A Beta Strain-Based Spike Glycoprotein Vaccine Candidate Induces Broad Neutralization and Protection against SARS-CoV-2 Variants of Concern

ABSTRACT The coronavirus disease 2019 (COVID-19) pandemic is still ongoing. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants of concern (VOCs) are circulating worldwide, making it resistant to existing vaccines and antiviral drugs. Therefore, the evaluation of variant-based expanded spectrum vaccines to optimize the immune response and provide broad protectiveness is very important. In this study, we expressed spike trimer protein (S-TM) based on the Beta variant in a GMP-grade workshop using CHO cells. Mice were immunized twice with S-TM protein combined with aluminum hydroxide (Al) and CpG Oligonucleotides (CpG) adjuvant to evaluate its safety and efficacy. BALB/c immunized with S-TM + Al + CpG induced high neutralizing antibody titers against the Wuhan-Hu-1 strain (wild-type, WT), the Beta and Delta variants, and even the Omicron variant. In addition, compared with the S-TM + Al group, the S-TM + Al + CpG group effectively induced a stronger Th1-biased cell immune response in mice. Furthermore, after the second immunization, H11-K18 hACE2 mice were well protected from challenge with the SARS-CoV-2 Beta strain, with a 100% survival rate. The virus load and pathological lesions in the lungs were significantly reduced, and no virus was detected in mouse brain tissue. Our vaccine candidate is practical and effective for current SARS-CoV-2 VOCs, which will support its further clinical development for potential sequential immune and primary immunization. IMPORTANCE Continuous emergence of adaptive mutations of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) continues to challenge the use and development of existing vaccines and drugs. The value of variant-based vaccines that are capable of inducing a higher and broader protection immune response against SARS-CoV-2 variants is currently being evaluated. This article shows that a recombinant prefusion spike protein based on a Beta variant was highly immunogenic and could induced a stronger Th1-biased cell immune response in mice and was effectively protective against challenge with the SARS-CoV-2 Beta variant. Importantly, this Beta-based SARS-CoV-2 vaccine could also offer a robust humoral immune response with effectively broad neutralization ability against the wild type and different variants of concern (VOCs): the Beta, Delta, and Omicron BA.1 variants. To date, the vaccine described here has been produced in a pilot scale (200L), and the development, filling process, and toxicological safety evaluation have also been completed, which provides a timely response to the emerging SARS-CoV-2 variants and vaccine development.

IMPORTANCE Continuous emergence of adaptive mutations of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) continues to challenge the use and development of existing vaccines and drugs. The value of variant-based vaccines that are capable of inducing a higher and broader protection immune response against SARS-CoV-2 variants is currently being evaluated. This article shows that a recombinant prefusion spike protein based on a Beta variant was highly immunogenic and could induced a stronger Th1-biased cell immune response in mice and was effectively protective against challenge with the SARS-CoV-2 Beta variant. Importantly, this Beta-based SARS-CoV-2 vaccine could also offer a robust humoral immune response with effectively broad neutralization ability against the wild type and different variants of concern (VOCs): the Beta, Delta, and Omicron BA.1 variants. To date, the vaccine described here has been produced in a pilot scale (200L), and the development, filling process, and toxicological safety evaluation have also been completed, which provides a timely response to the emerging SARS-CoV-2 variants and vaccine development.
T4 foldon was inserted at the C terminus of S-TM for self-trimerization (Fig. 1A). S-TM has a molecular weight of approximately 180 kDa according to SDS-PAGE analysis after expression and purification (Fig. 1B). By transmission electron microscopy (TEM) and two-dimensional class averaging, the ectodomain of S-TM, formulated in PB buffer (20 mM Phosphate Buffer, 150 mM NaCl, and 0.02% polysorbate 80 (PS80)), pH 6, showed a prefusion trimeric conformation with good uniformity and dispersion with a 19.6-nm length and a 13.23-nm width ( Fig. 1C and D). The stability of antigen stands for a critical factor that contributes to the viability of a vaccine candidate. The electron microscope photos showed that the purified S-TM still presented a homogeneous trimeric structure with good dispersion when stored at 4°C (for 87 days) and 280°C (for 112 days), respectively ( Fig. S1A to D).
S-TM + Al + CpG induced high binding antibodies and cross-neutralizing antibodies against SARS-CoV-2 VOCs. Mice were immunized intramuscularly (IM) with various doses of S-TM (1, 2.5, 5, 10, and 25 mg/dose) on days 0 and 21 with S-TM, S-TM 1 Al, or S-TM 1 Al 1 CpG. Blood samples were collected 14 days after the second immunization for S-TM binding and neutralizing antibody tests ( Fig. 2A). S-TM binding antibody titers in the S-TM 1 Al 1 CpG groups were significantly higher than those in the S-TM and S-TM 1 Al groups at the corresponding antigen dose levels (except for the 10 or 25 mg S-TM 1 Al 1 CpG group) (Fig. 2B). This phenomenon also appeared in neutralization results, which in the S-TM 1 Al 1 CpG groups were the highest at the corresponding antigen dose levels ( Fig. 2C to E), especially for the Beta strain, and significant differences were observed in all antigen-dose groups (geometric mean titer [GMT]: S-TM 1 Al 1 CpG 1,987 to 3,329, S-TM 1 Al 259 to 644, S-TM 135 to 228, P , 0.05) (Fig. 2C). Overall, the mice vaccinated with S-TM 1 Al 1 CpG groups had significantly higher humoral immunogenicity than that of nonadjuvanted and Al-adjuvanted groups. These results indicate the effect of CpG on humoral immunogenicity. To further analyze the spectrum of serum-neutralizing antibodies against VOCs, we compared the cross-neutralization ability in the S-TM 1 Al 1 CpG group toward WT and VOCs together. As shown in Fig. S2A, the GMTs of serum-neutralizing antibodies against the Beta strain was higher than against the Delta and WT strains (P , 0.05). Neutralizing antibody levels were similar for WT and Delta (P . 0.05). Immunization with the Beta-based S-TM 1 Al 1 CpG vaccine showed the highest neutralizing antibody against the parent strain, while reduced antibody against Delta VOC (Fig. S2A to F). Encouragingly, the serum of mice that received 5-or 10-mg doses maintained quite a high level of neutralizing antibodies against Delta, with degrees of neutralization that declined less than 3-fold compared to that against Beta ( Fig. S2D and E). According to the above-described experimental results, we immunized mice with 2.5, 5, and 10 mg S-TM 1 Al 1 50 mg CpG (one-fifth of the dose of the finished preparation) were immunized with 1, 2.5, 5, 10, and 25 mg S-TM that was nonadjuvanted or adjuvanted with 50 mg Al or 50 mg aluminum hydroxide (Al) 1 10 mg CpG twice on days 0 and 21 with primary analysis for humoral-and cell-mediated immunogenicity conducted on day 35 for blood and splenocyte samples. (B) S-TM binding antibody titers with ELISA. (C to E) Neutralizing antibody titers against Beta (C), wild-type (WT) (D), and Delta (E) strains with live SARS-CoV-2 assay. The points represent individual mice. For binding antibody (Ab), the bars represent the geometric mean with 95% CI of the antibody titer (EC 50 ), and comparisons were made using one-way analysis of variance (ANOVA) with Tukey's multiple-comparison test. For the neutralizing antibody, the limitation of detection (LD) was initial dilution fold (1:25), data lower than the LD were considered initial dilution and were used as fold limited detection for calculation and plotting. Bars represent the geometric mean with 95% CI of the antibody titer, and the comparisons were made using ordinary one-way ANOVA multiple-comparison test. P values , 0.05 were considered significant. S-TM, spike trimer protein; CpG, CpG Oligonucleotides; MN50, the neutralization titer for 50% neutralization of viral infection; IFN, interferon; IL, interleukin; ns, not significant. and continued to explore the neutralizing antibody change between Omicron and other strains. Another group of BALB/c mice (15 to 19 g, female) was randomly grouped (n = 6/group) and immunized intramuscularly on days 0 and 21 with S-TM 1 Al 1 CpG (Fig. 3A). Neutralizing antibody titers against the Beta strain (GMT 1,199 to 2,136) were significantly higher than those against the Delta strain (GMT 200 to 599) and the Omicron strain (GMT 100 to 266) at the corresponding antigen doses (Fig. 3B). Furthermore, neutralizing antibody titers against the Delta strain and the Omicron strain dropped 5.04-fold (3.57 to 7.55) and 11.17-fold (8.03 to 13.48), respectively, Bars represent the geometric mean with 95% CI of the antibody titer, and comparisons were made using two-way ANOVA with Tukey's multiple-comparison test. P values , 0.05 were considered significant. S-TM, spike trimer protein; Al, aluminum hydroxide; CpG, CpG Oligonucleotides; ns, not significant. GMT, geometric mean titer. compared to that of the Beta strain. Neutralizing antibody titers against the Omicron strain dropped 2.35-fold (1.79 to 2.99) compared to that of the Delta strain ( Fig. 3C to E). Overall, in all groups, the 5 mg S-TM 1 Al 1 CpG group showed the highest neutralizing antibody titers against the four strains, and they were significantly higher than those of the 10 mg S-TM 1 Al 1 CpG group (P , 0.05). The 5 mg S-TM 1 Al 1 CpG group showed the least neutralizing antibody titer drop against the Omicron strain compared to the Beta strain (8.03) (Fig. 3C).
S-TM + Al + CpG induced a Th1-biased cell-mediated immune response. After two IM immunizations with various doses of S-TM (1, 2.5, 5, 10, and 25 mg/dose) on days 0 and 21 with S-TM, S-TM 1 Al, or S-TM 1 Al 1 CpG, splenocytes were harvested at 14 days, followed by stimulation with spike peptide pool detection of Th1 (interleukin 2 [IL-2] and interferon g [IFN-g ]) and Th2 (IL-4) for the study of antigen-specific cell-mediated immunity ( Fig. 2A). The S-TM 1 Al 1 CpG group induced more IL-2-and IFNg -secreting cells than the Al-adjuvanted and CpG-adjuvanted groups, which indicates that S-TM 1 Al 1 CpG induced spike (S)-specific IL-2-and IFN-g -secreting cells, leading to antigen-specific cell-mediated immunity (Fig. 4A). We calculated the ratio between them to evaluate the type of cell-mediated immune response. The ratio of Th1 to Th2 cytokine-secreting cells from the same mouse (IFN-g /IL-4 and IL-2/IL-4) was higher in the S-TM 1 Al 1 CpG group than in the S-TM 1 Al group (1.5 versus 0.3 and 7.3 versus 1.1; P , 0.002 and P , 0.00001) (Fig. 4A). These results show that CpG can balance the Th2biased cell-mediated immune response and induce a Th1-biased cell-mediated immune response. In addition, only IFN-g showed an antigen dose-dependent change in the S-TM 1 Al 1 CpG group, and dose saturation was observed at 25 mg S-TM in IL-2-and IL-4-secreting cells (Fig. S3A).
For CD4 1 and CD8 1 T-cell-mediated immunity, the S-TM 1 Al 1 CpG group induced a higher ratio of mature CD4 1 and CD8 1 T cells than the S-TM group (Fig. 4B). The ratio of CD8 1 but not CD4 1 T cells showed an antigen dose-dependent change in the S-TM 1 Al 1 CpG group, and a saturated dose was observed at 25 mg S-TM (Fig. S3B).
S-TM + Al + CpG protects H11-K18-hACE-2 mice from beta strain infection. H11-K18-hACE-2 mice were vaccinated with 10 mg S-TM 1 Al 1 CpG or PB buffer to evaluate the induction of protective immunity against challenge with the SARS-CoV-2 Beta strain (Fig. 5A). In the PB buffer group, mice showed trembling, poor mobility and balance, weight loss, and death at 4 days postinfection (dpi); most of the mice experienced 20% weight loss at 6 dpi; and 100% of the mice were dead at 8 dpi (Fig. 5B). However, all mice in the S-TM 1 Al 1 CpG group maintained their body weight and survived until 11 dpi (Fig. 5B). In terms of virus load, PB buffer-treated mice had 10 3 copies/mL in the lungs and 10 6 to 10 7 copies/mL in the brain at 4 and 8 dpi. The mice immunized with S-TM 1 Al 1 CpG had almost undetectable virus copies in the brain and decreased 2 to 3 log in the lung at 4 and 8 dpi (Fig. 5C). Notably, 1 of 3 immunized mice at 4, 8, and 11 dpi almost had no virus detected in their lung tissue. Thus, S-TM 1 Al 1 CpG effectively stimulated H11-K18-hACE-2 mice at 13 days after the second vaccination and induced humoral immunogenicity, with high neutralizing antibody titers against the Beta strain (GMT 3,326), WT strain (GMT 933), and Delta strain (GMT 309) (Fig. 5D).
For pulmonary histology analysis, due to the death of mice in the PB buffer group after infection, the samples were collected for the analyses at 4 and 5 dpi. In the S-TM 1 Al 1 CpG group, the samples were collected at 4, 8, and 11 dpi. As seen from the pathological sections and scores, the lung pathology of mice infected by the Beta strain was mainly manifested by thickening of the alveolar wall and diffused alveolar damage interstitial pneumonia, accompanied by mild bronchitis and alveolitis (Fig. 5E). At 4 dpi, PB buffer-treated mice showed thickening of the alveolar septa and interstitial inflammation with hemangiectasis, hyperemia, interstitial edema, and infiltration of lymphocytes ( Fig. 6). At 5 dpi, the thickening of the alveolar septa and interstitial inflammation were aggravated, and there was more infiltration of lymphocytes and monocytes (Fig. 6). The S-TM 1 Al 1 CpG-immunized mice showed a significant reduction in pathological reactions and lung inflammation at 4 to 11 dpi (Fig. 6). These results indicate that S-TM 1 Al 1 CpG immunization can delay disease progression with only mild lung inflammation during Beta strain infection and may relieve pathological symptoms in the following days.

DISCUSSION
In this study, the S-TM protein expressed in CHO cells with a foldon (GYIPEAP RDGQAYVRKDGEWVLLSTFL) existed in the form of a stable and uniform trimer. The foldon, located at the C terminus of phage T4 fibrin, contains 27 amino acids (containing a large number of hydrophobic amino acids) and forms a trimer based on intermolecular hydrogen bonds and hydrophobic forces (11). This method has also been applied to research on the structure of the S protein in the early stages of COVID-19 (12). Consistent with the earlier observations, we also found that in low pH conditions, S-TM revealed more homogenous and stable trimer than at pH 7 to 8. There is a pH switch domain (residues 617 to 639) that undergoes dramatic conformational changes at different pH values and that may account for the enhanced stability of the spike trimer at lower pH (13). The biological mechanism of S-TM structural arrangement in different pH values still needs to be further studied. To date, our vaccine has been produced in a pilot scale (200L), and the development and filling processes have also been completed. The antigen produced on the pilot scale has good dispersibility, stability, and homogeneity.
We showed that two doses of S-TM 1 Al 1 CpG were effective in eliciting neutralizing activity against various SARS-CoV-2 strains, including WT, Beta, Delta, and Omicron. Although there was no specific antigen dose-dependent relationship in the immunization of BALB/c mice with S-TM 1Al 1 CpG, a 25-mg antigen dose did not show increased titer of both binding and neutralizing antibodies, indicating that 25 mg had reached saturation. In addition, 1 mg of S-TM 1 Al 1 CpG induced an even higher neutralizing antibody titer than S-TM alone or S-TM 1 Al at the 10-or 25-mg antigen doses, which indicates the potential for a more than 10-fold or greater dose-sparing effect H11-K18-hACE-2 mice (n = 9) were immunized with S-TM 1 Al 1 CpG (10 mg) with or without adjuvants (only PB buffer) twice on days 0 and 21, and serum samples were taken at day 34 for immunogenicity assays. Two weeks after the second immunization, the mice were challenged with the SARS-CoV-2 Beta strain. Body weight was tracked, and the animals were euthanized for tissue sampling at 4 to 11 days postinfection (dpi). (B) Survival rates and changes in body weight after virus challenge in H11-K18-hACE-2 mice. (C) RNA copies in the lung and brain were detected by reverse transcription (RT)-PCR. The detection threshold (0.5 copies/mL) is shown by the dotted line in each panel. RNA copies below the threshold were set to 1/5 threshold for calculation and plotting. (D) Neutralizing antibody titer against the Beta, WT, and Delta strains 13 days after the second immunization. (E) Lung pathology scoring in mice at 4, 5, 8m or 11 dpi. For RNA copies and lung pathology scores, the bars indicate mean 6 SEM, and statistical significance was calculated with two-way ANOVA with Tukey's multiple-comparison test. For statistical analysis of antibody titers, the results are presented as the geometric mean with 95% confidence interval and statistical significance calculated with one-way ANOVA with Tukey's multiplecomparison test. P values , 0.05 were considered significant. S-TM, spike trimer protein; Al, aluminum hydroxide; CpG, CpG Oligonucleotides; PB buffer: 20 mM Phosphate Buffer, 150 mM NaCl, and 0.02% polysorbate 80 (PS80), pH 6. provided by the Al 1 CpG adjuvant. This is consistent with other reports that CpG can enhance vaccine effectiveness through strong B-cell activation, resulting in enhanced antibody production (14).
Although our Beta variant-based vaccine candidate showed the highest neutralizing antibodies against the homologous strains, it was observed that mice produced neutralizing antibodies against Delta that were almost equal to or slightly higher than the WT strain and neutralizing antibodies against Omicron that were only 1.2 to 3.36 times lower than the WT strain. Studies have revealed that the current vaccines (adenovirus vectors, mRNA, and subunit vaccines) targeting Beta mutated SARs-CoV-2 RBD, or spike proteins elicited higher levels of binding and cross-neutralizing antibodies against variants than against the WT SARS-CoV-2 (15)(16)(17)(18). While vaccines based on the WT strain produced neutralizing antibodies against the Beta and Delta strains that were approximately 7 to 29 times lower than that of the WT strain, for Omicron cross-neutralizing antibodies, it was approximately 26 times lower or even provided no detectable titer against the Omicron variant (19). Therefore, we speculate that our S-TM based on the Beta strain has better immunogenicity and can stimulate mice to produce a broader cross-neutralizing antibody response to VOC strains (Beta, Delta, and Omicron strains) than the vaccines based on the WT strain. One study reported that the levels of neutralization antibodies induced by an Omicron-specific booster dose were lower, indicating that an Omicronspecific vaccine may not provide sufficient immunity or protection (20). The results of in vivo experiments show that those mutations decreased the antigenicity of Omicron in general (21,22). This raises the possibility and significance of exploring Beta-specific vaccines as an effective vaccine against Omicron. Our S-TM candidate induces strong broad neutralizing antibodies, potentially due to some shared mutation sites (K417, E484, N501, and D614G) in its sequence with Omicron. Some studies have shown that the cross-neutralizing antibodies of Beta and Omicron variants mainly depend on the 417N/T and 501Y mutations (23). In addition to the RBD regions, other conserved regions in spike proteins may provide broadly neutralizing antibodies (24).
Adjuvants can effectively enhance and regulate the immune response intensity and bias of vaccines. Aluminum adjuvants have been used in many vaccines to induce bias toward the Th2 immune response. Enhanced respiratory disease (ERD) induces a Th2-biased immune response induced by the FI-RSV vaccine combined with aluminum adjuvant in children. CpG is an oligodeoxynucleotide containing a nonmethylated cytosine guanine dinucleotide that binds to TLR9 receptors in cells to induce an immune response that is Th1 oriented. Synergistic use with aluminum adjuvant can induce the body to produce a balanced immune response, and a CpG library can be formed in vivo through the action of aluminum adjuvant, thus prolonging its action time in vivo. As expected from the known Th1 bias in the response to CpG-containing adjuvants, BALB/c mice that received S-TM 1 Al 1 CpG induced more IL-2-and IFNg -secreting cells than those who received S-TM 1 Al, and the same magnitude of immune responses was observed in another preclinical subunit protein vaccine based on the Beta strain (25,26). In addition, S-TM 1 Al 1 CpG induced a Th-1-biased cell-mediated immune response and inhibited an Al-directed Th-2-biased cell-mediated immune response, which reduced the possibility of a Th-2-biased cell-mediated immune response associated with vaccine-associated enhanced respiratory disease (VAERD), indicating the safety of our vaccines (27,28).
Furthermore, S-TM 1 Al 1 CpG induced more CD4 1 and CD8 1 T cells than S-TM in BALB/c mice, which is associated with better virus clearance and an improved clinical outcome (29)(30)(31)(32). Although Omicron has many more mutations than the other VOC strains, a study showed that 80% of conserved CD4 1 and CD8 1 T-cell epitopes were observed on vaccinees from different vaccine suppliers (33). In our study, we used the SARS-CoV-2 S peptide based on the WT strain to stimulate splenocytes, and we still observed a significant increase in S antigen-specific IL-2-, IFN-g -, and IL-4-secreting cells, which reconfirmed the conservativeness of T-cell epitopes against SARS-CoV-2 (34)(35)(36)(37). This gives us a more optimistic cross-reaction against other VOCs, especially for the Omicron strain.
In our study, we demonstrated that our vaccine candidate S-TM 1 Al 1 CpG protects against lethal infection of H11-k18 mice, a severe disease model (38,39). For the PB control group, inoculation of K18-hACE2 mice with a 50% tissue culture infective dose (TCID 50 )/mouse SARS-CoV-2 of 25 (based on doses leading to 100% mouse death in our preliminary study) resulted in 100% mortality, lung disease with signs of interstitial pneumonia, immune infiltration, and extensive spread of the virus to the central nervous system and the brain (Fig. 5 and 6; Fig. S4). Although the challenge dose was lower than that used in a K18 mouse study using a 100 TCID 50 /mouse challenge dose, the survival days, weight changes, and viral load after challenge were not significantly different from those used in our study ( Fig. 5; Fig. S4), suggesting minimal impact of the challenge dose on mouse model establishment (40). This finding indicates that a SARS-CoV-2 mouse model was successfully established in our study, which sets the foundation for subsequent virus challenge. Although SARS-COV-2 primarily targets the respiratory system, some patients with COVID-19 also develop extrarespiratory symptoms, including neurological manifestations. SARS-CoV-2 can infect neurons in human brain organoids (41). In our study, immunization with S-TM 1 Al 1 CpG not only significantly inhibited viral replication and reduced pathological lesions in the lung but also prevented severe symptoms and death in mice by almost completely blocking viral replication in the brain (Fig. 5; Fig. S4), which precluded potential antibody-dependent enhancement (ADE) effect, and we can further speculate that our vaccine candidate is effective not only in preventing SARS-CoV-2 infection but also in preventing virus entry into the central nervous system and is highly effective in preventing severe illness and death.
H11-k18 mice immunized with S-TM 1 Al 1 CpG produced a high titer of neutralizing antibodies against Beta SARS-CoV-2, which played an important role in resistance to virus infection. In our study, BALB/c mice immunized with S-TM 1 Al 1 CpG produced cross-neutralizing antibodies against the VOC Omicron BA.1 strain. Therefore, it can be inferred that our vaccine candidate is also protective against Omicron strain infection. All H11-K18 mice immunized twice with the finished dosage form of S-TM 1 Al 1 CpG survived challenge with the Beta, Delta, and Omicron BA.1 strains, while all the mice in the control group died. As Omicron BA.2 and BA.5 variants emerged, we further explored the protective effect against Omicron BA.2.12.1 and BA.5 variant nonlethal challenge with other hACE2-KI mice. Within expected, the mice immunized twice with the finished dosage form of S-TM 1 Al 1 CpG showed significant reduction in lung nucleic acid load and pulmonary pathology (unpublished data). In mice, the S-TM 1 Al 1 CpG elicits protection against SARS-CoV-2 variant infection with no evidence of ADE.
At the time of this article's composition, most of the population has received a COVID-19 vaccine as a primary series based on the WT strain (42,43). However, reduction or loss of detectible levels of neutralizing antibody against the variants at ;6 months might be indicative of waning protection and likely due to further affinity maturation of B cells and alteration of the available antibody repertoire (44). The value of variant-based vaccines that can induce a higher and broader immune response against SARS-CoV-2 at booster vaccination was being evaluated. A clinical trial showed that a heterologous boosting with Beta-based MVB.1.351 vaccine resulted in a higher neutralizing antibody titer against the Beta, Delta, and Omicron BA.1 variants than did the Wuhan-Hu-1 strain-based vaccine formulation (45); this demonstrated the high value of variant-adapted vaccines capable of inducing a higher and broader immune response against SARS-CoV-2 at booster vaccination. The long-term antibody level could be a critical factor that contributes to the application prospect of the present vaccine candidate. Our vaccine candidate is presumed to have great potential on providing broad and long-term protection as booster vaccination, which could be ideal for alleviating the burden of the global COVID-19 pandemic and is very promising for further development from the bench to the clinic.

MATERIALS AND METHODS
Animal studies, facilities, and ethics statements. Specific pathogen-free (SPF) BALB/c female mice (15 to 19 g) and H11-K18-hACE-2 female mice (5 weeks) were purchased from Beijing HFK Bioscience Co., Ltd., and Gempharmatech Co., Ltd., and were kept under standard pathogen-free conditions at Yikang Protein design, expression, and structure. To generate the recombinant spike construct based on the Beta variant (S-TM), the following point mutations were introduced: L18F, D80A, D215G, R246I, K417N, E484K, N501Y, D614G, A701V, and deletion of 242-244. To maintain the stability of the prefusion conformation, R682Q, R683Q, R685Q, K986P, and V987P were also introduced. For the trimer conformation, the foldon sequence, GYIPEAPRDGQAYVRKDGEWVLLSTFL, derived from the native T4 phage fibritin to stabilize a trimeric hairpin propeller and a flexible protein linker, GS, was substituted at the C-terminal domain. Full-length S genes were cloned between the EcoRI-HindIII sites in the ZY-5.0 vector, named ZYB0029. The expression vector ZYB0029 was transfected into CHO-K1 cells using a GenPulser (Bio-Rad, USA). With increasing concentrations of MSX, stable clones were screened by monoclonal sorting. After harvesting the clarified cell culture medium, we purified recombinant S-TM by consecutive chromatographic steps (cation exchange (POROS XS, Thermo), hydrophobic chromatography (Phenyl HP), compound anion chromatography (Mix-A), and cation exchange (SP Mustang)). The S-TM was formulated at PB buffer (20 mM Phosphate Buffer, 150 mM NaCl, and 0.02% polysorbate 80 (PS80), pH 6). The expression and purification of S-TM were performed in the good laboratory practice (GLP) laboratory at Beijing JOINN Biologics Co. Ltd.
Electron microscopy was performed by Shuimubio Services (Beijing, China) with a Thermo Scientific TalosL120C electron microscope operated at 120 keV. SARS-CoV-2 S-TM proteins were diluted to 0.3 mg/ mL in PB buffer. High-magnification images were acquired at nominal magnifications of 57,000Â (0.245 nm/pixel) and 73,000Â (0.192 nm/pixel), and the images were acquired at electron doses of 30e-/ A2. The images were collected by a TalosL120C electron microscope, and two-dimensional classification was conducted by cisTEM software.
BALB/c mouse immunization. To explore the effect of dose on immunity, 136 BALB/c mice (15 to 19 g, female) were randomly grouped (n = 8/group). The mice were immunized intramuscularly (IM) with various doses of S-TM (1, 2.5, 5, 10, and 25 mg/dose) that were nonadjuvanted or adjuvanted with 50 mg Al (manufactured by Croda) or 50 mg Al plus 10 mg CpG (Parr Biotechnology [Jiangsu] Co., Ltd.). The total injection volume of the mixed vaccines (antigen 1 adjuvant) was 100 mL per IM dose. Two IM doses were administered (on days 0 and 21). The animals were bled from the cheek on day 35 for humoral immune response analyses. Serum from day 35 was neutralized against the Beta, WT, and Delta strains. The spleens were removed after euthanization on day 35 for ELISpot assays.
Another group of BALB/c mice (15 to 19 g, female) were randomly grouped (n = 6/group). The mice were immunized intramuscularly with various doses of S-TM (2.5, 5, and 10 mg/dose) that was nonadjuvanted or adjuvanted with 50 mg Al plus 50 mg CpG ODN (Parr Biotechnology [Jiangsu] Co., Ltd.). The immunization scheme was the same as that mentioned above. The animals were bled from the cheek on day 35 for WT, Beta, Delta, and Omicron strain SARS-CoV-2 neutralization assays.
S-TM binding antibody ELISA. We coated 96-well plates (Thermo, Denmark, lot number 442404) with S-TM (5 or 100 mL/well) at 4°C overnight. The plates were washed three times with PBST (phosphate-buffered saline containing 0.05% Tween 20) and then blocked with 2% bovine serum albumin (BSA) (lot number: BSA-YCSB-500g) at room temperature for 1 h. After washing the plate three times with PBST, 2-fold dilutions from 1:100 of the antisera were added to the wells (100 mL/well) and incubated at room temperature for 2 h. After the plates were washed three times with PBST, 0.09 mg/mL rabbit anti-mouse F(ab)/horseradish peroxidase (HRP) (Jackson, lot number 315-035-047) was added to the well and incubated at room temperature for 1 h. The plates were then washed three times with PBST, and signals were developed using 3,39,5,59-tetramethylbenzidine (TMB) substrate (Thermo Scientific) at room temperature. The colorimetric reaction was stopped after 20 min by adding 2 M HCl. The optical density (OD) was measured at 450 nm by a microplate reader (Multiskan MK3, Thermo). The 50% effective concentration (EC 50 ) values were calculated by four-parameter fitting using GraphPad Prism software (GraphPad Software, Inc.).
Live SARS-CoV-2 neutralization assay. Heat-inactivated antisera (56°C for 30 min) from BALB/c mice and H11-K18-hACE-2 mice were serially diluted in cell culture medium in 2-fold dilutions starting from 1:25. The diluted sera (50 mL) were mixed with an equal volume of solution (50 mL) containing 100 TCID 50 live SARS-CoV-2 VOC strain (Beta, Delta, and Omicron) and WT strain in each well. After 2 h of incubation at 37°C in a 5% CO 2 incubator, Vero cells (1 Â 10 4 /well) were added to each well and cultured in a 5% CO 2 incubator at 37°C for 4 to 5 days. The cytopathic effect (CPE) on each well was recorded under a microscope, and the neutralization titer was calculated as the reciprocal of the maximum dilution required for 50% neutralization of viral infection by Karber's method.
Elispot assay and flow cytometry. Splenocytes from immunized mice were harvested 2 weeks after the second IM immunization for multifunctional cytokine and CD4 1 /CD8 1 T-cell analysis. To detect antigen-specific T-cell responses, ELISpot kits measuring Th1 cytokines (IFN-g and IL-2) and Th2 cytokines (IL-4) were used per the manufacturer's instructions (Mabtech: 3321-4AST-2, 3441-4APW-2, and 3311-4APW-2). In all, 2 Â 10 5 splenocytes were stimulated in vitro with 4 mg/mL SARS-CoV-2 spike peptide pool (Sinobiological, PP003). Phorbol 12-myristate 13-acetate (PMA) 1 1 mg/mL ionomycin (Dakewe Biotech Co., Ltd., 2030421) and 1 mg/mL ConA (Beijing Solarbio Science & Technology Co., Ltd., C8110) as nonspecific stimuli were added to the positive-control wells, whereas the negative-control wells received no stimuli. After 20 h of incubation at 37°C, the plates were washed, and 1 mg/mL antibodies (diluted with PBS containing 0.5% FBS) were added to each well (100 mL/well). After 2 h of incubation at room temperature, the plates were washed, and 100-fold diluted streptavidin-ALP was added for 1 h of incubation at room temperature. Blots were developed by the addition of BCIP/NBT substrate solution. Deionized water was added to each well until clear blots were observed. The plates were stored in the dark, and the IFN-g , IL-2, and IL-4 spot-forming units (SFU) were counted before the plate dried out.
H11-K18-hACE-2 mouse immunization and virus challenge. Female H11-K18-hACE-2 mice (12 weeks) were randomly grouped (n = 9/group) and immunized by intramuscular injection with two doses spaced 21 days apart (study days 0 and 21) containing 10 mg of S-TM with 50 mg Al plus 10 mg CpG or with only PB buffer at 100 mL/dose. Serum was collected for analysis on day 34. Vaccinated mice were then transferred into the BSL-3 animal laboratory at IVDC, China CDC, on day 35 and were intranasally challenged with live SARS-CoV-2 virus Beta on day 36. For the virus challenge, vaccinated H11-K18-hACE-2 mice were anesthetized by intraperitoneal injection with 0.2 mL/10 g of tribromoethanol and then intranasally inoculated with 25 TCID 50 of SARS-CoV-2 VOC Beta strain (GD2021) in 20 mL. The challenged mice were weighed on the day of infection and 4, 5, 6, 8, and 11 dpi. At 4, 8, and 11 dpi for the S-TM 1 Al 1 CpG-immunized group, three mice were sacrificed at each time point. For the PB buffer group, three mice were sacrificed only at 4 dpi, and the rest of the samples were collected according to the survival of the mice after 4 dpi. Lungs and brains were collected for viral load and pathology. Samples from natural death mice after infection in the PB group were collected only for viral load.
Real-time RT-PCR for SARS-CoV-2 RNA quantification. Tissue homogenization was performed at the BSL-3 animal laboratory at IVDC, China CDC. The tissue homogenates were clarified by centrifugation at 11,000 rpm for 5 min, and 50 mL supernatant was subjected to RNA extraction with a nucleic acid extraction kit (Xi'an Tianlong Science and Technology Co., Ltd., China). Reverse transcription-quantitative PCR (RT-qPCR) was performed using a novel coronavirus (2019-nCoV) nucleic acid detection kit (lot number: 20210810F, Shanghai BioGerm Medical Technology Co., Ltd., China) according to the protocol from the China CDC. The ORF1ab gene of SARS-CoV-2 was amplified by RT-qPCR. The gene copies indicating the viral load were quantified according to the standard (high concentration) SARS-CoV-2 nucleic acid (National Institute of Metrology, China, GBW [E] 091089).
Histopathology analyses. The tissues were fixed in 4% formalin and paraffin embedded. Lung sections were prepared and stained with hematoxylin and eosin (H&E). Sections were scored on a severity scale of 0 to 4 for the degree of inflammatory cell infiltration in different anatomical sites of the lung (scored 0 when no or only 1 inflammatory cell infiltration was visible, scored 1 when less than 5 infiltrations of inflammatory cells were visible, scored 2 when 5 to 10 infiltrations of inflammatory cells were visible, scored 3 when 10 to 20 infiltrations of inflammatory cells were visible, and scored 4 when more than 20 infiltrations of inflammatory cells were visible). Lung disease at different anatomical sites under light microscopy was assessed by the expansion of the parenchymal wall, interstitial pneumonia, alveolitis, and bronchiolitis.
Statistical analysis. Statistical analyses were performed using Prism 8.0 (GraphPad Software). Serum antibody titers were plotted for individual animals and the geometric mean titer (GMT) and 95% confidence interval (95% CI) or the means 6 standard error of the mean (SEM) as indicated in the figure. RNA copies, lung pathology scores, and splenocytes of different cytokines are presented as mean SEM. Comparisons were performed using ordinary one-way or two-way analysis of variance (ANOVA) with Tukey's multiplecomparison test. P values , 0.05 were considered significant. "ns" indicates no significance.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.7 MB.