SARS-CoV-2 RBD dimers elicit response comparable to VLPs in mice

Abstract We report the direct comparison of monomeric, dimeric and trimeric RBD protein subunit vaccines to a virus-like particle (VLP) displaying RBD. After two and three doses, a RBD dimer and trimer elicited antibody levels in mice comparable to an RBD-VLP. Furthermore, an Omicron (BA.1) RBD hetero-dimer induced neutralizing activity similar to the RBD-VLP. A RBD hetero-dimer and RBD-VLP also shows comparable breadth to other SARS-CoV-2 variants-of-concern (VOCs).

and trimers. We expressed genetic fusions of the RBD with SpyTag or SpyCatcher in yeast, and then covalently linked two or more complementary components by spontaneous overnight conjugation (Fig.  1a). This process yielded dimers and trimers of RBD or VLP's bearing RBD. All RBD molecules contained two mutations (L452K+F490W) that, in our previous work, improved both the manufacturability and immunogenicity of the ancestral RBD molecule 19 ; this engineered RBD also exhibited increased potency when displayed at high valency on a protein nanoparticle. We also constructed a heterotypic dimer of the RBD-L452K-F490W (RBD-J) and the Omicron (BA.1) RBD. Multimeric protein vaccines that include multiple RBD serotypes have exhibited increased breadth against SARS-CoV-2 VOCs 20,21 . We veri ed correct assembly of dimers, trimers, and VLPs by SDS-PAGE and biolayer interferometry (supplementary g. S1). As expected, the dimers, trimers and VLPs showed increased avidity compared to the monomeric RBD. (supplementary g. S1) We evaluated the immunogenicity of each design in BALB/c mice by immunizing groups of ve mice intramuscularly with 10µg of each RBD construct (Fig. 1a) formulated with 40µg of alum hydroxide. Separate groups of mice were immunized with either two doses at weeks 0 and 6, or three doses at weeks 0, 3, and 6. We collected and analyzed sera at weeks 0, 3, 6, 8, 10, and 24 ( Fig. 1b).
After three weeks, we observed that a prime dose of the RBD-VLP elicited a potent response and high titers of antibodies against the WA1/2020 RBD. Mice immunized with a prime dose of RBD dimers (both homo-and hetero-dimer) and trimer showed lower, though not statistically different, levels of Spikespeci c antibodies, compared to the RBD-VLP (Fig. 1c-d). The levels of Spike antibodies were signi cantly lower in mice immunized with a prime dose of either the RBD-monomer or SpyTag-RBD compared to the RBD-VLP at week 3.
A second dose of RBD led to seroconversion in all groups. At week 8 (after all doses), levels of Spikespeci c antibodies were not signi cantly inferior between mice immunized with the RBD dimers or RBD trimer compared to mice immunized with the RBD-VLP, for mice that received either two or three doses (supplementary Table 1-2). The levels of Spike-speci c antibodies in mice immunized with two doses of the RBD monomer or SpyTag-RBD monomer, however, remained signi cantly lower than the levels of antibodies in mice immunized with the RBD-VLP (supplementary Fig. 2, supplementary Table 1). When immunized with three doses, titers of Spike-speci c antibodies in mice immunized with the RBD monomer were statistically comparable to the RBD-VLP, although we observed a higher variance in response in this group.
We observed a similar trend for Omicron BA.1 Spike-speci c antibodies. Levels of Omicron BA.1 Spike antibodies in mice immunized with the RBD dimers or RBD trimer were not signi cantly inferior compared to mice immunized with the RBD-VLP, for mice that received either two or three doses ( Fig. 1c-d). Levels of antibodies against the Omicron BA.1 Spike in mice immunized with two doses of either RBD monomer were signi cantly lower compared to the RBD-VLP at week 8 (supplementary Fig. 2a). Mice immunized with three doses of the RBD monomers exhibited statistically comparable levels of antibodies against Omicron BA.1 Spike compared to the RBD-VLP.
We then measured the titers of antibodies that neutralized the WA1/2020 or Omicron BA.1 SARS-CoV-2 pseudoviruses. When immunized with two doses of the RBD dimers and trimer, nAb titers elicited against the WA1/2020 or Omicron BA.1 pseudovirus were not signi cantly inferior compared to the RBD-VLP group at week 8 (Fig. 1e). In contrast, nAb titers against the WA1/2020 or Omicron BA.1 pseudoviruses elicited by two doses of either RBD monomer were signi cantly lower compared to the RBD-VLP, as expected (Fig. 1e). When immunized with three doses, nAb titers against the WA1/2020 pseudovirus elicited by the RBD monomers, dimers, and trimer were all statistically similar to the RBD-VLP at week 8 (Fig.1f). In mice immunized with three doses, nAb titers against the Omicron BA.1 were only signi cantly different from the RBD-VLP in mice immunized with the SpyTag-RBD (Fig. 1f). These results are consistent with reports suggesting that two vaccine doses elicit poor neutralization against Omicron BA.1, and that a third dose is required to have neutralizing activity in humans 22 .
For all groups, Spike-binding antibodies were sustained through the end of the study at 24 weeks ( Fig. 1cd). Notably, neutralizing titers against the WA1/2020 and Omicron BA.1 pseudoviruses were also sustained for at least 6 months ( Fig. 1g-h). The antibody levels and neutralizing antibody titers sustained in mice immunized with the RBD dimers and trimer were not signi cantly inferior to the RBD-VLP.
Overall, Spike-speci c IgG levels had strong correlations with RBD-speci c IgG levels for both WA1/2020 Lastly, we evaluated breadth against several SARS-CoV-2 VOCs. We observed that levels of Spike binding antibodies against VOCs from two doses of the RBD dimers and trimer were statistically comparable to the RBD-VLP (Fig. 2a, Supplementary Table 5). Two doses of the monomeric RBD groups exhibited substantially lower coverage to VOCs (Fig. 2a). After three doses, we observed that the RBD dimers and trimer elicited similar coverage to VOCs compared to the RBD-VLP (Fig. 2b, Supplementary  Table 6). In this case, the Omicron (BA.1) Spike-speci c IgG response elicited by the hetero-dimer was highest among the three dose groups, although not signi cantly higher. Finally, we evaluated the durability of this response to more novel VOCs, such as Omicron sub-variants. We observed that RBD dimers, trimer and VLP sustained comparable levels of Spike protein speci c antibodies at week 24 ( Fig.   2c-d). Notably, the RBD hetero-dimer sustained the highest median IgG levels to Omicron sub-variants (Fig. 2d), although not signi cantly different from the RBD homo-dimer, trimer and VLP.
In this study, we have demonstrated that RBD dimers can elicit comparable responses to RBD-VLPs in mice after two or three doses with a conventional adjuvant (alum). These results support the utility of RBD dimers to elicit strong responses and breadth against SARS-CoV-2 variants. In particular, a heterodimeric presentation of RBDs including the Omicron BA.1 RBD, showed a slightly enhanced response against VOCs. Given the evolving viral landscape 23 , future studies should focus on evaluating these RBD dimers as booster-only alternatives in animal models with preexisting immunity. These RBD dimers can be modi ed or mixed to present RBD combinations that may lead to increased coverage to VOCs.
Furthermore, additional studies with novel adjuvants 24 could inform approaches to further potentiate the neutralizing response and durability of dimer-based antigens.
Additionally, questions remain about the durability of neutralizing responses to novel variants 2 afforded by currently approved vaccines to SARS-CoV-2. Alternative formulations for boosters could rely on candidates amenable to low-cost, scalable and transferable manufacturing that is accessible globally.
RBD dimers could also serve as a multivalent display platform for novel RBD variants either in heterotypic or cocktail forms 21 . Such candidates may also have utility for a low-cost seasonal 25 solution to decaying immunity and emerging variants.

Strains
Recombinant genes for RBD variant expression were codon optimized for Komagataella pha i expression, and synthesized cloned into a custom vector on a BioXP (Telesis Bio). Linear DNA was puri ed, and constructs were transformed into a modi ed wild-type K. pha i (NRRL Y-11430) as described previously 19 .

Cultivations
Strains producing material for further analytical characterization and mice immunization were cultivated in 400mL in 2L ba ed asks as described previously 19 . Cells were inoculated at 0.1 OD 600 , grown for 24 hours in complex media containing 4% glycerol, pelleted, resuspended, and grown for 24 hours in complex media containing 40 g/L sorbitol and 1.5% methanol.

Protein Puri cation
Puri cation of harvested cell culture supernatant containing the recombinant protein was performed using the downstream processing (DSP) module of InSCyT as described previously 26 . The supernatant pH was adjusted inline to pH 5.0 with 100mM citric acid. The supernatant was loaded into a 5mL prepacked CMM HyperCel column (Sartorius AG, Gottingen, Germany). The column was re-equilibrated with 20mM sodium citrate, pH 5.0, washed with 20mM sodium phosphate pH 6.5, and elution was initiated with 20mM sodium phosphate, 300mM NaCl, pH 8.5. Eluate from column 1 above 20mAu was loaded into a 1mL prepacked HyperCel STAR AX column (Sartorius AG, Gottingen, Germany). Flowthrough material above 20mAu was collected.
Approximately 0.1 µg of each sample was loaded on 12% Tris-glycine gel (Invitrogen, WXP01220BOX) run for 90 min at 140 V in 1X Tris-glycine SDS running buffer (Invitrogen, LC2675). Novex™ PageRuler prestained protein ladder (Invitrogen, 26616) was used as marker to estimate molecular weight of the proteins. The gel was stained using InstantBlue® Coomassie Protein Stain (Abcam, ab119211) for 60 minutes and destained with water for at least another 60 minutes Biolayer interferometry (BLI) Biolayer interferometry was performed using the Octet Red96 with Protein A (ProA) biosensors (Sartorius ForteBio, Fremont, CA) as previously described 26 . ProA biosensors were hydrated for 20 min in kinetics buffer prior to each run. Kinetics buffer comprising 1X PBS pH 7.2, 0.5% BSA, and 0.05% Tween 20 was used for all dilutions, baseline, and disassociation steps. ACE2-Fc was loaded at a concentration of 10 µg/mL. Samples were loaded in a 96-well black microplate (Greiner Bio-One, Monroe, NC) at starting concentrations of 15 µg/mL for ACE2-Fc binding. Association and dissociation were measured at 1000 rpm for 300 and 600 seconds, respectively. Binding a nity was calculated using the Octet Data Analysis software v10.0 (Pall ForteBio), using reference subtraction, baseline alignment, inter-step correction, Savitzky-Golay ltering, and a global 1:1 binding model.

Preparation of vaccine materials
Puri ed RBD-SpyTag was conjugated onto Hepatitis B surface antigen (HBsAg) SpyCatcher VLPs overnight at a 1.2:1 RBD to HBsAg molar ratio. For RBD dimers, RBD-SpyTag and RBD-SpyCatcher were conjugated overnight at a 1:1.2 molar ratio. For RBD trimers, SpyTag-RBD-SpyTag and RBD-SpyCatcher were conjugated overnight at a 1:2.5 molar ratio. Conjugated RBD-VLP was buffer exchanged and concentrated with 100K molecular weight cutoff Pierce™ protein concentrators (Thermo Fisher, 88524).
Conjugated RBD dimers and trimers were buffer exchanged and concentrated using 50K molecular weight cutoff Pierce™ protein concentrators (Thermo Fisher, 88539). RBD monomer and RBD-SpyTag were buffer exchanged and concentrated using 10K molecular weight cutoff Pierce™ protein concentrators (Thermo Fisher, 88517).
Materials were formulated with a 10mM Histidine, 20mM Sodium phosphate, 5mM Tris, 37.5mM NaCl, pH 7.4 buffer. All RBD groups were diluted to a nal total protein concentration of 100µg/mL. Alhydrogel (Invivogen, vac-alu-250) was added to a nal concentration of 400µg/mL.

Animals and study design
Female C57BL/6 mice (The Jackson Laboratory) 4-6 weeks of age were randomly assigned to experimental groups. Mice were immunized with 10µg of the indicated RBD formulation, adjuvanted with 40µg of alum hydroxide, via intramuscular injection (n = 5/group). Separate groups of mice were administered either a two-dose regimen, with injections at weeks 0 and 6, or a three-dose regimen, with injections at weeks 0, 3, and 6. At the indicated timepoints, peripheral blood was collected via the submandibular route and processed to isolate serum for immunologic or virologic assays. All animal studies were conducted in compliance with relevant local, state, and federal guidelines. All animal studies were reviewed and approved by the Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee.
Electrochemiluminescence assay (ECLA) ECLA plates (MesoScale Discovery SARS-CoV-2 IgG, Panels 22,23) were designed and produced for multiplex binding assays with up to 10 antigen spots in each well, including either Spike or RBD proteins from multiple SARS-CoV-2 variants. Plates were blocked with 150µL of Blocker A solution (1% bovine serum albumin (BSA) in distilled water) for at least 30m at room temperature with shaking at 700 rpm on a digital microplate shaker. The plates were then washed 3 times with 150µL of wash buffer (0.5% Tween in 1x PBS), blotted dry, and 50µL of serum samples, pre-diluted 1:5,000 in Diluent 100 (MesoScale Discovery), were added in duplicate and the plates were set to incubate with shaking at 700rpm at room temperature for 2h. Secondary antibody was prepared using Jackson Immuno Rabbit Anti-Mouse IgG detection antibody (Cat No: 315-005-045) conjugated to the MSD GOLD SULFO-TAG by NHS-Ester chemistry per the manufacturer's guidelines (Cat No: R91AO-1).
The plates were washed 3 times and 50µL of tagged secondary antibody diluted 1:500 in Diluent 100 was added to each well and incubated shaking at 700 rpm at room temperature for 1 h. Plates were then washed 3 times and 150µL of MSD GOLD Read Buffer B (MesoScale Discovery) was added to each well and the plates were read immediately after on a MESO QuickPlex SQ 120 machine. Antibody levels for each sample were reported as Relative Light Units (RLU) which were calculated as the average Sample RLU minus Blank RLU for each sample. 2x106 RLU was de ned as the upper limit of detection. For samples with signal exceeding this value, samples were pre-diluted further, assays were repeated; the resultant RLU value was multiplied by the additional dilution factor to correct for this dilution before analysis. 1000 RLU was de ned as the lower limit of detection for each assay.

Pseudovirus neutralization
The SARS-CoV-2 pseudotyped viruses were generated using human embryonic kidney (HEK) 293T cells (American Type Culture Collection CRL_3216) which were co-transfected with a spike protein expressing To measure neutralizing activity of the serum, HEK293T-human angiotensin-converting enzyme 2 (hACE2) cells were seeded at a density of 2.0 × 104 in 96-well tissue culture plates overnight. Threefold serial dilutions of heat-inactivated serum samples were prepared and mixed with 60 µl of pseudovirus and incubated at 37°C for 1 hour. This mixture was then added to the seeded HEK293T-hACE2 cells, and 48 hours later, Steady-Glo Luciferase Assay (Promega) was used to lyse cells according to the manufacturer's instructions. SARS-CoV-2 neutralization titers were determined as the sample dilution at which a 50% reduction (NT50) in relative light units (RLUs) was observed relative to the average of the virus control wells.