RBD-Modified Bacterial Vesicles Elicited Potential Protective Immunity against SARS-CoV-2.

The disease caused by SARS-CoV-2 infection threatens human health. In this study, we used high-pressure homogenization technology not only to efficiently drive the bacterial membrane to produce artificial vesicles but also to force the fusion protein ClyA-receptor binding domain (RBD) to pass through gaps in the bacterial membrane to increase the contact between ClyA-RBD and the membrane. Therefore, the load of ClyA-RBD on the membrane is substantially increased. Using this technology, we constructed a "ring-like" bacterial biomimetic vesicle (BBV) loaded with polymerized RBD (RBD-BBV). RBD-BBVs injected subcutaneously can accumulate in lymph nodes, promote antigen uptake and processing, and elicit SARS-CoV-2-specific humoral and cellular immune responses in mice. In conclusion, we evaluated the potential of this novel bacterial vesicle as a vaccine delivery system and provided a new idea for the development of SARS-CoV-2 vaccines.

The receptor binding domain (RBD) of the SARS-CoV-2 spike protein can bind to angiotensin-converting enzyme 2 (ACE2) in host cells, leading to viral invasion.[1] Therefore, the RBD is considered an effective subunit vaccine target.[2] The trimer structure of the RBD serves important functions based on the structure of SARS-CoV-2,[3] and vaccine forms based on RBD polymers have been shown to enhance vaccine-induced neutralizing effects.[4,5] However, current vaccines have multiple limiting issues, such as high manufacturing costs, low yields, and transportation difficulties. The Escherichia coli expression system can achieve highly efficient expression of exogenous proteins and improve vaccine yields with low costs,[6] but exogenous proteins in E. coli are poorly folded and tend to form misfolded inclusion bodies,[6,7] greatly limiting the advancement of prokaryotic expression systems in viral vaccines. Correct exposure and polymerization of the RBD in prokaryotic expression systems will facilitate the development of low-cost vaccines against SARS-CoV-2.

The use of bacterial outer membrane vesicles (OMVs) to display exogenous proteins can improve the folding ability of exogenous proteins and enhance the immune response capacity due to efficient uptake of OMVs by dendritic cells (DCs). While many studies have shown that OMVs can serve as an effective vaccine delivery system for bacterial,[8−10] viral,[11] and tumor vaccines,[12,13] unfortunately, the capacity of OMVs to present exogenous proteins is limited because high contents of the ClyA membrane-anchoring protein cannot be localized to the outer membrane. However, ClyA can be transported to the outer membrane only through its own bacterial regulatory mechanism, which considerably limits the efficiency of ClyA anchoring to the outer membrane to deliver antigens. To artificially produce modified bacterial vesicles with high protein contents, two requirements should be met: (1) A large number of bacteria should be artificially induced to bud to simulate the process of natural OMV release, and the sizes of artificial buds should be close to those of OMVs. (2) ClyA should be artificially forced to bind tightly to the outer membrane such that a large amount of ClyA is anchored to the outer membrane and undergoes conformational changes and assembly, thereby mediating the efficient delivery of foreign proteins. However, no technology has been found to achieve these two requirements to date.

In this study, we developed a technology that can achieve these two requirements simultaneously. We used high-pressure homogenization technology not only to efficiently drive the bacterial membrane to produce artificial vesicles but also to force the fusion protein ClyA-RBD to pass through gaps with the bacterial membrane to increase the contact between ClyA-RBD and the membrane. Therefore, the load of ClyA-RBD on the membrane is greatly increased. This technique is a next-generation bacterial membrane manipulation technique. Taking advantage of the highly efficient assembly of ClyA on bacterial outer membranes,[14,15] we utilized ClyA to mimic the support structure of SARS-CoV-2 S2 protein for the RBD and achieved outward exposure of the RBD on the self-assembled bacterial vesicles with high efficiency, leading to high exposure and polymerization of ClyA-RBD on bacterial membranes and the presentation of “ring-like” nanostructures on the vesicles. The presentation of the RBD on bacterial biomimetic vesicles (BBVs) was 28.16-fold greater than that on OMVs, and the yield of BBVs was 107-fold higher than the yield of OMVs. These RBD-BBVs injected subcutaneously can accumulate in lymph nodes, promote antigen uptake and processing, and elicit SARS-CoV-2-specific humoral and cellular immune responses in mice. In conclusion, we evaluated the potential of this novel bacterial vesicle as a vaccine delivery system and provided a new idea for the development of SARS-CoV-2 vaccines.

Results and Discussion

High-Pressure Homogenization Drove Bacteria to Form a Large Number of Biomimetic Vesicles and Increased the RBD Load

OMVs have been considered an excellent vector for vaccine or drug delivery. We have attempted to use OMVs to load the RBD to fight against SARS-CoV-2. However, low yields, a low load of exogenous proteins, and safety concerns greatly limit the use of OMVs as vectors for SARS-CoV-2 vaccines. In this study, we developed a next-generation self-assembly technique using a genetically engineered modified bacterial membrane. We compared the genetic modification technology of BBVs and OMVs and found that the RBD load on BBVs was significantly higher than that on OMVs (Figure A). The RBD load on BBVs was 28.16-fold higher than the load on OMVs (Figure A). Furthermore, the yield of RBD-BBVs was 107-fold higher than the yield of RBD-OMVs prepared from wild-type BL21 (Figure B). More importantly, almost no bacterial nucleic acids were detected in RBD-BBVs (Figure C), substantially improving the safety of the vaccine. After sample preparation, the assembled RBD-BBVs were isolated through iodixanol gradient centrifugation (Figure D). SDS-PAGE electrophoresis confirmed the high-efficiency loading of the RBD on BBVs. Western blotting using specific antibodies against S1 of SARS-CoV-2 further confirmed that the genetically engineered protein presented on BBVs was the RBD (Figure E). To prove that high pressure is the main factor driving the increase in RBD load, we analyzed the load on the bacterial membrane prepared by the RBD under different pressure conditions. The experimental results proved that the RBD load is positively correlated with pressure (Figure F). In addition, the bacterial membranes prepared using two different ultrasonic disintegration conditions could not effectively increase the RBD load (Figure G). Moreover, high-pressure homogenization can better drive bacteria to form vesicles than ultrasonic disruption (Figure H). We also used nanoparticle tracking analysis (NTA) to analyze the particle sizes of RBD-modified BBVs and wild BBVs, and the results showed that the particle sizes of RBD-modified BBVs were larger than those of wild BBVs, which may be related to the large amount of the RBD displayed on the surface (Figure I). Finally, we analyzed the formation mechanism of BBVs through iTRAQ. The cellular component (CC) results from Gene Ontology (GO) enrichment analysis proved that BBVs efficiently enrich bacterial plasma membrane proteins and substantially reduce intracellular proteins such as ribosomes (Figure J). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis also proved that compared with the whole bacterial protein, the upregulated signal was related to membrane function, and the downregulated signal was related to the biological process in the cell (Figure S1). Therefore, we believe that BBV production involves reassembly of the plasma membrane after the release of intracellular proteins following high-pressure extrusion. In summary, we developed a genetic engineering technique for BBVs that addressed the technical bottleneck of OMV modification and achieved highly efficient presentation of the RBD on BBVs.

Figure 1

High-pressure homogenization drove bacteria to form a large number of biomimetic vesicles and increased the RBD load. (A) Wild-type E. coli BL21 whole cells (WT-WC), bacteria expressing the RBD (RBD-WC), BBVs loaded with the RBD (RBD-BBVs), and OMVs loaded with the RBD (RBD-OMVs) were analyzed by SDS-PAGE followed by silver staining. The positions of the RBD are marked with red arrows. The RBD content was assessed with Image Lab software (Bio-Rad). The experiment was repeated three times. (B) Comparison of the yield of RBD-modified OMVs and BBVs using the Bradford assay to calculate the total protein amount. The experiment was repeated three times. (C) The DNA content in the samples was analyzed with EB-stained agarose electrophoresis. (D) Photos show the layering of BBV samples before and after gradient centrifugation. The black arrowheads indicate that the sample was distributed into various layers. (E) Protein expression in unmodified native BBVs, RBD-BBV samples, and whole bacterial cells before and after IPTG induction were analyzed with SDS-PAGE and silver staining. BBV and RBD-BBV samples are indicated by half arrows, while the complete arrow indicates the position of the ClyA-RBD fusion protein. The presence of the RBD was verified by Western blot using antibodies against S1 of SARS-CoV-2. The complete arrow indicates the position of the ClyA-RBD fusion protein. (F) SDS-PAGE and silver staining results of bacterial membranes prepared after exposure to different high-pressure homogenization conditions. Image Lab software was used to analyze the RBD content in each lane. The experiment was repeated three times. (G) Bacterial membrane prepared with 1200 bar high-pressure homogenization and two sonication conditions (ultrasonication time: 5 min on the left and 10 min on the right). SDS-PAGE and Image Lab software were used to analyze the RBD content in each lane. The experiment was repeated three times. (H) TEM analysis of vesicles formed by bacterial membranes prepared by 1200 bar high-pressure homogenization technology and sonication. (I) NTA particle diameter analysis (n = 3). (J) GO enrichment analysis of the cellular compartments of the differentially increased or decreased proteins. The top three proteins with the most significant upregulation or downregulation between BBVs and whole bacteria.

The RBD Was Exposed on the BBV Surface and Polymerized into a Ring-like Structure

Under transmission electron microscopy (TEM), we observed the ring-like structure formed by ClyA-RBD on the vesicle surface (Figure A left). We used the homology modeling method to construct the structural model of the recombined protein ClyA-RBD (Figure S2). As shown in Figure S2, the domains of ClyA and the RBD are connected by a short linker peptide, and these two domains do not overlap, confirming our hypothesis that the parental sequences in the recombinant can fold independently into their respective native conformations without spatial interference between the folded domains. Interestingly, when the recombinant model was mapped onto the self-assembled ClyA that spans the bacterial outer membrane (Figure A right), the RBDs polymerized into a ring-like 12-mer structure,[14,16] were close to each other and were located in the region distant and outside of the outer membrane, which may provide stabilization to the associated subunits and can explain the experimentally observed binding reaction of the RBDs in recombinant to hACE2 (Figure C). As proteinase K (PK) can remove proteins on the vesicle surface, most proteins were removed from RBD-BBVs after PK treatment, indicating that the proteins on RBD-BBVs mainly consisted of the RBD and membrane proteins, with no observable excess intracellular protein, and that the RBD was completely exposed on the BBV surface. Specific antibodies against SARS-CoV-2 S1 were used to confirm complete exposure of the RBD on the surface (Figure B). A binding assay for RBD-BBVs and hACE2 based on the specific binding of native RBD to hACE2 (Figure C) was performed, and the experimental results confirmed that the polymerized RBD on RBD-BBVs has stronger hACE2 binding ability than monomer RBD and that the binding of RBD-BBVs to hACE2 was comparable to that for S1-Fc proteins expressed in HEK293 cells (Figure C). To analyze the spatial conformation of the RBD on BBVs, we tested the binding level of RBD-BBVs with six monoclonal antibodies that have been shown to have virus neutralization effects. The experimental results proved that RBD-BBVs can bind to four of the six antibodies, and only two antibodies have no binding reaction (Figure D). Therefore, we believe that RBD-BBVs have a large number of spatial conformations similar to S1 expressed in eukaryotic systems. However, due to a lack of glycosylation modification, some spatial conformations may not be completely consistent with glycosylation modification of S1, and as a result, D001 and D002 antibodies cannot recognize them. However, the structures that can be recognized by neutralizing antibodies (MM43, MM57, R004, and R001) have the correct spatial conformation, and these spatial conformations are sufficient for the possibility of inducing neutralizing antibodies. In summary, we constructed RBD-BBVs that highly effectively exposed RBD polymers on the vesicle surface and displayed a ring-like vesicle structure.

Figure 2

The RBD was exposed on the BBV surface and polymerized into a ring-like structure. (A) Representative images of BBVs and RBD-BBVs as observed by TEM. The structural model of the recombined protein ClyA-RBD with ring-like structures is on the right. The linker is Gly-Ser-Gly-Ser (GSGS). (B) SDS-PAGE and silver staining results after PK treatment. RBD protein was analyzed via Western blot using antibodies against S1 of SARS-CoV-2. (C) Schematic diagram of the affinity analysis of RBD-BBVs and ACE2. S1-Fc is a protein expressed by HEK293 cells that has been shown to have the correct S1 conformation. The RBD content in RBD-BBVs was calculated based on RBD load%. The binding measurement is presented as the OD450 (n = 3). (D) The six monoclonal antibodies that can neutralize SARS-CoV-2 were serially diluted and reacted with RBD-BBVs (n = 3). The half maximal inhibitory concentration (IC50) of each neutralizing antibody is marked in brackets.

The BBV Delivery System Improved Vaccine Stability and Accumulated in Lymph Nodes

The in vivo local stability of vaccines may enhance the intensity and duration of antigen-induced immune responses.[17,18] We observed that the in vivo stability of the RBD delivered by BBVs was improved compared to that of the RBD protein (Figure A). In addition, RBD-BBVs accumulated in lymph nodes more than RBD protein (Figure B). We continued to analyze the cell types that take up RBD-BBVs in the lymph nodes. The experimental results proved that RBD-BBVs were mainly taken up by CD11c+CD11b- (DC cell marker)[12] (Figure C). In addition, we used a microarray analysis to analyze the immune response caused by RBD-BBV accumulation in the lymph nodes. The analysis results showed that gene expression in the lymph nodes of mice after RBD-BBV injection was significantly upregulated or downregulated (Figure D). After cluster analysis (Figure E), we found that the enriched gene expression changes mainly occurred in items related to the immune response (Figure S3), such as “defense response to virus”, “immune system process”, and “innate immune response” (Figure F). The results showed that BBVs have a strong immunostimulatory effect, suggesting that BBVs have strong adjuvant activity. In summary, these results indicate that the BBV vaccine delivery system can increase the stability of the vaccine, target the lymph nodes, and exert strong immune adjuvant activity.

Figure 3

The BBV delivery system improved vaccine stability and accumulated in lymph nodes. (A) The vaccine was injected subcutaneously from the base of the tail. Live imaging of mice to observe the time course of 5 μg of RBD-BBV-Cy7 and RBD-Cy7 in vivo. The relative fluorescence intensity was analyzed with Molecular Imaging software (n = 3). (B) Twelve hours after subcutaneous injection of different samples, inguinal lymph nodes of mice were collected and analyzed for fluorescence intensity. (C) Analysis of the fluorescence intensity of CD11c+CD11b-cell (DC marker) and CD11b+CD11c-cell (macrophage marker) uptake of RBD-BBV-Cy7 in lymph nodes (n = 5). (D) Microarray analysis was used to analyze gene expression in lymph nodes 12 h after subcutaneous injection of different samples. In this volcano map, each point in the map represents an mRNA. The abscissa represents the logarithmic value of the fold change (log2FC) of the difference in expression of a certain mRNA in the two sets of samples. (E) Cluster analysis (n = 2). (F) GO bubble chart. Different shapes correspond to different GO classifications, and the size of the dot corresponds to the number of different genes in the GO entry. The bubble color changes from purple-blue-green-red, and a smaller enrichment P value corresponds to greater significance.

The BBV Delivery System Promoted Antigen Uptake and DC Maturation

The high-efficiency uptake of antigens by DCs is the key to a successful vaccine. We observed highly efficient uptake of both RBD-BBVs and BBVs by DCs (Figure A). When DCs are stimulated by pathogen-associated molecular patterns (PAMPs) on BBVs, signaling pathways such as Toll-like receptors (TLRs) are activated to promote DC maturation.[17,19] In addition, proinflammatory cytokines (i.e., interleukin (IL) 6, IL1β, and tumor necrosis factor alpha (TNFα)) released by stimulated DCs will also increase, and these proinflammatory cytokines will promote DC maturation through a feedback mechanism.[20,21] Therefore, after RBD-BBV and BBV stimulation, we observed that the characteristics of DC maturation included not only upregulation of CD80 (Figure B) and CD86 (Figure C) expression but also increased secretion of IL6 and IL1β (Figure D). In summary, we observed that RBD-BBVs and BBVs were similar in terms of DC uptake and maturation stimulation, indicating that these effects stemmed from the nanostructure of BBVs[22] and the stimulation[23] and targeting of PAMPs on BBVs to DCs. Therefore, BBVs play a key role as RBD vaccine vectors.

Figure 4

The BBV delivery system promoted antigen uptake and DC maturation. (A) Immunofluorescent staining of bone marrow-derived dendritic cells (BMDCs) was analyzed with a confocal high-content imaging system. The rate of BMDC uptake was measured 4 h after the addition of 0.05 mg/mL BBVs or RBD-BBVs. The white boxes indicate the enlarged regions. Representative images are presented. (B) Flow cytometry analysis of the proportions of CD11c+CD80+ and (C) CD11c+CD86+ cells 24 h after stimulation with 0.05 mg/mL BBVs or RBD-BBVs (n = 3). (D) ELISA of cytokine levels in BMDC culture supernatant 24 h after stimulation with 0.05 mg/mL BBVs or RBD-BBVs (n = 3). (E) Lysosomal escape 12 and 24 h after treatment. Immunofluorescence detection of the RBD (green) and BBVs (green) is presented on the left and right, respectively; lysosomes are marked in red. The overlapping fluorescent signals are presented in yellow. The white arrowhead indicates fluorescent signals of successful lysosomal escape. Representative images are presented. (F) Immunofluorescence analysis of lysosomal escape 24 h after the addition of RBD-BBVs or BBVs to BMDCs to detect BBVs and (G) the RBD. White arrowheads indicate fluorescent signals of successful lysosomal escape. Representative images are shown.

Furthermore, we observed the highly efficient lysosomal escape of RBD-BBVs. Lysosomal escape of the RBD was concurrent with BBVs (Figure E), and both RBD-BBVs and the empty BBV vector were able to mediate lysosomal escape (Figure F). Hence, the RBD loaded on RBD-BBVs was able to achieve highly efficient lysosomal escape, along with BBVs (Figure G). Therefore, we suggest that the RBD loaded on RBD-BBVs achieved lysosomal escape utilizing the inherent features of BBVs. Lysosomal escape is key to induction of the T cell response.[24,25]

RBD-Modified BBVs Elicited Potential Protective Immunity against SARS-CoV-2

We immunized mice with RBD-BBVs to evaluate the induction of immune responses (Figure A). IgG responses specific to the SARS-CoV-2 spike protein were induced in mice after immunization with 5 μg or 0.5 μg of RBD-BBVs (Figure B), and both IgG1 (Figure S4A) and IgG2a (Figure S4B) responses were significantly induced. Similar IgM responses were also observed (Figure S4C). The titer of SARS-CoV-2 spike protein-binding antibodies induced by RBD-BBVs was analyzed with 2-fold serial dilutions of the antisera; the antibody titers were 1600 and 300 after immunization with 5 μg and 0.5 μg, respectively (Figure C). We determined the ability of RBD-BBV-induced antibodies to block the binding of S1 to hACE2 and found that antibodies induced by 5 μg of RBD-BBVs were able to block binding (Figure D). More importantly, we investigated the ability of RBD-BBV-induced antibodies to block the invasion of Vero cells by live SARS-CoV-2 (Figure S5A) and found that RBD-BBV-induced antisera inhibited the invasion of Vero cells by SARS-CoV-2, as reflected by the inhibition of pathological changes in Vero cells by antibodies (Figure S5B). Antibodies generated after three immunizations with RBD-BBVs at doses of both 5 μg and 0.5 μg were able to block cell invasion of SARS-CoV-2, and the 50% effective concentration (EC50) titers for 5 μg and 0.5 μg of RBD-BBVs were 117.3 and 74.6, respectively (Figure E). In summary, RBD-BBVs induced neutralizing antibodies against SARS-CoV-2. In addition, T cells are essential for clearing SARS-CoV-2 infection and providing long-term immune memory.[26] Therefore, we also evaluated the ability of RBD-BBVs to induce specific systemic cellular immune responses against SARS-CoV-2 in the spleen. The two main subsets of T cells (CD4+ T and CD8+ T) participate in the immune response against SARS-CoV-2 infection in different manners. Activated CD8+ T cells can directly kill infected cells, while CD4+ T cells can drive and help CD8+ T cell responses and participate in the process of antigen-specific B cells producing neutralizing antibodies.[27] Therefore, we first analyzed the ability of RBD-BBVs to induce CD4+ T and CD8+ T cell responses. The experimental results showed that RBD-BBVs induced CD4+ T (Figure F) and CD8+ T (Figure G) responses. In addition, type 1 helper T (Th1) cells are essential for inducing neutralizing antibodies and cellular immunity,[28,29] and we therefore further confirmed that RBD-BBVs can induce Th1 cell responses (Figure H). Similarly, both the experimental and clinical evidence proved that cytotoxic T lymphocytes (CTLs) play a key role in protecting against acute viral infections.[30−32] Mice immunized with RBD-BBVs produced virus-specific CTL responses (Figure I). The results of ELISpot experiments also showed that RBD-BBVs induced an antigen-specific T cell population (Figure J). In summary, these results indicate that RBD-BBVs can induce a wide range of cellular immune responses against SARS-CoV-2.

Figure 5

RBD-modified BBVs elicited potential protective immunity against SARS-CoV-2. (A) Schematic diagrams of the immunization and blood sampling procedures. (B) The levels of IgG against SARS-CoV-2 spike protein in sera after 50× dilution were determined with ELISA, and the results are presented as OD450 values (n = 3). (C) The titers of IgG antibodies against SARS-CoV-2 spike protein were determined using serially diluted sera (n = 5). (D) Schematic diagram of RBD-BBV-induced antibody blocking of the binding of S1-Fc protein and hACE2. Sera after three immunizations with 5 μg of BBVs or RBD-BBVs were diluted 50-fold and used to block the binding between S1-Fc and hACE2. The results are presented as OD450 values (n = 3). (E) SARS-CoV-2 EC50 titers calculated based on the results of antibody neutralization assays (n = 3). (F) Flow cytometry analysis of the ability of RBD-BBVs to induce SARS-CoV-2-specific CD4+ and (G) CD8+ T cells in the spleens of mice (n = 3). (H) Statistical data for the flow cytometry analysis of SARS-CoV-2 S1-specific Th1 (CD4+IFN-γ+) cells and (I) CTLs (CD8+IFN-γ+) induced by RBD-BBVs in mouse spleens (n = 3). (J) Representative images from the ELISpot analysis of the number of spleen lymphocytes secreting IFN-γ after stimulation with S1-Fc protein following vaccine immunization. Statistical analysis of the number of spots from the ELISpot analysis of the spleen (n = 3). (K) Statistical analysis of flow cytometry results for CD4+ TCM cells (CD4+CD44hiCD62Lhi) and (L) CD8+ TCM cells (CD8+CD44hiCD62Lhi) in the spleen and (M) CD4+ TEM cells (CD4+CD44hiCD62Llo) and (N) CD8+ TEM cells (CD8+CD44hiCD62Llo) in the spleen (n = 3).

The ability of vaccines to induce T cell immune memory reflects the potential sustainability of protection, which may be very important for the development of a SARS-CoV-2 vaccine.[33] We used flow cytometry to evaluate systemic levels of SARS-CoV-2-specific memory T cells. We observed the induction of CD4 (Figure K) and CD8 (Figure L) central memory T (TCM) cells after RBD-BBV immunization. These results indicate that RBD-BBVs induced the proliferation and differentiation of immune memory cells resistant to SARS-CoV-2 infection. We also observed the induction of CD4 (Figure M) and CD8 (Figure N) effector T cells (TEM) by RBD-BBVs, indicating that the protective immune cells induced by RBD-BBVs can migrate and cause an antiviral effect immediately during pathogen encounter.[34] Finally, for vaccines against respiratory diseases, the level of the cellular immune response in lung tissue should also be regarded as an evaluation indicator. The presence of a virus-specific cellular immune response in the spleen does not necessarily guarantee that the lung tissue (the area directly involved in the pathogenesis of SARS-CoV-2) has an adequate level of immunity.[35] Therefore, we also evaluated the level of the cellular immune response caused by RBD-BBVs in the lungs. The experimental results proved that RBD-BBVs can also cause a virus-specific cellular immune response in the lung consistent with the systemic response (Figure S6). In summary, we confirmed that RBD-modified BBVs elicited potential protective immunity against SARS-CoV-2.

In this study, we developed and evaluated a novel E. coli-based vaccine delivery system using BBV technology. However, this study did not challenge vaccinated mice against SARS-CoV-2 for two reasons: (1) This article focuses on the first development of BBVs as a new vaccine delivery vector, and SARS-CoV-2 is just one example. (2) The neutralizing antibody of SARS-CoV-2 is currently believed to reflect the level of protection conferred by the vaccine. Of course, we will continue to evaluate the vaccine in animal models in future work.

Due to the lack of protein glycosylation and the formation of inclusion bodies in prokaryotic expression systems, most vaccine development efforts targeting SARS-CoV-2 have focused on mammalian expression systems instead of the highly efficient system.[5,36] However, studies have reported successful development of antiviral vaccines with the use of for expression, such as the expression of virus-like particle (VLP) structures of human papillomavirus (HPV) and hepatitis E virus (HEV)[37,38] and the use of OMVs to deliver viral antigens of dengue virus, influenza virus, and Middle East respiratory syndrome (MERS) coronavirus.[11,39,40] Therefore, when using an expression system, vaccines should form stable polymers and regular nanostructures, and RBD-BBVs designed by us meet those requirements. However, we also found that the spatial structure of the RBD presented on RBD-BBVs is not completely consistent with that of S1 from the HEK293 expression system. This result can be explained by the difference in the expression system or whether it has glycosylation modification and also reflects the limitations of the expression system. However, the structures that can be recognized by neutralizing antibodies (MM43, MM57, R004, R001) have the correct spatial conformation (Figure D) and are sufficient for inducing neutralizing antibodies. Notably, according to the analysis of the IgM results, the IgG antibody level can continue to increase, which also suggests that we can further optimize the vaccine immunization program to increase the IgG antibody level in the future.

Finally, this genetic engineering technique for bacterial membranes will be applied in various fields for the development of bacterial vaccines, tumor vaccines, antibacterial agent delivery, RNA delivery, photothermal therapies, photoacoustic therapies, and tumor immune regulation.[8,12,13,41−46] In summary, based on technical innovations, we constructed ring-like BBVs that presented the RBD with high efficiency. This RBD-BBV vaccine may provide a new concept for the development of SARS-CoV-2 vaccines.

Concluding Remarks

We used high-pressure homogenization technology to drive bacterial membranes to produce artificial vesicles. ClyA-RBD was forced by high pressure to pass through narrow gaps together with the membrane, thereby increasing the contact between ClyA and the membrane. Therefore, the load of the RBD on the membrane was increased. Finally, a ring-like BBVs loaded with polymerized RBD (RBD-BBV) was constructed. Immunization with RBD-BBVs elicited SARS-CoV-2-specific neutralizing antibodies and cellular immune responses in mice.