We focused on sarbecovirus spikes that bind human ACE2 (hACE2), as the ability of coronaviruses to bind this receptor has been linked to the risk of efficient zoonotic transmission to humans6. Also, existing surrogate neutralization assays for SARS-CoV-2 spike can be directly applied to other hACE2-binding spike proteins. Therefore, a subset of hACE2-binding spikes, namely BANAL-20-52, Pangolin-GX, Pangolin-GD, SHC014, Bat-WIV1, and Bat-SARSL sarbecoviruses, were selected for analysis.
To illustrate the homology between these spikes and their receptor binding domains (RBD), the amino acid sequences were aligned using Clustal Omega7 and the Percent Identity Matrix was plotted in a Heatmap (Fig. 1A). There is a clear divergence of SARS-CoV-1 from SARS-CoV-2, with the sarbecovirus spikes chosen for the current study separating into two groups, being more similar to either of these two SARS viruses. The phylogenetic trees (Fig. 1B) highlight the evolutionary relationship among the tested viral spikes and their receptor binding domains (RBD): based on sequence identity, variants of SARS-CoV-2, such as BA.1, have become more dissimilar from the Reference SARS-CoV-2 spike than some animal-origin viral spikes, such as BANAL-20-52 (97.4% vs 98.7% identity, respectively). These differences are more noticeable in rapidly evolving regions, such as the RBD (93.3% vs 97.3% identity, respectively).
We produced the sarbecovirus spikes using our CHO-expression platform and trimeric antigen design developed previously for the SARS-CoV-2 spike (SmT1)8; purity was excellent (> 95%) for all constructs with SDS-PAGE/Coomassie staining showing the presence of a single major species at ~ 170 kDa with low levels of high-molecular-weight bands (Fig. 1C). We confirmed the ability of these purified spike antigens to bind to HEK293T-hACE2 cells (Fig. 1D), allowing for the assessment of serum cross-neutralization capacity using a flow cytometry-based surrogate neutralization assay established previously9–11.
Next, using mouse serum generated in previous preclinical studies10,11, we evaluated the cross-protective ability of antibodies induced by SARS-CoV-2 spike protein subunit vaccination using equivalent antigen doses (3 µg) of Reference, Beta, Delta, a Trivalent (1 µg each of Reference + Beta + Delta) or Omicron (BA.1) in AddaS03-adjuvanted formulations. Notably, while Omicron (BA.1) has over 30 mutations in the spike protein compared to Reference, the Beta and Delta variants differ by fewer than 10 amino acids. The immunogenicity of each of these vaccines against SARS-CoV-2 Reference strain and the aforementioned variants has previously been confirmed10,11. Using a cell-based spike-hACE2 binding assay, which has been previously shown to correlate strongly with the SARS-CoV-2 neutralization activity as determined with other widely-used assays (e.g., PRNT, pseudolentiviral neutralization and in vivo viral challenge9–11), the neutralization activities of these serum samples against the selected spikes were measured (Fig. 2). Consistent with our previously published data, neutralization activity for a given spike was highest with serum from animals immunized with the matching antigen: Reference and Omicron (BA.1) SARS-CoV-2 were most neutralized by serum from mice vaccinated with Reference (81.3%) and Omicron (BA.1) spike (56.6%), respectively10,11. The inverse was also true, whereby Reference or Omicron (BA.1) vaccinated mice had the lowest Omicron (BA.1) (22.6%) or Reference (21.6%) neutralizing titer, respectively.
Following this trend, the neutralizability of each spike is linked to their respective relationship to SARS-CoV-2. This is highlighted with BANAL-20-52 and Pangolin-GD, which have a nearly identical neutralization profile to Reference SARS-CoV-2. This correlates strongly with the observed homology within the RBD and not the entire spike protein sequence, as both BANAL-20-52 and Pangolin-GD have ~ 97% identity to Reference SARS-CoV-2 in this region, while Pangolin-GD has lower (91.2%) homology to the entire Reference spike. Interestingly, serum from the Omicron (BA.1) spike-immunized mice showed much lower neutralization against a number of sarbecoviruses than seen following immunization with Reference and sometimes Beta or Delta spike, which often induced an intermediate neutralizing profile. For example, ~ 10% neutralization of binding of BANAL-20-52 spike to hACE2 with following immunization with the Omicron (BA.1) vs. >90% when the immunization antigen is Reference Spike. Significant differences in the neutralization activity between the sera of Reference and Omicron-immunized mice were also seen against the spike protein based on other SARS-CoV-2-Like viruses (Pangolin-GX & Pangolin-GD) as well as SARS-CoV-1 and the SARS-CoV-1-Like virus, Bat WIV. Despite all of these proteins binding the same receptor (hACE2), there was much lower neutralizing activity detected against sarbecovirus spikes that are closely related to SARS-CoV-1. The highest neutralization was observed with Bat-WIV1 spike, with neutralization activity of ~ 30% seen with serum from mice vaccinated with Reference, Beta or Trivalent SARS-CoV-2 spike, and only ~ 10% with Omicron (BA.1), at the dilution tested.
To confirm the impact of the vaccine platform on these observations, we next evaluated the neutralizing potential of antibodies invoked by vaccination with mRNA / lipid nanoparticles (mRNA / LNPs) vaccine formulations encoding different SARS-CoV-2 spikes. We generated DNA templates encoding SARS-CoV-2 Reference and Omicron (BA.4/BA.5) spike proteins, with sequences based on the design used by Pfizer-BioNTech12. The mRNA was transcribed in vitro and includes N1-methyl-pseudouridine in place of the canonical uridine, as is typical for the clinically approved mRNA / LNP vaccines12,13. After encapsulation into commercially available lipids, Balb/c mice (n = 10) were vaccinated with 1µg of mRNA / LNPs on Day 0 and Day 21. As before, the serum from these animals was assessed 7 days post-boost for sarbecovirus spike neutralizing potential (Fig. 3). As expected, the binding of Reference and Omicron (BA.4/BA.5) spike proteins were most strongly neutralized by serum from mice vaccinated with mRNA / LNPs encoding Reference (61.8%) and Omicron (BA.4/BA.5) spike (71.1%), respectively10,11, while the bivalent vaccine induced a more intermediate neutralizing profile. Interestingly, despite containing equivalent doses of each spike mRNA, the Bivalent vaccine induced a more potent neutralizing titer towards the BA.4/BA.5 (56.3%) than the Reference (27.9%) spike. The serum from mice vaccinated with Reference, Omicron (BA.4/BA.5) and Bivalent spike all similarly neutralized (35.8% – 47.4%) the hACE2-binding of Omicron (BA.1) spike, which has a similar degree of homology to the Reference and Omicron (BA.4/BA.5) spike proteins. As seen with the protein subunit vaccines in Fig. 2, the reference-based mRNA/LNP vaccines induced significantly higher neutralization profiles of BANAL-20-52 and Pangolin-GD than vaccines designed to target Omicron (BA4/BA5). Meanwhile, no significant differences were observed in the neutralization profile of the other spike proteins, including those which had shown less pronounced, but still significant, differences with the protein vaccine formulations (i.e., Pangolin-GX, Bat WIV1 and SARS-CoV-1. Altogether, these data highlight the superior ability of SARS-CoV-2-based vaccine formulations based on the Reference sequence to generate neutralizing antibodies to certain animal-derived sarbecoviruses, whether using protein subunit or mRNA / LNP vaccine platforms.