Meningococcal Detoxified Outer Membrane Vesicle Vaccines Enhance Gonococcal Clearance in a Murine Infection Model

Abstract

Background

Despite decades of research efforts, development of a gonorrhea vaccine has remained elusive. Epidemiological studies suggest that detoxified outer membrane vesicle (dOMV) vaccines from Neisseria meningitidis (Nm) may protect against infection with Neisseria gonorrhoeae (Ng). We recently reported that Nm dOMVs lacking the major outer membrane proteins (OMPs) PorA, PorB, and RmpM induced greater antibody cross-reactivity against heterologous Nm strains than wild-type (WT) dOMVs and may represent an improved vaccine against gonorrhea.

Methods

We prepared dOMV vaccines from meningococcal strains that were sufficient or deleted for PorA, PorB, and RmpM. Vaccines were tested in a murine genital tract infection model and antisera were used to identify vaccine targets.

Results

Immunization with Nm dOMVs significantly and reproducibly enhanced gonococcal clearance for mice immunized with OMP-deficient dOMVs; significant clearance for WT dOMV-immunized mice was observed in one of two experiments. Clearance was associated with serum and vaginal anti-Nm dOMV immunoglobulin G (IgG) antibodies that cross-reacted with Ng. Serum IgG was used to identify putative Ng vaccine targets, including PilQ, MtrE, NlpD, and GuaB.

Conclusions

Meningococcal dOMVs elicited a protective effect against experimental gonococcal infection. Recognition and identification of Ng vaccine targets by Nm dOMV-induced antibodies supports the development of a cross-protective Neisseria vaccine.

Meningococcal outer membrane vesicle vaccines confer protection against gonococci in a murine infection model. Vaccines that induced higher serum and vaginal antibodies and cross-reactivity against a greater number of gonococcal proteins were associated with greater rates of clearance.

Responsible for an estimated 87 million new infections annually [1], Neisseria gonorrhoeae (Ng) is a significant cause of global morbidity. Isolation of multidrug-resistant gonococci have been reported within recent years, including identification of isolates resistant to third-generation cephalosporins that constitute the last remaining antibiotic class available for monotherapy treatment [1–3]. The potential for the development of “untreatable gonorrhea” and the lack of vaccines available to prevent disease has led the Centers for Disease Control and Prevention to designate Ng an “urgent threat” to public health.

Surveillance data gathered following administration of Nm detoxified outer membrane vesicle (dOMV) vaccines in Cuba [4] and Norway [5] and retrospective case-control studies conducted following a widespread immunization campaign with the MeNZB dOMV vaccine in New Zealand [6, 7] suggest that decreases in gonorrhea may be associated with Neisseria meningitidis (Nm) dOMV vaccination. Similarly, a study conducted in Canada in which participants were immunized with the 4CMenB vaccine (BEXSERO, containing recombinant proteins and dOMVs produced from the same Nm strain used for MeNZB) suggested that vaccination was associated with an estimated 59% reduction in the Ng incidence rate, though the results were not statistically significant [8]. In laboratory settings, parenteral immunization with the 4CMenB vaccine enhances gonococcal clearance in a mouse model of infection [9]. Likewise, nondetoxified outer membrane vesicle (OMV) vaccines isolated from gonococcal strains exhibit a protective effect when administered intranasally [10] or intravaginally [11], indicating that Neisseria OMV vaccines may prove effective at preventing gonorrhea.

Previously, we described dOMV vaccines prepared from isogenic Nm strains that were deleted for major outer membrane proteins (OMPs), including PorA, PorB, and RmpM [12]. Deletion of the immunodominant Nm protein PorA alone or in combination with PorB and RmpM elicited antibodies with bactericidal activity against a greater number of heterologous Nm strains relative to antibodies from wild-type (WT) dOMV-immunized animals. In this study, we sought to determine whether the enhanced cross-reactivity induced by OMP-deficient Nm dOMVs could extend to Ng strains. Using a murine model of gonococcal infection, we provide evidence that dOMV vaccines engineered from OMP-deficient meningococcal strains reproducibly enhance clearance of gonococci. Faster and higher clearance rates were associated with recognition of an increased number of gonococcal antigens by immune sera, allowing for identification of potential targets for vaccine development.

MATERIALS AND METHODS

Bacterial Growth Conditions

Neisseria meningitidis strains were cultured on Brain Heart Infusion agar supplemented with 5% equine serum or in Tryptic Soy Broth medium with antibiotics as needed: erythromycin (3 µg/mL), kanamycin (50 µg/mL), chloramphenicol (5 µg/mL). Neisseria gonorrhoeae strains were cultured on Gonococcal Base (GCB) agar or in GCB broth with Kellogg’s supplements [13]. Cytidine-5′-monophospho-N-acetylneuraminic acid (CMP-NANA, Nacalai) was added where indicated at a concentration of 50 µg/mL.

Mouse Immunization/Challenge Model

Construction of OCh and ΔABR strains from WT strain MC58 and preparation of dOMVs were described previously [14]. Female 4-week-old BALB/cAnNCr mice (Charles River) were immunized in 2 independent experiments (n = 20 mice per group per experiment) with a 1:1 mixture of 12.5 µg dOMVs and Alhydrogel (InvivoGen) adjuvant by intraperitoneal injection at 2-week intervals (days 0, 14, and 28); control groups included unimmunized mice and mice that received Alhydrogel in phosphate-buffered saline (PBS) alone (Alum). Venous blood was collected on days 24 and 38, and vaginal lavages were collected on day 24. Three weeks after the final immunization, mice were staged to determine the phase of the estrus cycle, and those in the diestrus or anestrus were entered into the challenge stage of the study. 17β-estradiol pellets were implanted subcutaneously and mice were vaginally inoculated with 10⁶ colony-forming units (CFUs) of Ng strain F62 [9]. Vaginal swabs were quantitively cultured for Ng up to 10 days postchallenge and used to prepare stained smears to examine influx of polymorphonuclear leukocytes (PMNs). All mouse experiments were conducted according to guidelines established by the Association and Accreditation of Laboratory Animal Care using a protocol approved by the Uniformed Services University’s Institutional Animal Care and Use Committee.

Human Complement Serum Bactericidal Assays

Bacteria were cultured to mid-log phase, centrifuged, and suspended in 0.5% bovine serum albumin in Hank’s Balanced Salt Solution containing calcium and magnesium, and 10⁵ CFUs were incubated with heat-inactivated anti-Nm dOMV antisera generated in 4 different rabbits (identified as B5, B7, B14, and B8) [12] that were either untreated or preabsorbed against whole-cell Nm strain MC58. Serum samples pooled from 4–6 mice immunized with dOMVs or Alum were tested similarly. The active complement source was pooled immunoglobulin G (IgG)/immunoglobulin M–depleted human serum (generous gift of Pel-Freez Biologicals) at 25% final concentration; heat-inactivated complement was tested in parallel as a negative control. After 45 minutes, aliquots were plated in triplicate on GCB agar and incubated overnight. Mean CFUs of test sera were determined relative to the mean CFUs of bacteria grown in the presence of active complement alone to determine percent survival.

Immunoblots

Neisseria whole-cell lysates were fractionated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to iBlot nitrocellulose membranes (ThermoFisher), blocked with 10% milk, and probed with anti-LpdA (ab80913, Abcam), anti-PorA (MN14C11.6, National Institute for Biological Standards and Control [NIBSC]), or anti-PorB (8B5-5-G9, NIBSC) monoclonal antibodies. Alternatively, pooled mouse serum or vaginal lavage samples or unabsorbed or preabsorbed rabbit sera were used to probe blots. Blots were developed by incubation with horseradish peroxidase–conjugated anti-mouse or anti-rabbit IgG or immunoglobulin A (IgA) secondary antibodies (Bio-Rad), followed by Pierce ECL Western Blotting Substrate (ThermoFisher).

Enzyme-Linked Immunosorbent Assays

Microtiter plates (384-well) were coated with 80 ng/well of Ng strain F62 OMVs, and total IgG, IgG1, IgG2a, and IgA antibodies were measured from serum and vaginal lavage samples as endpoint titers determined by standard enzyme-linked immunosorbent assay [15].

Identification of Cross-Reactive Antigens

Antibodies from serum pools of WT dOMV–, ΔABR dOMV–, and Alum-immunized mice (diluted 1:20) were bound to Protein G Dynabeads (ThermoFisher) and were used to immunoprecipitate antigens from Ng strain FA1090 OMVs (100 µg) pretreated with 1% Elugent (Millipore) according to the manufacturer’s instructions. Antigen-bound Dynabeads were captured via magnet and washed. Peptide identification was conducted by the University of California, Davis Proteomics Core Facility via liquid chromatography–tandem mass spectrometry using the UniProt Database.

RESULTS

Meningococcal Vesicle-Specific Antibodies Cross-React With Gonococcal Strains

To test the impact of OMP deficiency on cross-reactivity of anti-Nm dOMV antibodies, we first fractionated whole-cell lysates of 4 Ng laboratory strains (FA19, FA1090, MS11, and F62, Supplementary Table 1) by SDS-PAGE and probed with previously characterized antisera obtained from dOMV-vaccinated rabbits [12]. Vaccines were prepared from 3 isogenic Nm strains: (1) WT strain MC58; (2) OCh, expressing a hybrid PorB type and deleted for PorA, which is not present in Ng [14]; (3) ΔABR, deleted for PorA, PorB, and RmpM. The antisera recognized antigens common among all 4 Ng strains tested, and consistent differences were observed in the protein recognition patterns when the antisera were compared (Supplementary Figure 1). Antisera from WT dOMV-immunized rabbits B1 and B5 recognized 1 or 2 approximately 11–13 kDa antigens expressed by Ng strains FA19, MS11, and F62 that were previously identified as Nm L3,7,9 and L8 lipooligosaccharide (LOS) species [12]. B3/B7 and B13/B14 antisera from OCh- and ΔABR dOMV-vaccinated animals, respectively, primarily recognized higher molecular weight (~43–95 kDa) antigens common to all 4 Ng strains. The gonococcal RmpM homologue (~34 kDa) was recognized by rabbit anti-WT (B1/B5) and anti-OCh (B3/B7) dOMV antisera, but not anti-ΔABR (B13/B14) antisera, which were raised against dOMVs deficient in RmpM. Serum antibodies from Alum-immunized rabbits (B4/B8) did not bind Ng at comparable serum dilutions.

Antisera were next assessed for cross-reactive functional activity in hSBAs. Ng strains were grown in the presence of CMP-NANA to allow LOS sialylation and enhance serum resistance as occurs in vivo [16–19]. B5, B7, B14, and B8 antisera were either untreated or pre-absorbed against meningococcal strain MC58 and tested for bactericidal activity (Figure 1 and Supplementary Table 2). Antibodies in the unabsorbed sera from rabbits immunized with WT (B5), OCh (B7), or ΔABR (B14) dOMVs killed ≥50% of F62 gonococci at respective titers (Titer₅₀) of 1:40, 1:80, and ≥1:160. Background killing of F62 by B7 and B14 preabsorbed sera (Titer₅₀ of 1:5 and 1:20, respectively) was associated at least in part with incomplete absorption of antisera against MC58 (Supplementary Figure 2), with a corresponding reduction in Titer₅₀ equivalent to ≥8-fold. Antiserum from an Alum-immunized control rabbit (B8) was also bactericidal (Titer₅₀ of 1:40), consistent with preexisting Ng-specific IgG antibodies by immunoblot. In contrast to serum from vaccinated animals, preabsorption with MC58 only reduced B8 bactericidal activity by 2-fold (Titer₅₀ of 1:20), suggesting that bacteriolysis of Ng by the B5, B7, and B14 sera was primarily induced by Nm dOMV immunization. Strain MS11 was also susceptible to serum-dependent bacteriolysis, with reductions in Titer₅₀ of ≥32-fold and ≥4-fold for B5 and B7 unabsorbed/adsorbed serum pairs, respectively, while B8-mediated bacteriolysis was only apparent at a 1:5 serum dilution. Serum-resistant strains FA19 (Figure 1) and FA1090 (data not shown) were not killed by any of the antisera.

Figure 1.

Bactericidal activity of B5, B7, and B14 serum antibodies from rabbits immunized with meningococcal wild-type (WT), OCh, or ΔABR detoxified outer membrane vesicle vaccines against 3 gonococcal strains, FA19, MS11, and F62; B8 antiserum from an Alum-immunized rabbit was also tested as a negative control. Unabsorbed sera (blue squares) and sera preabsorbed against meningococcal strain MC58 (green circles) were diluted 1:5 through 1:160 and tested for induction of bacteriolysis in the presence of active (closed symbols) or heat-inactivated (HI; open symbols) human complement. Hashed lines represent Titer₅₀, Titer₃₀, and Titer₁₀ values. *P ≤ .05, **P ≤ .01, ***P ≤ .001, and ****P ≤ .0001 represent significant differences between unabsorbed and absorbed sera by 2-tailed unpaired Student t test.

Meningococcal Vesicle Vaccines Enhance Gonococcal Clearance

Antigen binding patterns and bactericidal activity of Nm dOMV rabbit antisera against Ng strains suggested that the Nm WT, OCh, and ΔABR dOMV vaccines may confer differing levels of protection against Ng. To examine this hypothesis directly, WT, OCh, and ΔABR dOMV vaccines were produced (Supplementary Figure 3) and tested in a murine lower reproductive tract infection model. Mice (n = 20/group) were administered 3 doses of vaccine or Alum at 2-week intervals and challenged intravaginally with gonococcal strain F62 3 weeks after the final immunization. A decrease in the percentage of infected, vaccinated mice began at day 3 postinoculation, and by day 7, a significantly higher percentage of OCh-immunized (56%) and ΔABR dOMV-immunized (77%) mice, but not WT dOMV-immunized mice (39%), were cleared of Ng relative to the Alum-immunized (17%) and unimmunized (19%) controls (Figure 2A). The ΔABR dOMV-immunized group alone exhibited a significant decrease in mean bacterial burden relative to the Alum control group at day 7 (P = .01), though an area under the curve (AUC) analysis showed no difference (data not shown).

Figure 2.

Dynamics of gonococcal clearance in response to vaccination with meningococcal detoxified outer membrane vesicles. Percent of colonized mice (left panel), percent of mice cleared at the end of the time course (middle panel), and the average bacterial burden (right panel) are shown for the first (A) and second (B) challenge studies. For A, the number of infected mice included 18 for the wild-type (WT), OCh, and Alum groups, 17 for the ΔABR group, and 16 for the unimmunized (Unim.) group. For B, the number of infected mice included 18 for the ΔABR and unimmunized groups, 17 for the WT and OCh groups, and 16 for the Alum group. P_WT, P_OCh, and PΔ ABR represent P values of WT-, OCh-, and ΔABR dOMV-immunized groups relative to Alum-immunized (black text) and unimmunized (blue text) control groups, respectively, as determined by the log-rank (Mantel–Cox) test (percent clearance) and 2-way analysis of variance with Bonferroni correction (mean bacterial burden). Hashed line represents the lower limit of detection (20 colony-forming units [CFU]/mL of vaginal lavage) for colony counts.

We repeated the challenge experiment, extending the period of observation from 7 days postinoculation to 10, using new, freshly prepared vaccines. The dynamics of colonization were somewhat altered, with clearance beginning at day 1 for both WT- and ΔABR dOMV-immunized mice, and steady clearance continuing through day 10 for all 3 vaccinated groups (Figure 2B). When compared to the Alum-immunized and unimmunized groups, the percentage of F62-infected animals was significantly lower in the vaccinated groups over the time course, and by the study endpoint (day 10), 71%–72% of dOMV-immunized mice had cleared infection compared to 25% and 33% of mice in the Alum-immunized and unimmunized groups, respectively. Despite increased clearance among the vaccinated groups, the WT dOMV-immunized mice alone exhibited a significant decrease in mean bacterial burden (P = .03 by analysis of variance) relative to the Alum-immunized mice at day 10; no difference in bacterial burden among the groups was calculated by AUC (data not shown).

Characterization of Murine Anti-meningococcal Vesicle Antibody Responses

To better understand the immune factors that contributed to F62 clearance, we compared Ng-specific IgG and IgA titers in vaginal lavages and sera collected 10 days after the second immunization (day 24) and in sera collected 10 days after the final immunization (day 38). In the first challenge study, significantly higher titers of total IgG antibodies were detected in the vaginal lavages of mice administered OCh and ΔABR dOMVs compared to WT dOMVs or Alum-immunized and unimmunized controls (Figure 3). Elevated vaginal IgG titers in the ΔABR dOMV group were associated with significantly higher total serum IgG titers relative to the WT dOMV group at the same timepoint, though all vaccinated mice produced significantly higher total serum IgG titers relative to controls at day 38. In the second challenge study, few of the mice produced any detectable total vaginal IgG. Significantly higher total serum IgG titers were only observed in the WT dOMV group at day 24 relative to negative controls, with all vaccinated groups producing significantly higher titers by day 38. Vaccination of mice with dOMVs was associated with elevated vaginal and serum IgG1 titers at both timepoints independent of challenge study (Supplementary Figure 4); vaginal and serum IgG2a and IgA titers differed according to challenge study.

Figure 3.

F62-specific vaginal and serum total immunoglobulin G (IgG) antibody titers. Total IgG titers in vaginal lavages at day 24 (top panel) and sera at day 24 (middle panel) and day 38 (bottom panel) from each independent challenge study are shown. Purple, red, blue, green, and black symbols represent antibody titers of individual mice that cleared infection on days 1, 3, 5, 7, and ˃7, respectively. *P ≤ .05, **P ≤ .01, ***P ≤ .001, and ****P ≤ .0001 by Kruskal–Wallis test with Dunn multiple comparison. Abbreviations: IgG, immunoglobulin G; Unim., unimmunized; WT, wild-type.

When pooled, serum antibodies from mice immunized with OCh and ΔABR dOMVs in the first challenge study bound to Ng proteins that were not recognized by anti-WT dOMV antibodies in immunoblots (Figure 4A). In contrast, pooled serum from mice immunized with WT dOMVs in the repeat study recognized largely the same Ng-specific antigens as the anti-OCh and anti-ΔABR antibodies (Figure 4B). A similar finding was observed when gonococcal lysates were probed with pooled vaginal lavage samples, with anti-OCh and anti-ΔABR dOMV antibodies from the first challenge study binding high-molecular-weight (~150 kDa to ~200 kDa) antigens that were not detected with anti-WT dOMV antibodies (Supplementary Figure 5). No differences in IgA recognition patterns by immunoblot were observed.

Figure 4.

Serum immunoglobulin G antibodies induced by immunization with wild-type (WT), OCh, and ΔABR detoxified outer membrane vesicle (dOMV) vaccines cross-react with FA19, FA1090, MS11, and F62 gonococcal lysates. Binding of antibodies to the parental meningococcal vaccine strain MC58 is also shown. Serum dilutions for samples from the (A) first and (B) second challenge studies are as follows: WT and OCh, 1:40 000; ΔABR, 1:80 000; Alum, 1:10 000. Closed arrows indicate immunogenic antigens unique to OCh or ΔABR dOMV immunization relative to WT dOMV-specific antisera. Open arrows indicate binding of anti-RmpM antibodies. No bands were observed upon probing with the Alum-specific serum pool at comparable dilutions.

Functional Antibody Responses

To examine whether dOMVs induced bactericidal activity against Ng, we tested pooled serum from each experimental group against sialylated Ng to better mimic LOS sialylation that occurs in mice during vaginal challenge [17]. Little bactericidal activity was detected at a 1:5 dilution against the serum-sensitive challenge strain F62, or strains MS11 and FA1090 (Figure 5A); there was no difference compared to control sera. Opsonophagocytosis was not measured due to limitations in serum volumes. However, an AUC analysis of the number of PMNs per mouse observed during the infection period suggested that an increased presence of vaginal PMNs was associated with clearance regardless of vaccination status (Figure 5B), despite no apparent correlation between PMN influx and rate of clearance (Supplementary Figure 6).

Figure 5.

Bactericidal antibody and murine polymorphonuclear leukocyte (PMN) responses. A, Pooled sera from mice immunized with wild-type, OCh, and ΔABR vesicle vaccines or Alum alone in the first challenge study were tested for bactericidal activity at a titer of 1:5 against gonococcal strains FA1090, MS11, and F62. Closed and open circles indicate percent survival upon incubation with active and heat-inactivated complement, respectively. B, Vaginal PMN influx over the course of infection of mice that cleared F62 or remained infected by the day 7 timepoint. Datasets from the first (upper panel) and second (middle panel) challenge studies and the combined dataset (bottom panel) are shown. Bars represent median ± 95% confidence interval. **P ≤ .01 by 2-tailed unpaired Student t test. Abbreviations: AUC, area under the curve; HI, heat-inactivated; PMN, polymorphonuclear leukocyte; WT, wild-type.

Identification of Cross-Reactive Neisseria Antigens

To identify the Nm dOMV antigens responsible for the elicitation of cross-reactive IgG antibodies, we incubated detergent-treated FA1090 OMVs with anti-WT and anti-∆ABR dOMV serum pools from the first challenge study, which showed distinct, diverse antigenic immune responses. FA1090 was chosen as an established sequenced reference strain used routinely in Ng pathogenesis and vaccine development research, including the human challenge model [20]. Antigen-antibody complexes were captured on Protein G–conjugated beads, and mass spectrometry analysis was performed. Thirteen gonococcal proteins that were significantly enriched in at least 1 of the experimental groups (WT- and ΔABR dOMV-immunized) relative to the Alum-immunized control were identified (Supplementary Table 3). Eight of the proteins reached the cutoff for consideration (mean number of spectra in experimental group ≥1.0 more than the mean number of spectra in the Alum control group; Table 1). Alum samples were significantly reduced (~2-fold) for mouse IgG peptides, which is consistent with lower cross-reactive antibody levels in control samples (Supplementary Figure 7). Background binding to the column was also observed, with samples precipitated with anti-ΔABR dOMV antibodies trending toward enrichment of gonococcal PorB peptides, though the difference was not significant. Rmp-specific peptides were only significantly enriched in the WT samples.

Table 1.

Proteins Identified in Immunoprecipitation Screen

		Probability Score, %			No. of Total Spectra			Sequence Coverage, %
Protein	Function	WT	ΔABR	Alum	WT	ΔABR	Alum	WT	ΔABR	Alum
Spg	IgG-binding protein G	100.0 ± 0.0	100.0 ± 0.0	100.0 ± 0.0	1088.0 ± 42.4	1112.4 ± 30.1	1205.2 ± 55.0	25.0 ± 1.0	23.2 ± 1.4	24.4 ± 1.5
Ighg1	Ig γ-chain C region	100.0 ± 0.0	100.0 ± 0.0	100.0 ± 0.0	22.4 ± 4.4	19.8 ± 1.6	9.0 ± 2.3	18.0 ± 2.7	14.6 ± 0.4	14.6 ± 1.8
Por	Major OMP P.IB	100.0 ± 0.0	100.0 ± 0.0	60.0 ± 24.5	45.2 ± 1.7	24.4 ± 17.4	1.4 ± 0.7	60.4 ± 0.9	29.4 ± 9.7	5.9 ± 2.8
Rmp	OmpA family protein	100.0 ± 0.0	60.0 ± 24.5	40.0 ± 24.5	6.2 ± 0.5	1.6 ± 0.8	0.6 ± 0.4	20.0 ± 2.4	5.9 ± 3.0	2.4 ± 1.7
PilQ	Type IV biogenesis and competence protein	60.0 ± 24.5	100.0 ± 0.0	20.0 ± 20.0	9.2 ± 5.7	27.0 ± 8.8	1.2 ± 1.2	9.7 ± 5.4	21.4 ± 5.3	1.7 ± 1.7
MtrE	Outer membrane efflux protein	100.0 ± 0.0	100.0 ± 0.0	0.0 ± 0.0	12.4 ± 2.0	19.6 ± 2.3	0.0 ± 0.0	32.8 ± 4.7	45.4 ± 3.2	0.0 ± 0.0
GroL	60 kDa chaperonin	100.0 ± 0.0	100.0 ± 0.0	60.0 ± 24.5	6.2 ± 1.2	8.8 ± 0.7	1.2 ± 0.6	9.6 ± 1.1	13.4 ± 0.7	2.8 ± 1.3
AceF	Acetyltransferase component of pyruvate dehydrogenase complex	80.0 ± 20.0	100.0 ± 0.0	20.0 ± 20.0	1.6 ± 0.4	8.6 ± 3.3	0.2 ± 0.2	2.2 ± 0.6	10.3 ± 3.7	0.3 ± 0.3
GuaB	Inosine-5′-monophosphate dehydrogenase	0.0 ± 0.0	100.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	6.2 ± 0.9	0.0 ± 0.0	0.0 ± 0.0	15.8 ± 2.5	0.0 ± 0.0
NlpD	Peptidase	100.0 ± 0.0	19.8 ± 19.8	0.0 ± 0.0	1.4 ± 0.2	0.2 ± 0.2	0.0 ± 0.0	3.7 ± 0.6	0.5 ± 0.5	0.0 ± 0.0

Values are expressed as the mean of the 5 biological replicates ± the standard error of the mean.

Abbreviations: IgG, immunoglobulin G; OMP, outer membrane protein; WT, wild-type.

Two previously identified Ng vaccine candidates, the multidrug-efflux protein MtrE [21] and the pilin protein PilQ [9, 22], were identified in the screen, with PilQ-specific peptides significantly enriched in the anti-∆ABR dOMV group alone. Probing of a ∆pilQ knockout strain with anti-∆ABR dOMV IgG antiserum (Supplementary Figure 8) verified that PilQ was the approximately 70 kDa antigen commonly recognized by IgG/IgA antibodies in immunoblots conducted with serum (Figure 4) and vaginal lavage (Supplementary Figure 5) pools isolated from the vaccinated groups. Peptides from the acetyltransferase AceF and the inosine-5′-monophosphate dehydrogenase GuaB, which promotes macrophage phagocytosis and is expressed in the cerebrospinal fluid of Nm-infected mice [23], were additionally only enriched in the anti-ΔABR dOMV group; enrichment of the membrane protein NGO0788 was observed, though the mean number of identified spectra did not reach the cutoff threshold (Supplementary Table 3). The anti-WT dOMV group alone was significantly enriched for NlpD, a membrane lipoprotein that promotes Ng cell separation [24] and virulence in a murine model of Yersinia pestis [25]. Enrichment for peptides specific to the 60 kDa chaperonin GroL was observed in both the anti-WT and anti-ΔABR dOMV groups.

DISCUSSION

Despite early trial failures [26–28] and the lack of protection afforded by primary infection [29], Ng vaccine design has been a topic of renewed focus within recent years. Epidemiologic data provide evidence that Nm dOMV vaccines may confer some degree of protection to vaccinated individuals against gonococcal infections [4–7]. However, the immunologic basis for protection remains obscure. Immunization of animals and human subjects demonstrates cross-reactivity of Nm dOMV-specific antibodies with gonococcal antigens [9, 30–32], and in vivo studies with Nm or Ng OMV vaccines show elicitation of antibodies corresponding with faster clearance of mice relative to PBS-immunized controls [9–11]. Intravaginal vaccination of mice with nondetoxified Ng OMVs induces protection against infection and production of gonococcal-specific local and systemic responses when delivered with QuilA [10] or microencapsulated interleukin 12 [11].

In our studies, parenteral administration of Nm dOMV vaccines formulated with Alum was sufficient to enhance gonococcal clearance, with associated increases in systemic and local IgG and IgA levels. In the first challenge experiment, accelerated clearance of Ng occurred in mice immunized with OMP-deficient dOMVs, but not WT dOMVs, with the ΔABR dOMV-immunized group exhibiting the highest clearance rate. In the repeat study, all 3 dOMV vaccines were protective. While several factors may be responsible for the different outcomes of these experiments, we detected significantly higher total vaginal IgG titers relative to the WT dOMV-immunized group in the first challenge study, a finding that was not observed in the repeat experiment. The ΔABR dOMV-specific vaginal antibodies also bound a wider breadth of Ng antigens in immunoblots, corresponding with binding of serum IgG to a larger number of Ng antigens. Binding of cross-reactive IgG antibodies in the second challenge study was similar for the dOMV-vaccinated groups, all of which exhibited similar levels of clearance. Based upon the respective clearance profiles, we speculate that vaginal and systemic IgG antibody to multiple cross-reactive antigens contributed to the protection we observed against Ng. However, because we did not collect vaginal lavage samples beyond day 24 in our model, a more thorough investigation of the association between local immune responses and gonococcal clearance is warranted.

The importance of antibody responses in the murine challenge model is highlighted by in vivo studies showing that Ng OMV-mediated gonococcal clearance is abrogated in µMT mice [11], implicating a major role for B cells in establishment of long-lived protection. Likewise, the protective effect of OMVs is completely or partially abrogated in interferon-γ– and CD4⁺ T-cell knockout mice, respectively, suggesting that Th1-driven CD4⁺ T-cell responses contribute to elicitation of protective antibody [11]. Gonococcal PorB can suppress dendritic cell–mediated CD4⁺ T-cell proliferation [33], and meningococcal PorB can promote Th2 skewing [34], suggesting that the absence of PorB in our ΔABR dOMV vaccine may promote B-cell maturation and antibody secretion. While antibody appears essential, the importance of bacteriolysis in dOMV-mediated protection remains unclear. Elicitation of murine antibodies exhibiting bactericidal activity against unsialylated Ng strains has been reported for 4CMenB [9] and the genetically detoxified Nm NOMV-FHbp vaccine [31, 32]. However, immunization of macaques or human study participants elicits antibodies with little-to-no anti-gonococcal bactericidal activity [9, 31, 32], comparable to our observations with sialylated bacteria.

Although the mechanism of antibody-mediated clearance has yet to be elucidated, it is clear that Nm dOMV vaccines stimulate complex immune responses that result in varying levels of protection from Ng. In particular, the similarity of the dynamics of clearance observed for the ΔABR dOMV group and the different responses observed for the WT dOMV group between the first and second challenge studies provide a basis for further exploration. As confirmed in our immunoprecipitation experiments, both the WT and the ΔABR dOMV vaccines were immunogenic and cross-reactive against gonococcal OMVs, with antibody responses elicited to different antigens. Considering the high variability of PorB, the elicitation of anti-PorB antibodies by the WT dOMV vaccine is unlikely to be productive. Similarly, PorA is not produced in Ng [35], while the immunodominance of PorA in Nm dOMV vaccines is well-characterized [36–38], rendering a substantial portion of the anti-dOMV antibodies nonfunctional. In addition, the ability of Ng anti-Rmp to impede gonococcal clearance in vivo [39] and to block the bactericidal function of antibodies directed against other surface-expressed structures has been reported [40], suggesting that the maintenance of Nm RmpM may diminish dOMV vaccine effectiveness.

The presence of PorA, PorB, and RmpM in licensed Nm dOMV vaccines may function not only to induce antibody responses that are not protective, but may also obscure immune recognition of less abundant cross-reactive antigens exposed upon porin deletion. In support of this theory, serum antibodies from Alum-immunized and unimmunized mice recognized a single 70 kDa antigen, which was of the same approximate size as PilQ. All vaccinated mice produced antibodies to PilQ, though enrichment was observed in ΔABR vs WT immunoprecipitation serum samples, suggesting enhanced boosting of preexisting anti-PilQ antibodies in the absence of PorA, PorB, and RmpM, an effect that likely extends to other cross-reactive OMPs as well. The fact that the WT dOMV vaccine promoted significant clearance in only 1 of the 2 challenge studies is noteworthy and lends credence to an alternative paradigm in vaccine development, one in which breadth of coverage of dOMVs is enhanced by removal of antigenically diverse and immunodominant OMPs rather than targeting of cross-protective antigens for overexpression.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors acknowledge Kristina Riebe for providing technical assistance in the performance of enzyme-linked immunosorbent assays.

Disclaimer. The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the US Food and Drug Administration (FDA), the Uniformed Services University of the Health Sciences, Duke University School of Medicine, the Department of Defense, the Department of Health and Human Services, or the National Institutes of Health (NIH).

Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases (interagency agreement number AAI14024 to A. E. J.). Construction of the Duke Regional Biocontainment Laboratory in which work was conducted received partial support from the National Institute of Allergy and Infectious Diseases (grant number UC6-AI058607 to G. D. S.).

Potential conflicts of interest. K. A. M. and M. C. B. are named as inventors on patent applications focusing on meningococcal outer membrane vesicle vaccines. Rights to these inventions have been assigned to the FDA. All other authors report no potential conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Presented in part: 21st International Pathogenic Neisseria Conference, Pacific Grove, California, 23–28 September 2017 (abstract O13); and STI & HIV 2019 World Congress (Joint Meeting of the 23rd International Society for Sexually Transmitted Diseases Research and 20th International Union Against Sexually Transmitted Infections congresses), Vancouver, Canada, 14–17 July 2019 (abstract 001.6).