Vaccination with B.1.1.7, B.1.351 and P.1 variants protects mice from challenge with wild type SARS-CoV-2

Vaccines against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have been highly efficient in protecting against coronavirus disease 2019 (COVID-19). However, the emergence of viral variants that are more transmissible and, in some cases, escape from neutralizing antibody responses has raised concerns. Here, we evaluated recombinant protein spike antigens derived from wild type SARS-CoV-2 and from variants B.1.1.7, B.1.351 and P.1 for their immunogenicity and protective effect in vivo against challenge with wild type SARS-CoV-2 in the mouse model. All proteins induced high neutralizing antibodies against the respective viruses but also induced high cross-neutralizing antibody responses. The decline in neutralizing titers between variants was moderate, with B.1.1.7 vaccinated animals having a maximum fold reduction of 4.8 against B.1.351 virus. P.1 induced the most cross-reactive antibody responses but was also the least immunogenic in terms of homologous neutralization titers. However, all antigens protected from challenge with wild type SARS-CoV-2 in a mouse model. Author Summary The emergence of variants of SARS-CoV-2 has led to an urgency to study whether vaccines will lead to cross-protection against these variants. Here, we demonstrate that vaccination with spike proteins of various variants leads to cross-neutralizing responses, as well as protection in a mouse model against wild type SARS-CoV-2.


41
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in late 2019 in Wuhan, China. 42 Since then, the virus has caused the coronavirus disease 2019 (COVID-19) pandemic leading to 43 approximately 4 million official deaths globally (as of July 2021). While coronaviruses usually mutate more 44 slowly than other RNA viruses due to the proof-reading activity of their replication machinery [1], viral 45 variants started to emerge in the summer of 2020 in humans and mink in Europe [2][3][4]. In late 2020, 46 additional variants, termed variants of concern (VoC) emerged in the UK [5], in South Africa [6] and in 47 Brazil [7]. These variants, B.1.1.7, B.1.351 and P.1, are more infectious than wild type SARS-CoV-2 and 48 feature extensive changes in both the receptor binding domain (RBD) and the N-terminal domain (NTD) 49 of the spike protein. These two domains harbor the vast majority of neutralizing epitopes [8][9][10][11][12][13][14] and 50 consequently it has been observed that -especially for B.1.351 -the neutralizing activity of wild type 51 post-infection and post-vaccination sera is reduced [15][16][17][18]. In addition, efficacy and effectiveness of 52 vaccines against B.1.351 has been shown to be somewhat reduced, depending on the type of vaccine 53 platform used [19,20]. For one currently licensed vaccine, the efficacy against B.1.351 was lost [21]. 54 Updated vaccines based on variant spike sequences are currently being tested by vaccine producers and 55 may be licensed in the future if variants emerge that escape vaccine-induced immunity to an even larger 56 degree. However, the process of updating vaccine antigens to match circulating variants is not as straight 57 forward as it seems. Several variants might circulate simultaneously, making it difficult to choose the right 58 antigen for optimal protection. Of course, multivalent vaccines that include more than one variant antigen 59 can be formulated, but this increases complexity and decreases the amount of vaccine doses that can be 60 manufactured. Understanding the antigenic relationship between variants is therefore of high 61 importance.

62
Here, we vaccinated mice with recombinant spike proteins from the wild type Wuhan-1 strain, B. derived, all animals mounted strong neutralizing antibody responses ( Figure 1A), while negative controls 75 showed no neutralizing activity ( Figure 1B). The negative control group received an irrelevant control 76 protein, influenza virus hemagglutinin. However, there was a trend towards B.1.1.7 vaccinated animals 77 showing higher neutralizing capacity against homologous virus as compared to the other spike antigens. 78 P.1 seemed to induce the lowest neutralizing activity against homologous viruses. These differences were 79 small and only significant for B.1.1.7 versus P.1.

80
As expected, when testing for cross-reactivity, the different spike proteins induced the highest 81 neutralization titers against the homologous viruses. Sera from wild type spike vaccinated animals 82 neutralized WA1 best, followed by B.1.1.7, P.1 and B.1.351 ( Figure 1C). Sera from B.1.1.7 vaccinated 83 animals neutralized B.1.1.7 best, followed by wild type, P.1 and B.1.351 ( Figure 1D). For B.1.351 84 vaccinated animals, we detected the highest titers against B.1.351 followed by wild type, B.1.1.7 and P.1 85 ( Figure 1E). P.1 induced a surprisingly uniform level of immunity with the lowest drop to wild type virus 86 followed by B.1.351 and P.1 ( Figure 1F). The steepest drops in neutralization were detected for B. approximately equal distance from WA1 in opposite directions. The sera loosely cluster in the vicinity of 92 the antigen they were raised against.

93
Antibody binding is less affected than neutralization 94 We repeated our analysis using an ELISA with the respective spike proteins as substrates. While 95 neutralization requires binding of antibodies to a limited number of epitopes mostly on RBD and NTD, 96 many more binding epitopes exist on the spike protein. Therefore, more even reactivity was expected. 97 We did detect differences in reactivity when binding was tested against the respective matched spikes 98 ( Figure 3A) but while these differences were statistically significant in three cases, they were relatively 99 small. However, it seemed that vaccination with B.1.351 induced slightly more homologous binding 100 antibodies compared to the other immunogens. Low background reactivity was detected in sera of the 101 control animals ( Figure 3B).

102
Both wild type spike and B.1.1.7 spike induced relatively even binding antibody responses ( Figure 3C and 103 D) with maximum fold-reduction of 1.2 and 1.3-fold respectively. A stronger reduction was detected when 104 B.1.351 was used as immunogen with 3.2-fold and 3.8-fold reduction in binding to wild type and B.1.1.7 105 spike respectively ( Figure 3E). The drop for P.1 was smaller (1.8-fold). P.1 also induced comparable binding 106 antibody response with a maximum fold-reduction of 1.5-fold against B.1.1.7 ( Figure 3F).

107
These discrepancies between neutralization and binding antibody profiles allowed us to calculate ratios 108 between binding and neutralizing antibodies. The best (higher) ratios (indicating higher neutralization) 109 were found in sera from wild type and B.1.1.7 vaccinated mice (Supplementary Figure 1A). For each 110 vaccination group, the ratio was always best against the homologous virus and dropped with antigenic 111 distance (Supplementary Figure 1B-E). The most stable ratio was observed for P.1 vaccinated animals 112 (Supplementary Figure 1E).

113
All spike vaccinated animals are protected against challenge with wild type SARS-CoV-2

114
Finally, we wanted to assess if the induced neutralizing antibody responses can protect animals from 115 challenge with prototypic SARS-CoV-2 strain WA1. Since BALB/c mice are not susceptible to this virus, they 116 had to be pre-sensitized via intranasal transduction with adenovirus expressing human angiotensin 117 converting enzyme 2 (hACE2) before challenge, as previously described. The main readout for the 118 challenge experiment were virus titers in the lungs of infected animals. On day 2 post challenge, control 119 animals showed high viral loads in their lungs (approximately 10 6 plaque forming units per ml of lung 120 homogenate) ( Figure 4A). In contrast, no virus was detected in wild type and P.1 spike vaccinated animals. 121 For B.1.1.7 and B.1.351 spike vaccinated animals, one animal per group showed traces of virus replication 122 in the lung, but titers were barely above the limit of detection. On day 5 post infection, no virus was 123 detectable in the lungs of vaccinated individuals while control animals still showed high virus loads ( Figure  124 4B).  Liu et al. [26]. This suggests that mice may be a good model system to study antigenic 142 variability among variants. Such model systems are of importance, as they ensures the continued 143 availability of first-infection sera for the characterization of novel variants. However, there may be subtle 144 differences between mouse strains and certainly between mice and humans, that need to be further 145 explored. Interestingly, binding was much less affected than neutralization, likely due to many more 146 conserved binding epitopes present on the spike outside of RBD and NTD (which harbor most of the 147 neutralizing epitopes). Importantly, all spike antigens, independently of the lineage, provided robust 148 protection against challenge with the prototypic WA1 strains, suggesting that 'updated' vaccines -149 especially if they induce high neutralizing antibody titers -would sufficiently protect against most other 150 circulating variants as well as prototypic SARS-CoV-2 strains.

151
Our work here has focused on neutralizing and binding antibodies, which have been implicated as 152 correlates of protection for SARS-CoV-2 vaccine induced immunity [27,28] and reduction in neutralizing 153 antibodies in sera from convalescent individuals and vaccinees against variants has been observed. 154 However, T-cell responses very likely contribute to protection from COVID-19 as well. We have not 155 analyzed T-cell responses in our experimental animals but others have shown that the impact of variants 156 on these responses is minimal [29]. Another caveat of our study is that we were not able to include 157 B. SARS-CoV-2 intranasally with 1x10 5 PFU. Mice were humanely sacrificed on day 2 and day 5 for assessment 198 of virus in the lungs. Lungs were homogenized using special tubes and a BeadBlaster 24 (Benchmark)  199 homogenizer [39,40]. Viral load in the lung was quantified via a classic plaque assay [41].

200
ELISA. Ninety-six-well plates (Immulon 4 HBX; ThermoFisher Scientific) were coated with 2 μg/mL of each 201 respective protein with 50 μL/well overnight at 4°C. The next morning, the coating solution was discarded, 202 and each plate was blocked with 100 μL/well of 3% non-fat milk (AmericanBio; catalog no. 203 AB10109-01000) in phosphate buffered saline containing 0.01% Tween (PBS-T). Blocking solution was 204 kept on the plates for 1 hour at room temperature (RT). Serum samples were tested starting at a dilution 205 of 1:50 with 1:5 fold subsequent serial dilutions. Serum samples were added to the plates for 2 hours at 206 RT. Next, the plates were vigorously washed thrice with 200 μL/well of PBS-T. Anti-mouse IgG-HRP 207 (Rockland; catalog no. 610-4302) was used at a dilution of 1:3000 in 1% non-fat milk in PBS-T and 100 μL 208 of this solution was added to each well for 1 hour at RT. The plates were washed thrice with 200 μLwell 209 of PBS-T and dried on paper towels. Developing solution was prepared in sterile water (WFI; Gibco) using 210 SigmaFast OPD (o-phenylenediamine dihydrochloride, catalog no. P9187; Sigma-Aldrich), and 100 μL was 211 added to each well for a total of 10 min. To stop the reaction, 50 μL/well of 3 M hydrochloric acid was 212 added and the plates were read in a plate reader, Synergy 4 (BioTek), at an absorbance of 490 nanometers. 213 Data were analyzed in GraphPad Prism 7.

214
Neutralization assay. Twenty-thousand Vero.E6 cells were seeded per well in a 96-well cell culture plate 215 (Corning; 3340) 1 day prior to performing the assay. Serum samples were heat-inactivated at 56°C for 1 216 hour prior to use. Serum dilutions were prepared in 1× minimal essential medium (MEM; Gibco) 217 supplemented with 1% FBS. Each virus was diluted to 10,000 TCID 50 s/mL and 80 μL of virus and 80 μL of 218 serum were incubated together for 1 hour at RT. After the incubation, 120 μL of virus-serum mixture was 219 used to infect cells for 1 hour at 37°C. Next, the virus-serum was removed and 100 μL of each 220 corresponding dilution was added to each well. One hundred μL of 1X MEM was also added to the plates 221 to make a total volume of 200 μL in each well. The cells were incubated at 37°C for 3 days and then fixed 222 with 10% paraformaldehyde (Polysciences) for 24 hours. The next day, cells were stained using a rabbit 223 anti-N antibody (Invitrogen; PA5-81794) as primary and a goat anti-rabbit secondary conjugated to 224 horseradish peroxidase (Invitrogen; 31460). This protocol was adapted from an earlier established 225 protocol [34,35,42].

226
Antigenic cartography. A target distance from a serum to each virus is derived by calculating the 227 difference between the logarithm (log 2 ) reciprocal neutralization titer for that particular virus and the 228 log 2 reciprocal maximum titer achieved by that serum (against any virus). Thus, the higher the reciprocal 229 titer, the shorter the target distance. As the log 2 of the reciprocal titer is used, a 2-fold change in titer will 230 equate to a fixed change in target distance whatever the magnitude of the actual titers. Antigenic 231 cartography [23] (Smith et al 2004) was then used to optimize the positions of the viruses and sera relative 232 to each other on a map, minimizing the sum-squared error between map distance and target distance. 233 Each virus is therefore positioned by multiple sera, and the sera themselves are also positioned only by 234 their distances to the viruses. antigen they were raised against. The X and Y axes both correspond to antigenic distance, with one grid 426 line corresponding to a two-fold serum dilution in the neutralization assay. The antigens and sera are 427 arranged on the map such that the distances between them best represent the distances measured in 428 the neutralization assay.