Oil-in-water emulsion adjuvant A-910823, A-910823 with various proportions of squalene and α-tocopherol, and A-910823 at different average diameters were prepared at Shionogi & Co., Ltd. (Osaka, Japan) with a high-pressure homogenization technique as previously described (19, 20), with some modifications, which are summarized in the Supplementary Methods . The LNP-formulated mRNA vaccine (LNP-mRNA) was prepared using ionizable Lipid8 and mRNA encoding the full-length SARS-CoV-2 spike protein from the Pango lineage A strain, as previously described (17). Lipid8, distearoylphosphatidylcholine (Nippon Fine Chemical Co., Ltd, Tokyo, Japan), cholesterol (Nippon Fine Chemical Co., Ltd), and 1,2-dimyristoyl-sn-glycerol methoxypolyethylene glycol 2000 (NOF Corporation, Osaka, Japan) were dissolved in ethanol. The lipid solution and mRNA in acetate buffer (pH 4.0) were mixed at a flow ratio of 1:3 (v/v) and a total flow rate of 12 mL/min. The mixture was dialyzed with PBS (pH 7.4) at 4°C, and then passed through a 0.45-μm filter. The LNP preparation was analyzed for particle size and mRNA encapsulation. Empty-LNP (eLNP) was prepared in the same way, using acetate buffer (pH 4.0) without mRNA.

To observe physicochemical characteristics of A-910823, 50 µL of the sample diluted 10-fold in phosphate buffered saline (PBS) was pipetted onto a copper grid (TED PELLA, Redding, CA, USA) with formvar film and incubated for 5 min. After removing excess liquid, negative staining was performed with 1% uranyl acetate for 5 min. The grid was observed with JEM-1400 Flash (JEOL, Tokyo, Japan).

Female Balb/cAJcl mice (6–8 weeks old) were purchased from CLEA Japan, Inc. (Tokyo, Japan). All mouse experiments were performed in accordance with the appropriate laws and guidelines approved by the Institute of Medical Science at the University of Tokyo, Tokyo, Japan (approval no. PA21-46) and the Institutional Animal Care and Use Committee of Shionogi & Co., Ltd (approval no. S19017D and S20073C).

Modified adjuvants, including those with one or more constituents omitted and those with differing particle sizes (described in the Supplementary Methods ), were administered IM as two doses (1 µg of S-910823 mixed with or without 50% v/v of each modified adjuvant per dose) at 14-day intervals. In detail, after anesthetization, on day 0 and day 14, Balb/c mice received an IM dose of the following to the femoral muscle: recombinant S-910823 (1 µg) in 50 µL PBS solution (antigen-alone group); recombinant S-910823 (1 µg) in 25 µL PBS solution mixed with A-910823 (25 µL), A-910823 without α-tocopherol (25 µL), A-910823 without squalene (25 µL), A-910823 without α-tocopherol and squalene (25 µL), A-910823 of different particle sizes (25 µL), QS21 (5 µg; Creative Biolabs, Shirley, NY, USA), AddaVax (25 µL; InvivoGen, San Diego, CA, USA), alum (250 µg; InvivoGen) or eLNP (equivalent to lipid content of 5 µg mRNA-LNP); or mRNA-LNP (5 µg) in 50 µL PBS solution.

In a heterologous booster study, Balb/c mice received an IM dose (to the femoral muscle) of 3-µg mRNA-LNP as the first and second immunizations and 3-µg mRNA-LNP or S-910823 (1 µg) mixed with A-910823 (25 µL) as a booster dose, on days 0, 20, and 42, respectively.

Mice were euthanized and whole blood, draining lymph nodes, spleens, femurs, tibias, and femoral muscles at the injection site were obtained. Sera were prepared from the blood samples and stored at −80°C until the assays were performed. The spleens were dissociated into cell suspensions using a gentleMACS Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany). The red blood cells were lysed with Red Blood Cell Lysis Buffer (Roche, Basel, Switzerland), and the splenocytes and bone marrow cells were suspended in Roswell Park Memorial Institute-1640 medium (Nacalai Tesque Inc., Kyoto, Japan) supplemented with 10% fetal bovine serum (Sigma-Aldrich, Burlington, MA, USA). The draining lymph nodes were collected, and single-cell suspensions were prepared using BiomusherII (Nippi Research Institute of Biomatrix, Ibaraki, Japan). Femoral muscle samples were fixed with 10% neutral buffered formalin (Mildform 10N; FUJIFILM Wako Pure Chemical, Osaka, Japan) for 24 hours at room temperature (RT), paraffin-embedded, sectioned, and stained with hematoxylin and eosin (H&E).

Histopathological changes, such as inflammation and necrosis of muscle fiber, were independently evaluated by two pathologists using the H&E-stained specimens.

The Bio-Plex Pro Mouse Cytokine 23-plex Assay kit (Bio-Rad, CA, USA) was used to measure cytokines in mouse serum following the manufacturer’s instructions.

To measure the immunogenicity after IM dosing, the antigen-specific total immunoglobulin G (IgG), and IgG1 and IgG2a titers were measured by ELISA. In brief, 96-well, half-area plates (Corning Inc., Corning, NY, USA) were coated with 1 μg/mL of SARS-CoV-2 S-protein receptor-binding domain (RBD) from Pango lineage A, His Tag (multi-angle light scattering verified; ACROBiosystems, Newark, DE, USA) overnight at 4°C. After washing with PBS containing 0.05% Tween 20 (PBS-T) (Nacalai Tesque Inc.), the plates were blocked with 1% bovine serum albumin (Sigma-Aldrich) in PBS-T for 1 hour at RT. Sera were 5-fold serially diluted in the range of 1/100 to 1/7,812,500 with PBS-T containing 0.1% bovine serum albumin. The samples were incubated for approximately 2 hours at RT, followed by incubation with horseradish peroxidase-conjugated anti-mouse IgG, IgG1, or IgG2a antibody (Southern Biotech, Birmingham, AL, USA) for approximately 1 hour at RT. TMB Microwell Peroxidase Substrate (KPL, Gaithersburg, MD, USA) was added to initiate the color reaction for approximately 5 min at RT. Finally, sulfuric acid (Nacalai Tesque Inc.) was added to stop color development, and the optical density was measured at 450 nm using a PowerWave HT microplate reader (BioTek, Winooski, VT, USA).

Neutralizing antibody levels in the sera collected at day 28 and day 60 were analyzed against ancestral SARS-CoV-2 WK-521 using VeroE6/TMPRSS2 cells as described previously (16) and summarized in the Supplementary Methods .

Draining lymph nodes were collected at day 28, day 60, and at 4, 16, 24, and 72 hours post-first immunization to analyze Tfh cells and germinal center B (GCB) cells, and inflammatory cells, respectively. Draining lymph node cells were then washed and incubated with anti-CD16/32 (clone 2.4G2; BD Pharmingen, San Diego, CA, USA), before being stained for viability with the Zombie NIR™ Fixable Viability Kit (BioLegend). The cells were then immunostained with the antibodies outlined in Supplementary Table 1 , before being examined using the BD LSR II flow cytometer (BD Bioscience, San Jose, CA, USA) or the NovoCyte 3000 flow cytometer (Agilent Technologies, Inc., Santa Clara, CA, USA). Data were analyzed by FlowJo software (Tree Star, Ashland, OR, USA) or NovoExpress 1.4.0 software (Agilent Technologies, Inc.).

Antigen-specific cytokine production was measured using enzyme-linked immunospot (ELISpot) assays, which were performed using ELISpotPLUS kits for mouse interferon (IFN)-γ, and IL-4 (Mabtech, Cincinnati, OH, USA), following the manufacturer’s instructions with slight modifications, which is summarized in the Supplementary Methods . The number of antigen-specific antibody-secreting cells (ASCs) was measured using a B-cell ELISpot assay with ELISpot polyvinyl difluoride plates and a mouse IgG ELISpot BASIC Kit (Mabtech), following the manufacturer’s instructions with some modifications (18), as summarized in the Supplementary Methods .

A one-way analysis of variance followed by Tukey’s multiple comparison test was used for groups of three or more. Statistical analyses were performed using GraphPad Prism 9 (GraphPad Software Inc, San Diego, CA, USA).

Although clinical trial data demonstrated the efficacy of S-268019-b in humans, detailed knowledge of the characteristics of the induced immune response is still limited. A comparison between the adaptive immune response profile enhanced by A-910823 and that of other adjuvants was conducted to more fully characterize A-910823. AddaVax, QS21, and alum, which have been widely used in adjuvant research, were used for the comparison, and each dose was set as described in previous reports (21–25). In addition to those adjuvants, eLNP was also evaluated. LNP are novel adjuvant components that can be incorporated into mRNA vaccines. LNP-encapsulated nucleoside-modified mRNA vaccines have shown great efficacy against SARS-CoV-2 (26), and animal experiments have demonstrated that LNP has intrinsic adjuvant activity (27). The dose of eLNP was set as lipid content equivalent to 5 µg mRNA-LNP, the dose at which mRNA-LNP induces an acquired immune response ( Supplementary Figure 1 ).

Mice received two IM doses of S-910823 plus A-910823, AddaVax, QS21, alum, or eLNP on days 0 and 14. Anti-RBD total IgG, IgG1, and IgG2a titers were measured 13 days after the second dose on day 28 ( Figures 1A–C ). Compared with the antigen-alone group, the geometric mean titers (GMTs) of total IgG and IgG1 were significantly higher in the A-910823 adjuvant group (shown in red; P=0.0050 and P=0.0017, respectively). Although there was no significant difference in IgG2a GMT between the antigen-alone group and the A-910823 adjuvant groups, the GMT values increased with A-910823. With the exception of alum, there were no significant differences in the GMTs of total IgG, IgG1, and IgG2a induced by A-910823 and other adjuvants.

The GMT of neutralizing antibodies against SARS-CoV-2 WK-521 (Wuhan strain) were also validated ( Figure 1D ). The GMT of the S-910823 only group was 10.0, and the GMT of the S-910823 plus A-910823 group was 160.0. Although no statistically significant differences were identified, the GMT values increased with A-910823. Furthermore, compared with other adjuvants, the S-910823 plus A-910823 group had the second highest GMT value after AddaVax (S-910823 plus AddaVax: 211.1; S-910823 plus QS21: 139.3; S-910823 plus alum: 20.0; and S-910823 plus eLNP: 80.0).

Tfh cells, a helper T cell subset present in germinal centers in LNs, facilitate the generation of GCB, memory B cells, and long-lived plasma cells, which are required for long-term antibody responses (28). The generation of Tfh and GCB in draining lymph nodes was investigated to further understand the effect of A-910823 on those cells. We assumed the Tfh and GCB cell responses were antigen-specific based on the timing of the sampling with respect to vaccination. Tfh and GCB cells were induced at significantly higher levels in the A-910823 adjuvant group compared with the antigen-alone group (P<0.0001 and P=0.0056, respectively; Figures 1E, F ; Supplementary Figure 2 ). Tfh cell induction was highest in the A-910823-adjuvant group ( Figure 1E ). The mean %GCB in B cells was higher than that in other adjuvants ( Figure 1F ). Furthermore, to quantify long-lived plasma cells (29, 30), we measured the number of SARS-CoV-2 spike-protein-specific ASCs in bone marrow cells on day 109 using an ELISpot assay. ASCs were induced at significantly higher levels in the A-910823 adjuvant group compared with the antigen-alone group (P<0.0001; Figure 1G ). A-910823 induced significantly higher ASCs than did the other adjuvants, with the exception of eLNP. These results indicate that the adjuvant A-910823 has a high ability to induce humoral immunity. To investigate the cellular immune responses stimulated by the adjuvants, we measured the production of Th1 (IFN-γ) and Th2 (IL-4) cytokines from splenocytes collected on day 28 using an ELISpot assay ( Figures 1H, I ). Although there was no significant difference between the antigen-alone group and A-910823 adjuvant group in the induction of IFN-γ- and IL-4-producing cells, A-910823 slightly enhanced the induction of IFN-γ-producing cells. The highest induction of IFN-γ-producing cells was observed in the eLNP group.

Adjuvants elicit systemic and local immune responses after injection, and those responses are different depending on the characteristics of the adjuvants. We next investigated systemic inflammation and local reaction induced by A-910823. Compared with A-910823, induction of tumor necrosis factor (TNF)-α was particularly high in the eLNP group ( Supplementary Figure 3A ). Compared with A-910823, induction of IL-6 was particularly high in the mRNA-LNP and QS21 groups at 6 hours after injection ( Supplementary Figure 3A ). At 24 hours after injection, S-910823 plus A-910823 induced mild inflammatory cell infiltration in the femoral muscles compared with antigen alone but did not induce necrosis of muscle fiber. Conversely, S-910823 plus QS21 induced not only mild inflammatory cell infiltration but also necrosis of muscle fiber ( Supplementary Figure 3B ), as previously reported in rats (31, 32). Furthermore, at 24 hours after injection, recruitment of neutrophils, eosinophils, and monocytes into the draining lymph nodes was comparable or significantly higher in the A-910823 group compared with other adjuvants ( Supplementary Figure 3C ).

Overall, these results indicate that adjuvant A-910823 enhanced humoral immune responses in mice, following potent Tfh and GCB cell induction, without inducing a strong systemic inflammatory cytokine response.

When given as a booster dose at >6 months after a two-dose BNT162b2 (tozinameran; Pfizer, New York, NY, USA) regimen, both BNT162b2 and S-268019-b elicited anti-spike-protein IgG antibodies and produced T-cell response and neutralizing antibodies in Japanese adults without prior SARS-CoV-2 infection (14). Therefore, we evaluated detailed immunological characteristics, including Tfh and GCB cell induction, using a booster protocol.

Anti-RBD total IgG titers and neutralizing antibodies against SARS-CoV-2 WK-521 were measured 18 days after the booster dose (day 60; Figures 2A, B ). Total IgG and neutralizing antibody titer of the S-268019-b boost group were higher than in the vehicle group (P<0.0001 and P=0.0197, respectively), and were comparable with those of the mRNA-LNP boost group (P=0.4574 and P=0.9539, respectively). We also investigated whether a booster dose of S-268019-b induced Tfh cells and whether it induced cellular immune responses characteristic of mRNA-LNP. The frequency of Tfh ( Figure 2C ) and GCB ( Figure 2D ) cells in draining lymph nodes in the S-268019-b boost group were higher than that in the vehicle group (P=0.0014 and P<0.0001, respectively) and in the mRNA-LNP boost group (P=0.0012 and P=0.0003, respectively).

In the ELISpot assay, compared with the vehicle group, the S-268019-b-boost group had significantly more IFN-γ- and IL-4-producing cells following splenocyte stimulation (P=0.0260 and P=0.0046, respectively; Figures 2E, F ). Compared with the mRNA-LNP boost group, the S-268019-b-boost group had a lower number of IFN-γ-producing cells (P=0.0192), but a comparable number of IL-4-producing cells (P=0.6515).

These results indicate that S-268019-b containing A-910823 adjuvant enhanced humoral immune responses with potent Tfh and GCB cell induction, even when given as a booster dose following mRNA vaccine administration.

To identify which components of A-910823 are important for the adjuvant effect (i.e., inducing humoral immune responses and Tfh cells), we prepared modified adjuvants excluding squalene and/or α-tocopherol, and analyzed the physicochemical properties of the modified adjuvants using transmission electron microscopy (TEM) and dynamic light scattering (DLS) techniques. The results revealed that A-910823 formed 147.0-nm particles, which are of comparable size to AddaVax (160.7 nm) ( Figures 3A–C ). α-tocopherol and squalene in A-910823 were found to affect the particle size of A-910823; modified adjuvant lacking α-tocopherol or squalene formed particles with a smaller (110.2 nm) or larger (183.8 nm) diameter than A-910823, respectively ( Figures 3A–C ).

Following administration of the modified adjuvants to mice, we observed that compared with the group that was administered modified A-910823 without α-tocopherol, antigen-specific total IgG and IgG1 ( Figures 4A, B ) GMTs measured on day 28 were significantly higher in the A-910823 adjuvant group (P=0.0008 and P=0.0027, respectively). Conversely, A-910823 without squalene induced total IgG and IgG1 GMTs comparable to those induced by A-910823. There was no significant difference in IgG2a GMT between A-910823 and modified A-910823 without α-tocopherol or without squalene ( Figure 4C ).

Neutralizing antibody titers were also validated. The A-910823 adjuvant group and the group that was administered modified A-910823 without squalene each had statistically superior neutralizing antibody titers compared with the S-910823 only group (P=0.0048 and P=0.0497, respectively). However, there was no significant difference between the S-910823 only group and the group that was administered modified A-910823 without α-tocopherol ( Figure 4D ).

Consistent with the results of humoral immune responses, the induction of Tfh and GCB cells by A-910823 on day 28 was highly dependent on α-tocopherol (Tfh, P<0.0001 and GCB, P=0.0006, both A-910823 versus A-910823 without α-tocopherol) ( Figures 4E, F ). Tfh cell induction was significantly reduced by excluding α-tocopherol (P<0.0001) and, to a lesser extent, by excluding squalene (P=0.0046). The induction of ASCs on day 77 was also significantly reduced by excluding α-tocopherol (P=0.0387) ( Figure 4G ). Antigen mixed with PS80 did not enhance the induction of IgG, neutralizing antibody titers, Tfh cells, GCB cells, or ASCs.

Given the variation in particle size of each modified A-910823 demonstrated by the TEM and DLS results, we also evaluated the effect of particle size on humoral immune responses and Tfh and GCB cell induction. We prepared modified A-910823 with the same composition as A-910823 but with different particle sizes ( Supplementary Methods 1.3 ; Supplementary Figures 4A–C ). Measurement of anti-RBD total IgG, and Tfh and GCB cells on day 28 after administration indicated that although particle size influenced Tfh cell induction, it did not affect IgG and GCB cell induction ( Supplementary Figures 4D–F ).

These findings suggest that both the presence of α-tocopherol and the formation of appropriate particle sizes allowed A-910823 to function as a potent Tfh-inducing adjuvant.

Next, we investigated the effect of α-tocopherol on the innate immune response, in terms of recruitment of inflammatory cells to the draining lymph nodes and induction of serum cytokines and chemokines.

The absolute number of neutrophils, eosinophils, and monocytes in the draining lymph nodes at 16 hours after the dose of A-910823 mixed with S-910823 was significantly higher compared with S-910823 alone (P=0.0018, P=0.0386, and P=0.0367, respectively). In contrast, these cells were not significantly increased with A-910823 without α-tocopherol. No significant differences were observed in dendritic cells ( Figure 5A ).

Cytokines and chemokines were measured by multiplex analysis, and an A-910823-dependent increase was observed for each molecule ( Figure 5B ). The induction of granulocyte-colony stimulating factor, tumor necrosis factor-α, IL-1α, IL-6, IL-12 (p70), and keratinocyte-derived chemokines (KC) were dependent on α-tocopherol. IFN-γ, IL-5, macrophage inflammatory protein 1b, and monocyte chemoattractant protein-1 were induced by both α-tocopherol and squalene.

These results indicate that recruitment of inflammatory cells to the draining lymph nodes and induction of serum cytokines and chemokines by A-910823 is dependent on α-tocopherol.