Nanoparticle display of prefusion coronavirus spike elicits S1-focused cross-reactive protection across divergent subgroups

Abstract Multivalent antigen display is a fast-growing area of interest towards broadly protective vaccines. Current nanoparticle-based vaccine candidates demonstrate the ability to confer antibody-mediated immunity against divergent strains of notably mutable viruses. In coronaviruses, this work is predominantly aimed at targeting conserved epitopes of the receptor-binding domain. However, targeting other conserved non-RBD epitopes could further limit the potential for antigenic escape. To further explore new potential targets, we engineered protein nanoparticles displaying CoV_S-2P trimers derived from MERS-CoV, SARS-CoV-1, SARS-CoV-2, hCoV-HKU1, and hCoV-OC43 and assessed their immunogenicity in mice. Monotypic SARS-1_S-2P nanoparticles elicited cross-neutralizing antibodies against MERS_S and protected against MERS-CoV challenge. MERS and SARS-I53_dn5 nanoparticles elicited S1-focused antibodies, revealing a conserved site on the NTD. Moreover, mosaic nanoparticles co-displaying distinct CoV_S-2P trimers elicited antibody responses to distant cross-group antigens while protecting against MERS challenge despite diminished valency of MERS_S-2P. Our findings will inform further efforts towards the development of pan-coronavirus vaccines.


19
Coronaviruses (CoVs) comprise a broad family of viruses with diverse strains known to 20 infect mammals and birds. Betacoronaviruses (-CoVs) in particular thrive in animal reservoirs 21 and represent a constant threat to human health. Inclusive of the most recently emerged -CoV,

22
SARS-CoV-2, there are seven CoVs known to infect humans (human CoVs or hCoVs); four of 23 which circulate endemically 1, 2 . Although endemic hCoV infections typically manifest as mild 24 respiratory disease, zoonotic spillover of β-CoVs into human populations has been associated with high disease morbidity and mortality, economic burden, and widespread global epidemics 3 .

26
Since the emergence of SARS-CoV-2 in late 2019, over 600 million infections and nearly 7 27 million deaths worldwide 4 have been attributed to COVID-19, the disease caused by SARS-

28
CoV-2, underscoring the need for broadly protective CoV vaccines. In addition to the four 29 endemic human coronaviruses-hCoV-OC43, hCoV-HKU1, hCoV-229E, and hCoV-NL63-the 30 emergence of the epidemic-and pandemic-causing SARS-CoV-1, MERS-CoV, and SARS-CoV-31 2 from zoonotic spillover events has resulted in outbreaks with substantial morbidity and 32 mortality 2, 5 . Evidenced by the ongoing pandemic, SARS-CoV-2 will continue to evolve and is 33 likely to become the next endemic human coronavirus. The emergence of SARS-CoV-2 variants 34 of concern (VOC), particularly the delta and omicron VOC (Pango naming of B.1.617.2 and 35 B.1.1.529, respectively), highlights the threat of viral escape from antibodies induced by the 36 currently available COVID-19 vaccines and the need for next-generation vaccines that are 37 capable of inducing broadly protective immunity against a wide range of CoVs 6 . Recently, there 38 has been documentation of broad cross-reactivity either deriving from natural infection 7 , or from 39 immunogens delivering multiple sarbecoviral antigens 8,9,10 . There is much interest in designing 40 immunogens to target antibody responses to domains of the S1 subunit at the apex of the spike 41 (S) protein 11, 12 -particularly the Receptor Binding Domain (RBD), as this region is targeted by 42 potently neutralizing antibodies (nAbs) 13 . However, as evidenced by the omicron VOC, the RBD 43 is particularly susceptible to mutation and recombination leaving room for immune evasion via 44 antigenic drift and shift. Others have previously shown elicitation of broad protection from 45 influenza through the multivalent display of hemagglutinin on the I53_dn5 nanoparticle 46 platform 14 . Applying lessons learned from these studies here, we describe nanoparticles that 47 display spike trimer from diverse coronaviruses. We show that CoV prefusion-stabilized S 48 (CoV_S-2P) trimers displayed on I53_dn5 self-assembling nanoparticles are able to elicit 49 broadly cross-reactive and protective antibody responses. Furthermore, by co-displaying these diverse spikes in a mosaic antigen array, we induce robust and protective immunity even at low 67 trimers and nanoparticles were tested for antigenicity by ELISA using monoclonal antibodies 68 specific to each CoV S. Antibody binding was comparable between soluble trimer and 69 nanoparticle in each case indicating that antigenicity is similarly intact in each formulation ( Fig.   70 1b). The purified nanoparticles were also imaged by negative stain electron microscopy (NS-71 EM) showing that the nanoparticles were well-assembled and homogeneous, displaying highly-72 ordered S proteins (Fig. 1d). These characterization data show that, we were able to efficiently 73 express, and assemble antigenically intact nanoparticle immunogens. We hypothesized that multivalent display of CoV S-2P trimers would elicit a more robust 78 antibody response than soluble S-2P, improving potency of potentially cross-reactive antibodies 79 specific to broadly conserved yet potentially subdominant epitopes. To test this, we immunized a Computer-generated models of prefusion-stabilized spike trimers (S-2P) from MERS, SARS-1, and SARS-2, and their homotypic display on I53-dn5 nanoparticle. Icosahedral nanocage displays 20 trimers. b Trace profiles of S-2P_dn5B trimer and _dn5 nanocage purification by size exclusion chromatography. c Antigenicity ELISA comparing binding of antibodies specific for MERS, SARS-1, or SARS-2_S-2P to soluble trimer (triangles) or dn5 assembly (circles) respectively. d Representative images of CoV-S-2P_dn5 at 50,000x magnification and 2D class averages. Scale bars correspond to 100 nm (representative images) and 20 nm (2D class averages).

91
To assess the quality of the vaccine-matched and cross-reactive antibody responses, we 92 generated correlation plots of neutralizing-to-binding antibody titers for SARS-1, SARS-2, and 93 MERS, in which the slope indicates the ratio of neutralizing to binding activity. Because SARS-

96
1_S-2P binding titers while neutralization remained intact. This suggests that particle assembly

108
We also performed a dose escalation study in which C57BL/6 mice were immunized at a-f Groups of 10 female BALB/c were immunized at weeks 0 and 3 with 10 µg of SARS-1_S-2P as a soluble trimer or displayed on I53_dn5 particles, MERS_S-2P trimer or H1 trimer displayed on I53_dn5 nanoparticles with SAS adjuvant and bled at week 5 for serology. Control mice were immunized with H1_dn5. a-c Serum was screened for binding by ELISA to SARS-1_, SARS-2_, and MERS_S-2P. d-f Serum was then assessed for its capacity to neutralize SARS-1, SARS-2, and MERS pseudoviruses. g-i To plot the potency of neutralizing antibodies, correlation plots of binding (x-axis) to neutralization (y-axis) where the slope (neutralization/binding) indicates the ratio of neutralizing to binding antibody titers were generated. b-g Boxes and horizontal bars denote the interquartile range (IQR) and medians, respectively. Whisker endpoints are equal to the minimum and maximum values. Statistical analysis was performed using non-parametric Kruskal-Wallis test with Dunn's multiple comparisons. *P<0.05, **P < 0.01, ****P < 0.0001.

153
We hypothesized that after nanoparticle assembly, crowding of the CoV spike antigens  Fig. 5b), suggesting that access to S2 is restricted on I53_dn5 and may bias 163 towards selection of B cells specific to the S1 domain. To see if antibodies cross-reactive to S1 164 could be visualized by EM, sera from mice immunized with SARS-1_I53_dn5 were pooled to 165 isolate IgG, which was then digested with papain to generate antigen-binding fragments (Fabs).

166
We assembled immunocomplexes of SARS-1_I53_dn5-elicited polyclonal Fabs bound to

202
indicating that despite reduced valency of OC43_S-2P per particle, the mosaic was able to 203 retain its contribution of cross-reactive breadth 14 .

206
To determine the protective efficacy of CoV S-2P immunogens using the I53_dn5

231
Nonetheless, these trends indicate that display of heterotypic hCoV_S-2P on the I53_dn5  prefusion coronavirus spikes, we suspect that the apparent S2-occlusion/S1-targeting may also 264 occur with other CoV-S_dn5 formulations, potentially eliciting potent S1-directed cross-265 neutralization in other coronaviruses as well-particularly in the -mosaic that may exploit 266 vulnerable S1 epitopes. However, we propose that it may be possible to improve access to S2 267 epitopes, with nanoparticle scaffolds that appropriately space the displayed spike antigens.

292
MERS S1 depletion did not appear to impact binding to HKU1_S-2P, suggesting that any

299
Other studies have reported intrasubgroup cross-reactivity through natural infection 7 or 300 via nanoparticle-based immunogens using mosaic co-display of RBDs from diverse SARS-like 301 CoV S proteins 9, 10 . We chose to display the full-length S ectodomain because it comprises the 302 full complement of epitopes exposed on the spike on the viral surface, including the highly 303 conserved S2 fusion machinery, and therefore provides a greater chance of identifying cross-304 reactive epitopes between diverse CoVs. However, the finding that S2-specific antibodies were 305 not induced suggests that BCR access may have been limited to the apex of S. New designs 306 with fewer S-2P displayed or larger nanoparticles to change the geometry of antigen 307 presentation and improve access to S2 could increase the level cross-reactivity induced.

308
Additionally, employing an S2-only display of diverse hCoV spikes and employing our mosaic 309 approach may increase recognition of conserved epitopes on this subdomain. Nonetheless, our 310 results confirm the presence of conserved protective epitopes between divergent subgroups on 311 the NTD and potentially either RBD or other subdomains that can be targeted by next-312 generation immunogens.

313
The β-CoV_I53_dn5 mosaic nanoparticle, which has a relatively low valency of 314 MERS_S-2P trimers, induced protective immunity against MERS-CoV lethal challenge in mice, 315 comparable to protection elicited by monotypic MERS_I53_dn5 (Fig. 6). Further studies will 316 need to evaluate the extent to which this mosaic nanoparticle protects against other CoV 317 challenges and establish the minimum valency of S-2P content needed to confer protection in 318 both homotypic and heterotypic challenge models. Future research will also be needed to map 319 additional cross-reactive epitopes and B cell responses following CoV_I53_dn5 immunization, 320 and additional experimentation will be required to define the optimal spacing and composition of

332
The ongoing COVID-19 pandemic, caused by SARS-CoV-2, highlights the need for 333 protection against pre-emergent zoonotic CoV threats in addition to continually emerging SARS-

334
CoV-2 VOC. The mosaic display of S-2P antigens on nanoparticles may serve as the basis for  Paired VH and VL in a 1:1 ratio were co-transfected transiently into FreeStyle293F cells as 364 previously described. The supernatant was harvested six days post-transfection and IgGs were