Bactericidal Antibody Responses Elicited by A Meningococcal Outer Membrane Vesicle Vaccine with Over-Expressed fHbp and Genetically-Attenuated Endotoxin

Abstract

Background

Outer membrane vesicle (OMV) vaccines from mutant N. meningitidis strains engineered to over-express factor H-binding protein (fHbp) elicited broadly protective serum antibody responses in mice. The vaccines investigated were not treated with detergents to avoid extracting fHbp, which is a lipoprotein. Because of their high endotoxin content, the vaccines would not be safe to administer to humans.

Methods

We prepared a native OMV vaccine from a strain engineered to over-express fHbp and in which the gene encoding LpxL1 was inactivated, which reportedly decreased endotoxin activity.

Results

The OMV vaccine from the mutant had similar or lower ability to induce proinflammatory cytokines by human peripheral blood mononuclear cells as a detergent-extracted wildtype OMV, and 1000- to 10,000-fold lower activity than a native wildtype OMV. In mice, the OMV vaccine from the mutant elicited higher serum bactericidal antibody responses against a panel of heterologous N. meningitidis strains than a control multicomponent recombinant protein vaccine, or a detergent-extracted OMV vaccine that previously had been demonstrated to confer protection against meningococcal disease in humans.

Conclusions

The data illustrate the potential to develop a broadly immunogenic native OMV vaccine with decreased endotoxin activity that is potentially suitable for testing in humans.

Introduction

No broadly effective vaccine is available against Neisseria meningitidis group B strains, which account for half of meningococcal cases in the United States [1, 2], and greater than 80 percent in Europe [3, 4]. The group B capsule is structurally similar to antigens expressed by neural tissues and, therefore, is a poor immunogen, which also has the potential to elicit autoantibodies. Thus, a polysaccharide-protein conjugate vaccine is unlikely to be feasible for prevention of group B disease [5].

Novel antigens discovered by “genome mining” are currently under investigation as group B vaccines. One highly promising candidate is factor H-binding protein (fHbp), which was also known as Genome-derived Neisserial Antigen 1870 [6]or LP2086 [7, 8]. FHbp is a surface-exposed lipoprotein present in all N. meningitidis strains [6]. This protein can be subclassified into three variants based on sequence similarity and antigenic cross-reactivity. In general, antibodies prepared against fHbp variant 1 (v.1) were bactericidal against strains expressing fHbp from the v.1 group but not against strains expressing v.2 or v.3 proteins (and vice versa) [6, 9]. The variant 1 antigen is part of a promising investigational meningococcal vaccine consisting of three recombinant proteins, two of which are fusion proteins expressing two antigens each (i.e., a total of five antigens) [10]. In humans, this vaccine elicited serum bactericidal antibody responses against genetically diverse N. meningitidis strains [11].

Outer membrane vesicle (OMV) vaccines are safe [12, 13] and efficacious against meningococcal disease [14, 15]. An OMV vaccine was licensed in New Zealand and controlled a long-standing group B epidemic [16-19]. However, serum bactericidal antibodies elicited by OMV vaccines are directed primarily at a major porin protein, PorA [20], which is immunodominant [21], and antigenically variable [22, 23]. OMV vaccines are treated with detergents to extract lipopolysaccharide (LOS) and decrease endotoxin activity. This procedure also removes detergent-soluble antigens such as fHbp or GNA2132, which in mice elicited broadly protective serum antibody responses [6, 24, 25].

To increase protective activity, we previously prepared “native” OMV vaccines from N. meningitidis strains engineered to over-express fHbp v.1 [26, 27]. The sera from immunized mice conferred broader bactericidal activity against genetically diverse N. meningitidis strains than sera from control mice immunized with recombinant fHbp v.1, or a native OMV vaccine prepared from the corresponding wildtype strain [26, 27]. The native OMV vaccines were prepared without the use of detergents to avoid extracting fHbp. Thus the endotoxin activity was too high for the vaccine to be administered safely to humans.

In the present study, we prepared a native OMV vaccine from a N. meningitidis mutant strain engineered to over-express fHbp and in which the LpxL1 gene encoding a late functioning acyl transferase also was inactivated. The deletion resulted in penta- instead of hexa-acylated Lipid A, which in previous studies decreased endotoxin activity while retaining adjuvant activity [28-30]. Our hypothesis was that this OMV vaccine would be less toxic than a native OMV prepared from a wildtype strain while retaining the ability of the mutant OMV to elicit serum anti-fHbp antibodies with broad bactericidal activity.

Materials and Methods

Meningococcal strains

Meningococcal strains used in this study are described in table 1. Strain H44/76 and mutants derived from this strain were used to prepare the OMV vaccines. This strain expresses a fHbp v.1 protein with an amino acid sequence identical to that of strain MC58 [6], which provided the gene to over-express fHbp v.1 (referred to in Table 1 as v. 1.1). The other six strains expressed heterologous PorA proteins to that of the H44/76 vaccine strain and also expressed different subvariants of fHbp v.1 (Table 1).

Table 1. Neisseria meningitidis strains.

Strain1	Genetic Lineage (ST complex)	Porin Protein		LOS Immunotype2	fHbp Subvariant (Sequence type)3	Anti-fHbp mAb reactivity4
		PorB	PorA			JAR 1	JAR 5	JAR 10
H44/76	ST 32	15	P1.7,16	L3,7,9	1.1	1	1	0
NZ98/254	ST 41/44	4	P1.7-2,4	L3,7,9	1.10	0	1	1
4243	ST 11	2a	P1.5,2	L3,7,9	1.3	0	1	0
Z1092	ST 1	4, 21	P1.5-2,10	L8,10	1.4	0	1	0
CA0408	ST 11	2-2	P1.5,2	L3,7,9	1.2	0	0	0
CA0501	ST 41/44	3-1	P1.19-1, 26	L3,7,9	1.8	0	1	1
CA0017	ST 60	2-99	P1.19-3, 15	L3,7,9	1.9-3	0	1	1

¹H44/76 was used to make the OMV vaccines. The other six strains expressed heterologous PorA serosubtypes to that of the vaccine strain and also expressed sub-variants of variant 1 fHbp.

²As determined by whole cell ELISA with immunotyping MAbs 9-2-L379, 17-1-L1, 2-1-L8 and 14-1-L10 [48].

³Subvariants of variant 1 fHbp have small differences in amino acid sequence to that of fHbp of strain MC58 (<12 percent overall). In this classification system of Maurizio Comanducci (personal communication), fHbp of MC58 was designated 1.1; and proteins with more differences in sequence were given higher designations (i.e., NZ98/254, 1.10) than proteins with smaller differences in sequence with that of MC58 (i.e., CA0408, 1.2).

⁴JAR 1 and 5 mAbs were from hybridomas from a mouse immunized with rfHbp v.1 (gene from MC58). JAR 10 mAb was from a mouse immunized with rfHbp v.2 (gene from 2996). The epitope defined by JAR 5 is expressed by most strains expressing fHbp v.1 and sub-variants of v.1 [9, 25]. The epitope defined by JAR 1 is expressed by a subset of v.1 strains (nearly all ET 5/ ST 32 strains). The epitope defined by JAR 10 is expressed by most v.2 strains and by a subset of v.1 strains from the lineage 3/ ST 41/44 complex) [9].

Growth conditions

N. meningitidis strains were grown at 37°C on GC agar plates in an atmosphere containing 5% CO₂, or in Mueller-Hinton broth (MHB) containing 0.25% glucose and 5 μg/ml chloramphenicol or 80 μg/ml kanamycin as required.

Electrophoretic studies

SDS-PAGE was performed with 4-20% gradient gels (Invitrogen, Carlsbad CA). For Western-blots, proteins were transferred onto nitrocellulose membranes, and the secondary antibody was HRP conjugated goat anti-mouse IgG/A/M (Invitrogen). For resolution of LOS, the samples were separated by SDS-PAGE at 20 mA constant for approximately 1.5 hours. The gel was agitated for 1 hour in 40% ethanol, 5% acetic acid and treated for 5 min with 0.7% (w/v) periodic acid dissolved in 40% ethanol and 5% acetic acid solution. After washing with H₂O, the gels were stained with 0.67% (w/v) silver nitrate in 0.019 M NaOH and 0.4% NH₄OH for 10 minutes, washed with H₂O, and developed in a solution containing 50 mg/l citric acid and 0.015% (v/v) formaldehyde. The reaction was stopped with 50% methanol.

Recombinant DNA methods

Transformation of N. meningitidis and over-expression of the fHbp v.1 encoded on plasmid pFP12-fHbp was performed as described previously [26, 27]. To generate the LpxL1 knock-out (KO) mutant, the LpxL1 gene from strain MC58 was amplified by PCR using primers LpxL1_for HindIII 5′-CCCAAGCTTATCCTTCGGGGATGCAGGTC-3′ and LpxL1_revXbaI: 5′-GCTCTAGAGCCGTCTGAACGTAGTCAGTAAAAATCGGGGC-3′. The lpxL1 fragment was cloned into Hind III and Xba I digested plasmid pUC18 resulting in plasmid pUCLpxL1. An internal 204 base pair fragment of the LpxL1 gene was deleted by inverse PCR with plasmid pUCLpxL1 as template using primers LpxL1_del1: 5′-AACTGCAGCGGTGAAGTGCGGATACAGG-3′ and LpxL1_del2: 5′-ACGCGTCGACAGGATTTCGGACGCAACG-3′. A kanamycin resistant cassette was ligated with the product from the inverse PCR reaction resulting in plasmid pUCLpxL1kan.

Outer membrane vesicles

The OMV vaccine consisted of membrane blebs released by the bacteria into the supernatant and isolated as previously described [21]. Relative amounts of fHbp in the different OMV preparations were visualized by Western blot using a murine anti-fHbp mAb, JAR 3 [25] (Figure 1). As shown in Panel A, 0.013 μg of native OMV protein prepared from the LpxL1 KO mutant with over-expressed fHbp contained slightly more fHbp than 0.05 μg OMV protein from the wildtype strain. Thus, the mutant OMV contained approximately 4- to 5-fold more fHbp than the wildtype OMV. Silver staining (Figure 1, panel B) showed a slightly lower molecular mass of the LOS from the LpxL1 mutant as compared with that of the wildtype. This observation was consistent with a structural change in the LOS from hexa- to penta-acylated Lipid A caused by deletion of the gene encoding LpxL1.

Figure 1.

Panel A. fHbp v.1 in OMV preparations as detected by Western-blot. WT, native OMV prepared from H44/76 wildtype strain; fHbp KO, OMV from mutant of H44/76 in which the gene encoding fHbp was inactivated; LpxL1 KO OE fHbp, OMV from mutant of H44/76 with inactivated LpxL1 and over-expressed fHbp; rfHbp, recombinant v.1 protein. Primary antibody was anti-rfHbp monoclonal antibody JAR3. Panel B. Silver stained SDS-PAGE of LOS in OMV preparations from strain H44/76. Amounts of vesicles loaded in each lane were standardized based on total protein content.

Extraction of LOS from OMV was performed based on the published deoxycholate procedure for production of the OMV vaccine by the Norwegian Institute of Public Health, Oslo Norway [31]. Based on silver staining, 0.2 μg of detergent-extracted OMV contained less LOS than 0.02 μg of native OMV (Figure 1, panel B, right). These data were consistent with removal of the majority of the LOS by the detergent treatment. As expected, the detergent treatment also removed most of the fHbp (Figure 1, panel A, right).

Serology

Total antibody responses to fHbp, NadA, GNA2132 and LOS were measured by ELISA as described [25]. The recombinant protein antigens on the plate consisted of purified His-tagged fHbp, NadA [32] or GNA2132 [24] (encoded by genes from strains MC58, 2996 or NZ98/254, respectively). LOS was purified from the wildtype and LpxL1 mutant strains of H44/76 using the procedure described by Westphal [33]. Secondary antibody was goat-anti mouse IgG+A+M conjugated to alkaline phosphatase.

Complement mediated bactericidal antibody responses were measured as described previously using washed, log phase bacteria [34]. The buffer was Dulbecco's phosphate buffered saline (Sigma-Aldrich) containing 0.9 mM CaCl₂ × 2 H₂O, 0.5 mM MgCl₂ × 6 H₂O and 1% (w/v) BSA. The complement was human serum from a healthy adult with no detectable intrinsic bactericidal activity. Bactericidal titers were defined as the serum dilution resulting in a 50% decrease of cfu after 60 min incubation at 37°C as compared with cfu at time 0 in negative control reactions.

Absorption of anti-fHbp antibodies

The absorption was performed as previously described [27] using a column containing His-tagged recombinant fHbp, or as a negative control, recombinant His-tagged NadA bound to Sepharose (GE Healthcare, Piscataway, NJ). Non-specific binding sites were blocked with normal mouse serum and the column was washed to remove unbound rfHbp. To compensate for dilution and incomplete recovery of antibodies during the absorption, the anti-LOS titers of the absorbed serum pools were adjusted to match the respective titers before absorption.

Cytokine assay

After obtaining informed consent, blood was obtained from healthy donors and the buffy coats were fractionated by Ficoll density centrifugation. The peripheral blood mononuclear cell (PBMC) layer was recovered, washed three times with RPMI 1640 medium and resuspended in complete medium (RPMI 1640 + L-glutamine + 25 mM HEPES, containing 10% FCS (HyClone, Logan UT) and 1% penicillin/ streptomycin/ glutamine). PBMCs were cultured in 96-well flat bottom plates at a density of 4×10⁵ cells per well. Serial 10-fold dilutions (1 μg/ml to 0.00001 μg/ml final concentration) of native OMV prepared from the H44/76 wildtype, or the LpxL1 KO mutant with over-expressed fHbp, or detergent-treated wildtype OMV were added. The samples were incubated for 4 hours at 37°C. Cytokine secretion was measured by Bio-Plex analysis (BioRad, Hercules CA) according to manufacturer's instructions using the human 27-plex panel. The following soluble proteins were assayed: IL-1β, IL-1ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p70, IL-13, IL-15, IL-17, eotaxin, basic FGF, G-CSF, GM-CSF, IFN-γ, IP-10, MCP-1, MIP-1α, MIP-1β, PDGF-BB, RANTES, TNF-α and VEGF. The concentrations of cytokines were analysed using the Bioplex System and Bioplex Manager 4.1 software.

Immunization

Mice were immunized with native OMV vaccine prepared from the LpxL1 KO strain of H44/76 with increased expression of fHbp. Control vaccines consisted of a clinical lot (FMEN0405) of detergent-extracted OMV vaccine prepared from strain H44/76, and a multi-component recombinant protein vaccine containing GNA2091 fused with fHbp v.1; GNA2132 fused with GNA1030, and NadA [10]). The Norwegian Institute of Public Health provided the detergent-extracted OMV vaccine pre-adsorbed with aluminum hydroxide [15]. Novartis Vaccines, Siena, Italy, provided the recombinant proteins. The recombinant antigens (300 μg/ml) were formulated in our laboratory with 3 mg of aluminum hydroxide/ml of water containing 10 mM histidine and 9 mg/ml NaCl.

Groups of 4-6 weeks old female CD-1 mice (Charles River Breeding Laboratories) were immunized intraperitoneally (10 to 15 mice per group). For each injection, the mice received a dose of 2 μg protein of native or detergent-extracted OMV vaccine, which were absorbed with 600 μg of aluminum hydroxide. The total dose of the recombinant protein vaccine was 60 μg (20 μg of each protein), which was absorbed with 600 μg of aluminum hydroxide. Three injections of vaccine were given, separated by three weeks. Blood was collected three weeks after the third injection. Sera were separated by centrifugation and stored frozen. The animal experiments complied with the relevant guidelines of Italy and institutional policies of Novartis Vaccines.

Results

Cytokine release from human PBMCs

Figure 2 shows the dose-responses of four proinflammatory cytokines, TNF-α, IL-1β, IL-6 and IL-8, as measured in experiments with PBMCs from two different donors. Small doses of native OMV from the wildtype strain elicited high levels of each of the cytokines while much higher doses of native OMV prepared from the LpxL1 KO mutant with over-expressed fHbp, or the detergent-treated OMV from the wildtype strain, were required for stimulation of cytokines. For example, with donor 2 (Panel A, right), a dose of 4 × 10⁻⁶ μg/ml of native OMV from the wildtype strain elicited the same concentration of TNF-α as did 3 × 10⁻² μg/ml of native mutant OMV (ratio 1:7500), or 2.5 × 10⁻³ μg/ml of detergent-extracted wildtype OMV (ratio 1:625). For each of the four cytokines, the native OMV from the mutant elicited lower or similar cytokine responses with both donor PBMCs as that of the detergent-extracted OMV from the wildtype strain.

Figure 2.

Release of proinflammatory cytokines IL-1β, IL-6, IL-8 and TNF-α after incubation of PBMCs with different concentrations of OMV for 4 hours. Left, PBMCs from Donor 1, experiment 1. Right, PBMCs from Donor 2, experiment 2. OMV vaccines tested were native OMV prepared from the H44/76 wildtype (WT native, open squares with solid line) or a H44/76 mutant with inactivated LpxL1 and over-expressed fHbp (mutant, native, closed squares with solid line), or a detergent-extracted OMV from H44/76 wildtype strain (WT extracted, open circles with dashed lines).

Of the 23 other cytokines measured, six (IL-1ra, G-CSF, IFN-γ, MCP-1, MIP-1α and MIP-1β) were above background levels after incubation of PBMCs with OMV from the wildtype strain. As shown in Figure 3 (supplemental data, to be accessed online), for each of these cytokines the native OMV from the mutant had much lower stimulating activity than that of native OMV from the wildtype strain (>500- to 10,000-fold), and gave similar or lower stimulation as that of the detergent-extracted OMV from the wildtype strain.

Figure 3.

(Supplementary for website). Release of cytokines in response to incubation of human PBMCs for four hours with different concentrations of OMV. Of 27 cytokines tested, 10 were stimulated above background levels: four proinflammatory cytokines (Figure 2) and the remaining six, IL-1ra, G-CSF, IFN-γ, MCP-1, MIP-1α and MIP-1β, are shown in this figure. Left, data from PBMCs of Donor 1, experiment 1. Right, PBMCs from Donor 2, experiment 2. Symbols for OMV vaccines tested are as described for Figure 2.

Serum antibody responses by ELISA

Figure 4 summarizes the serum antibody responses to recombinant fHbp, NadA, and GNA2132, which are the three antigens in the recombinant protein vaccine responsible for eliciting bactericidal antibodies [10]. We also measured titers to LOS from the wildtype and LpxL1 KO mutant N. meningitidis strains, which had been used to make the OMV vaccines; the respective anti-LOS titers were nearly identical. Therefore, the data are shown only for the mutant LpxL1 KO LOS.

Figure 4.

Serum antibody responses to fHbp (v.1), NadA. GNA2132 and LOS (LpxL1 KO mutant) as measured by ELISA. Left to right, mice immunized with aluminium hydroxide (Bar 1); the Norwegian OMV vaccine (OMV, Norway, Bar 2); a recombinant protein vaccine (Bar 3); or native OMV prepared from H44/76 with inactivated LpxL1 and increased expression of fHbp (OMV, mutant, Bar 4). The error bars show the 95% confidence intervals of the geometric mean titers.

Mice immunized with the recombinant proteins had high serum antibody titers to each of the respective antigens in the vaccine, and had no detectable serum antibody responses to LOS. Mice immunized with the Norwegian OMV vaccine had low anti-fHbp titers (GMT of 1:80), and undetectable antibody titers to the other antigens tested. In contrast, mice immunized with the native OMV vaccine prepared from the LpxL1 KO mutant with over-expressed fHbp had high antibody responses to fHbp and to LOS. The anti-fHbp GMT was 1:70,000, which was similar to that of the mice given the recombinant proteins (GMT of 1:100,000).

The titers shown in Figure 4 were measured in sera that had been incubated with the antigen at 4°C for 18 hrs. The respective titers to rfHbp were similar if the incubation were performed at 37°C for 2 hrs (Figure 5). However, the anti-LOS GMT decreased by more than 2 log10 if the incubation were performed at 37°C for 2 hrs (Figure 5).

Figure 5.

Effect of temperature on measurement of antibody titers to LOS and fHbp (v.1) by ELISA in sera from mice immunized with mutant OMV. Sera were incubated for 2 hrs at 37°C, or for 18 hrs at 4°C. After washing the plates, bound Ig was measured by goat-anti-mouse IgG+A+M conjugated to alkaline phosphatase.

Serum bactericidal antibody responses

The serum bactericidal antibody responses of each of the vaccine groups were high when measured against strain H44/76 (GMTs >1:1000, Figure 6). This strain was used to prepare the OMV vaccines and it expressed fHbp with an identical amino acid sequence to that of fHbp in the recombinant protein vaccine. Interestingly, mice given the mutant native OMV with increased fHbp had a 10-fold higher GMT against H44/76 than that of mice given the recombinant protein or detergent-extracted OMV vaccines. The higher bactericidal antibody responses to the mutant OMV may have reflected the combined bactericidal activity of homologous anti-PorA and anti-fHbp antibodies.

Figure 6.

Serum bactericidal activity (GMT) of immunized mice as measured against N. meningitidis strain H44/76 (used to prepare the OMV vaccines, see text) and six additional strains with heterologous PorA molecules to that of the vaccine strain and expressing subvariants of v.1 fHbp (see Table 1). Bars represent GMT of assays of 3 serum pools from each vaccine group (serum samples from 5 mice per pool). There were 2 serum pools for the Norwegian OMV vaccine group and the bars represent the GMT and range of the respective values. Symbols are identical to those used in Figure 4.

The six other test strains had a heterologous PorA to that of vaccine strain, and expressed subvariants of fHbp v.1 (Table 1). The serum bactericidal titers of mice given the Norwegian detergent-extracted OMV vaccine were <1:10 against all of these strains. The titers of mice immunized with the recombinant protein vaccine were high against some of these strains (e.g., CA0408 or 4243), but were low (e.g., NZ98/254) or undetectable (e.g., Z1092) against others. In contrast, mice immunized with the native OMV vaccine prepared from the mutant developed high bactericidal titers against all six of heterologous strains. The GMTs against the six heterologous strains were <1:10 for mice given the detergent-extracted Norwegian OMV vaccine, 1:82 for mice given the recombinant protein vaccine, and 1:797 for mice given the native OMV from the LpxL1 KO mutant with over-expressed fHbp (P<0.001).

To investigate the contribution of anti-fHbp antibodies to bactericidal activity of sera from mice immunized with the mutant OMV, we absorbed the serum pools using a column that contained recombinant fHbp-His bound to Ni-Sepharose. After absorption the sera were no longer bactericidal against the two heterologous stains tested whereas the control sera absorbed on a Ni-Sepharose column containing recombinant His-tagged NadA retained bactericidal activity (Table 2).

Table 2.

Effect of absorption of anti-fHbp antibodies on bactericidal activity of sera from mice immunized with OMV from mutant LpxL1 KO strain of H44/76 with over-expressed fHbp¹.

	1/Serum Antibody Titer1
	Before Absorption		After Absorption on Column
			rNadA		rfHbp
Assay	Pool 1	Pool 2	Pool 1	Pool 2	Pool 1	Pool 2
Anti-fHbp ELISA	40,000	20,000	30,000	30,000	640	300
Anti-LOS ELISA2 Bactericidal, Strain NZ98/254 Bactericidal, Strain Z1092	17,500	5000	17,500	5000	17,500	5000
	140	170	225	180	<12	<12
	800	850	750	900	<12	<12

¹Two serum pools from mice immunized with OMV with over-expressed fHbp were absorbed on columns containing a Ni-Sepharose matrix complexed with recombinant His-tagged fHbp or, as a control, His-tagged rNadA. The flow throughs were collected and assayed by ELISA for antibody to fHbp and LOS after incubation for 18 hrs at 4°C, and for bactericidal activity.

²The dilutions of the absorbed serum pools were adjusted to have the same anti-LOS titers as in the respective pools before absorption to account for dilution of the pools after passing over the columns and incomplete recovery of antibody.

Discussion

Detergent-extracted OMV vaccines are safe [13] and effective (reviewed in [35]). Their major limitation is strain-specific serum bactericidal responses. To broaden bactericidal antibody responses, OMV vaccines have been prepared from more than one strain [36], or from multiple strains engineered to express more than one PorA molecule [37-40]. However, it is difficult to provide broad coverage against endemic disease in large geographic areas with a vaccine that elicits bactericidal antibodies directed against PorA, since the strains are genetically diverse and express many different PorA molecules. Also, certain PorA serosubtypes were reported to be poorly immunogenic [41].

Recently, Weynants et al prepared detergent-treated OMV vaccines from mutant N. meningitidis strains engineered to over-produce several surface-exposed outer membrane proteins that normally are present in low copy number (TbpA, Hsf, NspA and Omp85) [42]. They found that overproduction of more than one minor protein was required to elicit serum bactericidal antibodies against heterologous strains. They theorized that a critical density of antibodies was required on the surface of the organism to engage C1q and activate classical complement pathway bactericidal activity. We have postulated a similar mechanism to explain why two non-bactericidal anti-fHbp mAbs that recognized non-overlapping epitopes became bactericidal when tested in combination [25, 43]. However, mice immunized with native OMV vaccines prepared from mutant N. meningitidis strains with overproduction of fHbp developed high serum bactericidal antibody responses against heterologous strains [26, 27]. Therefore, with a native OMV vaccine and over-production of fHbp, a single antigen was sufficient to elicit polyclonal bactericidal antibodies.

In the present study, we constructed a mutant N. meningitidis strain that both over-expressed fHbp and contained the LpxL1 knock-out mutation. Although inactivation of the LpxL1 gene of N. meningitidis was reported to decrease endotoxin activity [28-30], other published data indicated that non-LOS components of the bacteria can be potent inducers of cytokines [44]. Therefore, it was not clear how much attenuation we would achieve with a native OMV vaccine prepared from a LOS mutant with over-expressed fHbp. It was also unclear whether the OMV vaccine from the LOS mutant would retain sufficient immunogenicity since some of the adjuvant properties of LOS may depend on the presence of hexa-acylated LOS [45].

Our data showed that a native OMV vaccine from the mutant strain had 1000- to 10,000-fold lower activity in stimulating human PBMCs to release proinflammatory cytokines than a control native OMV from the wildtype strain. Further, the native OMV vaccine from the mutant had similar or lower activity in stimulating proinflammatory cytokines than a detergent-treated OMV in which most of the LOS had been extracted. Although additional studies are needed, these data suggest that native OMV vaccines prepared from LpxL1 knock-out strains may be safe to administer to humans.

The native OMV from the LpxL1 KO mutant with over-expressed fHbp elicited serum bactericidal responses to strain H44/76, from which the vaccine was prepared, that were 10-fold higher than those elicited by a clinical lot of the Norwegian detergent-extracted OMV vaccine, which was also prepared from strain H44/76, and which protected humans from meningococcal disease [15]. Mice immunized with the native OMV prepared from the mutant strain also developed higher and broader bactericidal antibodies than control mice immunized with a multicomponent recombinant protein vaccine containing as one of its antigens, fHbp.

Neither the detergent-treated nor native OMV vaccine elicited serum antibodies to NadA or GNA2132 (Figure 4). NadA was immunogenic during infection in humans [46], and also was immunogenic in mice immunized with an OMV vaccine prepared from a different N. meningitidis strain than H44/76 [47]. The H44/76 vaccine strain does not have the gene encoding NadA (authors' unpublished data), and natural expression of GNA2132 may have been too low for the OMV vaccines to elicit anti-GNA2132 antibody responses.

Mice immunized with the mutant OMV vaccine developed high titers of both anti-LOS and anti-fHbp antibodies (Figure 4). However, the serum bactericidal antibodies appeared to be directed at fHbp since depletion of anti-fHbp antibodies resulted in loss of serum bactericidal activity (Table 2). Antibodies to fHbp both activate complement-mediated bactericidal activity and inhibit binding of fH to the bacteria [43]. The latter enhances susceptibility of N. meningitidis to complement-mediated bactericidal activity, and it is difficult to separate the two functions of the anti-fHbp antibodies. Therefore, we can't exclude the possibility that the anti-LOS antibodies elicited by the native OMV vaccine contributed to bactericidal activity in the presence of anti-fHbp inhibition of fH binding to the bacteria. This possibility, however, seems unlikely since the anti-LOS antibodies bound poorly at 37 C (Figure 5), which was the temperature used to measure bactericidal activity. Also, in our previous study, mice immunized with a native OMV vaccine prepared from a fHbp KO mutant that was given with recombinant fHbp developed high titers of anti-fHbp antibodies but these antibodies were not bactericidal and did not activate C3b deposition on the bacterial surface of heterologous strains [26]. Thus, the combination of anti-LOS antibodies elicited by the native OMV vaccine and anti-fHbp antibodies elicited by recombinant fHbp was not sufficient for bactericidal activity. In contrast, sera from mice immunized with a native OMV with over-expressed fHbp were bactericidal and activated C3b deposition, and these activities were eliminated by absorption of the anti-fHbp antibodies. Conceivably, certain conformational fHbp epitopes were important for eliciting antibodies with broad bactericidal activity are expressed better by fHbp in the OMV than by recombinant fHbp. While the actual mechanism remains to be defined, the data illustrate the potential to develop a broadly immunogenic native OMV vaccine with decreased endotoxin activity that is potentially suitable for testing in humans.