Neisseria meningitidis GNA2132, a heparin-binding protein that induces protective immunity in humans

GNA2132 Is Immunogenic During Disease and Induces Bactericidal Antibodies in Humans.

The GNA2132 protein has a signal peptide with a typical lipobox motif (-LXXC-) and in the Nm MC58 strain has a predicted molecular weight of 50,553 Da. The protein contains an Arg-rich region (-RFRRSARSRRS-) located at position 296–305 that is highly conserved among different Nm strains. The protein is specific for Neisseria species, as no homologous proteins were found by searching nonredundant prokaryotic databases. GNA2132 antigen is present and expressed in all Nm strains analyzed so far, and analysis of the putative promoter and the coding sequence did not reveal any nucleotide sequence that might be involved in phase variation.

To investigate whether GNA2132 is expressed in vivo during infection we used a quantitative dot-blotting technique to probe the recombinant antigen with sera from 22 patients convalescing after meningococcal disease. All of the sera tested recognized GNA2132 protein as well as the tetanus toxoid used as positive control, but they did not recognize an unrelated his-tagged protein used as negative control (Fig. 1). Having established that the protein is expressed and is immunogenic during infection, we asked whether GNA2132 is able to induce bactericidal antibodies in humans immunized with the recombinant MenB vaccine that contains GNA2132, NadA, and fHBP.

Given the complex nature of the vaccine, to detect only the bactericidal antibodies against GNA2132, we constructed a target strain susceptible only to anti-GNA2132 antibodies. Starting from the Nm isolate named 5/99, we first deleted the gene coding for NadA, then we deleted the gene coding for the endogenous GNA2132, obtaining the strain 5/99ΔnadAΔgna2132 (5/99ΔΔ). The fhbp gene was not removed as 5/99 strain expresses a low level of variant 2.8 of fHBP that is not recognized by the antibodies induced by the vaccine. In a second step, we introduced in the 5/99ΔΔ strain the gna2132 gene from the NZ98/254 strain under the control of an isopropyl-1-thio-β-D-galactopyranoside (IPTG)–inducible promoter. The final recombinant strain was named 5/99ΔΔC2132. The recombinant strain expressed the GNA2132 antigen and the level of expression was proportional to the concentration of IPTG added (Fig. 2A). Sera from mice immunized with the individual recombinant antigens showed that the 5/99ΔΔC2132 strain, expressing levels of GNA2132 comparable to those found in several clinical isolate, was susceptible to bactericidal killing by sera raised against GNA2132 and the MenB vaccine but was not susceptible to killing by sera raised against all of the other vaccine components (Fig. 2B). However, the new strain was still susceptible to anti-capsular, anti-PorA, anti-PorB, and anti-OMV antibodies (Fig. 2B), confirming that the genetic manipulation did not affect the sensitivity to complement-mediated killing. Finally, the two strains with and without GNA2132 were tested in the bactericidal assay, using human complement and five sera from adults immunized with the recombinant MenB vaccine. The strain 5/99ΔΔ, not expressing GNA2132, was not killed by human sera and showed negative titers comparable between the preimmune and the immune sera. Conversely, when GNA2132 was expressed, the 5/99ΔΔC2132 strain was killed by all of the sera tested with titers ranging from 16 to 64 (Fig. 2C). When increasing concentrations of IPTG were used, the bactericidal titer increased in parallel with the GNA2132 expression, confirming that the measured bactericidal titers correlate with the expression of GNA2132 (Table S1). We conclude that GNA2132 induces bactericidal antibodies in humans and that the bactericidal activity depends on the amount of the antibodies and the level of expression of the GNA2132 protein.

GNA2132 Binds Heparin Through the Arg-Rich Region.

Arg and Lys residues are present in the heparin-binding sites of different proteins (9), where they are able to interact with negatively charged sulfonyl groups of heparin or other sulfated polysaccharides. We therefore tested the ability of GNA2132 to mediate a direct interaction with heparin-like molecules via the Arg-rich region using a recombinant his-tagged form of the protein. Affinity chromatography with heparin as the ligand showed that the full length rGNA2132^MC58-his molecule bound heparin. To define the role of the Arg-rich region in the interaction, we generated a deletion mutant of the Arg-rich region (ΔRR-his) and a mutant wherein all Arg residues were substituted with a Gly (mRR-his; Fig. 3A). Affinity chromatography showed that neither of these mutants bound heparin confirming the key role of this motif in the interaction (Fig. 3 A and B). Furthermore, native gel electrophoresis was used to confirm the ability of GNA2132 to bind heparin. Incubation with heparin resulted in a different migration of the GNA2132 and the same behavior was observed when heparan sulfate proteoglycan (HSPG) was used as the ligand (Fig. 3C).

The interaction of GNA2132 with heparin was further analyzed by surface plasmon resonance (SPR). To this end, heparin was biotinylated and the resulting product was characterized by ¹H NMR (Fig. S1). Biotinylated heparin was immobilized onto a streptavidin Biacore chip. The WT rGNA2132^MC58-his and the mutant mRR-his were injected at different concentrations and the resonance units (RU) of binding on immobilized heparin was detected and reported in sensorgrams (Fig. 3D and Table S2). A clear interaction was evident between immobilized heparin and the rGNA2132^MC58-his. The binding RU values indicated that the presence of the Arg-rich region gives a significantly higher heparin binding capacity to the protein (Table S2). Overall these results prove that the binding of GNA2132 to heparin is mediated by a specific heparin-binding domain constituted by the Arg-rich region.

GNA2132 Increases Serum Resistance in the Presence of Heparin.

The binding of heparin to bacteria has been reported to increase resistance to the bactericidal activity of normal human serum (10). To determine if the heparin binding protein GNA2132 plays a role in resistance of Nm to normal human serum, we selected the unencapsulated ∆siaD mutant of Nm strain Y2220 (Y2220ΔsiaD). This strain was selected because it does not bind detectable amounts of factor H (11), and therefore we could measure increasing serum resistance in the absence of capsule and factor H, which are already known to be involved in serum resistance. The survival of the unencapsulated Y2220 strain lacking GNA2132 (YΔ2132) was compared with the survival of a complemented strain expressing GNA2132 from an IPTG inducible promoter (YΔ2132-C2132). The role of heparin in serum resistance was assessed by incubating the strains with 50 U of heparin before performing the bactericidal assay (BCA). In the absence of heparin, none of the strains survived in normal human serum (NHS) (Fig. 4A). In the presence of heparin, >50% of Y2220 expressing GNA2132 survived while there was not increased survival in the Y2220 strain lacking GNA2132.

The Neisseria gonorrheae opacity proteins (Opa) have been reported to bind to heparin and to increase serum resistance (12). To confirm that the increased serum resistance of YΔ2132-C2132 observed in the presence of heparin was due to expression of GNA2132 and not to Opa proteins, we used IPTG to modulate expression of GNA2132 and then measured the ability of these strains to resist killing by NHS in the presence of heparin. The YΔ2132-C2132 strain was resistant to killing by NHS only in the presence of IPTG when GNA2132 was expressed. Addition of IPTG did not affect serum resistance of YΔ2132, which does not express GNA2132 under any growth condition. These data establish a clear relationship between expression of GNA2132 and increased serum resistance in the presence of heparin (Fig. 4B). In support of this relationship, increased expression levels of GNA2132 resulted in increased serum resistance in experiments in which the concentration of IPTG was varied (Fig. 4C).

It is worth mentioning that the increased serum resistance could not be observed in the capsulated Y2220 strain, likely because the capsule has a dominant effect in increasing serum resistance and masks the lower effect of GNA2132.

The GNA2132 Antigen Is Processed by NalP in N. meningitidis.

Western blot analysis performed on outer membrane proteins (OMPs) showed the presence of two GNA2132-specific bands in strain MC58, which were absent in the mutant strain (MCΔ2132; Fig. 5A). The first band migrated with an apparent molecular weight of approximately 60 kDa, corresponding to the full-length GNA2132 protein, and the second band at approximately 45 kDa, corresponding to a fragment of the protein. A specific band at approximately 22 kDa was identified in the supernatant, suggesting the processing of the protein and the release of a fragment (Fig. 5A). Purification of the 22-kDa protein fragment from the supernatant and subsequent N-terminal sequence analysis showed that this protein fragment starts with Ser293 and hence corresponds to the C-terminal region of GNA2132.

To evaluate whether the processing was characteristic of other Nm strains, we analyzed the expression of GNA2132 in a panel of diverse meningococcal strains. Western blot analysis revealed that GNA2132 is expressed by all strains tested. However the protein was cleaved and the C-fragment released in the supernatant in only five of 20 strains tested (Fig. 5A and Table S3). It is interesting to notice that all of the strains that expressed and cleaved GNA2132 belong to the hypervirulent clonal complex 32 (Table S3). This analysis suggests that processing of GNA2132 is a strain-dependent event, perhaps reflecting the GNA2132 amino acidic sequence or the presence of a specific protease. To further investigate these possibilities, subsequent studies were done using two strains: MC58, wherein GNA2132 is cleaved, and the NZ98/254 strain, which has a nonprocessed GNA2132. Starting from the MCΔ2132 strain, we created a strain expressing the NZ98/254 variant of GNA2132 (MCΔ2132-CNZ). Western blot analysis of OMP and S/N prepared from strains MCΔ2132, MC58, NZ98/254, and MCΔ2132-CNZ showed that the NZ98/254 variant of GNA2132 is cleaved when expressed in the MC58 genetic background, confirming the strain specificity of the processing (Fig. 5B). These data suggest that the processing of GNA2132 results from a distinct Nm protease, likely located on the bacterial surface of some, but not all, strains.

We considered NalP, a phase variable autotransporter protein with serine protease activity, to be a strong candidate for the processing of GNA2132 because NalP has been shown to modulate the processing of other surface exposed Nm proteins (13, 14).

We generated a NalP deletion mutant in strain MC58 (MCΔnalP). GNA2132 expression and processing were evaluated by immunoblotting of OMP and supernatants. In the MCΔnalP strain, a higher amount of the GNA2132 full-length protein can be detected, whereas the N- and C-fragments were not detectable, indicating that GNA2132 is no longer cleaved in MCΔnalP strain (Fig. 5C). GNA2132 was processed in only some Nm strains, and this might correlate with the finding that the nalP gene is prone to phase variation (13). Sequence analysis of the nalP gene in the panel of Nm tested confirmed a phase off gene in the strains that did not show the processing of GNA2132 (Table S3). Overall, these results confirm that NalP mediates the processing of GNA2132 upstream of the Arg-rich region and the release of the C-terminal fragment of the protein.

GNA2132 Antigen Is a Target of Human Lactoferrin.

To investigate whether human anti-infective proteases cleave GNA2132, such as lactoferrin (hLf), that recognize and cleave Arg-rich sequences to inactivate surfaced exposed virulence factors of Gram-negative bacteria (15, 16), we incubated rGNA2132^MC58-his with hLf purified from human milk. SDS/PAGE analysis showed that GNA2132 is cleaved into two fragments of approximately 43 kDa (N1) and approximately 21 kDa (C1; Fig. 5D, lanes 1 and 2). The 21-kDa fragment was subjected to N-terminal sequence analysis. The sequence obtained (245-SLPAEMPL-252) showed that the cleavage mediated by hLf occurs immediately downstream of the Arg-rich region. As the C-terminal fragment released after the cleavage of NalP contains the Arg-rich region, we asked whether this fragment might be a target of hLf. We prepared recombinant proteins C-his and C1-his and incubated them with hLf. SDS/PAGE analysis showed that, after incubation with hLf, the 22-kDa C fragment is converted to the 21-kDa C1 fragment (Fig. 5D, lanes 3 and 4). No effect was observed on the C1 fragment (Fig. 5D, lanes 5 and 6). This demonstrated that the recombinant C-his fragment containing the Arg-rich region is also a target of hLf and suggests that hLf can act on the full-length GNA2132 as well as on the secreted C fragment.

We used recombinant his-tagged forms of the N-terminal and the C-terminal regions generated by the NalP (N-his and C-his) and by the hLf cleavages (N1-his and C1-his) to evaluate their ability to bind heparin. The data showed that only fragments containing the Arg-rich region (N1-his and the C-his) are able to bind heparin, confirming the key role of the region in this interaction (Fig. S2). Analysis of the same recombinant fragments by quantitative dot-blot experiments showed that they were recognized by convalescent human sera.

Processing of GNA2132 by NalP and hLf Does Not Affect the Bactericidal Antibody Response.

We explored whether the GNA2132 processing described earlier impairs the bactericidal antibody response induced by GNA2132. BCA assays were performed using two different sera against GNA2132 in a strain containing NalP (MC58) and a strain where NalP had been deleted (MCΔnalP). Bactericidal titers observed were within one dilution, suggesting that NalP cleavage does not significantly affect the GNA2132-mediated bactericidal activity (Table 1). Then, we explored whether the action of hLf impairs the bactericidal activity of anti-GNA2132 antibodies. After the treatment of Nm with hLf or with PBS solution, BCA assays were performed. Also in this case, the treatment with hLf did not result in a significant decrease of the GNA2132-mediated bactericidal activity (Table 1). Similar results were obtained treating 5/99ΔΔC2132 with hLf and using human sera.

Bacterial Strains and DNA Manipulation.

Constructs and strains used in this study are listed in Table S1. Detailed description of bacterial strains, growth conditions, and methods for DNA manipulation and construction of Nm recombinant strains are provided in SI Materials and Methods.

Binding of Recombinant GNA2132 to Heparin.

The rGNA2132^MC58-his protein and the mutants were analyzed with a 1 mL HiTrap heparin HP column (GE Healthcare) for their ability to bind heparin. Preparation and characterization of biotinylated heparin is reported in SI Materials and Methods. Biotinylated heparin was immobilized on a streptavidin chip (SA chip; Biacore/GE Healthcare). GNA2132 and mutants were injected at different concentrations (from 0.054 μM to 1.7 μM). The analysis was developed on Biacore X100 instrumentation (Biacore/GE Healthcare). Detailed methods and further details are reported in SI Materials and Methods.

Serum Bactericidal Assay.

Susceptibility of Nm to complement mediated killing was determined using a serum bactericidal assay as described previously (5). The serum bactericidal assay with human complement and human sera was performed as described previously (37). Detailed methods are reported in SI Materials and Methods.