Protective prototype-Beta and Delta-Omicron chimeric RBD-dimer vaccines against SARS-CoV-2

Summary Breakthrough infections by SARS-CoV-2 variants become the global challenge for pandemic control. Previously, we developed the protein subunit vaccine ZF2001 based on the dimeric receptor-binding domain (RBD) of prototype SARS-CoV-2. Here, we developed a chimeric RBD-dimer vaccine approach to adapt SARS-CoV-2 variants. A prototype-Beta chimeric RBD-dimer was first designed to adapt the resistant Beta variant. Compared with its homotypic forms, the chimeric vaccine elicited broader sera neutralization of variants and conferred better protection in mice. The protection of the chimeric vaccine was further verified in macaques. This approach was generalized to develop Delta-Omicron chimeric RBD-dimer to adapt the currently prevalent variants. Again, the chimeric vaccine elicited broader sera neutralization of SARS-CoV-2 variants and conferred better protection against challenge by either Delta or Omicron SARS-CoV-2 in mice. The chimeric approach is applicable for rapid updating of immunogens, and our data supported the use of variant-adapted multivalent vaccine against circulating and emerging variants.


In brief
Vaccine immunogens designed as chimeras of the spike protein receptorbinding domain of two distinct SARS-CoV-2 variants elicit broad serum neutralization and protection from infection by the Beta, Delta, and Omicron variants in mice and macaques. The chimeric approach is applicable for rapid updating of immunogens against both circulating and emerging variants.

INTRODUCTION
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants are continually emerging, and they become the circulating strains in the world (www.who.int). Several highly transmis-sible variants of concern (VOCs) showed altered pathogenicity and transmission (Karim and Karim, 2021;Krause et al., 2021;Viana et al., 2022) bringing new waves of infection worldwide. The immune escape of the variants to the current approved vaccines and, concomitantly, the breakthrough infections have become Figure 1. Antigenic and structural characterization of homotypic and chimeric RBD-dimers (A) Schematic diagram of prototype RBD-dimer, Beta RBD-dimer, prototype-Beta chimeric RBD-dimer, and Delta-Omicron chimeric RBD-dimer. Two SARS-CoV-2 RBDs were dimerized as tandem repeat (SP, signal peptide). (B) Footprint of hACE2 and five classes of antibodies on SARS-CoV-2 RBD. Prototype RBD is shown as gray surface (PDB: 6LZG). Residues that mutated in SARS-CoV-2 Beta and Delta variants are colored in red and blue, respectively. Footprints of hACE2 (PDB: 6LZG), CB6 (PDB: 7C01), CV07-270 (PDB: 6XKP), C110 (PDB: 7K8V), S309 (PDB: 6WPT), and CR3022 (PDB: 6W41) are highlighted in pink, blue, yellow, orange, cyan, and green, respectively. (C) Footprint of hACE2 and five classes of antibodies on SARS-CoV-2 RBD. Prototype RBD is shown as gray surface. Residues that mutated in SARS-CoV-2 Omicron variants are colored in red. (D) Binding affinities of antigens bound to hACE2 and representative mAbs targeting five major sites. Red indicates no binding (N/B). The ones with affinity reductions more than 100-fold are colored in yellow.
(legend continued on next page) ll OPEN ACCESS the global challenge to end the pandemic (Liu et al., 2021a). Currently, all approved vaccines are based on prototype SARS-CoV-2. Given the circulating strains of SARS-CoV-2 are fast changing, it is debated yet in the field that what our vaccine strategy is going forward? Should it be the prototype boosting, or using variant-specific vaccine, or using multivalent vaccine with broader coverage of both circulating and emerging variants?
Coronavirus spike protein mediated virus entry via its receptor-binding domain (RBD) Wang et al., 2020). RBD is the major target for neutralizing antibodies (NAbs) and a favorable antigen for vaccine development because of its immunodominance . We previously designed novel vaccines against Beta coronavirus using tandem repeat dimeric RBD as antigens, which substantially enhanced immunogenicity in animal model (Dai et al., 2020). Based on this strategy, we developed protein subunit vaccine, ZF2001, against coronavirus diseases 2019 (COVID-19) , which has received conditional marketing authorization in China and emergency use authorization in Uzbekistan, Indonesia, and Columbia. ZF2001 is rolling out in the vaccination campaign with more than 250 million doses administrated globally. Phase 3 clinical trials showed that ZF2001 has more than 80% efficacy in preventing symptomatic COVID-19 . ZF2001 vaccine uses RBD sequence from Wuhan-Hu-1 strain (prototype) (Wei et al., 2020) and elicits antibody responses in human with varying neutralizing activities to SARS-CoV-2 variants Zhao et al., 2021). Before the emergence of Omicron variant, Beta variant contains the most mutations in the spike protein and showed the largest reduction in sensitivity to vaccines or neutralizing monoclonal antibodies (mAbs) (Abdool Karim and de Oliveira, 2021;Garcia-Beltran et al., 2021;Liu et al., 2021b;Lucas et al., 2021;Shen et al., 2021;Tegally et al., 2021;Wall et al., 2021;Wang et al., 2021aWang et al., , 2021bWibmer et al., 2021;Wu et al., 2021). Similar conclusion was also observed in vaccinees who received ZF2001 vaccine . The recent occurring Omicron variant showed more severe immune escape than the Beta variant (Cameroni et al., 2022;Cele et al., 2022;Dejnirattisai et al., 2022;Liu et al., 2022;VanBlargan et al., 2022), and accordingly, the neutralizing activities of ZF2001-elicited human sera were 3.1-to 10.6-fold lower against recently circulating Omicron variant than prototype (Zhao et al., 2022). Given the positive correlation between neutralizing titer and protection efficacy (Khoury et al., 2021), the sera neutralization against SARS-CoV-2 variants is believed to impact vaccine effectiveness.
When the Beta emerged as the most resistant VOC in early 2021, we first generated prototype-Beta chimeric RBD-dimers to cover both prototype and Beta linages. The antigenic characterization and structural analysis of RBD-dimer immunogens were performed to validate the major epitopes displayed. The neutralizing activities of vaccine-elicited mice sera against a panel of SARS-CoV-2 variants were further tested. The protection efficacy was evaluated in mice. Given the outperformance of the prototype-Beta chimeric RBD-dimer in comparison with its homotypic counterparts, we produced the prototype-Beta chimeric antigen in pilot scale for preclinical efficacy study in rhesus macaques. Vaccine-elicited humoral and cellular immune responses were tested, and subsequently, the protective efficacies against challenge from prototype, Beta and Delta SARS-CoV-2, were further evaluated. Over the next few months, Delta and Omicron became the major circulating variants one after another. Vaccines inducing broader immune responses, with particularly high activity against both Delta and Omicron variants, are urgently needed. Therefore, using the chimeric approach, we rapidly developed the Delta-Omicron RBD-dimer vaccine. The antigenic integrity and cryoelectron microscopy (cryo-EM) structure were determined; the immunogenicity and protective efficacy were further evaluated, demonstrating the advantage and generalizability of the chimeric vaccine approach in countermeasure of SARS-CoV-2 variants. This is a proof of concept for rapid updating of immunogens based on the chimeric approach, and our data supported the WHO's advice to develop multivalent and variant-adapted COVID-19 vaccines ensuring the breadth of the immune response against circulating and emerging variants (WHO, 2022).

RESULTS
Development of prototype-Beta vaccine based on the chimeric RBD-dimer vaccine design strategy The chimeric RBD-dimer vaccine design strategy was first tested in the surge of Beta variant in early 2021. The RBD of Beta variant has three mutations: K417N, E484K, and N501Y ( Figure S1A). To adapt Beta variant, we designed two immunogens based on the original prototype RBD-dimer: (1) two copies of Beta RBD and (2) the C-terminal prototype RBD linked N-terminal Beta RBD as a prototype-Beta chimera (Figure 1A). These immunogens were expressed in Expi293F cells and purified as single dimer-sized proteins (molecular weight $60 kDa) as verified by analytical gel filtration and gel electrophoresis (Figures S1B-S1D). Next, surface plasmon resonance (SPR) assay was performed to verify the exposure of both receptor-binding motif (RBM) and major antigenic sites, using receptor protein of human angiotensin converting enzyme (hACE2)  and SARS-CoV-2 RBD-specific mAbs as probes. Representative mAbs target five major antigenic sites with available structures (Figure 1B;Barnes et al., 2020;Kreye et al., 2020;Pinto et al., 2020;Shi et al., 2020;Yuan et al., 2020). Monomeric prototype and Beta RBD proteins were tested for comparison. As expected, all these immunogens showed similar binding affinities to hACE2 (apparent K D ranging from 1.13 to 2.66 nM) ( Figures 1D and S2A). Compared with the prototype RBD, both Figure 2. Immunogenicity and protection efficacy of homotypic and prototype-Beta chimeric RBD-dimers Two batches of 8-to 10-week-old female BALB/c mice (n = 8 each batch) were immunized with two doses of 0.5-mg prototype, Beta or prototype-Beta chimeric RBD-dimer using AddaVax as adjuvant, 21 days apart. PBS plus adjuvant was given as the sham control. The first batch of BALB/c mice (n = 8) receiving two doses of vaccine or sham were challenged with Beta variant or prototype SARS-CoV-2 via the i.n. route at 46 and 61 days, respectively, post the second vaccination. (A) Sera collected at 14 days post the second immunization were tested for neutralization of a panel of pseudotyped viruses displaying prototype, Alpha, Beta, Gamma, Delta, Delta plus, Kappa, Lambda, and Omicron spikes. The values are the GMT ± 95% confidence interval (CI). The horizontal dashed line indicates the lower limit of detection (LLOD). (B) Radar plot demonstrating the neutralization profile of sera elicited by prototype, Beta or prototype-Beta chimeric vaccine against eight SARS-CoV-2 pseudotyped viruses. (C-H) Random selection of four mice in each group were challenged with 5 3 10 5 TCID 50 of prototype SARS-CoV-2 (GISAID: EPI_ISL_514256-7) (C-E) and the other four were challenged with 1 3 10 6 TCID 50 of Beta variant (GDPCC-nCoV84 strain) (F-H). Mice challenged with prototype SARS-CoV-2 had received Ad5-hACE2 intranasally 5 days before.
(legend continued on next page) ll OPEN ACCESS monomeric and dimeric forms of Beta RBD showed dramatically decreased binding affinities to mAb CB6 (no detectable binding), CV07-270 (>100-fold reduction), and C110 (>100-fold reduction), respectively, demonstrating the antigenic changes. By contrast, both prototype and prototype-Beta chimeric RBDdimer preserved the binding activities to almost all tested mAbs ( Figures 1D and S2A). This is because in the prototype-Beta chimeric design, the prototype domain can bind well with all these mAbs, including CB6.
Cryoelectron microscopy (cryo-EM) structural analysis of RBD-dimers For further structural characterization, we determined the cryo-EM structures of fragment antigen binding (Fab) of CB6 in complex with either prototype or prototype-Beta chimeric RBDdimer. Since the two copies of RBD tethered by the flexibly N and C termini, both the homotypic and chimeric RBD-dimers showed a variety of conformations, precluding the structure analysis at high resolution ( Figures S3A-S3F). After image reconstruction by two-dimensional (2D) classification and 3D refinement, the prototype and prototype-Beta chimeric RBD-dimers bound to CB6 Fabs were determined to 11.5-and 11.6-Å resolution of mass density maps, respectively, into which the previously published atomic models of SARS-CoV-2 RBD and RBD/ CB6 complex were fitted and rebuilt ( Figures S3A-S3F). The rebuilt models showed their overall conformations. Like MERS-CoV RBD-dimer structure as we reported earlier (PDB: 7C02) (Dai et al., 2020), both prototype and prototype-Beta chimeric SARS-CoV-2 RBD-dimers arranged as ''bilateral lung''-like structures with axial symmetry ( Figures 1E, 1F, S4D, and S4E). Two RBD subunits stack together via core domains of each other with their external domains exposed. As expected, prototype RBD-dimer is in complexed with dual CB6 Fabs at both arms ( Figure 1E). By contrast, prototype-Beta chimeric RBD-dimer only engages one CB6 Fab at one arm, which is consistent with the fact that Beta RBD with K417N mutation is resistant to CB6 binding (Figure 1F;Starr et al., 2021;Wang et al., 2021b). In summary, the antigenic characterization and structural analysis demonstrated that the recombinant RBD-dimers correctly presented RBM and major neutralizing epitopes.
Immunogenicity and protection efficacy of prototype-Beta chimeric RBD-dimer vaccine in mice To assess the vaccine immunogenicity, groups of BALB/c mice (two batches pooled together) were administrated with two doses of 0.5 mg prototype, Beta or prototype-Beta chimeric RBD-dimer, adjuvanted with AddaVax (a squalene-based adjuvant), 21 days apart. PBS plus adjuvant was given as sham con-trol. Mouse serum samples were collected 14 days after the second dose. We used a panel of pseudotyped viruses (vesicular stomatitis virus backbone) displaying SARS-CoV-2 spikes to test the sera neutralizing activities against variants, including the prototype (Wuhan-Hu-1 strain), all five VOCs (Alpha [B.1.1.7 [C.37]) ( Figure S1A). Two doses of prototype vaccine elicited high levels of NAbs against pseudovirus displaying prototype spike, with geometric mean titer (GMT) of 50% pseudovirus neutralization titer (pVNT 50 ) up to 1,779 (Figures 2A and S5A). The pVNT 50 GMT exhibited varying reductions against pseudovirus displaying variant spikes (Figures 2A and S5A). Antisera showed slight decline in neutralization of pseudovirus displaying Alpha spike (GMT, 1,479) but exhibited more pronounced reductions in neutralization of pseudoviruses displaying Beta (GMT, 487), Gamma (GMT, 699), Delta (GMT, 1,100), Delta plus (GMT, 1,119), Kappa (GMT, 1,205), Lambda (GMT, 807), and Omicron (GMT, 122) spikes ( Figure 2A). Omicron variant contains the most mutations in RBD (15 sites) ( Figure S1A) and is the most resistant to sera neutralization (14.6-fold reduction). Beta and Gamma variants, containing similar three mutations at RBD (K417N/T, E484K, and N501Y) ( Figure S1A), were the other two with large (2.5-to 3.7-fold) reductions. On the contrary, Beta RBD-dimer vaccine induced enhanced neutralizing responses against pseudotyped viruses displaying Beta and Gamma spikes (GMT 809 and 1,650, respectively) ( Figure 2A) but largely reduced responses against prototype and other variant pseudoviruses (GMTs ranging from 104 to 385) ( Figure 2A). Encouragingly, prototype-Beta chimeric RBD-dimer vaccine elicited balanced antibody responses to neutralize Omicron (GMT 434) and all other pseudotyped viruses tested (GMTs ranging from 1,140 to 2,964). It outperformed prototype vaccine to elicit neutralizing responses against Beta, Gamma, and Omicron pseudotyped viruses (Figure 2A) and was also superior to the Beta vaccine to elicit antisera in neutralization of pseudotyped virus displaying prototype, Alpha, Delta, Delta plus, Kappa, Lambda, or Omicron spikes ( Figure 2A). More intuitionally, the radar plot showed that prototype-Beta chimeric RBD-dimer vaccine induced the broadest and most balanced neutralizing activity against SARS-CoV-2 variants, compared with homotypic RBD-dimer vaccines ( Figure 2B).
To further explore the protective efficacy of these RBD-dimers, the first batch of eight mice in each group were evaluated for protection against SARS-CoV-2 infection. BALB/c mouse is not sensitive to prototypic SARS-CoV-2 infection because of the low binding affinity between mouse ACE2 and S protein (C) Pulmonary viral gRNA levels were detected by qRT-PCR. For (C, D, F, and G), shown are the box and whiskers plots of 25th to 75th percentile with median as center and whiskers of minimum to maximum percentile. p values were analyzed with two-tailed unpaired t test (ns, p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). For (E and H), red and gray lines indicate linear regression line and 95% CI, respectively. r and p values represent Spearman's correlation coefficients and corresponding two-sided p values, respectively. Symbols represent individual mouse and may overlap for equal values. Horizontal dashed lines indicated the LLOD. but becomes sensitive to Beta variant due to the promoted affinity by N501Y mutation at S protein (Niu et al., 2021). Therefore, four mice in each group were transduced via intranasal (i.n.) route with adenovirus (Ad5) expressing hACE2 and, 5 days later, challenged intranasally with prototypic SARS-CoV-2 (hCoV-19/ China/CAS-B001/2020 strain). The other four mice in each group were challenged directly with Beta SARS-CoV-2 variant (GDPCC-nCoV84 strain) (Meng et al., 2021) via i.n. route. Mice were euthanized and necropsied at 5 days postinfection (DPI) to quantify viral genomic (g)RNA and subgenomic (sg)RNA, an indicator of viral replication (Wö lfel et al., 2020), in lung. For mice challenged with prototype SARS-CoV-2, high levels of both viral gRNA (average: 1.72 3 10 9 copies/g) and sgRNA (average: 1.9 3 10 8 copies/g) were detected in sham-immunized mice ( Figures 2C and 2D). By contrast, significantly reduced viral loads (p < 0.0001) were detected in vaccine-immunized mice. Averages of pulmonary viral gRNA were 4.61 3 10 5 , 2.58 3 10 6 , and 2.66 3 10 5 copies/g in prototype, Beta and prototype-Beta chimeric vaccine groups, respectively, with 2-4log 10 reduction compared with the sham group ( Figure 2C). In line with the trends in neutralization, both prototype and prototype-Beta chimeric vaccine groups showed significantly lower viral gRNA compared with the Beta vaccine group (p = 0.016 and 0.0045, respectively) ( Figure 2C). All vaccine groups showed undetectable pulmonary viral sgRNA, indicating the complete control of viral replication ( Figure 2D). Analysis of immune correlates of protection following vaccination showed that NAb titers correlated strongly with the reduction of pulmonary prototypic SARS-CoV-2 gRNA based on a linear model (r = À0.8967, p < 0.0001) ( Figure 2E).
To provide a further assessment of protection following vaccination, mice in each group were assessed for virus-related pathology in lung at 5 DPI. Tissue sections were stained with hematoxylin and eosin (H&E) to examine histopathology. Sham recipients challenged with either prototype virus or Beta variant showed moderate-to-severe histopathological changes in lung, including vanishment of alveolar cavities, pulmonary vascular congestion, and diffuse inflammatory cell infiltration ( Figure S6A). In contrast, mice vaccinated with prototype, Beta or prototype-Beta chimeric RBD-dimer vaccine exhibited relieved lung injury ( Figure S6A).
Moreover, histopathology showed that both Beta and prototype-Beta chimeric vaccines provided better protection compared with prototype vaccine when mice were challenged with SARS-CoV-2 Beta variant ( Figure S6A). The histopathology result was consistent with the tendency of pulmonary viral gRNA shown above and demonstrated the more balanced protection conferred by prototype-Beta chimeric RBD-dimer vaccine.
Immunogenicity of prototype-Beta chimeric RBD-dimer vaccine in rhesus macaques We next selected the prototype-Beta chimeric RBD-dimer immunogen for further preclinical study. Pilot scale immunogen protein was produced in clinical-grade CHO cell line and formulated with alum-based adjuvant as vaccine. Alum-only buffer was produced as the sham control.
Twenty-four healthy young rhesus macaques (Table S1) were immunized intramuscularly with three jabs of 25 mg vaccine (n = 12), a dose used in human , or sham (n = 12), 21 days apart. Serum samples were collected before (day 0) and after (days 14, 35, and 52) priming ( Figure S5B). The average endpoint titers of RBD-binding antibody raised from 512 after one dose to 46,341 after two doses and further up to 73,562 after three doses in the vaccine group ( Figure 3A). The sera after three doses were tested for neutralization of pseudotyped viruses expressing four VOCs (Figures 3B and S5C). Sera from vaccinated animals showed robust and balanced neutralizing activities against prototype, Alpha, Beta, and Delta pseudoviruses (GMT ranging between 2,607 and 3,200). All vaccine-elicited sera neutralized Omicron pseudovirus (GMT 496), with a 6.5-fold reduction of titer compared with prototype neutralization ( Figure 3B). All vaccine-elicited sera neutralized authentic prototype SARS-CoV-2 (GMT 837), variant Alpha (GMT 583), Beta (GMT 393), and Delta (GMT 295) ( Figure 3C). As the authentic Omicron variant (BA.1 subvariant, National Pathogen Resource Center of China (NPRC): 2.192100005 strain) was available in Chinese Center for Disease Control and Prevention (China CDC) currently, macaque sera were tested for neutralization of Omicron and showed 1.6-fold reduction of activity against Omicron than against prototype SARS-CoV-2 ( Figure 3D). In addition, peripheral blood mononuclear cells (PBMCs) collected at day 6 after the third dose were tested for cytokine production using enzyme-linked immunospot (ELISPOT) assay. In consistent with trends of prototype RBD-dimer vaccine ZF2001 in rhesus macaque and human Yang et al., 2021), prototype-Beta chimeric vaccine elicited moderate but balanced T H 1 (IFN-g and IL-2) and T H 2 (IL-4) cytokine production ( Figure 3E).

Protective efficacy of prototype-Beta chimeric RBDdimer vaccine in rhesus macaques
The 12 rhesus macaques receiving prototype-Beta chimeric vaccine were challenged (n = 4) with prototype SARS-CoV-2 (GDPCC-nCoV27 strain), Beta variant (GDPCC-nCoV84 strain), and Delta (CCPM-B-V-049-2105-8), respectively, via both the upper and lower respiratory tracts. The other 12 sham-immunized animals were also challenged in parallel as controls (n = 4) ( Figure S5B). Body weight and temperature were monitored daily until euthanasia at 7 DPI ( Figure S5D). Swabs were ll OPEN ACCESS collected at 0, 1, 3, 5, and 7 DPI from nose, throat, and anus for viral gRNA quantification. At 7 DPI, lung tissues were collected from 7 lopes for histopathology examination. Lung tissues from each lope (a sample mixture of six sites) were quantified for viral gRNA.
Viral loads (gRNA) were detected in many lung lopes of control animals, with the average 1.66-log 10 copies per gram in prototype challenge group, 2.93-log 10 in Beta challenge group, and 6.95-log 10 in Delta challenge group ( Figures 4A, 4D, and 4G). In contrast, significantly lower viral loads were detected in vaccinated animals, with only 1 and 3 lopes having detectable viral gRNA in prototype and Beta challenge group, respectively ( Figures 4A and 4D). In comparison to sham-immunized macaques, lower viral loads were also detected in both upper respiratory tract (nasal and throat swabs) and anus (anal swab) of the vaccinated animals in all days after infections with prototype or Beta SARS-CoV-2 ( Figures 4J-4L). However, the vaccine protection against Delta viral gRNA in upper airway is less pronounced ( Figures 4J and 4K).
At 7 days post inoculation of prototype, Beta or Delta SARS-CoV-2, all sham-immunized macaques developed moderate-tosevere pneumonia, with thickened alveolar septa, vanishment of pulmonary alveolar, congestion, and massive inflammatory cell infiltration in alveoli (Figures 4B,4C,4E,4F,4H,and 4I). In contrast, none of vaccinated animals developed severe pneumonia, with the pathological scores significantly lower than that of the sham group ( Figures 4B, 4C, 4E, 4F, 4H, and 4I). Histopathology examination demonstrated that prototype-Beta chimeric RBD-dimer vaccine can dramatically relieve the lung injury by either infection of prototype, Beta or Delta SARS-CoV-2.

Development of Delta-Omicron chimeric RBD-dimer vaccine
In the late 2021, Omicron variant were circulating globally and gradually substituted its predecessor Delta (Karim and Karim, 2021). The mutations in the spike protein of Omicron variant lead to the resistance for humoral immune responses induced by early SARS-CoV-2 infection and vaccination (Cameroni et al., 2022;Dejnirattisai et al., 2022;Hoffmann et al., 2022). In the context of cocirculation of Delta and Omicron variants, we developed the Delta-Omicron vaccine with our chimeric RBDdimer approach ( Figure 1A).
The Delta-Omicron chimeric RBD-dimer protein was expressed and purified with high yield and purity, indicating the capacity for Twenty-four rhesus macaques (see Table S1) were immunized with 3 doses of prototype-Beta chimeric RBD-dimer vaccine or sham (n = 12), 21 days apart. Serum samples were collected according to the study schedule shown in Figure S5B. Article large-scale production ( Figure S1E). Next, the exposure of RBM and major antigenic sites of RBD-dimer vaccine were verified through SPR assay using hACE2 protein and major classes of RBD-specific mAbs as probes (Figures 1C and 1D). Monomeric RBD proteins from prototype, Delta, and Omicron variants were used for comparison. Delta-Omicron chimeric RBD-dimer showed similar binding affinity (8.44 nM) to hACE2 in comparison with the affinities of monomeric prototype (6.53 nM), Delta (5.09 nM), and Omicron RBD (6.59 nM) (Figures 1D and S2B). For the mAb binding, the Delta RBD did not bind to mAb C110 and exhibited decreased binding affinity to mAb CV07-270; the Omicron RBD did not bind to mAb CB6 and displayed reduced binding affinities to mAbs CV07-270 and C110 (Figures 1D and  S2B). In contrast, Delta-Omicron chimeric RBD-dimer bound all the tested representative mAbs, but like a combination of Delta and Omicron RBD with the decreased affinities to mAbs CV07-270 and C110, respectively ( Figures 1D and S2B).
The cryo-EM structure of Delta-Omicron chimeric RBD-dimer in complexed with CB6 Fab was determined at resolution of 12.8 Å . It reassembles the structure of prototype-Beta chimeric RBD-dimer in complexed with CB6 Fab as the ''bilateral lung''-like structure engaging one CB6 Fab at the Delta arm ( Figures 1G, S4A-S4C, and S4F). In summary, the antigenic characterization and structural analysis indicated that the Delta-Omicron chimeric RBD-dimer protein was correctly folded, presenting the RBM and major antigenic epitopes.
Immunogenicity and protection efficacy of Delta-Omicron chimeric vaccine in mice To evaluate the immunogenicity of Delta-Omicron chimeric vaccine, BALB/c mice were immunized with two doses of AddaVax adjuvanted Delta-Omicron chimeric protein or homotypic prototype RBD-dimer protein, 21 days apart. Delta-Omicron vaccine elicited high titers of serological NAbs against pseudotyped virus displaying Delta, Omicron (BA.1), and Omicron (BA.2) spike proteins, respectively, with 18.6, 19.2, and 19.1 times of those induced by homotypic prototype RBD-dimer vaccine ( Figure 5A). Meanwhile, the GMTs of neutralizing activities against prototype, Alpha, and Beta SARS-CoV-2 pseudovirus were all higher in the Delta-Omicron vaccine group in comparison with the homotypic prototype vaccine group, demonstrating the outperformance of Delta-Omicron chimeric vaccine to induce broader immune responses ( Figures 5A and 5B).
To evaluate the protection efficacy of Delta-Omicron chimeric vaccine, the immunized BALB/c mice (n = 10) were randomly divided into two batches (n = 5) and challenged with SARS-CoV-2 Delta (CCPM-B-V-049-2105-8 strain) and Omicron (BA.1 subvariant, CCPM-B-V-049-2112-18 strain) variants, respectively. Since wild-type BALB/c mice were sensitive to Omicron variant but not Delta variant, the mice for Delta variant challenge were transduced with Ad5 expressing hACE2 5 days before virus inoculation. At 3 DPI, all mice were euthanized and necropsied, and lung samples were collected for virus titration.
For mice challenged with Delta variant, the averages of pulmonary viral gRNA were 1.09 3 10 10 in the sham group but reduced to 1.43 3 10 8 copies/g in the prototype vaccine group, and additionally reduced (474-fold) to 2.37 3 10 7 copies/g in the Delta-Omicron chimeric vaccine group ( Figure 5C). In line with this, the pulmonary viral sgRNA were detected in all mice in the sham group with high levels (average: 1.70 3 10 8 copies/g), but only detectable in three mice receiving prototype vaccine with the average titer of 1.07 3 10 6 copies/g and undetectable in the all mice receiving Delta-Omicron vaccine, suggesting the complete control of Delta viral replication ( Figure 5D).
The histopathological analysis showed that at 3 days post either Delta or Omicron variant infection, sham-immunized mice developed pulmonary damage, including inflammatory cell infiltration, congestion, and vanishment of alveolar cavities ( Figure S6B). As a comparison, both prototype and Delta-Omicron chimeric vaccine-immunized mice showed alleviative lung injury, with the pathological scores significantly lower than those of the sham group ( Figures 5F and 5J). Delta-Omicron chimeric vaccine showed lower pathological scores in comparison with the homotypic prototype form, indicating a better protection (Figures 5G, 5H, and 5J).  . Protection of rhesus macaques by prototype-Beta vaccine immunization (A, D, and G) Box and whiskers plots of 25th-75th percentile with median as center and whiskers of minimum to maximum percentile prototype (A), Beta variant (D), and Delta variant (G) viral gRNA in lung tissues at 7 DPI. Each dot represents a lung lope. The 28 lopes from four animals in each group were analyzed together. p values were analyzed with two-tailed Mann-Whitney test (**p < 0.01; ****p < 0.0001). (B, E, and H) Pooled analyses of pathological scores for lung tissues of macaques challenged with prototype SARS-CoV-2 (B), Beta (E), and Delta (H). The values are means ± SEM. p values were analyzed with two-tailed Mann-Whitney test (**p < 0.01; ****p < 0.0001). (C, F, and I) Histological pathology analyses of lung sections of macaques challenged with prototype SARS-CoV-2 (C), Beta (F), and Delta (I). Black bar represents 100 mm. (J-L) Box and whiskers plots of 25th-75th percentile with median as center and whiskers of minimum to maximum percentile viral gRNA at 0, 1, 3, 5, and 7 DPI in nasal swabs (J), throat swabs (K), and anal swab (L). Each dot represents an animal sample. The horizontal dash lines represent limit of detection, with the upper and lower ones for highest and lowest limit of detection, respectively. (legend continued on next page) ll OPEN ACCESS against the SARS-CoV-2 variants, we developed a chimeric RBD-dimer vaccine approach, which induced broader immune responses than homotypic RBD-dimer vaccines. Based on this approach, we firstly designed the prototype-Beta chimeric RBD-dimer vaccine and then the Delta-Omicron chimeric vaccine.
In the first half of 2021, Beta variant was highly concerned because of its greatly decreased sensitivity to vaccine-elicited sera and its reduction of vaccine efficacy in clinical trials and effectiveness in the real world (Abu-Raddad et al., 2021;Chemaitelly et al., 2021;Garcia-Beltran et al., 2021;Hoffmann et al., 2021a;Li et al., 2020;Lucas et al., 2021;Madhi et al., 2021;Shinde et al., 2021;Tang et al., 2021;Tegally et al., 2021;Xu et al., 2021). Therefore, we present a chimeric vaccine design using RBD-dimer to cover both prototype and Beta variants. The prototype-Beta vaccine induced broader NAbs against SARS-CoV-2 variants than its antigenically homotypic counterparts, especially, with higher activities to cross-neutralize the recently circulating Omicron variant pseudotyped virus ( Figure 2A). Accordingly, prototype-Beta chimeric version showed better protection against SARS-CoV-2, compared with both homotypic prototype and Beta vaccines. The nonhuman primates study showed that high viral loads of Delta SARS-CoV-2 were detected in both upper respiratory tract and lung of all macaques. Our data supported the finding that Delta variant caused frequent breakthrough infections in the real world (Levine-Tiefenbrun et al., Singanayagam et al., 2022). Although Delta variant was more difficult to be cleared in the upper airway, prototype-Beta chimeric RBD-dimer vaccine significantly decreased viral loads of 3.4-log 10 in lung and prevented virus-induced lung lesion ( Figures 4G-4I).
In the second half of 2021, Delta and Omicron VOCs emerged one after another and became the circulating variants successively. The Omicron variant carries mutations at as many as 32 positions in the S protein, including 15 positions in the RBD, and has showed strongly reduced sensitivity to the approved vaccine-elicited sera (Cameroni et al., 2022;Cao et al., 2022;Cele et al., 2022;Dejnirattisai et al., 2022;Hoffmann et al., 2022;Liu et al., 2022;Planas et al., 2022;VanBlargan et al., 2022). In the present study, we also observed that pseudotyped virus displaying Omicron spike was the most resistant to prototype vaccine-elicited antisera (14.6-fold reduction in neutralization) (Figure 2A). However, the sera elicited by prototype-Beta vaccine showed 3.6-fold higher than that induced by homotypic prototype vaccine (Figure 2A). Recently, in the context of the global cocirculation of Delta and Omicron variants, we rapidly developed the Delta-Omicron vaccine with our chimeric RBDdimer approach. The Delta-Omicron vaccine outcompeted the homotypic prototype vaccine to prevent infections by either Delta or Omicron variants in mice and was also balanced to counter prototype and the other VOCs. Interestingly, recent studies showed the Omicron-specific mRNA vaccine induced mice sera that potently neutralized Omicron variant but showed limited neutralization of historical D614G or other SARS-CoV-2 variants, such as Beta and Delta variants Ying et al., 2022). Consistently, WHO advised to develop multivalent vaccines as next generation of COVID-19 vaccine candidates for induction of broader immune responses against both circulating and emerging variants (WHO, 2022). Our result supported the variant-adapted multivalent vaccines are superior to homotypic prototype or variant-specific vaccine in the context of the fast-changing variant circulation.
Previous studies described the use of the ancestral SARS-CoV-2 RBD scaffolded immunogens to induce robust NAbs against ancestral SARS-CoV-2 and, concomitantly, enhance the broadly reactive NAbs against VOCs in monkeys; however, the neutralizing titers elicited against VOCs declined when compared with that against the ancestral SARS-CoV-2 (Arunachalam et al., Saunders et al., 2021). In contrast, we used an alternative approach, heterotypic RBD-dimer connected in tandem, to induce broad and balanced neutralizing responses against both ancestral SARS-CoV-2 and variants. In view of the approval of homotypic RBD-dimer-based vaccine ZF2001 and the high efficacy of this vaccine in Phase 3 clinical trials , heterotypic RBD-dimer chimera designs would be a feasible approach to rapidly adapt SARS-CoV-2 variants or even other coronaviruses in future.
In the structural analysis of RBD-dimer antigens, both prototype-Beta and Delta-Omicron chimeric RBD-dimers are observed to be arranged as similar ''bilateral lung''-like (B) Radar plot demonstrating the neutralization profile of sera elicited by prototype vaccine or Delta-Omicron chimeric vaccine against five SARS-CoV-2 pseudotyped viruses. For (C, D, G, and H), shown are the box and whiskers plots of 25th-75th percentile with median as center and whiskers of minimum to maximum percentile. p values were analyzed with two-tailed unpaired t test (ns, p > 0.05; *p < 0.05; **p < 0.01; ****p < 0.0001). For (E and I), red and gray lines indicate linear regression line and 95% CI, respectively. r and p values represent Spearman's correlation coefficients and corresponding two-sided p values, respectively. Symbols represent individual mouse and may overlap for equal values. Horizontal dashed lines indicated the LLOD. For (F and J), the values are means ± SEM. p values were analyzed with two-tailed Mann-Whitney test (ns, p > 0.05; **p < 0.01; ****p < 0.0001). structures, exposing five major antigenic sites. This supports the versatility of the tandem RBD-dimer as the module to adapt different coronaviruses for the induction of broader immune responses. The present study was timely and encouraging for our fight against COVID-19 pandemic, especially for the Omicron rampage. The Delta-Omicron chimeric vaccine is under development by Anhui Zhifei Longcom Biopharmaceutical Co., Ltd. This design can also be applied in other vaccine platforms, such as mRNA, DNA, and viral vectors.
Limitations of the study First, the immunogenicity and protection efficacy of Delta-Omicron chimeric vaccine was not evaluated in the nonhuman primate model because of the limited animal supply at this moment. We are trying to arrange it. Second, the current vaccines used in humans are based on the ancestral prototype strain, and new variant-specific vaccines should be considered for their role as boosters as well as first-time immunizations. Our data demonstrated the variant-adapted multivalent vaccines were superior to the monovalent prototype-or variant-matched vaccine as first-time immunizations. However, it would be important to further explore whether these vaccines also outperform the others as boosters, given more than 10 billion doses of prototype-based COVID-19 vaccines have been administrated globally.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, George F. Gao (gaof@im.ac.cn).

Materials availability
All requests for unique/stable reagents generated in this study should be directed to and will be fulfilled by the lead contact author with a completed Materials Transfer Agreement.

Data and code availability
The cryo-EM maps have been deposited in the Electron Microscopy Data Bank with accession codes EMDB: EMD-33225 (SARS-CoV-2 prototype RBD-dimer bound to CB6 Fab), EMDB: EMD-33234 (SARS-CoV-2 prototype-Beta chimeric RBD-dimer bound to CB6 Fab) and EMDB: EMD-33235 (SARS-CoV-2 Delta-Omicron chimeric RBD-dimer bound to CB6 Fab). All the other data supporting the finding of this study are available within the paper and are available from the corresponding author upon request. This study did not generate unique code.    The monomeric Omicron (BA.1) variant RBD (S protein 316-534, GISAID: EPI_ISL_6795848) was purchased from ACROBiosystems, which was expressed from HEK293 cells.

Cells and viruses
The hACE2 protein (residues 1-740, Genbank: NP_068576) was purchased from Sino Biological Inc., China, which was expressed from HEK293T cells.
Surface plasmon resonance (SPR) assay SPR binding experiments were carried out using a BIAcore 8000 device (GE Healthcare) at 25 C. The buffers for all proteins used for kinetic analyses were exchanged to PBST (10 mM Na 2 HPO 4 ; 2 mM KH 2 PO 4 , pH 7.4; 137 mM NaCl; 2.7 mM KCl; 0.005% Tween 20). Purified RBD-dimer and -monomer proteins were immobilized on a CM5 chip with the standard EDC/NHS coupling method at about 1,000 response units (RU). Serial dilutions of Fabs were prepared and used to flow over the chip surface. Data were collected over time. After each cycle, the sensor surface was regenerated via a short treatment using 10 mM NaOH. The apparent equilibrium dissociation constants (apparent binding affinity, K D ) for each antibody were calculated using BIAcore 8000 analysis software (BIAevaluation v3.0). Each set of equilibrium binding responses was fitted to the 1:1 binding model.

Cryo-EM data collection and 3D reconstruction
For samples of prototype RBD-dimer bound to CB6 Fab and prototype-Beta chimeric RBD-dimer bound to CB6 Fab, an aliquot of 3.5 ml solution (0.4 mg/ml) was applied to glow-discharged Quantifiol R 1.2/1.3 holey carbon grids and blotted for 1.5 s with a humidity of 95% before being plunged into liquid ethane using a Vitrobot Mark IV (Thermo Fisher). The frozen grides were loaded onto a Titan Krios cryo-transmission electron microscope (Thermo Fisher) that is equipped with a BioQuantum energy filter (Gatan), operated at 300 kV for data collection. Automatic data collection was performed using Serial-EM software. Movies were recorded with a Gatan K2 direct electron in a super-resolution counting mode at pixel size of 0.68 Å . The exposure was performed with a dose rate of 15 e -/ pixel/s and an accumulative dose of $60 e -/Å 2 for each movie which was fractionated into 32 sub-frames. The final defocus range of the datasets was approximately-(1.2-3.0) mm.
The drift correction of all stacks were performed with MotionCor2 (Zheng et al., 2017) to generate 2 3 binned images. Initial contrast transfer function (CTF) values for each micrograph were calculated with CTFFIND4.1 (Rohou and Grigorieff, 2015). Micrographs with an estimated resolution limit worse than 6.0 Å were discarded in the initial screening. The subsequent image processing and reconstruction were performed using Relion-3.1 (Zivanov et al., 2018) and cryoSPARC (Punjani et al., 2017).
For the prototype-Beta chimeric RBD-dimer/CB6 Fab dataset, 302,365 particles were picked from 2036 micrographs. Then the picked particles were extracted and subjected to two rounds of reference-free 2D classification in Relion. A clean dataset with 162,269 particles from good 2D classes were selected and the initial model was generated by cryoSPARC ab initio. Then the model was used as reference in Relion 3D classification. After the third round of 3D classification without applying symmetry, the predominant class containing a subset of 31,507 good particles. These particles were subjected to 3D refinement, which yielded a reconstruction at 11.6 Å resolution as determined by the Fourier shell correlation (FSC) 0.5 cut-off value.
The prototype RBD-dimer/CB6 Fab dataset was processed similarly. Briefly, a total of 289,269 automatically picked particles were extracted in Relion for the following 2D and 3D classification. Three rounds of reference-free 2D classification were performed to remove the heterogeneous particles. A clean dataset with 214,464 particles from good 2D classes was selected and subjected to three rounds of 3D classification without applying symmetry. A single dominant class (26,842 particles) was identified and used to calculate the density map at 11.5 Å resolution by applying a 3D refinement with C2 symmetry.
The Delta-Omicron chimeric RBD-dimer/CB6 Fab dataset was processed similarly to those described above. The specific images processing and reconstruction were shown in Figures S4A-S4C.
Due to the fierce flexibility between two RBDs, we could only obtain low resolution maps as described above. However, we fitted the crystal structure of RBD/CB6 complex (PDB: 7C01) or its RBD part into the two density maps using CHIMERA (Pettersen et al., 2004), which showed a high degree of matching.

Pseudotyped virus neutralization assay
The pseudotyped viruses displaying SARS-CoV-2 spikes express GFP in infected cells. They were prepared as previously described . Mice sera were 2-fold serially diluted and incubated with pseudotyped virus at 37 C for 1 h. Then the mixture was transferred to pre-plated Vero cell monolayers in 96-well plates. After incubation for 15 h, the transducing unit numbers were calculated on a CQ1 confocal image cytometer (Yokogawa). Fifty percent pseudovirus neutralization titer (pVNT 50 ) was determined by fitting nonlinear regression curves using GraphPad Prism and calculating the reciprocal of the serum dilution required for 50% neutralization of infection. pVNT 50 below the limit of detection was determined as half the limit of detection.
Live SARS-CoV-2 neutralization assay Neutralizing antibody activities induced by vaccines in rhesus macaques against prototype, Alpha, Beta and Delta SARS-CoV-2 were titrated on basis of inhibition of cytopathogenic effect (CPE) in IMBCAMS. Briefly, equal volume of serially diluted serum and 100 TCID 50 SARS-CoV-2 was mixed and incubated for one hour at 37 C. The mixture of serum and virus was added to Vero E6 cells, Figure S5. Evaluation of the immunogenicity and protection efficacy of prototype-Beta chimeric RBD-dimer vaccine in animals, related to Figures 2, 3, and 4 (A) Antisera from mice vaccinated with prototype, Beta or prototype-Beta chimeric RBD-dimer vaccines were tested neutralization of a panel of pseudotyped viruses displaying prototype, Alpha, Beta, Gamma, Delta, Delta plus, Kappa, Lambda and Omicron spike, respectively. (B) Rhesus monkeys (n = 12) were immunized with clinical-grade prototype-Beta chimeric vaccine (25 mg antigen + aluminum hydroxide adjuvant) or placebo (aluminum hydroxide adjuvant). Three doses were administrated on days 0, 21, and 42. These monkeys were bled for humoral immunogenicity evaluation before immunization and 14, 35, and 52 days after receiving the first dose. Blood was collected for cellular immune responses on day 48. At the days 56, 57, and 58, four prototype-Beta chimeric RBD-dimer-vaccinated macaques and four sham-vaccinated macaques were challenged with total 1 3 10 6 TCID 50 prototype SARS-CoV-2, Beta variant and Delta variant, respectively. Nasal swabs, throat swabs and anal swabs were collected before challenging and at the days 1, 3, 5, 7 post infection for SARS-CoV-2 titration. Lungs were collected for virus titration and pathological examination on 7 DPI. (C) Antisera from vaccinated macaques were tested neutralization of a panel of pseudotyped viruses displaying prototype, Alpha, Beta, Delta and Omicron spike, respectively. (D) The macaques were challenged with prototype virus, Beta variant and Delta variant, respectively, and weight and temperature were monitored. The values are means ± SEM.