Prefusion-stabilized SARS-CoV-2 S2-only antigen provides protection against SARS-CoV-2 challenge

Structure-based design of prefusion-stabilized SARS-CoV-2 S2 subunits

We first generated a base construct for S2-only antigens (HexaPro-S2) by removing the entire S1 subunit from the HexaPro S protein (containing 6 stabilizing proline substitutions) and deleting a flexible region (residues 686–696) from the N-terminus of the S2 subunit. Given that the S1 subunit forms an extensive trimerization interface and belt-like structure that confines the membrane-distal apex of S2, we designed 8 interprotomer disulfide substitutions using the prefusion structure of HexaPro S (PDB ID: 6XKL)³⁰ as a guide (Fig. 1a). Based on the Cβ distances of residues at the interprotomer interface, the following three sets of substitutions were assessed in the base construct: (1) Y707C/P792C, S704C/K790C, A713C/L894C, and G891C/P1069C near the lateral face, (2) Q755C/N969C and G757C/S968C near the apex, and (3) S1030C/D1041C and G1035C/V1040C in the core of S2. The Y707C/T883C substitution from Olmedillas et al.³¹ was included as a comparison. Each construct was characterized for expression yield and interprotomer disulfide formation as assessed by non-reducing SDS-PAGE, monodispersity by size-exclusion chromatography (SEC) and thermostability by differential scanning fluorimetry (DSF).

a Side view of HexaPro (PDB ID: 6XKL). The S1 subunits are shown as a transparent molecular surface. The S2 subunit of each protomer is shown as a ribbon diagram. Each inset corresponds to a zoomed view of interprotomer disulfide designs. Side chains in each inset are shown as sticks with sulfur atoms in yellow. b Reducing (top) and non-reducing (bottom) SDS-PAGE analysis of each interprotomer disulfide variant. The molecular weight standards in kDa are indicated at the left. The position of monomer, dimer and trimer bands are indicated at the right. The SDS-PAGE analysis was performed once. c Size-exclusion chromatography and d, differential scanning fluorimetry analysis of each interprotomer disulfide variant. The vertical dotted line indicates (c) the peak retention volume and (d) the melting temperature for the HexaPro-S2 containing no disulfide substitution. Source data are provided as a Source Data file.

Four interprotomer disulfide bond substitutions had comparable protein expression relative to HexaPro S2, and one design—G757C/S968C—had a substantial decrease in protein yield (Fig. 1b). All five substitutions formed interprotomer disulfide bonds, but to different extents. Both Y707C/T883C and A713C/L894C showed detectable monomeric and dimeric fractions on the non-reducing gel. Notably, Y707C/P792C and S704C/K790C had 83% and 85% in the trimeric fractions, respectively. In contrast, the expression of Q755C/N969C, S1030C/D1041C, and G1035C/V1040C was completely abolished. Except for Y707C/P792C, all expressed constructs had a slight rightward shift of the SEC peak relative to the base construct, consistent with a more compact conformation (Fig. 1c). Both Y707C/T883C and Y707C/P792C exhibited a broader and less monodisperse peak than others, indicating heterogeneity, which was congruent with the DSF analysis that revealed multiple melting temperatures (Tm) (Fig. 1d). The substitutions S704C/K790C, A713C/L894C, and G757C/S968C showed substantial increases in Tm relative to the base construct (Fig. 1d and Supplementary Table 1), ranging from +7.1 to +13.7 °C. Next, we added A713C/L894C, G757C/S968C, or both disulfide substitutions on the background of S704C/K790C. The expression of the combinatorial disulfide constructs containing G757C/S968C was completely abolished. Although A713C/L894C/S704C/K790C showed comparable expression relative to S704C/K790C, it exhibited incomplete trimer formation on the non-reducing SDS-PAGE gel and was not pursued further (Supplementary Fig. 1a, b).

Based on the extent of trimer formation, expression yield, and thermostability, we prioritized S704C/K790C for further engineering through adding proline or interprotomer salt-bridge designs. Although the addition of Q895P decreased the expression relative to the parental construct, the Q957E substitution enhanced the expression with the majority of protomers covalently linked via the disulfide bond (Supplementary Fig. 1c, d). Given that several broadly protective antibodies have been demonstrated to target the helical stalk at the base of the globular S2 ectodomain^{16,17,18,19,20,21,32}, we hypothesized that transplanting multiple stalks (corresponding to SARS-CoV-2 spike residues 1142–1208) from related betacoronavirus spikes to this construct could have beneficial effects on immunogenicity (note that SARS-CoV and SARS-CoV-2 spikes have identical amino acid sequences in this region). The expression of the TriStalks construct (SARS-CoV-2, MERS-CoV, and HKU1) was lower and the SEC traces were more polydisperse than its parental S704C/K790C construct (Supplementary Fig. 1c, d). In contrast, the expression of PentaStalks (SARS-CoV-2, MERS-CoV, HKU1, OC43, and HKU9) was substantially higher than its parental construct. We further constructed a version of the S704C/K790C S2 antigen without the stem helix and remaining ectodomain stalk (residues 1142–1208), named Δstalk, which also expressed robustly and eluted as a monodisperse peak on SEC (Supplementary Fig. 1c, d). Large-scale expression of our best HexaPro-S2 variant that contains S704C/K790C/Q957E substitutions produced 2.5 mg of protein from 1 L of FreeStyle 293-F cells (Supplementary Fig. 2a). We renamed it HexaPro-SS (stabilized stem) and added the same modifications at the C-terminus of S2, generating HexaPro-SS-PentaStalks and HexaPro-SS-Δstalk. HexaPro-SS-PentaStalks and HexaPro-SS-Δstalk also expressed robustly in a large-scale format (Supplementary Fig. 2b, c). To confirm that our engineering did not alter the antigenic surface of S2, we examined the binding of various S2-specific antibodies to the antigens (Supplementary Fig. 3). As expected, RBD-targeting antibody N3-1³³ did not bind to HexaPro-SS or HexaPro-SS-Δstalk. Previous studies have reported that stem-helix-specific S2 antibodies show broadly neutralizing capability. Importantly, stem-helix antibodies IgG22²¹, S2P6¹⁸ and CC40.8²⁰ bound to HexaPro-SS at similar magnitudes as those observed for binding to the full S ectodomain S-2P and HexaPro, but not HexaPro-SS-Δstalk, which lacks the epitope. Interestingly, fusion-peptide-directed antibodies CoV44-79²³ and CoV91-27 bound strongly to HexaPro-SS and weakly to the full-length spikes (S-2P and HexaPro), suggesting that the fusion peptide is more accessible in the context of S2-only antigens. Collectively, we stabilized SARS-CoV-2 S2-only antigens through structure-based design and expanded the variety of S2-only antigens through additional modifications at the C-terminus to probe their influence on immunogenicity.

Prefusion-stabilized S2 structures reveal a range of flexibility at the trimer apex

To investigate whether our design effectively locked the S2-only antigen in the prefusion conformation, we determined the crystal structure of HexaPro-SS-Δstalk. The protein complex crystallized in space group R3 and an X-ray diffraction dataset was collected to a resolution of 3.2 Å. One protomer of the S2 subunit from the HexaPro spike structure (PDB ID: 6XKL) was used as a search model for molecular replacement. Iterative model building and refinement resulted in a structure with R_work and R_free values of 22.6% and 26.2%, respectively (Supplementary Table 2). The asymmetric unit contained two S2 protomers (each from a different trimer), packed in a head-to-head arrangement, resulting in the upper half of the central helices packing against each other. HexaPro-SS-Δstalk adopted a prefusion conformation and formed a trimer when crystallographic symmetry was applied (Fig. 2a). The cysteine substitutions (Cys704/Cys790) from symmetry mates are in proximity to each other to form disulfide bonds, but Glu957 could not form a salt bridge with Arg765 from the neighboring protomer due to separation of the protomers at the apex. In comparison with the S2 subunit from the HexaPro spike structure (PDB ID: 6XKL), HexaPro-SS-Δstalk is splayed apart at the trimer apex with the central helices ~15° further away from the 3-fold axis (Fig. 2b, c). Given that this partially open conformation could result from a crystal packing artifact, we set out to examine whether HexaPro-SS-Δstalk could sample multiple conformations as assessed by cryo-EM. After data collection and processing, we were able to obtain at least three distinct 3D reconstructions, with the apex of S2 adopting closed, semi-open and fully open conformations (Fig. 3). Although the low resolution of these 3D reconstructions prevented us from building atomic models, all three conformations are prefusion trimers, consistent with our crystallographic data. Taken together, the HexaPro-SS-Δstalk S2-only antigen presents multiple prefusion conformations that differ in the extent to which the apex is open.

a Side (left) and top (right) views of the crystal structure for HexaPro-SS-Δstalk are shown in ribbons. The cysteine substitutions (inset) from symmetry mates are in proximity to each other to form disulfide bonds, but Glu957 does not form a salt bridge with Arg765 from the neighboring protomer. b The side and top view of HexaPro-SS-Δstalk presents a prefusion conformation with the trimer apex splayed open. c The surface representation of HexaPro (PDB ID: 6XKL) with the S1 subunit removed, exhibiting a closed prefusion conformation.

a A representative micrograph (from 3350 collected) and b 2D class averages of HexaPro-SS-Δstalk. c 3D heterogenous reconstructions of HexaPro-SS-Δstalk demonstrate three distinct conformations with the trimer apex closed (left), semi-open (middle), and fully open (right).

Sera from S2-immunized mice neutralize diverse sarbecovirus rVSV-CoVs

To investigate whether our stabilized S2-only antigens could elicit broadly reactive antibody responses, we immunized mice with 10 µg of immunogen adjuvanted with Sigma Adjuvant System (SAS) using a prime-boost regimen. Serum was collected one week prior to boost (week 2) and infection (week 7), respectively (Supplementary Fig. 4). We then compared the ability of post-boost sera from these immunized mice to neutralize recombinant vesicular stomatitis virus displaying spike proteins (rVSV-CoVs) derived from a wide range of betacoronaviruses. All rVSV-CoVs displaying sarbecovirus-derived spike proteins (Wuhan-1, Omicron BA.1, SARS-CoV, and SHC014) were neutralized by sera from mice immunized with the full HexaPro spike protein ectodomain, which served as our positive control (Fig. 4a-e). Additionally, sera from mice immunized with S2 immunogens HexaPro-S2 (base construct), HexaPro-SS, HexaPro-SS-PentaStalks, and HexaPro-SS-Δstalk significantly neutralized rVSVs bearing spikes from SARS-CoV-2 Wuhan-1, SARS-CoV, and SHC014 compared to those mice immunized with PBS (Fig. 4a, c, d). Of mice immunized with S2 constructs, only HexaPro-SS-Pentastalk and HexaPro-SS-Δstalk significantly neutralized rVSV-SARS-CoV-2 Omicron BA.1 compared to mice immunized with PBS (Fig. 4b). No significant neutralization was observed for rVSV-MERS-CoV when tested with sera from mice immunized with HexaPro or any of the S2-only constructs (Fig. 4e).

a-e Pre-titrated amounts of rVSV-CoVs were incubated with serial 3-fold dilutions of sera from mice (n = 10/immunized group) immunized with respective S antigens or PBS at RT for 1 hr. Virus-sera mixtures were then added to monolayers of Vero cells. At 10 hr post-infection, cells were fixed, and nuclei were counterstained. Infected cells were scored by BioTek Cytation5 for presence of GFP. Sera from 10 mice are included in each group. Area under the curve (AUC) was calculated from normalized infectivity levels. One-way ANOVA with Tukey’s multiple comparisons test (a, c, e) or Welch’s ANOVA with Dunnett’s T3 multiple comparisons test (b, d) were run based on normality and homoscedasticity. **** p < 0.0001, *** p < 0.001, ** p < 0.01, * p < 0.05 a PBS:HexaPro, HexaPro-SS, HexaPro-S2, HexaPro-SS-PentaStalks, Hexapro-SS-Δstalk p = <1e-15, 2.86e-9, 2.61e-7, 1.48e-11, 1.11e-11, HexaPro:HexaPro-SS, HexaPro-S2, HexaPro-SS-PentaStalks, Hexapro-SS-Δstalk p = 5.64e-10, 3.01e-11, 1.04e-7, 1.34e-7; b PBS:HexaPro-SS-PentaStalks, Hexapro-SS-Δstalk p = 7.37e-4, 2.40e-3; c PBS:HexaPro, HexaPro-SS, HexaPro-S2, HexaPro-SS-PentaStalks, Hexapro-SS-Δstalk p = 5.14e-3, 2.75e-6, 9.77e-7, 6.54e-6, 6.68e-7; d PBS:HexaPro, HexaPro-SS, HexaPro-S2 p = 6.20e-3, 1.31e-2, 7.20e-3. Source data are provided as a Source Data file.

HexaPro-SS vaccinated mice are fully protected against lethal SARS-CoV-2 challenge and partially protected against lethal SARS-CoV challenge

To assess the ability of the various S2-only constructs to protect against lethal infection, we immunized 10-week-old female BALB/c mice with 10 µg of one immunogen and boosted with the same immunogen three weeks after prime immunization. Both immunizations used SAS as the adjuvant. Five weeks after the boost, animals were infected with either 10⁴ PFU SARS-CoV-2 MA10 or 10⁴ PFU of the heterologous SARS-CoV MA15^34,35,36 and monitored daily for changes in body weight and signs of morbidity. As soon as animals reach 80% of their starting weight, they are subject to more stringent observation with weights measured twice a day as well as a visual check-in between measurements. Animals that approach 70% of their starting weight are humanely euthanized. Dedicated groups of mice for each immunization regimen were sacrificed for tissue collection on days 2, 4, and 7 (Supplementary Fig. 4b). Only animals immunized with HexaPro spike ectodomain were fully protected against challenge with SARS-CoV-2 MA10, with no changes in body weight after infection (Fig. 5a). Importantly, even though some S2-only constructs provided minimal but statistically significant protection from weight loss (HexaPro-SS as well as HexaPro-SS-PentaStalks; Fig. 5a), they conferred significant protection from mortality with only one mouse succumbing to infection in the HexaPro-SS-Δstalk-immunized group. All mock-immunized but virus-challenged control mice succumbed to infection by day 4 (Fig. 5b). Gross evaluation of macroscopic changes in coloration of lung tissue at the time of sample collection (congestion score: 0–4; 0 = healthy pink lung, 1 = 25%, 2 = 50%, 3 = 75%, 4 = 100% of whole lung tissue exhibits dark red discoloration) confirmed this trend. HexaPro-immunized mice exhibited no changes in lung coloration, whereas all other vaccinated mice showed minimal changes (congestion scores: 0–1/4). In contrast, severe alterations in lung coloration were observed in the PBS-immunized control group (congestion scores: 2.5–4/4) on day 4 and no mice survived until day 7. Statistical differences were found between HexaPro-immunized mice and HexaPro-SS-Δstalk-immunized mice late during infection (day 7) (Fig. 5c). Inhibition of viral replication was obvious and statistically significant compared to PBS-immunized mice in the HexaPro-vaccinated group on day 2 after infection with a reduction of two orders of magnitude (10³ PFU) in viral titer in the upper respiratory tract compared to similar titers in all other groups (10⁵ PFU). Four days after infection, virus was cleared from nasal turbinates in the HexaPro group, whereas all other groups showed average titers around 100 to 1000 PFU. On day 7 after infection, virus was cleared in all experimental groups (Fig. 5d). The difference observed in the upper respiratory tract early during infection was even more pronounced than in the lower respiratory tract, with no detectable titer in the HexaPro group compared to lung titers between 10⁶ to 10⁷ PFU in all other immunized groups on day 2 after infection. PBS-vaccinated mice exhibited the highest lung titers at 10⁷ PFU on day 2 after infection. Viral lung titers on day 4 after infection were reduced in all immunized groups by roughly 3 logs, whereas the control group lung titer decreased by only 1 log. As observed in the upper respiratory tract, all animals surviving until day 7 after infection showed no detectable lung titers (Fig. 5e).

After being immunized with respective S antigens or PBS, BALB/cAnNHsd mice were challenged with a lethal dose of mouse-adapted SARS-CoV-2 (n = 14 mice for HexaPro and HexaPro-SS, n = 15 mice for all other groups). a, Mouse body weights were monitored daily following challenge, and the percentage of body weight loss over time is represented by a line plot summarized per strain per day by the mean ± SD. We analyzed the change in body weight using mixed models with repeated measures (exact P values for each term and the interaction is reported in the plot) followed by Tukey post-tests (n = 14 mice for HexaPro and HexaPro-SS, n = 15 mice for all other groups at day 0). Per-day comparisons with post-tests <0.05 are indicated with symbols defined in the key. b, The survival graph represents the probability of survival over two to seven days post-challenge (n = 5 mice per group). The Mantel-Cox log-rank test determined the survival curves were different (χ² = 47.18, df = 6, P = 1.7e-8) and the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli identified the pairwise differences of survival curves with a false discovery rate (Q) < 5%. c, The congestion score in lung and the viral titers in d, nasal turbinate and e, lung (n per group listed in the graphs) are represented by dot plot, summarized by the median ± IQR. We analyzed the congestion scores and titers using ANOVA on ranks per day with each P value reported in the figure. Dunn’s pairwise comparison post-tests for p < 0.05 and p < 0.001 are noted by * and ‡, respectively. Source data are provided as a Source Data file.

To test the breadth of protection provided by immunization with HexaPro and S2-only immunogens, we immunized 10-week-old female BALB/c mice as described above and challenged them with a lethal dose of SARS-CoV MA15. In contrast to the SARS-CoV-2 MA10 challenge, none of the constructs provided significant protection from weight loss after challenge with 10⁴ PFU of mouse-adapted SARS-CoV MA15 (Fig. 6a). However, all constructs conferred significant protection from mortality compared to the control group, in which all mice succumbed to infection by day 6. Among the groups of mice immunized with the different spike constructs, HexaPro, HexaPro-SS, and HexaPro-S2 showed similar protection with mortality rates of 20% (Fig. 6b). Gross evaluation of changes in lung coloration showed no significant differences by day 2 after infection (Fig. 6c). After day 2, there were no control animals left for comparison, however, the trends observed in mortality rates were confirmed via immunization with HexaPro, HexaPro-SS, and HexaPro-S2 leading to less discoloration of lung tissue compared to immunization with HexaPro-SS-Δstalk and HexaPro-SS-PentaStalks. Significant differences were found on day 4 between HexaPro-SS-Δstalk and mock-infected control groups and on day 7 between HexaPro-S2-immunized as well as HexaPro-SS-Δstalk-immunized mice and mock-infected control mice (Fig. 6c).

After being immunized with respective S antigens or PBS, BALB/cAnNHsd mice were challenged with a lethal dose of mouse-adapted SARS-CoV (n = 14 mice for HexaPro and PBS PBS, n = 15 mice for all other groups). a Mouse body weights were monitored daily following challenge, and the percentage of body weight loss over time is represented by a line plot summarized per strain per day by the mean ± SD. We analyzed the change in body weight using mixed models with repeated measures (exact P values for each term and the interaction is reported in the plot) followed by Tukey post-tests (n = 14 mice for HexaPro and PBS PBS, n = 15 mice for all other groups at day 0). One animal from the HexaPro-immunized group as well as one animal from the HexaPro-SS-immunized group reached 70% of their starting weight on day 7 after infection. Per-day comparisons with post-tests <0.05 are indicated with symbols defined in the key. b, The survival graph represents the probability of survival over two to seven days post-challenge (n = 5 mice per group). The Mantel-Cox log-rank test determined the survival curves were different (χ² = 25.36, df = 6, P = 3e-4) and the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli identified the pairwise differences of survival curves with a false discovery rate (Q) < 5%. c The congestion score in lung and the viral titers in d, nasal turbinate and e, lung (N per group listed in the graphs) are represented by dot plot, summarized by the median ± IQR. We analyzed the congestion scores and titers using ANOVA on ranks per day with each P value reported in the figure. Dunn’s pairwise comparison post-tests for p < 0.05 and p < 0.01 are noted by * and ***, respectively. Source data are provided as a Source Data file.

Viral titers in the upper respiratory tract (Fig. 6d) were similar across all groups at around 10⁵ PFU on day 2 after infection and 10⁴ PFU on day 4 after infection. All groups showed similarly high viral titers in the lower respiratory tract at 10⁷ PFU by day 2 after infection, which decreased to 10⁴ by day 4. The only exceptions were the two surviving mice in the HexaPro-SS-PentaStalks immunized group exhibiting slightly elevated titers (10⁵ PFU) in the upper respiratory tract on day 4 as well as the HexaPro-SS-immunized group showing reduced lung titers (10⁴ PFU) on day 4 after infection (Fig. 6e). By day 7 after infection neither upper nor lower respiratory tract samples showed any detectable titers. These results demonstrate that S2-only immunogens can provide protection against SARS-CoV-2 and partial protection against SARS-CoV despite not being able to elicit highly neutralizing RBD- and NTD-directed antibodies.

Double HexaPro boost increases protection against SARS-CoV in BALB/c mice

Based on the results of BALB/c mice immunized with different spike constructs showing protection from mortality but not weight loss in a lethal SARS-CoV MA15 challenge (Fig. 6), we altered our immunization strategy in an attempt to increase beneficial effects on cross-protection. Therefore, we immunized 10-week-old female BALB/c mice with 10 µg HexaPro, deglycosylated HexaPro-SS, or PBS (week 0) followed by a first boost (week 6) with either HexaPro, HexaPro-SS, deglycosylated HexaPro-SS ('degly'), or PBS followed by a second boost (week 10) with the same antigens as in the first boost (Fig. 7a). Given that glycosite-deleted spike protein elicited a broad-spectrum humoral and cell-mediated immunity³⁷, we hypothesized that deglycosylated HexaPro-SS could also provide a similar advantage. Thus, we treated HexaPro-SS with Endo H, leaving a single GlcNAc at each N-linked glycosylation site on S2. To see whether we could enhance the breadth of neutralization against a range of rVSV-CoVs, microneutralization assays were performed with mouse sera collected 2 weeks after the third dose of immunogen. We found that mouse sera from all four immunization strategies significantly neutralized sarbecovirus rVSV-CoVs rVSV-SARS-CoV-2 Wuhan-1, rVSV-SARS-CoV, and rVSV-SHC014 compared to mice immunized with PBS (Supplementary Fig. 5a, c, d). Both triple HexaPro immunization and HexaPro followed by 2 immunizations of deglycosylated HexaPro-SS significantly neutralized rVSV-SARS-CoV-2 Omicron BA.1 compared to PBS immunized mouse sera, with triple HexaPro immunization being more strongly neutralizing (p < 0.0001) (Supplementary Fig. 5b). There was a modest, but significant neutralization of rVSV-MERS-CoV seen with sera from mice immunized with HexaPro followed by 2 shots of deglycosylated HexaPro-SS (p < 0.05) compared to mice immunized with PBS (Supplementary Fig. 5e).

a After being immunized with respective S antigens or PBS, BALB/cAnNHsd mice were challenged with a lethal dose of mouse-adapted SARS-CoV MA15 (n = 9 mice for HexaPro-SS-SS, n = 10 mice for all other groups). The cartoons were created by BioRender.com. b Mouse body weights were monitored daily following challenge, and the percentage of body weight loss over time is represented by a line plot summarized per strain per day by the mean ± SD. We analyzed the change in body weight using mixed models with repeated measures (exact P values for each term and the interaction are reported in the plot) followed by Tukey post-tests (n = 9 mice for HexaPro-SS-SS, n = 10 mice for all other groups at day 0). Per-day comparisons with post-tests <0.05 are indicated with symbols defined in the key. c The survival graph represents the probability of survival over two to four days post-challenge (n = 5 mice per group). The Mantel-Cox log-rank test determined the survival curves were different (χ² = 14.48, df = 4, P = 5.9e-3). The two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli did not identify any pairwise differences of survival curves with a false discovery rate (Q) < 5%. d The congestion score in lung and the viral titers in e, nasal turbinate and f, lung (n per group listed in the graphs) are represented by dot plot, summarized by the median ± IQR. We analyzed the congestion scores and titers using ANOVA on ranks per day with each P value reported in the figure. Dunn’s pairwise comparison post-tests for p < 0.05 and p < 0.01 are noted by * and **, respectively. Source data are provided as a Source Data file.

To investigate whether this immunization strategy can protect the mice from SARS-CoV, double-boosted mice were infected 8 weeks after the second boost (week 18) with 10⁴ PFU mouse-adapted SARS-CoV MA15 and monitored daily for clinical signs of disease. Dedicated groups of mice for each immunization regimen were sacrificed for tissue collection on days 2 and 4 after infection (Fig. 7a). To reduce morbidity and mortality in our experimental cohort, we decided to end the experiment on day 4 instead of day 7 to keep the number of animals that reach experimental endpoints for humane euthanasia as low as possible. Only triple HexaPro-immunized mice showed significant protection from weight loss whereas all other groups followed the control group trajectory (Fig. 7b). However, all but the HexaPro-primed, HexaPro-SS-boosted immunization strategy conferred protection from mortality until the end of the study at day 4 (Fig. 7c). No significant differences were observed during gross evaluation of lung discoloration at the time of sample collection on day 2 after infection. Maximal congestion score was recorded for the one remaining control animal on day 4 after infection, whereas all immunized mice exhibited congestion scores of 0.5 to 2 with double HexaPro-boosted animals showing the least amount of change in lung coloration (Fig. 7d). Average viral titers in upper respiratory tracts were similar across all groups on both harvest days. Importantly, only two HexaPro double-boosted animals showed viral clearance on day 4 after infection (Fig. 7e).

Significant differences in viral titers in the lower respiratory tract were observed on day 2 after infection with lower titers in the HexaPro-HexaPro-HexaPro and degly-degly-degly immunized groups in comparison to mock-immunized animals. However, on day 4 we detected a clear difference in lung titers between the one surviving control animal (10⁶ PFU) and all immunized groups (10² to 10⁴ PFU). Again, 3 out of 5 animals from the double HexaPro-boosted group were able to clear virus by day 4. Additionally, 1 out of 5 animals from the double deglycosylated boosted group exhibited no detectable lung titer (Fig. 7f). Collectively, these data show that only a double boost with HexaPro provides increased protection not only from mortality but also from changes in body weight after SARS-CoV infection compared to a single boost (Fig. 6 and Fig. 7). Vaccination approaches with HexaPro in combination with two boosts of each construct did not protect from clinical signs of disease.