Research ArticleVaccine Design

Rational Design of a Meningococcal Antigen Inducing Broad Protective Immunity

See allHide authors and affiliations

Science Translational Medicine  13 Jul 2011:
Vol. 3, Issue 91, pp. 91ra62
DOI: 10.1126/scitranslmed.3002234


The sequence variability of protective antigens is a major challenge to the development of vaccines. For Neisseria meningitidis, the bacterial pathogen that causes meningitis, the amino acid sequence of the protective antigen factor H binding protein (fHBP) has more than 300 variations. These sequence differences can be classified into three distinct groups of antigenic variants that do not induce cross-protective immunity. Our goal was to generate a single antigen that would induce immunity against all known sequence variants of N. meningitidis. To achieve this, we rationally designed, expressed, and purified 54 different mutants of fHBP and tested them in mice for the induction of protective immunity. We identified and determined the crystal structure of a lead chimeric antigen that was able to induce high levels of cross-protective antibodies in mice against all variant strains tested. The new fHBP antigen had a conserved backbone that carried an engineered surface containing specificities for all three variant groups. We demonstrate that the structure-based design of multiple immunodominant antigenic surfaces on a single protein scaffold is possible and represents an effective way to create broadly protective vaccines.


The development of vaccines against many pathogens is often limited by the sequence variability of their protective antigens. When the antigenic variability is limited, as in the case of poliovirus or Streptococcus pneumoniae, vaccines have been developed by including up to 23 different antigenic variants in a single vaccine vial. However, when the antigenic variability is high, as is the case for influenza viruses, a new vaccine is needed every year. In other cases exemplified by the bacterium Neisseria meningitidis serogroup B [meningococcus B (MenB)], the protozoan parasite causing malaria, African trypanosomes, rhinovirus, and HIV, the extreme antigenic variability of these pathogens has stymied the development of a successful vaccine (1, 2). In all of these cases, the challenge for modern vaccinology is to develop antigens that are able to induce broad protective immunity against all natural antigenic variants of the pathogen. To date, this has been approached principally by trying to increase the immunogenicity of the most conserved regions of the antigen; however, these attempts have met with limited success (35).

Here, we used a different approach to develop a vaccine against MenB, an obligate human pathogen that can cause severe, often fatal, septicemia and meningitis. Poor immunogenicity, as well as the risk of cross-reactivity of type B capsular polysaccharide with human tissues, has greatly limited the development of a glycoconjugated vaccine against serogroup B meningococcal strains. Mining the genome of this bacterium recently led to the discovery of factor H binding protein (fHBP), a powerful protective antigen that binds to human factor H. Human factor H is a serum regulatory protein that provides the first line of defense of the innate immune system for protecting the human body against invading organisms. Antibodies elicited by injecting fHBP into mice are able to kill the bacterium in the presence of complement, a property that is known to correlate with protection in humans (6, 7). Sequencing the fhbp genes in almost 2000 strains (8, 9) of MenB revealed that more than 350 different nucleotide sequences exist, coding for almost 300 different polypeptides ( The sequence types with less than 10% sequence variability were found to induce some cross-protective immunity, whereas no cross-protection could be observed among more divergent sequences (10). Using this information, we divided the entire family of fHBPs into three major antigenic variants (variants 1, 2, and 3), which are present in 65, 25, and 10% of the MenB global population, respectively (8, 9, 11). The fHBP variant 1 antigen is an important component of a vaccine currently in phase III clinical trials containing five different antigens (4CMenB) recently discovered by reverse vaccinology (that is, a method for selection of vaccine candidates based on in silico analysis of bacterial genomes) (6). A second vaccine containing a mixture of the main fHBP antigenic variants 1 and 2 (or families A and B according to the different nomenclatures) is in phase II clinical trials (12).

Here, we designed a panel of chimeric fHBP antigens with the aim of combining in a single molecule the complete antigenic repertoire of the three major variant groups. When injected into mice, one of these engineered fHBP antigens was able to elicit cross-protective immunity against meningococcal strains carrying different fHBP alleles. Further engineering of this candidate led to a final molecule with increased efficacy.


Rational design of fHBP mutants

The three-dimensional (3D) structure of fHBP determined by nuclear magnetic resonance (NMR) (13, 14) and x-ray crystallography (15) showed that the protein is essentially composed of two β barrels connected by a short linker (Fig. 1). Sequence comparison of the three different variants suggested that the protein architecture is well conserved among different members of the fHBP family. Analysis of the epitopes recognized by variant-specific monoclonal antibodies (mAbs) (1620) revealed that fHBP amino acids important for recognition by the antibodies against variant 1 were Asp25, His26, Lys27, Thr56, Tyr57, Gly121, Glu146, Gly147, Gly148, Arg149, and Arg204 (colored red in Fig. 1). In contrast, the amino acids important for recognition of variants 2 and 3 were Ala174 (Lys in variants 2 and 3), Lys180 (Arg in variants 2 and 3), Asp192 (Glu in variants 2 and 3), and Gln216 (Ser or Gly in variant 2, Ser in variant 3) (colored purple in Fig. 1). In the primary sequence Arg204, one of the most important residues for epitope recognition of variant 1 falls in the middle of a region also important for variants 2 and 3 that spans Asp192 to Gln216 (yellow ribbon in Fig. 1). From the analysis of the 3D structures, it is clear that amino acids contributing to the immunogenicity of variant 1 or variants 2 and 3 are located in non-overlapping regions. This important observation suggested that the immunodominant regions of the three variant groups are distinct and that an alternative immunogenic epitope featuring variant 2 or 3 residues could be grafted onto the variant 1 backbone.

Fig. 1

3D structure of fHBP from MenB strain MC58 (subvariant 1.1). Ribbon diagrams show the distribution of the amino acids recognized by mAbs elicited by variant 1 (red) and variants 2 and 3 (purple) as reported previously (1620). Amino acid positions 174, 180, and 192 in variants 2 and 3 correspond to Lys, Arg, and Glu, respectively. In position 216, strains of variant 2 have Gly and strains of variant 3 have Ser. Ribbons showing the N- and C-terminal domains are colored blue and green, respectively. The region containing Arg204 is shown in yellow.

With the aim of generating a single antigen that is able to induce a broad antibody response against all of the fHBP variants of MenB, we initially generated, purified, and immunized mice with mutants of variant 1 containing single, double, or triple amino acid mutations derived from the sequences of variants 2 and 3. In no case were the new antigens found to induce a broad immune response, indicating that changing just a few amino acids was not sufficient to create immunodominant epitopes that could induce immune responses to variants 2 and 3. Therefore, we decided to engineer portions of the antigen surface potentially recognizable by an antibody as a whole, instead of making single amino acid mutations. We focused on the C-terminal β barrel domain of fHBP, which contains most of the amino acids recognized by antibodies against variants 2 and 3 (17). Taking into account that conformational epitopes of proteins typically range in size from about 900 to 2000 Å2 (2123), we divided the entire C-terminal β barrel domain into 11 partially overlapping areas of size large enough to hold at least one conformational epitope (table S1). Within each distinct area, residues of the variant 1 serobase protein [that is, the protein of variant 1 included in 4CMenB vaccine (7) and selected as a reference molecule] were replaced with the corresponding amino acids of variants 2 and 3 regardless of their position in the primary sequence. To preserve folding, we introduced amino acid substitutions only for residues whose side chains were well exposed to the solvent, leaving the internal core of the protein unaltered.

The different surface areas created by this structure-based mutational approach were assigned letters A to K (Fig. 2). Within each zone, groups of point mutations were introduced using the multiple sequence alignment of 85 different alleles (9) as a guideline. Depending on the number of hypervariable residues, distinct sets of substitutions were introduced within each area to reproduce all of the concurrent variations observed in the natural bacterial population. A numeric suffix was used to indicate each single mutant characterized by a specific combination of such hypervariable substitutions. A total of 54 different antigenic variants were designed (figs. S1 and S2), expressed in Escherichia coli, purified, and then used to immunize mice, using protocols established previously (6). Mouse sera were collected and tested in vitro for their ability to kill bacteria in the presence of complement. The screening assay for measuring protection induced in the mice was performed against a panel of seven meningococcal strains expressing divergent sequences of fHBP. These divergent sequences included subvariants 1.10, 1.14, 2.1, 2.4, 2.10, and 3.1 that had 93.4, 91.6, 74.1, 73.7, 69.7, and 62.8% amino acid identity compared to the serobase protein, respectively. Strain MC58 of MenB expressing the serobase protein subvariant 1.1 was used as a positive control (Table 1).

Fig. 2

Mutation of fHBP. The surface of the variant 1 fHBP reference protein is shown in gray. Backbones of the N- and C-terminal domains are colored blue and green, respectively. The different engineered regions are represented as solid surfaces. Residues conserved among the three variants are colored light blue. Residues conserved between variants 2 and 3 are colored dark red. Hypervariable residues are colored yellow. A more extensive figure with fHBP shown in three different orientations to better visualize the mutated surfaces is provided in fig. S2. A detailed representation of the eight selected candidates is also reported in fig. S3, A to G. Amino acid sequences of all the mutants generated are reported in fig. S3. Protein rendering was realized using UCSF (University of California, San Francisco) Chimera software (

Table 1

Principal features of the meningococcal strains used in this study. Meningococcal strains were selected to sample the fHBP genetic diversity observed in the bacterial population. Expression and exposure of fHBP on the bacterial cell surface was tested by Western blotting and fluorescence-activated cell sorting (FACS) analysis, respectively (figs. S1 and S2). na, not assigned; Cpx, clonal complex; ST, sequence type; NZ, New Zealand; AUS, Australia; C, Canada.

View this table:

Of the 54 mutants tested, 18 resulted in a more than 10-fold reduction in bactericidal activity of the mouse sera against the MC58 strain and were therefore discarded. Among the remaining 36 molecules, 15 showed at least a 10-fold increase in bactericidal titer against the prototypic strain of variant 2 or were also positive against variant 3. This group of 15 proteins carried mutations within 6 of the 11 different engineered areas, and so, we consequently selected a representative panel of six mutants (D5, E2, F3, G1, G2, and H9; fig. S3, A to G). Two additional mutants, B4 and J1, showing wide coverage against strains of variant 1 were also included. The results of the complete screening are reported in table S2.

Biochemical characterization of selected candidates

To verify that the mutations introduced did not cause major alterations in protein folding or stability, we analyzed the selected mutants to evaluate different biochemical and biophysical properties when compared with the serobase fHBP subvariant 1.1. In particular, the tendency to aggregate, secondary structure, and folding were monitored by size exclusion high-pressure liquid chromatography (SEC-HPLC), circular dichroism (CD) spectroscopy, and NMR, respectively. Analysis of the SEC-HPLC retention times of the serobase protein and respective mutant proteins showed that all fHBP mutants eluted as single peaks co-migrating with the original serobase protein. Comparison with retention times of reference molecules indicated that the mutants were present as monodisperse monomers and did not form aggregates (table S3A). As a second approach to confirm the quality of the proteins overall, far-ultraviolet (UV) CD spectroscopy of the same samples suggested that all of the fHBP proteins have the same secondary structure (table S3B). Subsequently, protein folding and conformation were checked in detail by expressing the selected mutants in 15N-enriched medium and analyzing them by NMR spectroscopy, which enabled analysis at the residue-specific level. Heteronuclear single-quantum correlation (1H-15N HSQC) spectra for each mutant indicate folded proteins with well-dispersed amide chemical shifts and whose total peak number matched the predicted one (Fig. 3). Most of the 1H-15N cross peaks in the 15N HSQC spectrum of each mutant overlapped those of the wild-type protein. In contrast, the major chemical shift variations were confined to the region harboring mutations on the C-terminal β barrel (residues 138 to 255). The new peaks that appeared due to point mutations also had chemical shift values indicative of a well-folded environment. Collectively, these data indicated that the mutations introduced within each mutant did not alter the overall architecture of the protein nor did they introduce regions of disorder, suggesting that the designed conformational epitopes were successfully grafted within the correct molecular scaffold.

Fig. 3

NMR spectra of fHBP. Shown are 1H-15N HSQC NMR spectra acquired at 25°C for fHBP subvariant 1.1 and mutants. The two panels on the left show the spectra acquired for wild-type (wt) fHBP in 20 mM sodium phosphate at pH 7.2 (upper panel) and 4 M urea at pH 2 (lower panel). In the remaining panels, NMR spectra acquired for each fHBP mutant are overlaid on the spectrum of the fHBP wild-type protein. Each peak represents a signal from the NH group of the protein. The wild-type spectrum is colored blue and mutant spectra are colored pink (B4), red (D5), aquamarine (E2), cyan (F3), hot pink (G1), light green (G2), sable (H9), and green (J1). The concentration of the protein samples was 1.0 mM in a 90% H2O/10% 2H2O mixture containing 20 mM sodium phosphate buffer at pH 7.2.

Bactericidal activity of selected candidates

To further investigate the extent of cross-protection that could be induced by the mutants, we immunized mice with the selected mutants formulated with an adjuvant suitable for human use composed of a mix of aluminum hydroxide and IC31 (a Toll-like receptor 9 agonist) recently used in clinical trials (24, 25). The bactericidal activity of the resulting sera was tested against an enlarged panel of MenB strains containing the serobase protein; the subvariants 1.10, 1.12, and 1.14 that are the most divergent sequences of variant 1; and the subvariants 2.1, 2.4, 2.7, 2.10, and 3.1 (Table 2). Bactericidal antibody titers elicited in mice by each mutant were compared to those obtained by immunizing mice with the homologous fHBP molecule purified from each strain. As expected, the serobase subvariant 1.1 elicited bactericidal titers >1000 against subvariants 1.1, 1.10, and 1.12, and titers of 256 or lower against distantly related 1, 2, and 3 variants. In marked contrast, the wild-type subvariant 2.1 induced no bactericidal antibody titers against variant 1, titers >1000 against all variant 2 strains, and a titer of 512 against variant 3 strains. When mice were immunized with each of the fHBP natural subvariants, immune sera had titers of >1000 against the homologous strain. The mutated molecules induced broader immunity with respect to that elicited by the fHBP subvariant 1.1. In particular, sera raised by G1, G2, and H9 were bactericidal against all of the strains tested. G1 elicited titers >1000 against most of the strains tested. Arg204, originally described as part of a bactericidal epitope of variant 1 (17), was mutated in G1 to serine without destroying variant 1 immunogenicity. This suggested that it was the larger surface area and not the individual amino acids that was important for immunogenicity.

Table 2

Bactericidal titers in mice. Shown are the bactericidal titers elicited in mice by selected fHBP mutants formulated with aluminum hydroxide plus IC31 adjuvant. The recombinant antigens were administered intraperitoneally to groups of eight mice at days 1, 21, and 35. Bactericidal activities in mouse serum are defined as the reciprocal of serum dilution that resulted in a 50% reduction in numbers of CFUs surviving compared with the results for an equivalent negative control containing heat-inactivated complement. The serobase protein from the MC58 strain (fHBP variant 1.1 wild type) and the homologous alleles of each strain have been included as positive controls. Titers of 512 or higher are in bold.

View this table:

Crystal structure of G1

To characterize further the lead candidate antigen, we determined the x-ray crystallographic structure of the fHBP mutant G1 refined to 1.9 Å resolution (Fig. 4 and table S5). A comparison of the G1 structure with the crystal structure of fHBP variant 1, which was determined in complex with factor H, shows that the backbone structures are virtually identical (root-mean-square deviation of 0.67 Å for 239 equivalent main-chain carbon atoms sharing 89% sequence identity). Only small differences in a few loop regions were observed, most notably for the loops connecting β strands β5-β6, β7-β8, and β15-β16 (Fig. 4A). These minor differences may be due to an intrinsic flexibility in the exposed loop regions, differences in crystal packing, or to the absence of factor H in the G1 structure. Electron density was clearly visible for the side chains of most of the mutated residues (colored green in Fig. 4B), confirming that the substitutions changed the exposed surface without inducing any local disorder. Comparison of the G1 structure (Fig. 4B) to the published structure of fHBP variant 1 in complex with factor H (Fig. 4C) shows an extreme similarity of the global structure composed of conserved amino acids and a distinct surface structure containing variant 2 and 3 specificities (in green) grafted onto the C-terminal portion of the molecule.

Fig. 4

Structural comparison of wild-type fHBP and the G1 mutant. Comparison of the structures of wild-type fHBP variant 1 and the chimeric G1 mutant reveals that the conserved fold displays a different immunogenic surface. (A) Ribbon display of the superimposed α carbon traces of G1 (N-terminal domain in blue, C-terminal domain in green) and fHBP variant 1 (yellow; from PDB entry 2W80) (13) reveals minor variations restricted to a few exposed loop regions (labeled). Spheres show α carbon atoms of G1 where mutations were inserted. The alignment reveals a very slight difference in domain orientation, such that β strands 13, 14, 15, and 16 are shifted by 0.2 to 1.0 Å. (B) Surface display of the G1 structure showing the newly introduced specificities in green. The green sites mutated in G1 are I134→L*, A135→G, S140→A, D142→N*, K143→Q*, E146→D*, R148→K, T150→E, A172→T*, A173→K, D191→E, A194→S, D196→E*, P199→A*, R203→S*, S208→L*, S210→D*, L212→R*, N214→G, Q215→S, A216→E*, and K229→R* (asterisks mark residues visible in this image). The unchanged portion of the molecule is colored as follows: (i) residues conserved among all three variants, blue; (ii) residues conserved between variants 2 and 3 but differing from variant 1, red; (iii) hypervariable residues, yellow. The G1 mutation scheme eliminates a Gly at position 147, with a corresponding shift of “−1” in the numbering scheme from this point onward; consequently, the residue S203 corresponds to R204 in wild-type fHBP (see also fig. S1). (C) Structure of the complex between wild-type fHBP variant 1 and human factor H (13) showing the surface of fHBP and α carbon trace of human factor H (green). Regions of the fHBP surface are colored according to the variability profile using the same color code as in (B). Figures were prepared using PyMOL (

Analysis of the G1 structure also shows that the region interacting with factor H, covering ~1500 Å2 of the surface, was minimally changed in G1. Only three residues were substituted in this region (namely, I134L, S208L, and S210D), corresponding to a total surface-accessible area of 250 Å2. Indeed, the binding of G1 to factor H was found to be similar to that of the native molecule (table S6).

Further engineering of G1

To further optimize the immunogenicity of the engineered G1 molecule and generate a final vaccine candidate, we produced a fusion protein, a general strategy previously shown to improve vaccine performance (26). Thus, two copies of G1 were fused to GNA2091, a meningococcal antigen that enhances the stability and efficacy of fHBP variant 1, although it does not itself induce bactericidal antibodies (26). The new fusion antigen, GNA2091-G1-G1, was formulated with the same adjuvant as described above, and the bactericidal assays were performed in the presence of rabbit and human complement. When rabbit complement was used, bactericidal titers induced against MC58 and NZ98/254 strains by GNA2091-G1-G1 (Table 3A) were comparable to those previously reported for the 4CMenB vaccine currently in phase III clinical trials (7). Moreover, the fusion molecule was able to promote the killing of each meningococcal strain at levels comparable to those observed when mice were immunized with the homologous alleles. A positive bactericidal titer against all variants tested was also obtained by performing the assay in the presence of human complement (Table 3B), which is the most stringent test available for vaccine potency (27). In this assay, which typically gives lower titers, positive titers are predictive of efficacy in humans. Collectively, these data provide a solid rationale for the use of this molecule in a vaccine that recognizes all of the antigenic variants of fHBP and thus should provide efficacious and broad protection against infection with MenB.

Table 3

Bactericidal titers elicited by the GNA2091-G1-G1 fusion protein. Bactericidal titers elicited in mice immunized with the GNA2091-G1-G1 fusion protein in the presence of rabbit (A) or human (B) complement. Antigens were formulated in aluminum hydroxide and IC31 adjuvant. The bactericidal activities induced against each strain by wild-type strains of subvariants 1.1 and 2.1 and the homolog fHBP alleles are also shown. In the assay with rabbit complement, titers of 512 or higher are shown in bold. We also compared the bactericidal assay titers elicited by G1 and GNA2091-G1-G1 with rabbit complement, repeating the measurements three times (fig. S6). In the assay with human complement (B), which is much more stringent, sera dilutions greater than 1:8 are shown in bold.

View this table:


We have successfully engineered a single molecule of fHBP to display simultaneously two independent immunodominant regions that induce protective immunity against all fHBP antigenic variants of the MenB serotype. In so doing, we have demonstrated that the structure-based design and engineering of the entire conformational surface of the epitope is essential to increase the immunogenicity of the molecule, whereas a change of only one or a few amino acids of the epitope is not sufficient. Notably, the most efficacious antigens present in the initial group of 54 proteins, G1 and G2, carry substitutions within the same region, one of the largest among those modified. Therefore, among all of the regions modified, the area defined as G is the one that best accommodated heterologous epitopes within the serobase protein structure without detrimental effects on the original repertoire and without interfering with the ability of the molecule to bind to factor H. Moreover, modifications in this region seem to modulate optimally the immunodominance among variants 1, 2, and 3. Grafting new epitopes into a single protein to make a single vaccine molecule that is able to induce immunity against all antigenic variants of a pathogen is a challenging task in vaccinology. Here, we have proposed a general strategy that is summarized in Fig. 5. For this type of approach, the critical steps are (i) the structure of the antigen necessary to define surface amino acids; (ii) epitope mapping by variant-specific mAbs, which provides the initial indication that distinctive epitopes segregating on the protein surface can be recognized; (iii) structure-based design of “transplanted” amino acids mimicking the newly introduced epitopes; and (iv) testing the biochemical and structural integrity of the generated molecules.

Fig. 5

A structure-based approach for designing fHBP mutants. The first steps of rational structure-based design are as follows: identification of surface-exposed residues, mapping of sequence variability onto the protein structure, and planning of point mutation groups to be introduced into the native molecule. These steps are followed by the experimental evaluation of candidate antigen performance measured by an in vitro bactericidal assay using human or rabbit complement with serum from mice immunized with the candidate antigens.

Traditionally, vaccines have been developed using or mimicking natural antigens, but here, we have shown that natural antigens can be improved to provide better immunity. Previous approaches to making universal vaccines against pathogens showing a degree of antigenic variation relied on the identification of conserved epitopes (35). Although this approach would be the most logical, so far it has met with limited success, probably because the conserved regions of the antigens have been naturally selected during evolution to be poorly immunogenic. Our results suggest an alternative approach in which structural knowledge is applied to graft multiple immunodominant epitopes onto a single molecule. We have engineered the fHBP protective antigen of MenB to contain an artificial combination of epitopes that is able to expand markedly the coverage of the original antigen. This success suggests that a similar approach could, in principle, be used in other cases where sequence variability is the main obstacle to vaccine development.

Materials and Methods


E. coli DH5a and BL21 (DE3) star were used as cloning strain and expression host, respectively, and used as recommended by the manufacturer (Invitrogen).

Cloning, expression, and purification in E. coli

The starting sequence of the fhbp gene where mutations were introduced is a version of MC58 fhbp devoid of the sequence coding for the leader peptide and glycine stretch sequence, starting with the Val codon. The DNA sequence encoding each mutant was obtained by modifying the codons to encode for the new amino acids avoiding the use of two rare AGG and AGA arginine codons. Nde I and Xho I restriction sites were added at 5′ and 3′ ends, respectively. Synthetic genes encoding for the 54 mutants were purchased from GeneArt. Each gene was digested with Nde I and Xho I and cloned into the Nde I/Xho I sites of the pET21b(+) expression vector (Novagen). The use of Nde I and Xho I sites resulted in the insertion of a starting Met codon and the addition of leucine-glutamate amino acids at the end of the sequence, respectively. Recombinant proteins were expressed as His-tag fusions.

Antigen formulation

All formulations were performed in sterile conditions under flow hood. Each recombinant protein was adsorbed onto aluminum hydroxide at protein, aluminum (alum), and NaCl concentrations of 100 μg/ml, 3 mg/ml, and 9 mg/ml, respectively, in 10 mM histidine (pH 6.5). Water for injection and histidine buffer were premixed. Sodium chloride was added to result in a final formulation osmolality of 0.308 mosmol/kg. Alum addition was calculated on the basis of the concentration of the alum stock to obtain a final concentration of 3 mg/ml. Antigens at respective concentrations were added to the mix and left for 15 min under stirring at room temperature and then stored overnight at 4°C before the immunization. IC31-containing formulations were prepared by adding an equal volume of IC31 (1000 nmol KLK and 40 nmol ODN1a) to the alum-adjuvanted formulations just before immunization. In this case, the immunization volume was doubled to maintain a constant antigen dose (20 μg). Final formulations were isotonic and at physiological pH. All alum and alum-IC31 formulations were characterized soon after immunization, antigen adsorption was >90%, and adsorption profile was similar for all antigens and adjuvants tested.

Mice immunization

To prepare antisera, we used 20 μg each of recombinant protein to immunize 6-week-old CD1 female mice (Charles River). Each antigen was used to immunize a group of eight mice. The recombinant proteins were administered intraperitoneally, together with aluminum hydroxide (3 mg/ml) or IC31 plus aluminum hydroxide at days 0, 21, and 35. Blood samples of each group were taken on day 49 and pooled for serological analyses.

Heat-inactivated whole-cell preparations

N. meningitidis wild-type strains and ΔfHBP mutant strains were grown overnight on agar chocolate plates at 37°C in 5% CO2. Colonies from each strain were collected and used to inoculate 7 ml of Mueller-Hinton broth containing 0.25% glucose to an initial OD620 (optical density at 620 nm) of 0.05 to 0.06. The culture was incubated for ~2.5 hours at 37°C with shaking until an OD620 of 0.5 was reached and then centrifuged for 10 min at 3500 rpm. The supernatant was discarded and the pellet was resuspended in 500 μl of phosphate-buffered saline (PBS). Heat inactivation was performed at 56°C for 30 min.

SDS–polyacrylamide gel electrophoresis and Western blot analysis

For Western blot analysis, 15 μl of each total whole extract sample in 1× SDS–polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer (Invitrogen) was separated by SDS-PAGE and transferred onto a nitrocellulose filter by standard procedure. Filters were blocked for 1 hour at room temperature by agitation in blocking solution (10% skim milk, 0.1% Triton X-100 in PBS) and incubated for another hour with a 1:1000 dilution of the anti-fHBP protein serum in washing solution (3% skim milk, 0.1% Triton X-100 in PBS). After washing, the filters were incubated in a 1:2000 dilution of peroxidase-conjugated anti–mouse immunoglobulin (Dako) in blocking solution for 1 hour, and the resulting signal was detected with the SuperSignal West Pico chemiluminescent substrate (Pierce). Results are reported in fig. S4.

Fluorescence-activated cell sorting analysis

The ability of polyclonal anti-fHBP sera to bind to the surface of live meningococci was determined with a FACScan flow cytometer on strains representative of the subvariants. Antibody binding was detected with a fluorescein isothiocyanate (FITC)–conjugated secondary antibody to mouse (whole molecule) (Sigma). The positive control included SEAM 12, a mAb specific for the MenB capsular polysaccharide (9). Negative control was the unrelated cytoplasmic protein NMB1380. Results are reported in fig. S5.

Complement-mediated bactericidal activity

Serum bactericidal activity against N. meningitidis strains was evaluated as previously described (8) using as complement source 25% (12.5 μl of complement in 50 μl of total volume reaction) of pooled baby rabbit serum (Pel-Freeze) or human serum. Serum bactericidal titers were defined as the serum dilution resulting in 50% decrease in colony-forming units (CFUs) per ml after 60-min incubation of bacteria with reaction mixture compared to control CFU per ml at time 0. Typically, bacteria incubated with the negative control antibody in the presence of complement showed a 150 to 200% increase in CFU per ml during the 60-min incubation. Because of the high number of mutants tested during the initial screening, pooled sera were tested only once against each of the seven reference strains.

Data reported in Table 3 are instead representative on average of three different immunization experiments in which pooled sera were tested. The experiments were repeated on different days with the same complement source, and results were comparable and reproducible. Bactericidal titers elicited by G1 and GNA2091-G1-G1 were also compared by repeating the measurements three times. Results of comparisons are reported in fig. S6.

Analytical SEC

Recombinant fHBP wild type and mutants (in 10 mM KPPi, pH 7.5) were analyzed by SEC-HPLC to study their state of aggregation. Samples were applied onto a Superdex 200 PC 3.2/30 (GE Healthcare) column at a flow rate of 0.1 ml/min in PBS buffer system. For calibration, a gel filtration standard (670, 158, 44, 17, and 1.35 kD; Bio-Rad) was applied (table S2).

CD spectroscopy

Far-UV CD spectra from 195 to 260 nm of fHBP wild-type and mutant proteins containing equimolar mixtures of proteins in 10 mM KPPi (pH 7.5) buffer were acquired at 25°C on a CD spectrometer (Jasco J-810) equipped with a water-cooled Peltier System (PCB1500). KPPi (10 mM, pH 7.5) buffer was used as a blank, and respective spectra were subtracted from all recorded CD spectra. Analysis of secondary structure of the constructs was evaluated by deconvolution of the spectra using CDNN v.2.1 (11). Original CD data in millidegrees were converted to Δε units.

NMR structural characterization

The fHBP variant 1.1 wild-type and mutant clones E. coli BL21 (DE3) are grown in ISOGRO-15N provided by Sigma. In all cases, the protein expression was induced by isopropyl-β-d-thiogalactopyranoside (IPTG). All samples were first purified by nickel-chelating affinity chromatography followed by anion exchange chromatography, and then all the proteins were dialyzed in 20 mM sodium phosphate (pH 7.2) before being subjected to NMR analysis. NMR measurements were acquired at 900 MHz on an ADVANCE Bruker spectrometer.

Protein crystallization

Crystallization experiments were performed in a nanodroplet sitting-drop vapor-diffusion format with 480 condition screens using 96-well low-profile Greiner plates and an Art Robbins Instruments Phoenix liquid handling robot. Crystals were grown at 277 K using sitting drops formed by mixing equal volumes (0.25 μl) of fHBP-G1 (7 mg/ml) in crystallization buffer (50 mM tris, pH 8.0) and a reservoir solution consisting of 20% PEG 3350 (polyethylene glycol, molecular weight 3350) and 0.2 M ammonium formate (pH 6.6). Crystals of fHBP-G1 belong to space group P21, with unit cell dimensions of a = 47.5, b = 99.6, c = 58.0, and β = 97.8. The asymmetric unit contains two monomers with a solvent content of 53% (Matthews coefficient, 2.62 Å3 dalton−1). All crystals were mounted in cryoloops and cooled to 100 K for data collection without use of cryo-protectant.

Data collection and structure determination and analysis

Diffraction data were collected at beamline 5.0.3 of the Advanced Light Source and were indexed and integrated with iMosflm and reduced with Scala (1, 2). The structure of fHBP-G1 was determined at 1.9 Å by molecular replacement in Phaser (3) using as search model coordinates from Protein Data Bank (PDB) entry 2W80 (4). A single solution was obtained for the two monomers with RFZ = 16.5, TFZ = 12.5, LLG = −85, and RFZ = 20.7, TFZ = 34.8, LLG = 1652, respectively. Rigid body and restrained refinement were carried out with Phenix (5) and manual model building in COOT (6). All crystallographic manipulations were carried out with the CCP4 package (7). Data collection and refinement statistics are shown in table S5. The structural similarity of fHBP-G1 was compared with existing fHBP structures using the PDBe FOLD server ( The surface-accessible areas and/or interfaces were measured using the CCP4 package (7) and the PDBe PISA server (

Supplementary Material

Fig. S1. Amino acid sequence alignment of all the fHBP mutants designed.

Fig. S2. Overview of the engineered surfaces of fHBP analyzed in this study.

Fig. S3. (A to G) Molecular localization of point mutations introduced in the best candidates.

Fig. S4. Western blot analysis of the meningococcal strains used in this study.

Fig. S5. FACS analysis of N. meningitidis strains carrying the different fHBP subvariants.

Fig. S6. Comparison of bactericidal titers elicited by G1 and GNA2091-G1-G1.

Table S1. Sizes of surface-accessible areas subjected to mutagenesis.

Table S2. Bactericidal activity of all the fHBP mutants.

Table S3. (A and B) Analytical size exclusion chromatography and circular dichroism spectroscopy of fHBP wild-type and mutant proteins.

Table S4. Data collection and refinement statistics for G1.

Table S5. Association rate, dissociation rate, and equilibrium dissociation constants of G1 and fHBP wild-type to fH deduced by surface plasmonic resonance.



  • * These authors contributed equally to this work.

References and Notes

  1. Acknowledgments: We thank L. W. Mayer (Centers for Disease Control and Prevention, Atlanta) for the U.S. strains, R. Borrow (Health Protection Agency, Manchester) for the UK strains, E. R. Moxon (University of Oxford) for the MC58 and C11 strains, D. R. Martin (Institute of Environmental Science and Research, Porirua) for the NZ98/254 strains, and G. Hogg (University of Melbourne) for the 9615945 strain. We also thank P. Dormitzer, E. De Gregorio, I. Ferlenghi, and E. Settembre for critical reading of the manuscript; S. Marchi for the purification of the G1 protein; G. Spraggon (Genomics Institute of Novartis Research Foundation, San Diego, CA) for continuous support in crystallography and helpful discussions; C. Mallia for assistance in manuscript editing; and G. Corsi for artwork. Funding: L.B., F.C., and S.D. were supported by Ministero dell’Istruzione, dell’Università e della Ricerca (grant FIRB-Proteomica RBRN07BMCT). Author contributions: M.S. designed the fHBP mutants, analyzed the data, and contributed to the writing of the paper; B.A. designed the expression strategy of the mutants and analyzed the data; B. Brunelli and M.M.G. performed serological analysis and analyzed the data; S.S. coordinated the protein purification and biochemical characterization work on the recombinant proteins; F.D.M. and E.P. performed gene expression of the mutants; D.V., L.C., and E.C. purified the recombinant proteins; M.J.B. and E.M. solved the crystal structure of G1; P.L.S., B. Brogioni, and M.T. were responsible for surface plasmonic resonance experiments; F.C. and S.D. were responsible for NMR experiments; L.B. supervised the NMR characterization and contributed to the writing of the paper; M.C. was responsible for the meningococcal molecular epidemiology; A.C., F.D., P.G., and M. Pallaoro were responsible for antigen formulations; M.H. and V.N.-D. performed biochemical characterization of candidates; M. Pizza conceived the experiments and contributed to the writing of the paper; R.R. conceived the experiments and wrote the paper. Competing interests: The following patents are associated with this work: WO2009/104097 (meningococcal fHBP polypeptides) and WO2011/024072 (hybrid polypeptides including meningococcal fHBP sequences). Accession numbers: Atomic coordinates of G1 can be found as PDB 2Y7S code.
  • Citation: M. Scarselli, B. Aricò, B. Brunelli, S. Savino, F. Di Marcello, E. Palumbo, D. Veggi, L. Ciucchi, E. Cartocci, M. J. Bottomley, E. Malito, P. Lo Surdo, M. Comanducci, M. M. Giuliani, F. Cantini, S. Dragonetti, A. Colaprico, F. Doro, P. Giannetti, M. Pallaoro, B. Brogioni, M. Tontini, M. Hilleringmann, V. Nardi-Dei, L. Banci, M. Pizza, R. Rappuoli, Rational Design of a Meningococcal Antigen Inducing Broad Protective Immunity. Sci. Transl. Med. 3, 91ra62 (2011).

View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article