Mucosal adjuvanticity and mucosal booster effect of colibactin-depleted probiotic Escherichia coli membrane vesicles

ABSTRACT

There is a growing interest in development of novel vaccines against respiratory tract infections, due to COVID-19 pandemic. Here, we examined mucosal adjuvanticity and the mucosal booster effect of membrane vesicles (MVs) of a novel probiotic E. coli derivative lacking both flagella and potentially carcinogenic colibactin (ΔflhDΔclbP). ΔflhDΔclbP-derived MVs showed rather strong mucosal adjuvanticity as compared to those of a single flagellar mutant strain (ΔflhD-MVs). In addition, glycoengineered ΔflhDΔclbP-MVs displaying serotype-14 pneumococcal capsular polysaccharide (CPS14⁺MVs) were well-characterized based on biological and physicochemical parameters. Subcutaneous (SC) and intranasal (IN) booster effects of CPS14⁺MVs on systemic and mucosal immunity were evaluated in mice that have already been subcutaneously prime-immunized with the same MVs. With a two-dose regimen, an IN boost (SC-IN) elicited stronger IgA responses than homologous prime-boost immunization (SC-SC). With a three-dose regimen, serum IgG levels were comparable among all tested regimens. Homologous immunization (SC-SC-SC) elicited the highest IgM responses among all regimens tested, whereas SC-SC-SC failed to elicit IgA responses in blood and saliva. Furthermore, serum IgA and salivary SIgA levels were increased with an increased number of IN doses administrated. Notably, SC-IN-IN induced not only robust IgG response, but also the highest IgA response in both serum and saliva among the groups. The present findings suggest the potential of a heterologous three-dose administration for building both systemic and mucosal immunity, e.g. an SC-IN-IN vaccine regimen could be beneficial. Another important observation was abundant packaging of colibactin in MVs, suggesting increased applicability of ΔflhDΔclbP-MVs in the context of vaccine safety.

KEYWORDS:

• Administration route
• mucosal adjuvanticity
• immunogenicity
• heterologous prime-boost immunization
• respiratory tract infection
• Streptococcus pneumoniae
• capsular polysaccharide
• probiotic Escherichia coli
• membrane vesicles

Introduction

Streptococcus pneumoniae, also called pneumococcus, is a Gram-positive pathogen encapsulated with polysaccharides capsule (CPS) and is the leading cause of respiratory infection, not only in young children,¹ but also in elderly individuals.^2,3 Currently, two types of pneumococcal vaccines have been approved, polyvalent pneumococcal conjugate vaccines (PCVs) and polyvalent pneumococcal polysaccharide vaccines (PPSVs). However, neither type elicits mucosal immunity because they are given by injection. It is important to achieve sterilizing immunity against pneumococci in the upper respiratory tract, since nasal colonization is a prerequisite for both common respiratory tract infections, such as otitis, sinusitis, and community-acquired pneumonia, and invasive pneumococcal disease (IPD), including life-threatening septicemia and meningitis. Therefore, development of a novel mucosal vaccine that elicits both systemic and mucosal immunity against pneumococcal diseases is highly desirable.

Membrane vesicles (MVs) are bacteria-derived small particles, which range in size from 20 to 300 nm in diameter. MVs contain their outermost surface antigens and pathogen-associated molecular patterns (PAMPs), such as lipopolysaccharide,⁴ as well as lipoprotein, peptidoglycan, fimbriae, flagella, and DNA/RNA. Therefore, MVs are used as acellular nanoparticle vaccines against the pathogens from which they are derived, which is exemplified by a 4-component vaccine against meningococcal invasive diseases that contain MVs of the serogroup (4CMenB, Bexsero®).⁵ On the other hand, a probiotic Escherichia coli (EcN or Nissle 1917)⁶-derived MVs carrying a protein antigen have been shown to induce self-adjuvanted protective humoral immune responses in a mouse model.⁷ EcN also showed an immunomodulatory effect on mammalian cells via Toll-like receptor 2 (TLR2)⁸ and NOD-like receptor NOD1.⁹ We have recently characterized probiotic EcN chimera-derived MVs displaying serotype-14 pneumococcal capsular polysaccharide CPS14 at high density (CPS14⁺MVs), which elicit potent IgG-dominant immune responses when subcutaneously (SC) administered.¹⁰

Intranasal (IN) vaccines that mimic natural infection induced not only mucosal IgA, but also systemic IgG responses than those via the parenteral route, such as SC and intramuscular administration. Some in vivo studies have shown that vaccines via IN route but not via SC route conferred protection against not only viral but also bacterial pathogens, e.g., Bordetella pertussis,¹¹ Burkholderia pseudomallei,¹² and Mycobacterium tuberculosis,¹³ which are the etiological agents of pertussis, melioidosis, and tuberculosis, respectively. On the other hand, in an S. pneumoniae vaccine study, the SC vaccine induced higher levels of serum antibodies, whereas the IN vaccine limited the bacterial load in the lung and blood by inducing IgA.¹⁴ The CPS14⁺MV vaccine via SC route caused IgG production in blood, but failed to elicit IgA responses.¹⁰ Based on these thoughts, it is interesting to examine whether IN booster with CPS14⁺MVs elicits robust immune responses and/or triggers class-switch recombination (CSR) to not only IgG but also IgA in mice that have already been SC prime-immunized with the same MVs, when compared to the homologous SC administration.

Some strains of Enterobacteriaceae carry a 54-kb pathogenicity island termed pks, which has been shown to be associated with development of colorectal cancer.^15,16 Colibactin is responsible for the genotoxic action of these strains and generated by non-ribosomal peptide synthetase (NRPS)/polyketide synthase (PKS) enzymes, which are encoded by genes of the pks island. EcN has been used as a probiotic for intestinal disorders for a long time and considered to be safe, though two different studies noted that this bacterium also hosts the entire pks island.^17,18 A more recent investigation confirmed genotoxic and mutagenic activities in EcN,¹⁹ suggesting the potential for adverse events, such as inflammation and carcinogenesis. Given the possible serious health concerns associated with EcN-derived colibactin, it may be better to avoid its use in clinical applications. Moreover, whether colibactin is present in MVs remains unknown.

In the present study, we discovered that colibactin was extracellularly released as MV cargo. With performance of in vivo intranasal immunization study using OVA as a model antigen, MVs isolated from the ΔflhDΔclbP strain lacking both flagella and colibactin showed significantly stronger mucosal adjuvanticity as compared to those from the parental ΔflhD strain and two other clb mutant strains. Furthermore, an investigation of immunization with different combinations of intranasal/subcutaneous administration showed the mucosal booster effect of intranasal CPS14⁺MV vaccines not only on triggering CSR to IgA, but also on enhancing SIgA production in saliva. Our data showed the potential of heterologous three-dose administration for building protective systemic and mucosal immunity. The implications as well as limitations of these findings for building protective immunity are also discussed.

Materials and methods

Bacterial strains and growth conditions

A flagellar-deficient derivative of the probiotic Escherichia coli strain, Nissle 1917 (DSM 6601, serotype O6: K5:H1)²⁰ was previously constructed and named EcNΔflhD strain.²¹ This flagellar-deficient mutant was used as parental strain in this study, because MVs prepared from the EcN wild-type strain were heavily contaminated with flagella.²¹ To obtain a colibactin-depleted EcNΔflhD strain, two clb genes involved in synthesis of colibactin in the pks island were chosen as targets to be deleted; clbA, which encodes a phosphopantetheinyl transferase that primes the NRPS - PKS enzymes to allow ketide - peptide chain elongation,²² and clbP, which encodes a membrane-bound protein with a periplasmic D-amino peptidase activity.^23–25 In addition, the clbS gene, which encodes a colibactin resistance protein and not expected to be involved in the genotoxicity,²⁶ was chosen as a target. The triple deletion mutants (ΔflhDΔclbA, ΔflhDΔclbP, and ΔflhDΔclbS) were constructed from the parental ΔflhD strain, using a method described by Datsenko and Wanner.²⁷ For construction of a EcNΔflhDΔclbP strain expressing exogenous serotype-14 pneumococcal CPS (CPS14), pNLP80 harboring an entire locus responsible for CPS14 biogenesis²⁸ was introduced. pWSK129²⁹ was used as a vector control of pNLP80. We chose pneumococcal CPS14 because CPS14 is one of the multi-serotypes included in the current polyvalent pneumococcal vaccine formula, as well as one of the most frequent serotypes that cause invasive pneumococcal diseases.³⁰ A pneumococcal strain used for upper respiratory tract infection experiment was ATCC 700676, a serotype 14 erythromycin-resistant strain, which was obtained from American Type Culture Collection (ATCC®, Manassas, VA, USA). The strain ATCC 700676 was grown in Todd Hewitt broth (Becton Dickinson, Franklin Lakes, NJ, USA) or on Mueller Hinton II agar (Becton Dickinson) plate at 37°C in 5% CO₂.

MV isolation

MVs were isolated from supernatant from a 16-hour bacterial culture of EcNΔflhD derivatives using a glycine induction method,^21,31 as described previously.¹⁰ Isolated MVs were resuspended in 20 mM Tris-HCl (pH 8.0) and stored at -20°C. The protein concentration of MVs was determined using a Bradford assay³² with bovine serum albumin (BSA) as the standard.

Field emission scanning electron microscopy (FE-SEM)

Morphological analysis of MVs was performed using an FE-SEM with sub-nanometer resolution (Regulus8220, Hitachi High-Technologies, Tokyo, Japan), as described previously.^10,21

SDS-PAGE and Western blot

MV and whole-cell samples were separated using Tris-glycine SDS-PAGE with 12.5% polyacrylamide gels and stained with Coomassie brilliant blue (CBB). For Western blotting, gels were electroblotted onto PVDF membranes. Rabbit polyclonal antibodies against CPS14 (#16753, SSI) were used at 1:10,000 dilution. Mouse monoclonal antibodies against MBP (E8032, New England Biolabs, Ipswich, UK) were used as previously described.¹⁰ Rabbit polyclonal antibodies against OmpA and CPS14 were used as previously described.^10,33 HRP-labeled anti-rabbit IgG (GE Healthcare) was used as a secondary antibody at 1:200,000 dilution. Following addition of Western BLoT Hyper HRP Substrate (Takara-bio, Shiga, Japan), chemiluminescence was visualized with a Fusion solo imaging system (Vilber Lourmat). For detection of both OmpA and MBP on the same membrane, fluorescence western blotting was performed using StarBright Blue (SBB) 520-labeled goat anti-rabbit IgG (Bio-Rad, Hercules, CA, USA) and SBB700-labeled goat anti-mouse IgG (Bio-Rad) as secondary antibodies. Fluorescence was visualized with Amersham ImageQuant 800 Fluor system (Cytiva, Marlborough, MA, USA). Excitation was performed using a 460-nm blue light laser for both SBB520 and SBB700, then fluorescence emission was collected in the green range of the spectrum with a band pass filter of 525 ± 10 nm (emission maximum: 520 nm) for the SBB520 dye and in the infrared range of the spectrum with a band pass filter of 715 ± 15 nm (emission maximum: 700 nm) for the SBB700 dye.

Nano-flow cytometry

The size and concentration of MV samples were analyzed using a NanoFCM (NanoFCM Inc., Xiamen, China), as described previously.³⁴ MV concentration and size distribution were calculated using the nano-flow cytometry software package, NanoFCM ver. 2.0.

LC-high resolution mass spectrometry (LC-HRMS) and nuclear magnetic resonance (NMR)

Colibactin and colibactin-related metabolites were isolated and analyzed by a combination of high-resolution mass spectrometry (LC-HRMS) and nuclear magnetic resonance (NMR) as described previously.³⁵

Animal experiments

All animal experiments were approved by our institutional animal care and use committee (protocol nos. 118,149 and 121,046) and performed in compliance with the institutional guidelines. Experimental overviews are shown with the timelines for immunization and sample collection in Figures 2(a), 4(a), 5(a), and S2(a). Six to seven-week-old female BALB/c mice (Japan SLC, Inc., Hamamatsu, Japan) were subcutaneously or intranasally immunized in the back or the nostrils two or three times with a three-week interval. MVs isolated from different EcN derivatives were used at a dose of 1 µg. Ovalbumin (OVA) was used as a model antigen and administrated at a dose of 5 µg. The volume for SC injection was 0.09 mL and for IN administration was 0.01 mL. Saliva, serum, and bronchoalveolar lavage fluid (BALF) were collected as described previously and used for detection of OVA- and CPS14-specific antibodies by ELISA. The procedures used for OVA or CPS14 antigen coating in ELISA plates have been described previously.^10,21

Upper respiratory tract infection model

Timeline of upper respiratory tract infection model used in this study was shown in Figure S2(a). The seven-week-old female BALB/c mice (n = 6) were grouped into three different regimens. Mice were immunized with CPS14⁺MVs by three-dose regimens of different routes: via SC-IN-IN (Grp. 2) and via SC-SC-SC (Grp. 3). The control arm without immunization was Grp. 1 (No immunization). Three weeks after the 3rd immunization, all mice were infected via nostrils with bacterial suspension of strain ATCC 700767 standardized at 2 × 10⁵ CFU with 10 µL of PBS (5 µL to each nostril). Twenty-four hours after the upper respiratory infection, mice were euthanized, and nasal wash and BALF were obtained. The mouse nasal cavity was washed 3 times with 200 µL of PBS, to yield 600 µL of the nasal washes. BALF samples were collected by 2 mL of PBS with 0.1% BSA. The CPS14-specific antibodies in nasal washes and BALF were examined by ELISA. The number of pneumococci in the nasal cavity and BALF was enumerated by CFU counting on Mueller Hinton II agar 24 hours after inoculum.

Statistical analysis

Statistical analysis was performed with one-way analysis of variance (ANOVA), followed by Tukey’s multiple comparison test. p values <.05 were considered to indicate statistical significance.

Results

Characterization of MVs of a series of clb gene-deletion mutants

The pks genomic island consists of 19 clb (clbA-S) genes that are essential for production of colibactin. ClbA, a pantetheinyl transferase, regulates colibactin production, while ClbP cleaves a precursor of colibactin into the genotoxic final product. On the other hand, ClbS is a resistant protein blocking the genotoxicity of colibactin (Figure 1(a)). In the present study, ΔflhDΔclbA, ΔflhDΔclbP, and ΔflhDΔclbS strains were constructed from the parental ΔflhD strains. MVs were isolated from culture supernatants of ΔflhD, ΔflhDΔclbA, ΔflhDΔclbP, and ΔflhDΔclbS strains using a glycine induction method.^21,31 No difference in the morphology, protein profile, and particle-size distribution was observed among these MVs (Figure 1(b–d)). LC-HRMS analysis revealed that the amount of colibactin in the parental (ΔflhD) cells was much lower as compared to the clinical isolates, #50, C08, and C34 (Figure 1(e)), which was in good agreement with a previous report by Hirayama et al.³⁶ Deletion of the clbP gene in #50 cells resulted in a dramatic decrease in the colibactin (Figure 1(e)), as described previously.³⁵ The amount of colibactin in whole cells of the ΔflhDΔclbS strain was comparable to that of the parental strain, whereas that in cells of the ΔflhDΔclbA and ΔflhDΔclbP strains was substantially decreased (Figure 1(e)). The amount of colibactin in whole cells was correlated with that in the MVs. Of note, colibactin was abundantly packaged into the MVs as compared to whole cells.

Figure 1. Characterization of membrane vesicles of different clb mutant strains.

(a) Gene cluster of colibactin. Shown are the pks gene island that encodes proteins responsible for colibactin production. A series of clb genes that encode non-ribosomal peptide synthetase (NRPS), polyketide synthase (PKS), and hybrid NRPS-PKS are shown as light green arrows. Several clb genes encoding accessory proteins are shown as pink arrows. The three clb genes targeted in the present study are indicated by red circles. (b–d) Characteristics of MVs. Two biological replicates of MV preparations from ΔflhD, ΔflhDΔclbA, ΔflhDΔclbP, and ΔflhDΔclbS strains were tested. Shown are representative data from two independent experiments. Similar results were obtained in these experiments. (b) Morphology. MVs of ΔflhD, ΔflhDΔclbA, ΔflhDΔclbP, and ΔflhDΔclbS strains were observed using FE-SEM. (c) Protein profiles. MVs of ΔflhD, ΔflhDΔclbA, ΔflhDΔclbP, and ΔflhDΔclbS strains, and standardized at 4 μg/mL. Twenty microliters of each sample was applied to 12.5% polyacrylamide SDS-PAGE, followed by CBB staining. MW: molecular weight marker. (d) Particle distribution (DLS). Diameter histograms of the MVs of ΔflhD (parental, black), ΔflhDΔclbA (blue),ΔflhDΔclbP (red), andΔflhDΔclbS strains are shown. (e) Colibactin quantification. Colibactin amounts were estimated using the amount of N-myr-Asn (m/z⁺ = 346) calculated from the results of LC-MS analysis, with the authentic reference of N-myr-Asn as the standard. The Y-axis shows the amount of colibactin (ng) per 1 mg of whole cell (red) or MV (blue) samples.

Mucosal adjuvanticity of MVs against a model antigen ovalbumin (OVA) in mice

A previous study found robust mucosal adjuvanticity of the parental (ΔflhD)-MVs.²¹ Mucosal adjuvanticity of the three deletion mutant-derived MVs was compared with that of the parental ΔflhD-MVs using an intranasal immunization mouse model with OVA as the model antigen (Figure 2(a)). The OVA ELISA data showed that ΔflhDΔclbP-MVs elicited stronger salivary and nasal wash SIgA responses, as compared to ΔflhD-MVs (Figure 2(b)). The OVA-specific IgG level was comparable among the different arms, and IgE was not detected in any of the mice (Figure 2(b)).

Figure 2. Mucosal adjuvanticity of MVs of different clb mutant strains.

(a) Timeline of immunization. The mucosal adjuvanticity of four different MVs (ΔflhD, ΔflhDΔclbA, ΔflhDΔclbP, and ΔflhDΔclbS) was examined using an intranasal immunization mouse model. Female BALB/c mice aged at six weeks were intranasally immunized twice with a two-week interval using four different arms: (i) OVA (5 µg)+ΔflhD-MVs (1 µg), (ii) OVA (5 µg)+ΔflhDΔclbA-MVs (1 µg), (iii) OVA (5 µg)+ΔflhDΔclbP-MVs (1 µg), and (iv) OVA (5 µg)+ΔflhDΔclbA-MVs (1 µg). At two weeks after the second vaccination, serum, saliva, nasal wash, and BALF samples were collected from the mice. Two independent animal experiments were conducted, and in each independent experiment four mice per arm were used. Therefore, shown are results of in total eight mice per arm. (b) OVA ELISA. ELISA was performed to examine OVA-specific serum IgM, IgG, IgA and IgE, salivary and nasal wash SIgA, and BALF IgG. Values are shown as OD405 (mean ± SD). Samples of serum, saliva, and nasal wash were diluted to 1:100, then used as primary antibodies for ELISA. All the values except for serum IgG were obtained after a 120-minute incubation with an alkaline phosphatase (AP) substrate. The value for serum IgG was obtained after 60 minutes of incubation with an AP substrate. Statistical analysis was performed using ANOVA and Tukey’s multiple comparison test. *p < .05. **p< .01.

Characterization of glycoengineered EcNΔflhDΔclbP-derived MVs

Next, glycoengineered ΔflhDΔclbP-MVs that displayed serotype-14 pneumococcal capsular polysaccharide (CPS14⁺MVs) were characterized (Figure 3). In a series of analyses based on FE-SEM, SDS-PAGE, and Western blot, we confirmed that morphology, protein profiles, and expression levels of CPS14 antigen, OmpA (outer membrane marker), and MBP (periplasmic marker) were comparable to those of ΔflhD-MVs displaying CPS14¹⁰ (Figure 3).

Figure 3. Characterization of MVs of parental and clbP mutant strains expressing pneumococcal serotype 14 capsular polysaccharide.

Protein profiles of whole cells (WCs) and MVs (a) and MV morphology (b) from parental (ΔflhD) and clbP mutant (ΔflhD ΔclbP) strains harboring an empty vector (CPS14: −) and CPS14 expression vector (CPS14: +). Three biological replicates of each WC and MV preparation were tested. Shown are representative data from three independent experiments. Similar results were obtained in three independent experiments. (a) Protein profiles of WCs (left panels) and MVs (right panels) of ΔflhD and ΔflhDΔΔclbP strains with or without CPS14 expression vector. Left panel: CBB stain. Middle panel: Western blot probed with anti-CPS14 antibody. Right panel: Fluorescent Western blot probed with anti-OmpA and anti-MBP antibodies. Asterisks (*) denote MBP signals. Daggers (†) denote OmpA signals. (b) Morphological analysis of MVs. Shown are representative FE-SEM images with low and high magnification of MVs (vector control) and CPS14⁺MVs.

Mucosal booster effect with two-dose (prime-boost) and three-dose (prime-boost-additional boost) regimens

To examine whether an intranasal (IN) boost with the CPS14⁺MV vaccine contributes to building rigor systemic and mucosal immunity, both a two-dose and three-dose immunization regimens which included SC prime immunization were used in a BALB/c mouse model (Figures 4(a) and 5(a)).

With a two-dose regimen (Figure 4(b)), there were no difference for levels of serum IgG and all IgG subclasses between SC boost and IN boost. An SC boost (SC-SC) elicited a significantly higher level of serum IgM as compared to the SC-IN regimen, whereas IN boost (SC-IN) elicited significantly higher serum IgG and salivary SIgA responses, as compared to the SC-SC regimen, which failed to elicit serum IgA and salivary SIgA responses. There was a tendency toward an increase in serum IgA level in the SC-IN regimen, as compared to the mock-immunization and the SC-SC regimen, although no significant difference was observed between the groups (p = .067). These data showed that not SC but IN triggered class-switch recombination (CSR) to IgA. IgE was not detected in any of the groups.

Figure 4. Humoral immune responses with two-dose vaccination regimen (Prime-Boost).

(a) Timeline of immunization and sample collection. Seven-week-old female BALB/c mice were immunized twice, at weeks 0 and 3, with CPS14⁺MVs (n = 32) or PBS (n = 8). The prime (1st) immunization was conducted via the SC route (n = 32), then the booster (2nd) immunization at week 3 was given to 32 SC-primed mice via the IN (n = 16) or SC (n = 16) route. At week 5, serum and salivary samples were obtained from all mice (shown by red square). (b) Humoral immune responses to CPS14. Serum and salivary sample obtained at week 5 were analyzed. Serum IgM, IgG, IgA, and IgE (top panels), serum IgG subclasses (IgG1, IgG2a, IgG2b, and IgG3) (middle panels), and salivary SIgA (upper right panel) were examined. For all ELISAs except for serum IgG, samples were used at 1:100 dilution, while those for IgG detection were used at 1:1,000 dilution. CPS14-specific antibody responses were analyzed by ELISA, for which the wells were coated with purified CPS14. The results are expressed as OD₄₀₅ peak values (mean ± SD) after incubation with an AP substrate. Statistically significant differences are indicated by asterisks (*p ≤ .05, **p ≤ .01, ***p ≤ .001, ****p ≤ .0001). ND: no significant difference. One-way ANOVA followed by Tukey’s multiple comparison test was used for the statistical analysis.

With a 3-dose regimen (Figure 5(b)), homologous subcutaneous immunization (SC-SC-SC) elicited the highest IgM responses among all tested regimens, though no statistically significant difference was observed. On the other hand, the SC-IN-IN regimen induced significantly higher serum IgA response as compared with the SC-SC-SC and SC-SC-IN regimens. The levels of serum IgG, IgG1, IgG2b, and IgG3 were comparable among all tested immunization regimens, whereas SC-SC-SC elicited a low level of serum IgG2a response. No IgE was detected in any of the groups. Notably, the level of salivary SIgA following SC-IN-IN was significantly higher than any other groups. Production of serum IgA and salivary SIgA increased as the number of IN administration doses increased (Figure 5(c)), again demonstrating that an IN boost triggers CSR to IgA.

Figure 5. Humoral immune responses with three-dose vaccination regimen (prime-boost-additional boost).

(a) Timeline of immunization and sample collection. Seven-week-old female BALB/c mice were immunized twice at weeks 0, 3, and 6 with CPS14+MVs (n = 32) or PBS (n = 8). The prime (1st) immunization was conducted via the SC route (n = 32), then the booster (2nd) immunization at week 3 was conducted with 32 SC-primed mice, via the SC (n = 16) or IN (n = 16) route. Next, the homologous SC prime-SC boost and heterologous SC prime-IN boost groups were further divided into two subgroups. An additional booster (3rd) immunization was given to via the SC (n = 8) or IN (n = 8) route to mice in the SC-SC (n = 16) and SC-IN (n = 16) subgroups. At week 8, serum and salivary samples were obtained from all mice (shown by red square). Note that the number of mice in the SC-IN-IN subgroup analyzed by ELISA was 7, as one mouse in the subgroup died accidentally after the sample collection at week 5. (b) Humoral immune responses to CPS14. Serum and salivary sample obtained at week 8 were analyzed. Serum IgM, IgG, IgA, IgE and salivary SIgA (top panels), and serum IgG subclasses (IgG1, IgG2a, IgG2b, and IgG3) (bottom panels) were examined. For all ELISAs except for serum IgG, samples were used at 1:100 dilution, while those for IgG detection were used at 1:1,000 dilution. CPS14-specific antibody responses were analyzed by ELISA, for which the wells were coated with purified CPS14. The results are expressed as OD405 peak values (mean ± SD), after incubation with an AP substrate. Statistically significant differences are indicated by asterisks (*p ≤ .05, **p ≤ .01, ***p ≤ .001, ****p ≤ .0001). ND: no significant difference. One-way ANOVA followed by Tukey’s multiple comparison test was used for the statistical analysis. (c) Relationship between number of IN administration doses and the serum IgA or salivary SIgA immune response.

Bacterial clearance in nasal cavity after SC-IN-IN

To compare the vaccine effect after immunization with the SC-SC-SC and SC-IN-IN regimens, bacterial clearance in nasal cavity was investigated using a respiratory tract infection mouse model (Figure S2(a)). ELISA data showed that both nasal and BALF IgA were elicited by the SC-IN-IN regimen, but not by SC-SC-SC one (Figure S2(b)). In addition, the results of nasal colonization 24 hour after pneumococcal infection showed that the bacterial clearance was achieved only by the SC-IN-IN group but not by the SC-SC-SC and non-immunized groups (Figure S2(c)).

Discussion

Novel vaccines that elicit both systemic and mucosal immune responses may offer a solution for reducing the burden of respiratory tract infections, such as pneumococcal diseases. Generally, IN vaccines trigger a robust protective immune response at mucosal sites of the respiratory tract, the frontline of infection with many pathogens. On the other hand, parenteral vaccines, such as those given an SC or intramuscular injection, usually elicit a stronger humoral immune response in blood as compared to IN administration, whereas the parenteral vaccines do not induce a secretory antibody response. Lapuente et al. reported that intranasal boost with adenoviral vectors led to both systemic and mucosal protective immunity against a SARS-CoV-2 infection in mice that had already been vaccinated with systemic plasmid DNA or mRNA priming.³⁷ Intranasal influenza-based boost vaccination also induces mucosal and systemic immunity for effective prevention of SARS-CoV-2 in both the upper and lower respiratory systems.³⁸ Furthermore, respiratory mucosal delivery of an adenoviral vector multivalent vaccine was shown to provide protection against ancestral and variant SARS-CoV-2 strains.⁴ There is strong interest in rerouting immune responses induced peripherally by IN vaccination to mucosal sites.

Because polysaccharides are per se T-cell independent antigen, the low immunogenicity not only of the current pneumococcal polysaccharide vaccine, e.g., PPSV23, but also of the conjugate vaccine, e.g., PCV7, 13, 15, and 20 have been described previously.³⁹ In our recent in vivo study,¹⁰ whereas the SC immunization with PPV23 and PCV13 induced weak IgM and moderate IgM/G responses, respectively, the SC immunization with CPS14⁺MV vaccine triggered robust IgG production in blood. However, the injected CPS14⁺MV vaccine failed to elicit IgA response.¹⁰ The present study expanded our knowledge on MV vaccinology when boosted via IN route. Our findings clearly revealed that IN booster with CPS14⁺MVs was effective for inducing both systemic and mucosal immune responses with CSR to both IgG and IgA. It has been reported that the nasal virus-like particle (VLP) vaccine, an alternative nanoparticle-based vaccine modality, triggered effective CSR to IgG/A to not only virus,^40,41 but also bacteria⁴² or protozoa.⁴³ Given that the morphology of CPS14⁺MVs resembles that of native viral virion (approximately 100 nm in diameter), the MV-based vaccine might induce potent immune responses through a nanopaticle-dependent manner similar to that of the VLP vaccine displaying antigens.⁴⁰

In the present study, ΔflhDΔclbP-MVs exhibited better mucosal adjuvant activity than the parent strain, though the reason is unknown, whereas two possible explanations are proposed. First, N-myr-Asn generated as a metabolite of colibactin is expected to be highly hydrophobic. Considering that adjuvant activity is often relevant to the structure of the lipid domain (hydrophobic region) of some molecules, e.g., lipopolysaccharides, mycobacterial trehalose-6,6’-dimycolate,^44,45the hydrophobic part of N-myr-Asn may regulate the host cell immune response. Another possibility is that since deletion of the clbP gene triggers accumulation of some colibactin precursors, they might stimulate production of IgA-promoting cytokines, i.e., the B cell-activating factor of the TNF family (BAFF), the proliferation-inducing ligand (APRIL), and TGF-β.^46,47 Colibactin precursors as microbial ligands may stimulate some pattern recognition receptors, such as Toll-like receptors (TLRs), NOD-like receptors (NLRs), RIG-I-like receptors (RLRs), C-type lectin receptors (CLRs), and cytosolic dsDNA sensors (CDSs). Further study is needed to elucidate the mechanism behind enhanced mucosal adjuvanticity caused by depletion of the clbP gene.

In the present study, EcNΔflhDΔclbP MVs were chosen as a vaccine carrier with the prospective advantages in terms of safety. One μg of CPS14⁺MVs was IN administered to a mouse in in vivo study. The safety is based on the detoxified property by LC-HRMS/NMR analysis for colibactin quantification (Figure 1(e) and S1) and the limulus assay for lipid A quantification.¹⁰ We performed in vivo safety assessment using BALB/c mice, in which no adverse reaction was observed when the intranasal administration dose was used at least not more than 10 µg (data not shown). Given the need of multivalent MVs for serotype-dependent immunity, the findings imply that more different pneumococcal CPS antigens can be introduced to MVs in the context of the safety concern.

Recently, a novel vaccine using pneumococcal membrane particles (MPs) has been reported with the vaccine effect in a mouse model.⁴⁸ MPs, like MVs, were isolated from bacterial culture and used as an acellular nano-particle vaccine. Two lipoproteins, MalX and PrsA, have been identified as the major protective antigens of the vaccine.⁴⁸ The major advantage of the MP vaccine was that it achieved a serotype-independent cross-protection, whereas either the present glycoengineered MVs or the current pneumococcal vaccines need to target a lot of different serotypes’ CPSs. On the other hand, MV vaccine can be manufactured at a lower cost than MP vaccine because MVs can be efficiently produced by using a glycine induction method^21,31 at P1 facilities, whereas MPs must be purified from a lot of culture plates or liquids of pneumococci at P2 facilities. Furthermore, the MP vaccine requires adjuvant for inducing potent immune responses,⁴⁸ which is in contrast to the case of the MV vaccine that strongly induced immune responses without any adjuvant (Figures 4 and 5). It has been reported that the binding of antibody to a surface antigen, phosphocholine, was largely blocked by CPS⁴⁹ and that CPS varies the thickness during the life cycle⁵⁰ and among serotypes and strains,⁵¹ suggesting that common protein antigens may be less accessible to antibodies by the thickened capsule. An approach that mutually compensates for the merits/demerits of MP and MV vaccines may be necessary, for example, by developing a technique to introduce cross-protective protein antigens as well as polysaccharide antigens into the MV vaccine, and vice versa.

The present findings showed a mucosal booster effect of intranasal CPS14⁺MV vaccines not only to trigger CSR to IgA, but also for enhancing SIgA production in saliva. Furthermore, colibactin was found to be much more abundant in MVs as compared to whole cells based on the same weight, suggesting a possible risk of carcinogenesis when colibactin-expressing MVs are used in clinical settings. A more comprehensive investigation of colibactin-depleted MVs, including a series of preclinical in vivo tests, is an important requirement for confirming safety in humans. In addition, robust mucosal immunity was shown to be elicited depending on the number of intranasal administration doses. However, additional studies are required for better understanding of the effect of various administration route combinations on immune responses. For example, it would be particular interest to examine how and to what extent immune responses are induced by a homologous IN regimen or a heterologous IN prime-SC-boost regimen. In addition, the effectiveness of the IN booster was evaluated only at the immunoglobulin level. So, it is necessary to examine cellular immunity such as Th1/Th2/Th17 cells with long-lasting memory immune responses following the IN booster. A dramatic pneumococcal clearance in nasal cavity was achieved by three-dose regimen of SC-IN-IN but not SC-SC-SC, suggesting eliciting SIgA response in nasal cavity may be responsible for the sterilizing immunity. However, the number of mice per arm is too small to draw a definite conclusion at present. Further studies to complete the animal study using sufficient number of mice, as well as to compare all different route combinations are in progress. However, despites these limitations, our findings suggest that novel use of flagella and colibactin-free MVs can provide a safe and effective vaccine platform, and may pave the way for a potential heterologous three-dose administration for building protective immunity.

Author contributions

R.N. designed the study, wrote the first draft of the manuscript, and did the final edits. H.U., T.Y., N.O., K.W., K.A., H.M., and R.N. conducted experiments. H.U., T.K., T.Y., N.O., K.W., K.A., H.M., M.T., N.N., Y.A., and R.N. reviewed manuscript. All authors contributed to the article and approved the submitted version.

Acknowledgments

The authors thank Fumiko Takashima, Junko Sugita, Toru Kikuchi, Hirotaka Kobayashi (National Institute of Infectious Diseases, Japan), Hirayama Satoru (Niigata University, Japan) for their valuable assistance. We are also grateful for Bernt Eric Uhlin and Sun Nyunt Wai (Umeå University, Sweden) and Mario F. Feldman and M. Florencia Haurat (Washington University, United States) for providing the precious materials.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/21645515.2024.2337987.

Additional information

Funding

This work was supported by grants from KAKENHI of the Japan Society for the Promotion of Science (JSPS) [nos. 19K22644, 20K09943, 21KK0164, and 21K18284] and Ministry of Education, Culture, Sports, Science and Technology (MEXT) [no. 20H03861].