Lipidation of Haemophilus influenzae Antigens P6 and OMP26 Improves Immunogenicity and Protection against Nasopharyngeal Colonization and Ear Infection

ABSTRACT

Nontypeable Haemophilus influenzae (NTHi) causes respiratory infections that lead to high morbidity and mortality worldwide, encouraging development of effective vaccines. To achieve a protective impact on nasopharyngeal (NP) colonization by NTHi, enhanced immunogenicity beyond that achievable with recombinant-protein antigens is likely to be necessary. Adding a lipid moiety to a recombinant protein would enhance immunogenicity through Toll-like receptor 2 signaling of antigen-presenting cells and Th17 cell response in the nasal-associated lymphoid tissue (NALT). We investigated effects of lipidation (L) of recombinant proteins P6 and OMP26 compared to nonlipidated (NL) P6 and OMP26 and as fusion constructs (L-OMP26ϕNL-P6 and L-P6ϕNL-OMP26) in a mouse model. After intraperitoneal or intranasal vaccination, antibody responses were compared and protection from NP colonization and middle ear infection were assessed. L-P6 and L-OMP26 induced approximately 10- to 100-fold-higher IgG antibody levels than NL-P6 and NL-OMP26. Fusion constructs significantly increased IgG antibody to both target proteins, even though only one of the proteins was lipidated. NP colonization and middle ear bullae NTHi density was 1 to 4 logs lower following vaccination with L-P6 and L-OMP26 than with NL-P6 and NL-OMP26. Fusion constructs also resulted in a 1- to 3-log-lower NTHi density following vaccination. NALT cells from mice vaccinated with lipidated protein constructs had higher levels of interleukin-17 (IL-17), IL-22, and CD4⁺ T-cell memory. Passive transfer of sera from L-OMP26ϕNL-P6-vaccinated mice to recipient infant mice reduced NP colonization and ear bulla NTHi density. We conclude that L-P6, L-OMP26, and fusion constructs generate enhanced antibody responses and protection from NP colonization and middle ear infection by NTHi in mice.

INTRODUCTION

Nontypeable (noncapsulated) Haemophilus influenzae (NTHi) causes respiratory infections in children and adults that lead to high morbidity and mortality worldwide. NTHi has become the most common cause of acute otitis media (AOM) in children, following the introduction of pneumococcal conjugate vaccines (1). It is the predominant pathogen of persistent AOM, recurrent AOM, and AOM treatment failures (2–4) and is a common cause of acute sinusitis and conjunctivitis in children and adults, and acute exacerbations of chronic obstructive pulmonary disease (COPD) in adults (5–8). Invasive diseases are also caused by NTHi (9, 10), and antibiotic resistance to NTHi is increasing (11, 12).

Protein D from NTHi is included in a pneumococcal conjugate vaccine as a carrier protein (PhiD-CV [not approved by the U.S. Food and Drug Administration]). PhiD-CV produced a 15% reduction in AOM caused by NTHi in children (13) but did not reduce nasopharyngeal (NP) colonization (14, 15), a key goal to achieve herd immunity from vaccination. Early-phase clinical studies in adults have evaluated NTHi proteins D, E, and PilA for safety and immunogenicity (16, 17), and an efficacy trial in adults to prevent acute exacerbations of COPD is nearing completion (18). Our group has been studying protein D, P6, and OMP26, with a focus on natural immunity induced by colonization and AOM infection in young children, including those who are otitis prone to NTHi (19), Streptococcus pneumoniae (20), and Moraxella catarrhalis (21) and have deficiencies in immunity response (22). Several groups have evaluated potential NTHi proteins, including proteins P4, P6, D, E, F, OMP26, PilA, Hap, HMW, and ZnuA, as vaccine candidates in animals (23–26).

It is apparent from our prior studies in children that natural exposure to NTHi and the immune response that follows are not sufficiently protective in prevention of subsequent NP colonization or AOM infection in either otitis prone or nonprone children during early life (25, 27–30). Also, recent clinical trials of pneumococcal recombinant proteins PhtD, PhtE, and PlyD1 (NCT01446926) and PhtD and Ply (NCT01262872) in human subjects failed to demonstrate protection from NP colonization (31) or AOM (32, 33), a desired outcome from vaccines targeting respiratory bacteria in order to provide herd immunity protection. Therefore, we sought a novel method to enhance immunogenicity of NTHi protein antigens as vaccines by making lipidated proteins (acylation at the N terminus). A lipid moiety stimulates the immune response via TLR2 receptor with TLR1 or TLR6 depending upon the triacyl or diacyl lipid nature of the ligands, respectively, and then downstream signaling via NFKβ (34, 35). Lipidated protein antigens from Streptococcus pneumoniae have been shown to stimulate Th17 cell responses (36).

We selected two highly conserved outer membrane and surface-exposed NTHi proteins (P6 and OMP26) as vaccine antigens that play different roles in pathogenesis. Of various vaccine candidates, we have evaluated (P6, PD, OMP26, PF, and Eftu), and we selected P6 since it is the most immunogenic in young children following NP colonization and AOM infections (19, 27, 37) and elicits bactericidal antibody in otitis-prone children despite their generally poor immune responsivity (27). P6 is naturally lipidated and has been shown to be protective in a rat lung challenge model (38, 39). OMP26 is not a naturally lipidated protein. In chinchillas immunized with OMP26, NTHi was rapidly cleared from the NP, and clearance was enhanced from middle ears in a model using direct inoculation of NTHi into middle ear bullae (40).

In the present study, we constructed and purified the novel vaccine antigens lipidated P6 (L-P6), lipidated OMP26 (L-OMP26), and lipidated fusion constructs L-OMP26ϕNL-P6 and L-P6ϕNL-OMP26 with the intent to enhance immunogenicity by stimulation of TLR2 receptors and Th17 cells and consequently improve protection against NP colonization and middle ear infections caused by NTHi in mice. We present evidence that all four protein constructs were more immunogenic, including nonlipidated proteins fused with lipidated proteins, and provided greater protection from NP colonization and middle ear infections than purified recombinant nonlipidated (NL) P6 and NL-OMP26.

RESULTS

Production and characterization of recombinant lipidated proteins.

Different cell lines [BL21 * (DE3)], media (2X-YT versus M9 minimal media), temperatures (30°C versus 37°C), and times after induction (4 h versus 16 h) were optimized to increase the expression and yield of lipidated proteins. CD41(DE3) cells with M9 minimal medium and overnight induction at 30°C gave a better yield for all lipidated proteins, including lipidated fusions. Adding a lipid moiety to OMP26, a native nonlipidated NTHi protein, was successful. The purity of each recombinant protein (lipidated and nonlipidated version) and their fusion constructs was verified by SDS-PAGE (Fig. 1). The protein bands corresponding to the molecular weights (MWs) of the lipidated proteins were also confirmed by immunoblots with either antigen-specific antibodies or His tag antibodies (data not shown). Lower levels of endotoxin occurred in lipidated proteins (average, 3.4 endotoxin units [EU]/mL) compared to NL proteins (average, 15.9 EU/mL). The endotoxin levels of all purified proteins given to animals were between 0.06 and 1.45 EU/mL, which is below the acceptable levels of endotoxin in vaccine formulations during preclinical research (41).

FIG 1.

SDS-PAGE characterization of purified lipidated (L) and nonlipidated (NL) proteins of P6 and OMP26 and their fusion constructs (L-P6ϕNL-OMP26, L-OMP26ϕNL-P6, NL-P6ϕNL-OMP26, and NL-OMP26ϕNL-P6), along with molecular markers. The relative size of each recombinant protein is shown.

L-P6 is predominantly triacylated, and L-OMP26 is predominantly diacylated.

Toll-like receptor 2 (TLR2) forms a heterodimer with coreceptors TLR1 or TLR6, depending upon either the triacylation or diacylation nature of the lipid, respectively, during stimulation (34). To estimate the proportion P6 and OMP26 palmitoyl chain that was triacylated or diacylated, we stimulated two cell lines with the lipidated proteins; HEK-Blue hTLR2-TLR1 and HEK-Blue hTLR2-TLR6 cells. Figure 2 shows the results of stimulations by lipidated constructs of P6 and OMP26 proteins and their fusion constructs. L-P6 stimulated hTLR2-TLR1 cells to 88% of the positive control consistent with triacyl lipidation (Fig. 2A). L-OMP26 stimulated hTLR2-TLR1 cells to 24% of the positive control consistent with lower levels of triacyl lipidation (Fig. 2A). L-OMP26ϕNL-P6 fusion stimulated hTLR2-TLR1 cells to 6% of the positive control consistent with very low triacyl lipidation, which is even lower than unfused L-OMP26. L-P6ϕNL-OMP26 fusion stimulated hTLR2-TLR1 cells to 47% of the positive control, consistent with lower triacyl lipidation than unfused L-P6.

FIG 2.

HEK-Blue TLR2/TLR1 and TLR2/TLR6 cell stimulation by lipidated (L) proteins and their fusion constructs. (A) TLR2-TLR1 cell stimulation with different proteins indicating triacyl lipidation presence in the designated vaccine constructs. Bars are plotted with normalized mean values and standard errors. The P values determined by t testing compared the OD values of protein constructs L-P6, L-P6ϕNL-OMP26, and L-OMP26 to a negative control (Pam2CSK4) and triacylation comparison among L-P6 and L-OMP26 and their fusions. (B) TLR2-TLR6 cell stimulation consistent with diacyl or triacyl lipidation of the vaccine constructs. The P values compared the OD values with protein constructs L-P6, L-P6ϕNL-OMP26, L-OMP26, and L-OMP26ϕNL-P6. NL proteins stimulated HEK-Blue TLR2/TLR1 cells similar to negative-control Pam2CSK4 (data not shown). ODN2006 (TLR9 antagonists), PBS, and medium alone were used as negative controls for HEK-Blue TLR-2-TLR6 cell stimulation, and they gave no signal (data not shown). Note that PAM3CSK4 stimulated HEK-BlueTLR2-TLR6 cells; therefore, the cell line cannot be used to distinguish diacyl and triacyl lipidation. The y axis displays normalized OD values of positive-control Pam3CSK4 (A) and Pam2CSK4 (B).

Similarly, all lipidated proteins stimulated hTLR2-TLR6 cells, consistent with diacyl lipidation (Fig. 2B). The triacyl positive control (Pam3CSK4) also stimulated the cells; therefore, responses reflect either triacyl or diacyl lipids. Compared to HEK-Blue hTLR2-TLR1 cell line stimulation, a higher response with L-OMP26 was measured compared to L-P6 constructs (P = 0.009) in hTLR2-TLR6 cells. Considering the low stimulation of hTLR2-TLR1 cells by L-OMP26 (Fig. 2A) and higher stimulation with hTLR2-TLR6 cells (Fig. 2B), we interpreted the result to suggest predominant diacyl lipidation of L-OMP26. This was confirmed with mass spectrometry (data not shown, Kaur et al., manuscript in preparation). L-OMP26ϕNL-P6 fusion and L-P6ϕNL-OMP26 fusion stimulated hTLR2-TLR6 cells similarly to their individual L-OMP26 and L-P6 protein components.

Three different signal sequences (SS1, SS2, and SS4) were used to add lipid to P6, with an interest to vary the triacyl and diacyl composition. However, we found that use of differing signal sequences did not change the level of stimulation of HEK-Blue hTLR2-TLR1 cells. Figure S1 in the supplemental material shows the normalized optical density (OD) values (normalized with PAM3CSK4 value), along with positive-control PAM3CSK4 (triacyl lipidated peptide) and negative-control PAM2CSK4 (diacyl lipidated peptide) of stimulation performed at two different concentrations. SS4 signal attached to P6 showed higher stimulation than its natural signal sequence, suggesting that our choice of signal sequence was appropriate (see Fig. S1).

Lipidation enhances the systemic IgG and IgM antibody response to vaccination with L-P6, L-OMP26, L-OMP26ϕNL-P6 fusion, and L-P6ϕNL-OMP26 fusion proteins.

We vaccinated mice with lipidated and nonlipidated constructs of P6 and OMP26. IgG and IgM responses to L-P6, NL-P6, L-OMP26, and NL-OMP26 proteins were compared after two and three intraperitoneal (i.p.) doses of 10 and 25 μg (Fig. 3). Lipidation of P6 and OMP26 significantly increased the magnitude of induced systemic IgG levels following i.p. immunization at the 10-μg dose (Fig. 3). At 25 μg, the IgG antibody levels for L-P6 were significantly higher after two doses but not after three doses, and the levels induced by LOMP26 versus NLOMP26 did not differ. Higher IgM antibody levels were consistently induced after immunization with lipidated constructs (Fig. 3).

FIG 3.

Immunogenicity of lipidated and nonlipidated P6 and OMP26 proteins. ELISA IgG and IgM titers (relative μg/mL) are shown after two and three immunization doses with different amounts (10 and 25 μg/dose) of antigens formulated in alum hydroxide. The P values were calculated by t test after transforming the data into log values comparing lipidated versus nonlipidated constructs in each panel. ns, not significant. Bars indicate mean values with the standard errors.

Increased immunogenicity in the NL portion of fusion proteins.

Lipidation of proteins may increase reactogenicity (42). Therefore, we designed a fusion construct of P6 and OMP26, attaching a lipidated protein to a nonlipidated protein to reduce the amount of lipid in the vaccine. Figure 4 shows the IgG levels induced by the fusion proteins L-OMP26ϕNL-P6 and NL-OMP26ϕNL-P6 compared to mixtures of the individual proteins. Figure 4A shows that IgG OMP26 levels are higher when L-OMP26 is in a fusion construct than when NL-OMP26 is in the fusion and that fusion of OMP26 to NL-P6 does not diminish IgG levels. Figure 4B shows the IgG P6 levels. The level of IgG to P6 is higher in the L-OMP26ϕNL-P6 fusion than when the proteins are mixed and in NL-OMP26ϕNL-P6, demonstrating increased immunogenicity in nonlipidated protein portion of fusion.

FIG 4.

Immunogenicity of lipidated and nonlipidated fusion proteins (L-OMP26ϕNL-P6 and NL-OMP26ϕNL-P6) compared to the mixture of their individual proteins to OMP26 (A) and P6 (B) showing increased immunogenicity in nonlipidated protein portion of fusion. Mice were vaccinated with three doses of 10 μg of each antigen formulated in alum hydroxide. ns, not significant. P values were determined by t test after transforming the data into a log scale. Bars represent mean values with the standard errors.

To determine whether the enhanced immunogenicity in nonlipidated protein portion of fusion is also observed after the order of proteins is switched; an L-P6ϕNL-OMP26 construct was made. Figure 5 shows the IgG responses of the fusions L-P6ϕNL-OMP26 constructs compared to a mixture of individual proteins. The level of IgG to OMP26 is higher in the L-P6ϕNL-OMP26 fusion than when the proteins are mixed (L-P6 and NL-OMP26) and in NL-OMP26, confirming again the increased immunogenicity effect of lipidation in nonlipidated protein portion of fusion (Fig. 5A). IgG P6 levels to L-P6 did not change (no significant difference) in an L-P6ϕNL-OMP26 fusion compared to individually given components showing that fusion of L-P6 to OMP26 does not diminish IgG levels (Fig. 5B). We also observed enhanced immunogenicity to NL-P6 when L-OMP26 + NL-P6 is mixed compared to NL-OMP26 + NL-P6 (P = 0.011) (Fig. 4B), but the enhanced immunogenicity was not observed with a different mixture for NL-OMP26 in L-P6 + NL-OMP26 compared to NL-OMP26+NL-P6 (Fig. 5A).

FIG 5.

(A) IgG to OMP26 after vaccination with L-P6ϕNL-OMP26 in comparison to a mixture of the individual proteins, L-OMP26, and a mixture of NL-OMP26 and NL-P6 showing the effect of increased immunogenicity by lipidation on the fused protein which is not lipidated. (B) IgG to P6 after vaccination with L-P6ϕNL-OMP26 in comparison to the mixture of the individual proteins, L-P6 and mixture of NL-OMP26 and NL-P6 showing increase immunogenicity effect of lipidation on the fused protein which is not lipidated Mice were vaccinated with three 10-μg doses of each antigen formulated in alum hydroxide. ns, not significant. P values were determined by t test after transforming the data into log scale. Bars represent mean values with the standard errors.

The IgG and IgM response of both fusions L-OMP26ϕNL-P6 and L-P6ϕNL-OMP26, along with mixtures of individual proteins at a different dose (25 μg/dose), are shown in Fig. S2 in the supplemental material, which again showed enhanced immunogenicity for the nonlipidated protein components in the construct. There was no difference in IgM titers comparing two versus three doses (Fig. 5) and the NL versions of P6 and OMP26 showed significant increases (P < 0.01 and P < 0.05, respectively) when attached in fusion proteins compared to their nonlipidated proteins (see Fig. S2).

Lipidation enhances protection from nasopharyngeal colonization and middle ear infection of mice.

Our mouse model is presented in Fig. 6A. Compared to NL proteins, lipidated proteins provided greater protection from NP colonization (Fig. 6B) and middle ear infection (Fig. 6C) measured as bacterial load after i.p. immunization. The effect of lipidation on protection was evaluated independently for both antigens (P6 and OMP26) after vaccinating individually by comparing lipidated and nonlipidated proteins, which showed a marginal reduction in bacterial burden that was not statistically significant (data not shown). We determined that X-31 viral load in the NP increased after inoculation, peaked around day 5 and then decreased to below the inoculum but still detectable on day 10 (see Fig. S3). Minimal X-31 detection was observed in ear bullae (see Fig. S3). Overall, higher IgG serum antibody levels correlated with lower CFU burden (correlation graph; see Fig. S4), and IgG serum antibody levels in mice immunized with lipidated constructs were higher than in mice immunized with nonlipidated constructs (see Fig. S4). The IgG levels in NP secretions induced by both fusions were compared to the individual protein mixture. Lipidated fusion proteins induced higher IgG levels in NP secretions (mean ± the standard errors of the mean [SEM], 13.1 ± 3.7 ng/mL for P6 and 152.3 ± 44.27 ng/mL for OMP26) compared to NL proteins (0.29 ± 0.08 ng/mL for P6 and 24.1 ± 5.7 ng/mL for OMP26) (P < 0.02 for P6 and P = 0.04 for OMP26).

FIG 6.

Mouse model and animal protection study. (A) Diagram showing mouse coinfection model. After three 10-μg doses of protein as a vaccine, mice were infected on day 35 with NTHi (10⁶ CFU/mice, strain 86-028-NP) and then, 6 h later, with X31 influenza virus (10³ ED₅₀). (B and C) i.p. immunization. The CFU burdens in the nasal lavage (B) and middle ear bullae (C) after vaccination with lipidated proteins were compared to nonlipidated proteins. (D and E) i.n. immunization. The CFU burdens in the nasal lavage (D) and middle ear bullae (E) after vaccination with L-OMP26ϕNL-P6 in the presence of curdlan adjuvant were determined. P values were evaluated by using the t test or ANOVA after log transforming the CFU. Each point represents an individual mouse with mean values and standard errors shown.

Intranasal (i.n.) immunization with lipidated proteins and a possible role in protection via the i.n. route was also tested using one of the fusions, the L-OMP26ϕNL-P6 fusion, and its role in protection via i.n. route. We found that i.n. immunization of L-OMP26ϕNL-P6 with curdlan as adjuvant provided greater protection from NP colonization (Fig. 6D) and middle ear infection (Fig. 6E) measured as bacterial load compared to adjuvant alone. The IgG levels of OMP26 and P6 in serum of mice vaccinated i.n. with L-OMP26ϕNL-P6 fusion is shown in Fig. S5 in the supplemental material. The results of repeat experiments involving i.p. and i.n. vaccinations with mixtures of lipidated (L-P6 + L-OMP26) and L-OMP26ϕNL-P6, L-P6ϕNL-OMP26 fusions and nonlipidated protein mixtures are shown in Fig. S6. The results are consistent with those shown in Fig. 6B to D, demonstrating a significant reduction in NP and ear NTHi CFU achieved by lipidated protein vaccinations compared to controls.

Lipidation of protein enhances Th-17 response and T-cell memory in NALT.

We found that L-P6 + L-OMP26 immunization resulted in greater stimulation of interleukin-17A (IL-17A; P = 0.009) and IL-22 (P = 0.03) responses from NALT compared to NL-P6 + NL-OMP26 immunization (Fig. 7A). The gating strategy of flow analysis showing IL-17 and IL-23 from one experiment of each group is shown in Fig. S7. We found that L-P6 + L-OMP26 immunization resulted in higher numbers of memory T cells in NALT (P = 0.03) compared to NL-P6 + NL-OMP26 immunization (Fig. 7B).

FIG 7.

(A and B) Cytokine (IL-17, IL-22, and IFN-γ) responses (A) and memory T cell levels (B) after in vitro stimulation of NALT from immunized mice that had been infected with NTHi. P values are shown comparing the response from L-P6 + L-OMP26 versus NL-P6 + NL-OMP26 vaccinations. P values were determined by Mann-Whitney test. Bars represent mean values with the standard errors.

Passive transfer of serum and protection in infant and adult mice.

Adult and infant mice may exhibit different immune protective mechanisms during disease pathogenesis. We passively transferred doses of 7 μL/mouse serum (in a final volume of 100 μL in infants and 200 μL in adults) obtained from two groups of mice vaccinated i.p. to 10-week-old mice and 10- to 15-day-old naive infant mice. The first group were sera from L-OMP26ϕNL-P6-vaccinated mice (containing a total ∼10 μg of IgG antibody to P6 and OMP26), and the second was the adjuvant control (no anti-P6 and anti-OMP26 antibodies). A reduction in NP colonization density (2-log reduction, P < 0.003) and ear infection (∼1 log, P < 0.008) was measured in infant mice (Fig. 8). In adult mice, no significant protection was observed (Fig. 8).

FIG 8.

Passive transfer of antisera obtained from L-OMP26ϕNL-P6-vaccinated and adjuvant control mice. Naive infant mice (10 to 15 days old) and adult mice (8 to 10 weeks old) (n = 5 in each group) received 7 μl of antisera from L-OMP26ϕNL-P6-vaccinated mice containing 10 μg of anti-P6 and anti-OMP26 antibodies, as determined by ELISA, or 7 μl of serum from adjuvant-vaccinated mice. The mice were challenged with a coinfection model (57), and the NTHi (86-028-NP strain) bacterial burden in ear bullae and NP lavage fluid was determined. P values were determined by a t test comparing two groups after log transforming the CFU. Each point represents an individual mouse with the mean values and standard errors shown.

DISCUSSION

Vaccines against NTHi bacteria are of major interest. In the present study, we investigated the impact of lipidation on immunogenicity and protection induced by NTHi proteins, including a protein that is not naturally lipidated (OMP26). We studied recombinant proteins L-P6 and L-OMP26 as individual vaccine antigens and as fusion constructs (L-OMP26ϕNL-P6 and L-P6ϕNL-OMP26) compared to NL-P6 and NL-OMP26 in a mouse model. L-P6 and L-OMP26 immunization induced significantly higher antibody levels than NL-P6 and NL-OMP26. Fusion constructs of lipidated and nonlipidated proteins significantly increased IgG antibody to both target proteins (L-OMP26ϕNL-P6 and L-P6ϕNL-OMP26), even though only one of the proteins was lipidated. NP colonization and AOM middle ear bullae NTHi density was 1 to 4 logs lower (P < 0.05 in both sites) after vaccination with L-P6 and L-OMP26 and fusion constructs compared to NL-P6 and NL-OMP26. NALT cells from mice vaccinated with lipidated constructs had significantly higher Th-17 responses (higher IL-17 and IL-22 levels and higher numbers of memory T cells). Passive transfer of sera from L-OMP26ϕNL-P6-vaccinated mice to recipient infant mice reduced NP colonization and ear bulla NTHi density.

Based on in vitro studies with HEK-Blue-hTLR2-TLR1, HEK-Blue-hTLR2-TLR6 cells, we suggested that L-P6 is predominantly in a triacylated form and that L-OMP26 is predominantly in a diacylated form. The ability for TLR2 to form heterodimers has been shown to lead to divergent responses depending on the heterodimer (43). Current work by our group involves production of purer triacyl lipidated and diacyl lipidated NTHi protein constructs and their fusions to study whether triacyl and diacyl lipid modifications result in differences in immunogenicity and protection from NP colonization and ear infection.

A major finding of the present study is the enhanced P6 and OMP26 antibody levels following i.p. or i.n. immunization with lipidated proteins compared to nonlipidated NTHi proteins. Prominent differences in serum IgG levels were measured when we compared lipidated and nonlipidated proteins after two doses and with smaller doses (10 μg/mouse). Giving three doses and a larger quantity of L-P6 and L-OMP26 antigen (25 μg/mouse) diminished immunogenicity differences, measured as serum IgG. IgM levels were significantly higher with lipidated constructs after two and three injections and at both antigen doses.

Lipoproteins may induce a higher inflammatory response than NL proteins. For example, fever was observed in infants given a lipidated protein in the meningococcal B vaccine Trumenba (44). Therefore, we made protein fusions where only the N-terminal protein is lipidated, hence reducing the amount of lipid by 50%. Parenteral vaccination with L-Omp26ϕNL-P6 not only enhanced the immunogenicity of the lipidated portion of protein, it also enhanced the immunogenicity of the NL fused protein. The effect of the increased antibody was also demonstrated if the protein orientation was reversed (L-P6ϕNL-OMP26). Enhanced immunogenicity to NL protein was not observed if lipidated and nonlipidated proteins were simply mixed together. One likely explanation for the enhanced immunogenicity and protection by lipidated proteins is the activation of TLR2. Previous studies have shown that protein lipidation modulates the immune response in a TLR2-mediated manner, enabling a Th1-related IgG response to pneumococcal infection in mice (45). Similarly, Neisseria meningitidis lipidated antigen showed differential T-cell polarization toward a Th1 phenotype than nonlipidated antigen (46). As a follow-up to the current findings, we are investigating relative TLR2/TLR1 and TLR2/TLR6 activations by NTHi lipoprotein constructs.

After the NTHi coinfection challenge, mice showed significant protection with lipidated proteins against NP colonization and ear infection compared to nonlipidated proteins. Both parenteral and mucosal vaccination methods gave similar results regarding enhanced protection using lipidated vaccine antigens. Higher IgG serum levels to the NTHi proteins studied correlated with a lower CFU burden in the NP and middle ear. Higher IgG levels were measured in serum and NP secretions of mice vaccinated with lipidated constructs compared to nonlipidated comparators. Previous studies have also shown that lipidated proteins with adjuvant (pneumococcal proteins DacB and PnrA) provide protection where the vaccine was given i.n. (47, 48).

Higher levels of IL-17 and IL-22 were measured in NALT among mice vaccinated with lipidated protein vaccine constructs. Th17 responses have been found to be promoted by TLR2 activation in dendritic cells and CD4⁺ T cells, and Th17-mediated protection against pneumococcal bacterial carriage has been shown to be dependent on TLR2 and surface-exposed lipoproteins (49, 50). Previous studies have shown that lipidated pneumococcal proteins stimulate higher Th17 cell responses in mice (45). IL-17 and IL-22 modulates recruitment of neutrophils, enhancing the clearance of bacteria from the NP (51), consistent with our results showing significant reductions in NTHi NP colonization density. Since AOM pathogenesis requires NTHi to reach a pathogenic threshold in the NP before bacteria ascend the Eustachian tube to cause AOM, significant reductions in middle ear bulla NTHi density likely related to Th17-mediated protection. The IL-17- and TLR2-dependent effects of lipidated proteins will be studied in the spleen and further studied in NALT after restimulation of cells, as previously described (48, 49).

Serum-derived IgG translocated to the mucosal NP surface has been shown to correlate with protection from Haemophilus influenzae type b by colonization and subsequent disease (52, 53). To study the effect of antibodies, we transferred serum-containing anti-P6 and anti-OMP26 antibodies from mice vaccinated with lipidated L-OMP26ϕNL-P6 fusion into infant and adult mice. A significant decrease in colonization and ear infection in infant mice was observed, indicating a role of antibodies in protection. In adult mice, antibody infusions did not provide protection, suggesting that other mechanisms, such as cellular immune responses, are more important. The result is consistent with our previous work with pneumococcal protein antigens where we studied antibody-mediated versus CD4⁺ T cell mediated protection. In those passive transfer experiments in naive infant and adult mice, adult mice showed CD4⁺ T-cell-mediated protection but no antibody-mediated protection, whereas infant mice showed synergistic protection with antibody and CD4⁺ T cells (54). However, a higher dose of antibodies specific to NTHi will be studied in future experiments in adult mice to further assess a dose effect.

There are limitations to our studies. NTHi does not naturally colonize the NP regions of mice. A high-challenge dose of NTHi (10⁶ CFU) and influenza coinfection was necessary to establish colonization that evolves to cause AOM. It has been shown that influenza infection transiently hampers neutrophil recruitment, which might impact protection studies (55). Comparative studies with all the vaccine constructs were not pursued. Further studies of cellular immune responses and greater in-depth analyses of immune mechanisms activated by lipidated versus nonlipidated NTHi vaccine constructs are needed.

In conclusion, we sought to increase systemic and mucosal antibody levels, stimulate Th17 immunity and T cell memory in the NP in order to provide enhanced protection from colonization and ear infections caused by NTHi in mice. Our strategy was to add lipids to recombinant proteins. To reduce potential reactogenicity from lipids, we constructed fusion proteins and lipidated only one of the two proteins in the fusion product. Our approach succeeded. Therefore, lipidation of NTHi antigens represents a promising path forward in developing novel NTHi vaccine formulations.

MATERIALS AND METHODS

Cloning and purification of recombinant lipidated and nonlipidated proteins.

P6 and OMP26 genes were synthesized using strain 86-028-NP strain sequences from GenScript (Piscataway, NJ) and cloned into a pET-21a vector for recombinant protein expression. Lipidated versions of P6 and OMP26 and their fusion constructs were designed to contain a P4 signal sequence (MKTTLKMTALAALSAFVLAGC) and a His tag at the C-terminal end. P6 protein with its natural signal sequence (MNKFVKSLLVAGSVAALAAC) and signal sequence of E. coli Pal homolog protein (MQLNKVLKGLMIALPVMAIAAC) were also synthesized. All of the constructs were expressed in Escherichia coli C43(DE3) (Lucigen) (56) (for lipidated proteins) or E. coli BLR21(DE3) (for nonlipidated proteins). The C43(DE3) strain was chosen to fully achieve lipidation of P6 and OMP26 proteins, as shown by Leng et al. (56). The fusion proteins L-OMP26ϕNL-P6 and L-P6ϕNLOMP26 were constructed with the signal sequence attached at the beginning, and a linker sequence (GSGGGG) was designed between two proteins so that they could independently fold. The amino acid sequence of L-OMP26ϕNL-P6 and L-OMP26ϕNL-P6 fusions are shown in Table S1 in the supplemental material. E. coli BLR21(DE3) strains containing NL P6 and OMP26 constructs were cultured in 2X-YT media, and C43(DE3) strains containing L constructs were cultured in M9 media. The expression of His₆-tagged proteins was induced at an OD at 600 nm (OD₆₀₀) of ∼0.6 with 0.4 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 2 to 4 h for NL proteins, and for L proteins, CD41(DE3) cells were grown at 37°C until reaching an OD₆₀₀ of ∼0.6 prior to the induction of protein expression with 0.4 mM IPTG overnight at 30°C. After induction, E. coli cells were centrifuged to make pellets for purification. NL proteins were extracted by sonication of the pellet in 1% Triton X-100 in phosphate buffer and nutated for 10 to 20 min, followed by centrifugation. Proteins were purified from the supernatant. For L proteins, overexpressed CD41(DE3) cell pellets containing target proteins were suspended in lysis buffer (50 mM Tris, 300 mM NaCl, 100 μg/mL lysosome, 1 mM phenylmethylsulfonyl fluoride) and lysed by three sonication cycles, followed by incubation in a water bath at 37°C for 20 to 30 min. The lysates were spun using ultracentrifugation at 29,400 × g for 30 min at 4°C. The supernatants containing cellular debris were discarded, and L proteins were extracted from the pellet using extraction buffer (1% Triton X-100, 2% Zwittergent, 50 mM Tris, 300 mM NaCl, and DNase at 5 μg/mL) and by incubating the pellet for 10 min at room temperature, followed by another three cycles of sonication and water bath incubation at 37°C for 30 min. The extract was spun using ultracentrifugation at 111,000 × g for 30 min at 4°C. The supernatant containing extracted protein was saved, and the extraction process was repeated using the extraction solution and ultracentrifugation. The two extracts were combined for purification.

Bio-Rad Profinity IMAC Ni-charged resin was used to purify the extracted His tag proteins by equilibration with 10 mM imidazole–0.5% Zwittergent–phosphate-buffered saline (PBS) and elution with 250 mM imidazole. Proteins were concentrated and buffer exchanged using a 3-kDa Centricon tube (Millipore/Sigma), and the final storage condition was PBS–0.05% Zwittergent at −80°C. NL proteins were purified similarly using IMAC Ni-charged resin, except there was no Zwittergent added to the buffer. The purity of each recombinant protein (lipidated and nonlipidated constructs) and their fusion constructs was verified by SDS-PAGE and staining the gels with SimplyBlue SafeStain dye (Invitrogen). All purified proteins were tested for endotoxin content with a Pierce Chromogenic Endotoxin Quant kit.

In vitro HEK-Blue-hTLR2-TLR1 and HEK-Blue-hTLR2-TLR6 cell stimulations.

HEK-Blue-hTLR2-TLR1 and HEK-Blue-hTLR2-TLR6 cells carrying a SEAP reporter construct (InvivoGen) were grown in Gibco Dulbecco modified Eagle medium supplemented with 10% fetal calf serum, 2 mM l-glutamine, and 100 μg/mL Normocin in the presence of HEK-BLUE selective antibiotic according to the manufacturer’s recommendations. When the cells reached 50 to 80% confluence, they were detached using 0.05% trypsin-EDTA, washed once with PBS, suspended in HEK-Blue detection medium (provided in the kit), and used directly. In some repeated experiments, Quant Blue detection medium (InvivoGen) was used as an alternative detection medium with a different incubation time and not added simultaneously in the presence of stimulants in accordance with the manufacturer’s instructions. For the assay, 2.8 × 10⁵/mL cells were stimulated with three different concentrations (10, 1, and 0.1 μg/mL) of purified lipidated and nonlipidated proteins in a final volume of 200 μL. Pam3CSK4 and Pam2CSK4 (InvivoGen) were included as positive controls for HEK-Blue-hTLR2-TLR1 and HEK-Blue-hTLR2-TLR6 cells, respectively. After 6 to 15 h of stimulation at 37°C and 5% CO₂, the OD₆₂₀ was measured.

Animal model of colonization and ear infection: vaccination schedules, influenza virus coinfection, and challenge with NTHi bacteria.

Animal experiments were conducted according to U.S. law and NIH regulations and with Rochester General Hospital Committee for Animal Ethics approval at our accredited institute by the American Association of Laboratory Animal Care. Six- to eight-week-old male and female inbred C57BL/6 mice (Jackson Laboratory) were immunized in different groups (n = 5 to 10 in each group).

Vaccine schedule.

Mice were vaccinated by either intraperitoneal (i.p.) injection or intranasal (i.n.) inoculation at weeks 0, 1, and 3 with either 200 μL (i.p.) or 20 μL (i.n.)/animal. Mice were vaccinated with different quantities of each protein (10 or 25 μg/dose of purified proteins mixed with their respective adjuvants). For i.p. injections, aluminum hydroxide was used (Alhydrogel; InvivoGen) as an adjuvant or no adjuvant, and for the i.n. vaccinations, Curdlan (InvivoGen) was used. Vaccine formulations contained 10 or 25 μg of protein component in L-P6, L-OMP26, L-OMP26ϕNL-P6, and L-P6ϕNL-OMP26 and their respective nonlipidated constructs (NL-P6 and NL-OMP26). To determine the role of antibodies in protection, infant mice (2 weeks old) and adult mice (8 to 10 weeks old) received passive transfer of mouse sera containing relative amount (based on enzyme-linked immunosorbent assay [ELISA] results) of 10 μg of anti-P6 and anti-OMP26 antibodies (obtained from mice vaccinated with the L-OMP26ϕNL-P6 fusion) and then challenged similarly in our NTHi/influenza viral coinfection model on the same day. Control group mice received the same amount of sera obtained from alum-treated control mice.

NTHi/influenza virus coinfection model.

Previously, we reported the development of a NTHi/influenza virus coinfection model (57). In that model, PR8 influenza virus was administered 1 week before NTHi infection, and the efficacy of protein D vaccination in preventing ear infection was described. In the present study, we modified the model by changing to a less-virulent strain of influenza virus, X-31. We conducted an experiment to determine the viral loads at days 0, 3, 5, 7, and 10 after X-31 inoculation.

For the coinfection and challenge study, C57BL/6J mice were i.n. inoculated with NTHi strain 86-028-NP or strain 575 (1 × 10⁶ CFU per 15-μL dose diluted in PBS, half the volume per nostril with animals lightly anesthetized and on their back) and then at 6 h postinfection received the X31 influenza virus strain (1 × 10³ EID₅₀ per 20-μL dose given i.n.). At 5 days after coinfection, the mice were euthanized with isoflurane, followed by a cervical dislocation, and then nasal lavage (NL), ear wash (EW), ear bullae (EB), and nasal-associated lymphoid tissue (NALT) were collected. The nasal lavage fluid was collected from the nares using 300 μL of PBS by retrotracheal lavage. Ear wash fluid was collected by inserting a 25-gauge needle through the tympanic membrane and pipetting and removing 10 μL of PBS (at a time) for a total of 60 μL (six washes) per ear. NL and EW were 10-fold serially diluted in PBS, placed onto chocolate plates, and enumerated for NTHi recovery (CFU/mL). After a washing step, the EB was removed and homogenized in 1 mL of PBS, serially diluted with PBS, placed onto chocolate plates, and enumerated for NTHi recovery (CFU/mL). NALT was obtained by scraping the cells from nasal septum palates and storing 2-5e10⁵ cells obtained per mouse in liquid nitrogen freezing medium for later batch experiments.

Mouse P6 and OMP-26 serum IgG and IgM ELISA.

Microplates (Microlon medium binding; Greiner Bio-One) were coated overnight at 4°C with nonlipidated NL-P6 or NL-OMP26 proteins at 1 μg/mL in carbonate buffer. For the standard, the coating was done with goat anti-mouse IgG-Fc (Bethyl, catalog no. A90-131A) at 1 μg/mL, providing a means for the relative quantitation of mouse IgG and IgM. The coated plates were washed five times with PBS–0.1% Tween and blocked with 4% skim milk in 1× PBS–0.1% Tween (the same solution was used for primary and secondary incubations). Each mouse serum sample was serially diluted, starting with a prestandardized dilution of (100 to 1:50,000). For the standard, mouse reference serum (Bethyl, catalog no. RS10-101) was tested at 10 ng/mL, followed by eight twofold dilutions as a relative quantitative measurement. The primary incubation was for 1 h at room temperature, followed by 5× plate washes, and the addition of IgG- or IgM-specific horseradish peroxidase-labeled secondary antibody at 1:5,000 (Bethyl, catalog numbers A90-131P for IgG and A90-140P for IgM), followed by incubation for an additional hour. The plates were rewashed before adding TMP substrate (KPL) and then allowed to develop for 20 min at room temperature; the reaction was then stopped by the addition of 1 M H₃PO₄, and the plates were read at 450 nm. P6 and OMP-26 IgG and IgM titers were approximated relative to the mouse standard concentration.

NALT analysis.

NALT (n = 5/group) was collected from mice that received i.p. immunizations with L or NL P6 and OMP26, followed by an i.n. infection with NTHi 5 days prior to the collection of NALT. Pooled NALT cells obtained from five animals were phenotypically characterized by using a LSRII flow cytometer (Becton Dickinson, San Jose, CA). After overnight (∼16 h) stimulation of NALT cells with NL P6 and NL OMP26 combined (10 μg/mL of each antigen), intracellular and surface cell staining was performed with rat anti-mouse CD4, CD3, CD62L, CD44, IL-17, IL-22, and IFN-γ antibodies conjugated with different fluorophores (antibodies were titrated for optimal performance before assay). Data were analyzed for at least 50,000 events with BD FlowJo software version 10. Isotype-matched antibodies were used as negative controls. For NALT cell analysis, the gate was initially set on the lymphoid fraction selected by the forward angle and 90° light scatter from NALT lymphocytes and then further on CD4⁺ cells and cytokine markers.

Statistical analyses.

All statistical analyses were performed using Prism version 6.0 (GraphPad Software, San Diego, CA). A two-tailed t test on log-transformed data or a nonparametric Mann-Whitney test was used to compare two groups. One-way analysis of variance (ANOVA) with Bonferroni’s posttest was used to compare multiple protection groups if multiple comparisons were made between groups. Differences with a P value of <0.05 were considered statistically significant.