilm formation in a DNase-sensitive way (Fig. 4A). Thus, with an increased number of positively charged DNA-binding polypeptides at the cell surface, the initiation of biofilm formation is stimulated also in strain BB-1. In addition, after longer periods of biofilm development, eDNA is apparently also implicated in biofilm formation of the wild-type strain BB-1 as evidenced by the negative impact of DNase I treatment on biofilm formation (J. Arenas, unpubl. obs.) and lattice formation (Fig. 2B). Thus eDNA enhances biofilm formation also in strains from cc8 and cc11. From our results, we conclude that NalP serves as a regulator of the binding of eDNA to the cell surface and that this is relevant for biofilm formation, not only in derivatives of cc32, but also in those of cc11 and cc8.
In summary, our study makes an important contribution to the understanding of the mechanism of biofilm formation in the human pathogen N. meningitidis and its control through a serine protease. We have demonstrated that two positively charged DNA-binding proteins, the heparin-binding protein NhbA and the α-peptide of IgA protease, are involved in eDNA-dependent biofilm formation. In addition, we demonstrated that the autotransporter proteases NalP, whose expression is phase variable, regulates biofilm formation by releasing those positively charged proteins from the cell surface. The identification of positively charged proteins involved in eDNA-dependent biofilm formation may impact also on studies of biofilm formation in other pathogens, where, likely, positively charged surface proteins may play a similar role. In addition, it may help to develop strategies to prevent biofilm formation.
Bacterial strains, plasmids and growth conditions
All strains and plasmids used are described in the Table 1. Disease isolates were from the Netherlands Reference Laboratory for Bacterial Meningitis (NRLBM), Amsterdam. They were classified in clonal groups based upon sequence analysis of conserved house-keeping genes and comparison of the results with the data on the Neisseria Multi Locus Sequence Typing (MLST) website (http://pubmlst.org/neisseria/). All meningococcal strains were grown at 37°C on GC medium base (Difco) supplemented with IsovitaleX (Becton Dickinson) in a humid atmosphere containing ∼ 5% CO2 or in tryptic soy broth (TSB) (Beckton Dickinson) at 37°C. For shaking growth conditions, bacteria were diluted in TSB to an optical density at 550 nm (OD550) of 0.1, and 5 or 25 ml of the suspensions were incubated in 25 cm2 polystyrene cell-culture flasks (Corning) or 250 ml Erlenmeyer flasks, respectively, with constant shaking at 110 r.p.m. For static cultures, bacteria from shaking cultures were diluted in TSB to an OD550 of 1, and 3 ml of the suspensions were added to six-well polystyrene plates and incubated for various time periods as described above. E. coli strains DH5α and BL21(DE3) (Invitrogen) were grown in Luria-broth (LB) containing appropriate antibiotics as required and incubated at 37°C. To induce the expression of genes from plasmids, bacteria were grown in LB or TSB containing 0.1 mM IPTG.
The nhbA knockout construct was obtained as previously described (Serruto et al., 2010). Briefly, two DNA fragments located upstream and downstream of nhbA were amplified by PCR from chromosomal DNA from stain HB-1 using primer pairs 2132-up-for-BglII (gcgcgcagatctggcgtttatgccttctttac) with 2132-up-rev-NcoI (cgcgcgccatgggggcatcatctccttcatcgtatt) and 2132-dw-for-NdeI (gcgcgccatatggattgatgttgatgcc) with 2132-dw-rev-EcoRV (gcgcgcgatatcgaggggtatctactcgcaaag). Both PCR products were purified, digested with appropriate restriction enzymes, for which sites were included in the primers, and sequentially cloned into pUC21 with in between a kanamycin-resistance (kan) cassette derived from pKOnalP-kan to yield plasmid pKOnhbA-kan. The kan cassette was replaced by chloramphenicol-resistance (cat) cassette derived from pCRII-skp2-cat (Volokhina et al., 2011) via SalI digestion to yield the plasmid pKOnhbA-cat. For complementation studies, a DNA fragment containing the nhbA gene and its flanking regions was amplified by PCR from chromosomal DNA of strain HB-1 using the primer pair GNA2132-FW-PpuMI (gcgcgcgcaggacccggcggaataaaccaagctat) and GNA2132-REV-HpaI (gcgcgcgcgttaactgccgcagttgggcaataagcaata). The resulting PCR product was digested with PpuMI and HpaI and inserted into the hrtA locus contained on plasmid pCRT_A_hrtA. Next, the cat cassette was amplified from pEN300 using primers hrtA_CAM_For (gcgcgcgttaacttaagggatgcataaactgc) and hrtA_CAM_Rev (gcgcgcgttaacttacgccccgccctgccactc) and inserted immediately after the nhbA locus via HpaI digestion to yield plasmid phrtA-nhbA.
Universal primers M13F and M13R were used to amplify a DNA segment encoding the α-peptide of IgAp of strain H44/76 (amino acid residues 1005–1182) from plasmid pET_IgA-AP by PCR. Primers 2132FW-OE-NdeI (gcgcgccatatgtcaaaacctgccgccgcccctgtt) and 2132Rev-OE-XhoI (gcgcgcctcgagttaatcgccgctgtcgataatgc) were used to amplify a DNA fragment corresponding to amino acid residues 35–420 of NhbA from genomic DNA of HB-1. The PCR products were digested with NdeI and BamHI and with NdeI and XhoI, respectively, and ligated into pET16b digested with the same enzymes, yielding plasmids pET-H-IgApAP and pET-H-NhbA respectively. The correct insertion of the fragments was confirmed by PCR and subsequent sequencing.
Capsule synthesis was inactivated by transformation of meningococci either with a PCR product obtained from plasmid pRIT16845 resulting in the replacement of the complete capsule B locus by an erythromycin-resistance (ery) cassette, or with entire plasmid pHC10 resulting in inactivation of the siaD gene in the capsule C locus by a cat cassette as described (Ram et al., 2003). To replace chromosomal nalP, app or iga by a kan cassette, previously described knockout constructs were used (Table 1). To replace nalP on the chromosome by a cat cassette, the kan cassette on pKOnalP-kan was replaced by a cat cassette derived from pEN300 via SalI digestion, and the resulting construct pKOnalP-cat was used for gene replacement. To obtain nhbA mutants, strains were transformed with PCR products from knockout constructs. For introduction of an ectopic copy of the nhbA locus on the chromosome, the entire plasmid phrtA-nhbA was used with selection for chloramphenicol-resistant transformants.
Meningococci were transformed as previously described (Volokhina et al., 2011). For all mutants generated in this study, PCR assays and, when appropriate, different immunoassays were used to confirm that the correct knockouts were obtained.
Biofilm formation under static conditions
Biofilms were formed under static conditions as described previously (Lappann et al., 2010) with some modifications. Bacteria from 4 h shaking cultures in TSB were adjusted to an OD550 of 1. Samples of 0.1, 0.5 or 3 ml per well were seeded in 96-, 24- or 6-well polystyrene plates (Corning Incorporated) and incubated for different time periods at 37°C. After incubation, the culture was removed and the adherent bacteria were washed with de-ionized water. For quantification, the biofilms were stained with 0.25% crystal violet during 2 min, washed with de-ionized water, solubilized in 33% of acetic acid and quantified by measuring the OD630. The OD630 measured after a similar staining procedure of empty wells was used as blank in each experiment.
Biofilm formation under flow conditions
Biofilms were cultivated in three-channel flow cells for inverted microscopes (cover glass mounted at the underside) with channel dimensions of 1 × 4 × 40 mm (BioCentrum-DTU, Denmark). The flow cell system was assembled and sterilized as described (Weiss Nielsen et al., 2011). The substratum for biofilm growth was a 24 × 50 mm borosilicate cover glass, size 1.5 (VWR International BV, the Netherlands). For inocula preparation, bacteria from 6-h-old shaking cultures in TSB were adjusted to an OD595 of 0.1 in TSB and incubated in the same conditions. After 2 h, the growth culture was diluted in TSB to OD595 of 0.08, and 200 μl was injected directly into the flow chamber. The flow cell was transferred to the microscope and maintained at 37°C for 1 h without flow in order to allow the attachment of the bacteria to the substratum. Next, the medium was pumped at a flow rate of 0.33 mm s−1 by using a peristaltic pump (Ismatec IPC 16), resulting in a refresh-rate of the flow chamber of 2 min.
To compare the difference in attachment strength of the biofilms to the substratum, the flow rate was increased 50-fold after 14 h growth and disintegration of the biofilm in the channel was monitored visually.
Microscopy, image analysis and film preparation
Microscopic examination of bacterial biofilms and image acquisition was performed using a Leica TSC SP5 inverted microscope equipped with a HCX PL APO 40×/0.85 objective (Leica Microsystems, the Netherlands). The microscope was encased in a dark environment chamber that was maintained at 37°C. Biofilm development was monitored for each channel at four positions every 7 min using bright-field imaging. To create a time-lapse movie of biofilm development, the images were combined to a movie using Leica LAS AF software.
In some experiments, the flow was stopped after 14 h of biofilm development, channels were clamped on the inflow side, and the biofilm was stained by slowly injecting 200 μl of a preparation of LIVE/DEAD BacLight bacterial viability stain (Life Technologies Europe BV, the Netherlands) dissolved in Dulbecco's Phosphate-Buffered Saline (DPBS, Lonza, USA) into the channel. Biofilm architecture images containing vertical and horizontal cross-sections were generated from each channel using Leica LAS AF. After image acquisition, structural parameters of the biofilm were analysed using comstat program (Heydorn et al., 2000).
Determination of eDNA in planktonic cultures
eDNA samples of cultures were prepared as described (Lappann et al., 2010), and quantified using the PicoGreen double-stranded DNA quantification reagent (Invitrogen) according to the manufacturer's instructions. The samples were excited at 485 nm and the emission was measured at 535 nm using a fluorescence microtitre-plate reader (FLUOstar OPTIMA, BMG Labtech). The results of three independent cultures were considered for statistical analysis (t-test at P < 0.05).
Extraction and treatment of crude DNA and chromosomal DNA
Chromosomal DNA was extracted and purified using standard methods (Maloy, 1990).To obtain crude meningococcal DNA, bacteria were grown to an OD550 of ∼ 3 in TSB and then incubated without shaking for 48 h at 37°C. Then, the cells were gently resuspended and the cultures were centrifuged for 10 min at 10 397 g. The resulting supernatant was supplemented with NaCl to a final concentration of 0.25 M. The crude DNA was precipitated by addition of two volumes of ethanol and subsequent centrifugation. The precipitate was dissolved in fourfold diluted Tris/EDTA buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.4). The DNA concentration was determined spectrophotometrically at 260 nm. When required, crude DNA was diluted in Tris buffer without EDTA and incubated with 100 μg ml−1 of filtered DNase I, proteinase K or RNase A (all from Sigma-Aldrich), or, as control, with Tris buffer for 2 h at 37°C. Subsequently, the enzymes were inactivated at 95°C during 20 min. In other experiments, crude DNA was incubated with 10% Elugent (Calbiochem, USA) in Tris/EDTA buffer for 1 h at room temperature. The activity of the enzymes and the effect of the detergent were examined in SDS-PAGE and agarose gels.
DNase I treatment and biofilm restoration experiments
Bacteria were treated with DNase and biofilm formation of DNase-treated bacteria was restored with crude DNA extracts as described (Lappann et al., 2010) with some modifications. Briefly, filtered DNase I was added to a final concentration of 100 μg ml−1 to static cultures to degrade eDNA. For restoration of biofilm formation, DNase-treated bacteria were washed twice with TSB by centrifugation in a biofuge (9300 g, 10 min), adjusted to an OD550 of 1, and 100 μl of samples were placed in 96-well plates. Where appropriate, 100 μl of Tris/EDTA buffer containing or not 200 μg of crude DNA was added to the cultures. Biofilms were quantified after 1 h of incubation.
SDS-PAGE, immunoassays and antisera
Bacterial samples for SDS-PAGE analysis were prepared as previously described (van Ulsen et al., 2003; Volokhina et al., 2011). Before SDS-PAGE, samples were boiled for 10 min in sample buffer containing β-mercaptoethanol. SDS-PAGE was performed using 12% (w/v) acrylamide gels. After electrophoresis, proteins were stained with Coomassie Brilliant Blue G250 or transferred to nitrocellulose membranes, which were subsequently incubated with PBS containing 0.1% (v/v) Tween 20 (Merck) (PBS-T) and 0.5% (w/v) non-fat dried milk (Protifar; Nutricia) (PBS-T-M). All incubations with antisera were carried out for 1 h at room temperature in PBS-T-M and followed by extensive washing in PBS-T. After incubation with appropriately diluted primary antibodies, the blots were incubated with horseradish peroxidase-conjugated goat anti-rabbit or -mouse IgG antisera (Biosource International) and developed by chemiluminescence using the Pierce ECL Western Blotting Substrate. Polyclonal antisera directed against the β-barrel domains of NalP (anti-NalPβ) (Oomen et al., 2004), IgAp (van Ulsen et al., 2003) and App (van Ulsen et al., 2003) were used. Antiserum for the detection of NhbA was kindly provided by GlaxoSmithKline Biologicals. For the detection of His-tagged recombinant proteins, a commercial mouse anti-His mAb was used (Thermo Fisher Scientific, USA).
Dot blotting for the detection of capsular polysaccharide B was performed as described (Arenas et al., 2006), but using heat-inactivated bacteria (56°C during 1 h). For this assay, the anti-MBPS monoclonal antibody was used (Frosch et al., 1985).
Overproduction and purification of recombinant proteins
Escherichia coli strain BL21(DE3) containing pET-H-IgApAP or pET-H-NhbA was used for the production of recombinant His-tagged α-peptide of IgAp or NhbA respectively. Strains were grown at 37° in LB to an OD600 of 0.6, when expression was induced by adding IPTG at 0.1 mM. After further incubation for 2 h, cells were harvested by centrifugation and broken by sonication in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8.0). The resulting lysate was centrifuged (10 000 g, 30 min, 4°C) to remove cell debris, and recombinant proteins were isolated from the supernatant by Ni-NTA agarose chromatography (Qiagen, USA) according with manufacturer's instructions.
The purity and identity of isolated proteins was confirmed by SDS-PAGE gels stained with Coomassie Brilliant Blue G250 and Western blots using anti-His antibodies. Concentrations of purified proteins were determined with the Pierce protein assay kit (Pierce, Rockford, IL). Purified proteins were aliquoted and stored at −20°C until use.
Electrophoretic mobility shift assay
To test the DNA-binding properties of proteins, the target DNA, plasmid pKOnhbA-kan, was first digested with NdeI for 1 h at 37°C. After heat inactivation of the enzyme, 450 ng of the plasmid were incubated with purified His6-NhbA, His6-α-peptide of IgAp or BSA as a control for 30 min on ice in Tris/EDTA buffer. The DNA was visualized after electrophoresis on 1% agarose gels stained with ethidium bromide.
All data obtained from biofilm formation experiments under static conditions were expressed relative to wild-type values before statistical analysis. Data from three independent experiments performed in duplicate were considered for statistical comparison. To determine alteration of biofilm characteristics under flow conditions, at least five image stacks randomly generated of each sample in a representative experiment, in which all strains were grown simultaneously but in different channels, were considered. Statistical comparisons were made using an unpaired statistical t-test and statistically significant differences were marked at *P < 0.05 and **P < 0.001.
We would like to thank Martine Bos and Vincent van Dam (Utrecht University), Peter van der Ley (Netherlands Vaccine Institute), Arie van der Ende (Amsterdam Medical Center) and GlaxoSmithKline Biologicals for strains, plasmids and antibodies used in this study. We gratefully acknowledge Tom Zalm, Florian Putker, Jan Grijpstra and Leticia Villalba (Utrecht University) for assistance in some of the experiments and Peter van Ulsen (Vrije Universiteit Amsterdam) for discussion. This work was supported by the Netherlands Organization for Health Research and Development (ZonMw). T.N.P.B. was partially supported by the King Abdullah University of Science and Technology (Grant No. № KUK-C1-017-12).