Epitope-focused vaccine immunogens design using tailored horseshoe-shaped scaffold

Abstract

The continuous emergence of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) variants highlights the need to update coronavirus 2019 disease (COVID-19) vaccine components. Epitope-based vaccine designs targeting conserved and immunorecessive regions of SARS-CoV-2 are critically needed. Here, we report an engineered epitope-focused immunogen design based on a novel horseshoe-shaped natural protein scaffold, named ribonuclease inhibitor 1 (RNH1), that can multiply display of conserved neutralizing epitopes from SARS-CoV-2 S2 stem helix. The designed immunogen RNH1-S1139 demonstrates high binding affinity to S2-specific neutralizing antibodies and elicits robust epitope-targeted antibody responses either through homologous or heterologous vaccination regimens. RNH1-S1139 immune serum has been proven to have similar binding ability against SARS-CoV, SARS-CoV-2 and its variants, providing broad-spectrum protection as a membrane fusion inhibitor. Further studies showed that RNH1 has the potential to serve as a versatile scaffold that displays other helical epitopes from various antigens, including respiratory syncytial virus (RSV) F glycoprotein. Our proposed immunogen engineering strategy via tailored horseshoe-shape nano-scaffold supports the continued development of epitope-focused vaccines as part of a next-generation vaccine design.

Graphical abstract

Introduction

Coronavirus 2019 disease (COVID-19) no longer constitutes a public health emergency of international concern. However, the evolution of genetically divergent severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) variants may cause future outbreaks of respiratory diseases. Vaccines against variants have been approved as boosters to compensate for the potential escape from neutralizing antibody protection [1, 2]. These updated vaccines mainly recall pre-existing memory B cells induced by previous SARS-CoV-2 infection or vaccination and mask the de novo generation of variant-specific B cells due to immune imprinting [3,4,5,6,7,8]. These observations underscore the development of next-generation universal vaccines should consider evolutionarily conserved regions that can induce more targeted immune responses [9].

Epitope-based immunogen design is an innovative approach that elicits desired immune responses due to target selectivity [10]. It has been employed in various types of epidemic or pandemic vaccine development [11,12,13,14,15]. Using a small, thermally and conformationally stable protein scaffold, Correia et al. [16] designed a viral epitope-focused vaccine immunogen of RSV that maintains viral epitope structure and induce neutralizing antibody responses in macaques. Immunogen designs using an HIV-1 fusion peptide, a critical component of viral entry machinery, have been validated to elicit cross-clade neutralizing antibodies in different animal models [17].

However, peptide-based vaccines suffer from limitations concerning effective immunogenicity. Fusion of single or multiple epitopes onto a carrier protein such as Keyhole Limpet Hemocyanin (KLH), Cross-Reactive-Material-197 (CRM197) or tetanus toxoid (TT) is a common approach, but this approach may stimulate carrier protein-related immune responses simultaneously [18,19,20,21]. Covalent conjugation with T-helper epitope or Toll-like receptor agonists is also a common method to magnify humoral responses [22, 23]. Maintenance of native B-cell epitope conformation found in the immunogen is another concern. B cell epitope structure mimic is a prerequisite in stimulating epitope-specific antibodies, which is difficult to achieve via simple carrier protein conjugation or covalent linkage. Ribonuclease/ angiogenin inhibitor 1 (RNH1), a 50-kDa protein, exists in most types of human tissues and has been extensively study as a ribonuclease inhibitor to regulate RNA turnover and angiogenesis [24, 25]. It belongs to a protein family with leucine-rich repeats (LRR) and has unique horseshoe-shape feature that parallel beta-strands and alpha-helices constitute its inner and outer walls. These features make it a possible alternative to enhance the immunogenicity of helical epitopes via multiply display [26]. Indeed, key neutralizing epitopes for many pathogens adopt alpha-helical conformations [27,28,29,30,31,32].

In the present study, using the horseshoe-shaped protein RNH1 as part of the scaffold, we designed an immunogen named RNH1-S1139 that can reproduce an epitope located at the S2 stem-helix of SARS-CoV-2 for multiple times. We confirmed that it maintains a high binding affinity against S2-specific neutralizing antibodies. When used as an immunogen, RNH1-S1139 elicits robust epitope-specific antibody responses with a broad spectrum. In the cell–cell fusion assay, the syncytium formation mediated by spike could be inhibited by the RNH1-S1139 immune serum. High epitope-specific IgG titers were observed when RNH1 was used to present other helical epitopes from SARS-CoV-2 FP and RSV F. Together, the tailored horseshoe-shaped immunogens increase the immunogenicity of conserved epitopes, and elicit epitope-specific antibody responses, which unlocks the potency of RNH1 as a versatile scaffold to multiple display helical epitopes.

Materials and methods

Experimental model and study participant details

The goal of this study was to evaluate the protectivity of designed vaccine immunogen RNH1-S1139. To test the S1139-specific antibody titer elicited by vaccination or natural infection, blood was collected from convalescents who had recovered from SARS-CoV-2 infection after two doses of Ad5-nCoV (Convidecia) vaccination [33]. All participants gave written informed consent according to the approval of the Medical Ethics Committee, Academy of Military Medical Sciences (AMMS) with an approval (ethics number AF/SC-08/02.299). All manipulations were strictly conducted in compliance with ethics guidelines and approved protocols. Separation of PBMCs from collected blood samples was accomplished through density gradient centrifugation using Ficoll according to manufacturer’s instruction (DKW-LST-25050SK). Briefly, blood samples were slowly transferred above equal-volume lymphocyte separation medium. After centrifugation at 800 × g for 30 min, PBMCs were collected, washed twice with PBS, resuspended in cell freezing medium (90% FBS and 10% DMSO), and preserved at − 80 °C.

BALB/c mice Unless described otherwise, each specific-pathogen-free (SPF) female BALB/c mice (6–8 weeks old) were intramuscularly (i.m.) immunized with 5 μg immunogen and 50 μg aluminum hydroxide (InvivoGen) in 10 mM phosphate buffer (PB, including 7 mM NaH₂PO₄ and 13.8 mM Na₂HPO₄, pH = 7.2) on days 0, 14, and 28. Blood was collected from the tail vein on days 0, 14, 28, 42, 56 post first immunization. Mice were also immunized with phosphate buffer plus 50 μg Aluminum Hydroxide as the negative control. Blood samples collected from the tail were centrifuged at 3000 rpm to isolate serum for subsequent antibody titration. For safety test, 12 mice were randomized into groups and each mouse was immunized with 5 or 50 μg immunogen with 50 μg Aluminum Hydroxide. Half of them were sacrificed three days post vaccination and the other half were sacrificed at day 60 after completion of three doses immunization under 14-day interval. Tissues from heart, liver, spleen, lung, kidney and injection site of mice were collected from each group and stained with H&E, whole blood was gathered for blood biochemistry test using Chemistry analyzer AU5800 (Beckman Coulter) and hematology test using 3-part hematology analyzer TEK-II mini.

New Zealand rabbits Specific-pathogen-free (SPF) female New Zealand rabbits (12 weeks old) were intramuscularly (i.m.) immunized with 50 μg immunogen with 500 μg aluminum hydroxide (InvivoGen) in 10 mM PB on days 0, 14, and 28. Blood was collected from auricular artery on days 0, 14, 28, 42, 56 post first immunization.

Animals involved in the experiment were approved and the experiments were carried out according to the guidelines of the Animal Care and Use Committee of the Laboratory Animal Ceter (IACUC-DWZX-2023-P045).

HEK293T, HKE293T-hACE2 and Expi293F (ThermoFisher) cells are maintained in our laboratory. HEK293T cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin solution at 37 °C, 5% CO₂ incubator. HEK293T-hACE2 cells were cultured in the above medium supplemented with 1 μg/mL puromycin under the same conditions. Expi293F cells were cultured in Expi293 expression medium (Gibco) at 37 °C under 5% CO₂ in a Multitron-Pro shaker. All cell lines used in this study were routinely tested for mycoplasma contamination.

Immunogen expression and purification

All the proteins mentioned below were purified from E.coli Origami B (DE3). Cells transduced with plasmids were induced by 0.1 mM IPTG at 37 °C and collected by centrifugation after 4 h. S1139X3, S1139X12, S1139X24, or RNH1-S1139 was applied to Nickle affinity column (HisTrap HP, Cytiva) preequilibrated with PBS, 200 mM NaCl, 20 mM imidazole, pH = 7.4 after sonication and centrifugation. Column was washed using PBS, 200 mM NaCl, 50 mM imidazole, pH = 7.4 with 5 X column volume, and immunogen with His-tag was eluted by PBS, 200 mM NaCl, 500 mM imidazole, pH = 7.4. Buffer was exchanged to PBS, 200 mM NaCl, pH = 7.4 by ultrafiltration. For RNH1-S1139, protein was further purified by Superdex75 column (GE Healthcare) in PBS with 200 mM NaCl, pH = 7.4. The peak fraction is eluted at about 65 mL and size was determined by SDS-PAGE. For RNH1-FP or RNH1-RSV-F purification, pellet was collected after sonication followed by centrifugation, and was denatured in 8 M urea. Supernatant was applied to Nickle affinity column (HisTrap HP, Cytiva) preequilibrated with PBS, 8 M Urea, 20 mM imidazole, pH = 7.4. Column was washed using PBS, 8 M Urea, 50 mM imidazole, pH = 7.4 with 5X column volume, and immunogen with His-tag was eluted by PBS, 8 M Urea, 500 mM imidazole, pH = 7.4. Eluted sample was further purified by HiTrap Q HP column with PB, 8 M Urea, pH = 7.4 equilibration buffer and PB, 8 M Urea, 0–1 M NaCl, pH = 7.4 elution buffer. Immunogen was renaturation and buffer was exchanged to PBS, 300 mM NaCl, 500 mM Arg, pH = 7.4 by ultrafiltration. Immunogen purity and integrity was determined by SDS-PAGE. Immunogen concentration was determined by BCA assay. RBD, S2, and S Trimer (WT, Beta, Delta, Omicron, Omicron BA.2, Omicron BA.4/5, Omicron XBB and Omicron BF.7) of SARS-CoV-2, and RSV-preF were purchased from Sino Biological Inc. Lyophilized CRM197 was purchased from Ligand Pharmaceuticals. Conjugation of S1139-CRM197 and S1139-KLH was performed by GL Biochem Ltd. (Shanghai, China).

mAb expression and purification

Variable region sequences of reported antibodies B6, 76E1 and D25 were retrieved from the literature [31, 34, 35]. Variable region sequences of top nine paired heavy and light chains of each single-cell sequencing immunization group were collected, codon optimized and conjugated to human Fc. Full-length antibody sequences were cloned into pcDNA3.1 using EcoRI and HindIII restriction sites and signal peptide and Kozak sequences were added to the N-terminal of expressed sequences. Plasmids of paired heavy and light chain genes were co-transfected into the Expi293F cells according to the manufacturer’s protocol. After 5 days, antibodies were purified from cell culture supernatants using a HiTrap rProtein A column (Cytiva). Buffer was exchanged to PBS by ultrafiltration, and antibody concentration was determined by BCA assay.

Circular dichroism (CD)

CD spectra were collected on MOS-500 spectropolarimeter using a quartz cell with a path length of 0.1 cm. Samples were buffer exchanged to PBS and diluted to 0.01 mg/mL before loading. The experimental parameters were: bandwidth 1 nm, repeat 1 time, acquisition period 1 s/point. Data was collected from190 to 260 nm. CD spectra were background corrected and scaled to mean residue ellipticity based on the absorbance at 205 nm.

Enzyme-linked immunosorbent assay (ELISA)

S1139-specific IgG detection 96-well plates (Corning) were coated overnight at 4 °C with S1139 peptide at final concentration of 8 μg/mL using PBS [36]. For KLH and RNH1 IgG titer determination, plates were coated overnight at 4 °C with respective protein at final concentration of 1 μg/mL using coating buffer (0.05 M Phosphate Saline, pH = 9.6). Plates were blocked with a 2% w/v solution of Bovine Serum Albumin (BSA; Sigma) in PBS-Tween 20 (PBS-T) for 2 h at 37 °C. Plates were washed three times and sera were serially diluted and incubated at 37 °C for 1 h. Plates were washed three times using PBS-T, and an anti-Ms HRP conjugated antibody (Abcam ab97265)/ IgG1a (Abcam ab97240)/IgG2 (Abcam ab97245) was used to determine binding titer at 1:10000 dilution ratio for 1 h at 37 °C, followed by incubation with TMB substrate (Solarbio) for 6 min, which was stopped by adding 1 M HCl. The absorbance at 450 nm–630 nm was recorded by a microplate reader (Spectra Max 190, Molecular Devices). Cut-off value was determined based on OD value of 2.1-fold day 0 serum.

SARS-CoV-2 S-specific IgG detection SARS-CoV-2 S Trimer (W.T. or BA.4/5) was coated overnight at 4 °C with respective protein at final concentration of 1 μg/mL using coating buffer (0.05 M Phosphate Saline, pH = 9.6). Plates were blocked with a 2% w/v solution of Bovine Serum Albumin (BSA; Sigma) in PBS-Tween 20 (PBS-T) for 2 h at 37 °C. The plates were washed three times and sera were serially diluted and incubated at 37 °C for 1 h. Then plates were washed three times. IgG titer from human serum were revealed using an anti-human IgG HRP conjugated antibody (Abcam ab97225) at final concentration of 100 ng/mL. After incubating for 1 h, plates were washed three times and TMB substrate was added for 6 min. The reaction was stopped by adding 1 M HCl. The absorbance at 450–630 nm was recorded by a microplate reader (Spectra Max 190, Molecular Devices). Area under the curve was calculated to represent binding strength.

Ag-Ab binding evaluation

Antigen of interest were coated overnight at 4 °C at final concentration of 1 μg/mL, using coating buffer (0.05 M Phosphate Saline, pH = 9.6). Plates were blocked with a 2% w/v solution of Bovine Serum Albumin (BSA; Sigma) in PBS-Tween 20 (PBS-T) for 2 h at 37 °C. The plates were washed three times and monoclonal antibodies were serially diluted from 100 ng/mL and incubated at 37 °C for 1 h. Plates were washed three times. An anti-human IgG Fc HRP conjugated antibody (Abcam ab97225) was applied to the plates at final concentration of 100 ng/mL for 1 h at 37 °C. After washing, TMB substrate was added to the plates and incubated for 6 min at room temperature, and the reaction was stopped using 1 M HCl. The absorbance was read at 450–630 nm using a microplate reader (Spectra Max 190, Molecular Devices).

Surface plasmin resonance (SPR)

Binding affinity of RNH1-S1139 and S Trimer to mAb B6 was measured in the BIAcore T200 (GE Healthcare, UK). Running buffer was Cytiva HBS-EP + (pH 7.4). All measurements were performed at 25 °C. B6 was immobilized on a protein A chip (Cytiva, USA) at final concentration of 1 μg/mL. RNH1-S1139 or S Trimer was then loaded from 100 to 0.78 nM with twofold serial dilution, and kinetic parameters (K_on and K_off) and affinities (KD) were calculated. Binding affinity of immunogens against B6 was calculated from all the binding curves based on their global fit to a 1:1 binding model using BIA evaluation 4.1 (GE Healthcare, UK).

Biolayer interferometry binding (BLI)

GatorPrime was used to perform binding affinity experiments of the mAbs to RNH1-FP and RNH1-RSV-F. MAb 76E1 for RNH1-FP and D25 for RNH1-RSV-F were loaded onto Protein A sensor (20–5006 GatorBio) at 40 μM for 300 s. These sensors were then first dipped into PBS for 120 s to establish a baseline and then exposed to each designed antigen or reference protein for 300 s to measure association. The sensors were then returned to PBS for 500 s for dissociation. For reverse loading order, His-tagged RNH1, RNH1-S1139, RNH1-FP, and RNH1-RSV F were immobilized onto the anti-his probe (20–5047 GatorBio) at 10 μg/mL and serially diluted Fab fragment of mAbs were loaded as analytes, respectively. Binding affinity was calculated based on their global fit to a 1:1 binding model using GatorOne (v2.15). All the samples were diluted using PBS containing 0.02% Tween 20 (PBST).

Antibody competitive binding assay by BLI

GatorPrime was used to perform the competitive binding experiment. His-tagged RNH1-S1139 was immobilized onto anti-his probe (20–5047 GatorBio) at 50 μg/mL for 60 s. After washing with buffer for 30 s, the probes were immersed in the wells containing the primary antibody B6 at a final concentration of 50 μg/mL for 240 s, which was followed by incubating with the secondary antibody for 480 s. All the samples were diluted using PBS containing 0.02% Tween 20 (PBST), and PBST was used as the buffer during the whole process.

Pseudovirus neutralization assay

Detection of neutralizing antibodies in sera from immunized mice was done using HIV-1 Env-pseudotyped viruses luciferase-expressing system. SARS-CoV wildtype and variants (BQ.1, XBB, BF.7, XBB.1.5) of SARS-CoV-2 pseudoviruses were purchased from Vazyme company. Briefly, serum samples from each group were threefold-serial diluted in 50 μL of DMEM, and co-incubated with 50 μL diluted pseudovirus for 1 h at 37 °C. 100 μL (2 × 10⁵/mL) HEK293T-hACE2 cells were added subsequently and incubated for 48 h to express the luciferase. Finally, 100 μL medium was removed, and 100 µl lysis reagent was added with luciferase substrate from a luciferase kit (Vazyme) and relative light unit (RLU) was detected by a multi-mode microplate reader (Tecan Spark).

Authentic virus neutralization assay

SARS-CoV-2 ancestral virus was isolated from lung lavage fluid of a patient by our institute (SARS-CoV-2/human/CHN/Wuhan_IME-BJ01/2020, genebank: MT291831.1). The authentic SARS-CoV-2 plaque assay combining with Reed-Muench method was performed to determine TCID₅₀. For authentic virus neutralizing assay, purified polyclonal antibodies from immunized rabbits (20 mg/mL) were twofold (PB, S1139, and RNH1-S1139) or fourfold (S Trimer) serially diluted at the initial dilution factor of 3 (PB, S1139, and RNH1-S1139) or 12 (S Trimer) in 120 μL DMEM medium supplemented with 2% FBS. After incubating with 120 μL TCID₅₀ of SARS-CoV-2 at 37 ℃ for 1 h, 200 μL of the mixture were transferred to the Vero-E6 cells (25000 cells/well seeded one day before) and incubated for 3 days in a cell culture incubator (5% CO₂ at 37 ℃). The crystal violet was used to stain the cells for 30 min at room temperature, and the absorbance at 570 nm/630 nm was then measured. Cells without viruses or antibodies were used as blank controls, and cells with viruses but without antibodies were used as virus controls. IC₅₀ was calculated using Reed-Muench method. All processes involving authentic SARS-CoV-2 virus were done in a biosafety level 3 (BSL-3) laboratory.

In vivo pseudovirus neutralizing assay

K18-hACE2 transgenic mice (6–8 weeks old) kept in our lab were immunized with 5 μg S Trimer or RNH1-S1139 with 50 μg Al(OH)₃ adjuvant (InvivoGen) three times (i.m.) at two-week intervals. 50000 TCID₅₀ HIV-1 Env-pseudotyped SARS-CoV-2 ancestral strain viruses (Vazyme) were inoculated (i.p.) 14 days after the third immunization. 3 mg D-Luciferin (PerkinElmer #122799) in 100 μL DPBS was used to detect the luminescence signal 4 days after pseudovirus inoculation by intraperitoneal injection. In vivo experiments were performed using murine fluorescence imaging (IVIS^® Spectrum, PerkinElmer, USA). The average radiance was calculated from the same area of each biological replicate.

Single cell VDJ sequencing

B cell enrichment Three spleens of mice immunized with S trimer and RNH1-S1139 were collected one week after the third dose. Cells passed through the 70 μm strainer were treated with red blood cell lysis buffer (Sigma) according to the manufacturer’s introductions and washed with 2% FBS 1640 medium twice. After centrifuging at 500 g for 5 min, viable cells were adjusted to 1 × 10⁷ cells/mL. B cells were enriched using a B cell isolation kit according to the manufacturer’s instructions (Miltenyi Biotec 130–042-201).

Single-cell V(D)J-seq library construction and sequencing Single-cell V(D)J-seq libraries were prepared using SeekOne® DD Single Cell 5’ library preparation kit and SeekOne DD Single Cell V(D)J Enrichment Kit (SeekGene Catalog No. K01201). Briefly, ten thousand cells were mixed with reverse transcription reagent and then loaded to the sample well in SeekOne DD Chip S3. Subsequently, Gel Beads and Partitioning Oil were dispensed into corresponding wells separately in Chip S3. After emulsion droplet generation, reverse transcription was performed at 42 ℃for 90 min and inactivated at 85 ℃ for 5 min. Next, cDNA was purified from a broken droplet and amplified in the PCR reaction. Then, the amplified cDNA product from 5ʹ cDNA was enriched for V(D)J amplification with specific primers. Next, the V(D)J amplification product was fragmented, end-repaired, A-tailed, and ligated to the sequencing adaptor. Finally, the indexed PCR was performed to amplify the V(D)J product, which contained the Cell Barcode and Unique Molecular Index. The indexed sequencing libraries were cleanup with SPRI beads, quantified by quantitative PCR (KAPA Biosystems KK4824), and then sequenced on Illumina NovaSeq 6000 with PE150 read length. The single-cell RNA library preparation and BCR immune repertoire sequencing were performed on the platform of SeekGene Biosciences (Beijing, China).

BCR mapping BCR sequences for each single B cell were assembled by seeksoul tools vdj (v1.0.0) for the identification of CDR3 sequence and the rearranged BCR gene with refdata-cellranger-vdj_GRCh38_alts_ensembl-5.0.0 as reference. Using barcode and raw clonotype ID information, productive BCR clonotypes were filtered. Moreover, processed FASTA sequences from scVDJ-seq were annotated using IgBlast v1.21.0 against the Mus musculus IMGT reference database. BCR analysis was performed in R v.4.2.2. Result visualization was performed using base R, venn v.1.12, pheatmap v.1.0.12, and circlize v.0.4.15. BCRs from identical immunization groups were grouped, clonal diversity calculation of IGH CDR3 was performed using the estimateAbundance function, and the smooth curve of diversity was generated using the plotAbundanceCurve function from alakazam v.1.3.0.

Real-time PCR

Single-cell suspensions from mouse spleens were gathered seven days post third immunization. Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s instruction. cDNA synthesis and quantitative real-time PCR (qPCR) was performed using One Step TB Green PrimeScript RT-PCR Kit (TaKaRa RR066A) on QuantStuidio3 (Applied Biosystems) according to the manufacturer’s instruction. Each group contains three biological replicates and the experiment was repeated twice.

Primer sequences for qPCR.

Bach2: 5′-GTCGAAAGAGGAAGCTGGACTG-3′ and 5′-GAGGCAGGAAAAGTTGTCCAGG-3′.

Bcl6: 5′-CAGAGATGTGCCTCCATACTGC-3′ and 5′-CTCCTCAGAGAAACGGCAGTCA-3′.

IRF8: 5′-CAATCAGGAGGTGGATGCTTCC-3′ and 5′-GTTCAGAGCACAGCGTAACCTC-3′.

Pax5: 5′-TGACGCAGGTGTCATCGGTGAG-3′ and 5′-ATTCGGCACTGGAGACTCCTGA-3′.

GAPDH: 5′-CATCACTGCCACCCAGAAGACTG-3′ and 5′-ATGCCAGTGAGCTTCCCGTTCAG-3′.

Cell–cell fusion assay

For testing inhibition of spike-mediated cell–cell fusion, HEK-293 T cells were seeded in 24-well plates at 20,000 cells/ well in 1 mL 10% FBS DMEM. After 16 h, cells were transfected with CMV-SARS-CoV-2-S-GFP plasmid as follows: 0.5 µg plasmid were diluted in 100 μL DMEM and mixed with 2 μL Turbofect (Thermo Scientific). After 15 min incubation, the mixture was plated. Transfection reagent with DMEM was added as non-transfected control. 1 mL fresh 10% FBS DMEM was changed 4 h after transfection to each well. For cell–cell fusion assay establishment, 20000 GFP-positive cells were seeded to 96-well plate and mAb B6 was added to the wells at a final concentration of 20000, 50, 0.125, 0 ng/mL and co-cultures were further incubated at 37 °C for 1 h with 5% CO₂. 20000 HEK-293 T-hACE2 cells were added to each well and incubated at 37 °C for 4 h for syncytia formation. Cells were stained with 1 μg/mL Hoechst 33342 for 10 min at 37 ℃. Three images of treatment well were acquired in DAPI, GFP and BrightField channels with a Cytation 5 equipment for fusion analysis.

Cell fusion quantification HEK-293 T cells transiently transfected with codon-optimized SARS-CoV-2 S were used as the donor cells. And 293 T cells stably expressing human ACE2 receptor were utilized as effector cells. Expression plasmids pCMV-α and pCMV-ω encoding the α and ω parts of β-galactosidase were transfected to donor and effector cells via lentiviral transfer system and named α/S and ω/ACE2, respectively [37]. A total of 40000 per well effector cells were incubated with purified and diluted polyclonal antibody from rabbit serum (initial concentration 1 mg/ml) (HiTrap rProtein A FF) at 37 °C for 4 h. Then, 40000 per well target cells were added and cocultured with effector cells at 37 °C for 48 h. The HEK-293 T cells only transfected with ω encoding plasmid incubated with target cells were used as the negative control. β-galactosidase activity was determined using Galacto-Star^™ System (Invitrogen T1012) according to the manufacturer’s instruction. Inhibition rate was calculated by normalizing to positive (100%) and negative (0%) controls.

Sequence alignment of coronavirus spike stem helix regions

Amino acid sequences of SARS-CoV-2 and its variants were downloaded from GISAID. Amino acid sequences of hCoV 229E, HKU1, HKU5, HKU4, MERS, OC43, SARS, and MHV were downloaded from NCBI. Sequence alignment of stem helix region was performed in Geneious Prime (v2022.2.2).

Quantification and statistical analysis

Data analysis was performed using Excel and GraphPad Prism 9.0. Antibody titer data were shown as log10 transformed before analysis. One-way ANOVA followed by Dunnett’s multiple comparison test was used to compare significant changes between groups. Data were considered statistically significant at ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001.

Results

Generation and characterization of RNH1-S1139

Broad neutralizing antibodies (bnAbs) play important roles in blocking SARS-CoV-2 variant infection. Immunogen designs based on these bnAb-targeting epitopes hold promise for inducing broad-spectrum antibody production. Antibodies isolated from COVID-19 convalescents that target the stem helix region of SARS-CoV-2, such as S2P6 and CC40.8, protected against viral challenge by inhibiting S-mediated membrane fusion [36, 38]. A linear epitope (S1139-₁₁₃₉DPLQPELDSFKEELDKYFKNHTSP₁₁₆₂) located at the S2 stem helix was chosen for further study. High conserveness of S1139 among Sarbecoviruses indicates that this epitope is a suitable target for universal vaccine design (Fig. S1).

We performed rational structural-based vaccine design to obtain immunogens capable of eliciting robust S1139-directed antibodies. Inspired by the identical secondary structure of S1139 epitope and RNH1 outer wall, we replaced the alpha helixes of RNH1 with S1139 epitope, and scaffold in the inner skeleton was repeated to accommodate the 24-aa length S1139 epitope (Fig. 1A). Thus, we successfully engineered a RNH1-S1139 immunogen displaying 16 repeats of the S1139 epitope. RNH1-S1139 was produced in E.coli Origami with high yield (about 5 mg from 7.5 g E.coli pellet) but low cost (Fig. 1B). Circular dichroism (CD) spectrum of RNH1-S1139 is similar to that of RNH1 (Fig. 1C).

Fig. 1

Generation and characterization of the engineered epitope-based immunogen RNH1-S1139. A Left, location of S1139 in S Trimer (PDB 6XR8) and its binding to mAb B6 (PDB 7M53) and S2P6 (PDB 7RNJ). Right, schematic illustration of RNH1 (PDB 1Z7X) and RNH1-S1139 (predicted by AlphaFold2 v.1.5.3.) and their structural comparison from different views. B SDS-PAGE verification of RNH1-S1139 molecular weight after His-tag affinity purification followed by size exclusion chromatography. RNH1-S1139 was expressed in E.coli Origami B (DE3). C Circular dichroism (CD) evaluating the secondary structural consistency between RNH1 and RNH1-S1139. D Binding ability of designed immunogens to epitope-specific antibody mAb B6 determined by ELISA. E Binding affinity of S trimer or RNH1-S1139 to B6 determined by surface plasmon resonance (Biacore). F S1139-specific IgG titers among S1139-based immunogens (S1139/S1139-KLH, S1139X3, S1139X12, RNH1-S1139 and S1139X24, having 1, 3, 12, 16 and 24 S1139 repeats, respectively) and controls (PB, RBD, S Trimer) at day 14 (after dose 2) and day 28 (after dose 3) determined by ELISA. N = 6 mice per group. Data are presented as mean values with standard deviation (SD) in D and F. D and F share the same figure legend. p values were determined by One-way ANOVA followed by Dunnett’s multiple comparison test. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001

Conjugation of peptide or protein antigens to carrier protein KLH is a common approach used to enhance the immunogenicity of antigens [39]. S1139 was covalently linked to KLH through an appended C-terminal cysteine. At the same time, multiple S1139 tandem repetitions (3, 12, and 24-times, named S1139X3, S1139X12 and S1139X24, respectively) were designed for comparisons (Fig. S2, Additional file 1; Table S1). Monoclonal antibody B6 is a broad-neutralizing antibody that cross-reacts with linear stem helix within S2 of S proteins among the beta-coronavirus genus [34]. We measured the binding ability of designed antigens to B6 by ELISA and the results show that S1139 epitope can be displayed correctly among various designs (Fig. 1D). More accurate binding affinity to epitope-specific broad-neutralizing antibody was tested using SPR for the RNH1-S1139 design that exhibited tight binding to epitope-specific antibodies by ELISA assay, and results show that it binds to B6 with less than picomolar avidities, which is almost twenty times higher than that of S Trimer (Fig. 1E). We measured equilibrium dissociation constant (K_D) of 0.175 nM for the binding between RNH1-S1139 and B6 mAb, while in the same assay, the K_D value of spike for B6 was 96.4 nM. S1139 epitope and RNH1 alone had very weak or no binding affinity to B6 mAb (Fig. S3). We also measured binding affinity of RNH1-S1139 with B6 using Fab of B6 as mobile phase. The results showed that RNH1-S1139 held similarly high binding affinity against mAb B6 as S Trimer did (Fig. S4).

To access the immunogenicity of RNH1-S1139, we immunized mice three times intramuscularly at two-week intervals. RNH1-S1139 elicited higher anti-S1139 IgG titer four weeks after initial immunization than any other design, including S1139-KLH (Fig. 1F). We then immunized mice with low- (0.5 μg), moderate- (5 μg), and high- (50 μg) dose of RNH1-S1139 and found that 5 and 50 μg induced similar levels of antibodies titers, which indicates that 5 μg dosage is sufficient for RNH1-S1139 to induce humoral immunity (Fig. S5). In addition to KLH, we also compared the efficacy between RNH1 and another commercially used carrier protein CRM197, and confirmed that RNH1 has a better epitope display effect (Fig. S5). Taken together, these data indicate that the design of RNH1-S1139 is suitable for mass production, shows strong bindings to epitope-specific antibodies, and more importantly, enhances immune recognition via epitope multivalent display.

RNH1-S1139 elicit durable and broad-spectrum immune responses, and is safe as a vaccine immunogen

Next, we evaluated in detail the properties of RNH1-S1139 as immunogen. It has been reported that host immune responses induced by various types of SARS-CoV-2 vaccines wane gradually over time [40, 41]. To assess the durability of RNH1-S1139-induced humoral responses, S1139-specific antibody was measured until 180 days post first exposure showing a long-lasting binding titer (Fig. 2A). We measured IgG1/IgG2a ratio among the antibodies produced by RNH1-S1139 immunization. This ratio suggested that RNH1-S1139 immunization gives rise to a T helper type 2 (Th2)-biased humoral response (Fig. 2B). This result is consistent with previous findings that S1139 contains partial or entire of predicted B cell epitope mapped in SARS-CoV-2 [42].

Fig. 2

RNH1-S1139 elicits long-term and cross-reactive antibodies and is safe in animal models. A Immunization strategy and bleeding schedule of mice. S1139-specific IgG titer measured by ELISA. B Titers of IgG1 and IgG2a antibody isotypes in immune serum measured by ELISA. C Binding antibody titers of RBD or RNH1-S1139 immunization group against S trimers from MERS-CoV, SARS-CoV and SARS-CoV-2 and its variants. Values were detected by ELISA and normalized to the SARS-CoV-2 S trimer group. D ELISA assay showing changes of binding antibody titers elicited by RNH1-S1139 against S1139 epitopes with single point mutation in S1139 or epitopes from different coronavirus strains. Each line represents a biological replicate. E Left, the crystal structure of mAb B6 Fab in complex with SARS-CoV-2 stem helix peptide (part of S1139). Middle, zoomed-in view showing how epitope forms salt bridge with Fab heavy and light chain (PDB 7M53). Structural model was prepared by PyMol. Right, superposition of S1139 epitope and D1153Y substitution. F Tissues of mice heart, liver, spleen, lung, kidney, brain, and muscle collected from PB, low-dose RNH1-S1139 (5 μg/Ms) and high-dose RNH1-S1139 (50 μg/Ms) groups 3 and 60 days post initial immunization were stained with H&E or Sirius red. N = 6 mice per group; data are presented as mean values with SD in A and B. p values were determined by One-way ANOVA followed by Dunnett’s multiple comparison test. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001

The emergence of SARS-CoV-2 Omicron variants challenges the protective ability of currently available vaccines [43]. Moreover, before SARS-CoV-2, there were two highly pathogenic β-coronaviruses crossing from animals to humans, MERS-CoV and SARS-CoV, and causing severe respiratory diseases. To analyze to which extent RNH1-S1139 immune serum recognized MERS-CoV, SARS-CoV, and SARS-CoV-2 variants, we incubated this serum with different S trimers. Comparable amounts of S-specific IgG were detected among SARS-CoV, SARS-CoV-2 and its variants (Beta, Delta, Omicron, Omicron sublineages BA.2, BA.5, XBB and BF.7). Less conserved RBD immune group showed decreased binding titers to MERS-CoV, SARS-CoV, and SARS-CoV-2 variants (Fig. 2C).

Of note, sequences of S1139 epitope are conserved among Sarbecoviruses; there are still amino acid mutations occurring in SARS-CoV-2 variants within S1139 epitope from worldwide sequenced strains (http://www.oncoimmunobank.cn/bcedb/item/jbrowse); and S1139 epitopes of beta-coronavirus or coronavirus show slight differences in amino acid sequence (Fig. S1). Thus, to determine how point mutations in S1139 or how other coronavirus-derived S1139 affect the binding ability of RNH1-S1139 elicited antibodies, we synthesized mutated S1139 peptides gathered from GISAID, including P1143L that occurs in Omicron subvariant BA.2.86, S1139 sequence of SARS-CoV-2 (W.T., Beta, Delta, Omicron), SARS-CoV, MERS-CoV, and more distantly related human and murine coronaviruses HCoV-HKU4, HCoV-HKU5, MHV, HCoV-OC43, HCoV-HKU1, HCoV-229E [44]. Antibody binding property is almost unaffected by point mutations within Sarbecovirus except for D1153Y (Fig. 2D). Superposition of S1139 and D1153Y shows clearly that the side chain of Tyr exhibits a phenyl ring, which may abrogate the salt bridge triad formed by D1153, B6 CDRH3 residue R104 and CDRL1 residue H33 (Fig. 2E) [34].

Safety is one of top concerns in vaccine development. Biophysical and biochemical testing to evaluate preliminary safety of RNH1-S1139 as vaccine immunogen was conducted in mice. No adverse effect was observed in either the moderate- or high-dose group (Additonal file 2; Table S2). Furthermore, no abnormal lesions were found in the heart, liver, spleen, lung, kidney, brain, and injection sites 3 days (moderate- and high-dose group) and 60 days (moderate-dose group) after injection during histopathological evaluation using hematoxylin and eosin (H&E) staining (Fig. 2F Left). No fibrosis formed in the heart, lung and liver stained with Sirius red (Fig. 2F Right). Collectively, these data suggest that RNH1-S1139 is safe as a vaccine immunogen candidate that elicits sustained and broadly cross-reactive antibody response.

RNH1-S1139 immune serum prevents SARS-CoV-2 entry by inhibiting S-mediated membrane fusion

Antigenicity of RNH1-S1139 was further validated in New Zealand rabbits, and a similar pattern to mice was observed for serum anti-S1139 IgG responses, confirming the robustness of RNH1-S1139 in eliciting antibody responses (Fig. 3A). Notably, the background antibody binding to RNH1 skeleton is kept at low level (Fig. S6). To study the neutralization potency of antibodies elicited by the epitope-based immunogen, a pseudovirus neutralizing assay was conducted. We found that RNH1-S1139 immune serum has weak but broad-spectrum neutralizing activities to SARS-CoV, SARS-CoV-2 and its variants BQ.1, XBB, BF.7 and XBB.1.5 (Fig. 3B). Next, the live SARS-CoV-2 virus was evolved in the neutralizing assay. Consistent with the results of the pseudovirus neutralizing assay, pAbs from the S Trimer immunization group inhibited viral infection even at high dilution ratio, and the RNH1-S1139 group showed mild neutralization activity in high pAb concentration. For the PB and S1139 groups, we did not observed any neutralization activity (Fig. 3C).

Fig. 3

RNH1-S1139 immune serum blocks SARS-CoV-2 entry into the cell through S-mediated membrane fusion. A Upper, timeline of vaccination and serum collection. Lower, levels of S1139-specific IgG titer in sera collected on days 14, 28, 42, and 56 post initial immunization. B Neutralizing antibody titers against pseudotyped SARS-CoV, SARS-CoV-2 and its variants elicited by S trimer (pink), S1139 (light blue) or RNH1-S1139 (blue). C Reciprocal IC₅₀ titers of authentic SARS-CoV-2 ancestral strain in immunized rabbit samples. Bar chart presents the mean value of reciprocal IC₅₀ with SD from four biological replicates calculated according to the Reed-Muench method (D) Representative images of syncytia formation (white circles) in the SARS-CoV-2 W.T. S-mediated cell–cell fusion in the presence or absence of immunized sera. Scale bar = 200 μm. E Quantitative assay of cell fusion inhibition. Co-culture of S-expressing and hACE2-expressing cells were set as positive controls. HEK-293 T cells only transfected with α fragment of β-galactosidase were used as negative control to fuse with hACE2-expressing target cells that carry the ω-fragment. Serum samples used in B, C, D and E were collected on day 42 after first immunization. N = 4 rabbits per group. Data are presented as mean values with SD in A, B, C and E. p values were determined by One-way ANOVA followed by Dunnett’s multiple comparison test. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001

As the S2 stem helix region of SARS-CoV-2 mainly functions as a cell-fusion mediator, a surrogate method was built to elucidate the RNH1-S1139 immune serum-mediated neutralization mechanism [45]. For donor cell construction, SARS-CoV-2 S and GFP co-expression plasmid was transiently transfected to HEK293T cells. HEK293T cells that stably overexpress ACE2 on cell membrane were used as acceptor cells. The model was validated using epitope-specific monoclonal antibody B6. Cell fusion was inhibited by B6 mAb and the inhibition rate was positively correlated with mAb concentration (Fig. S8). In our study, cell fusion occurred when donor and acceptor cells were mixed with polyclonal antibody purified from the PB group. Cell fusion was fully or partially halted after adding ten-fold diluted S trimer, S1139, or RNH1-S1139 polyclonal antibody purified from immune serum (Fig. 3D). Higher dilution fold increases S-mediated syncytia formation but at different rates among three immunization groups (Fig. 3D). An α-complementation assay based on β-galactosidase was adapted to quantify S-mediated cell fusion [37, 46]. Consistent with qualification results, fusion rate of S1139 group is almost identical to PB control, while for S trimer group, fusion rate showed at least fifty percent increase when compared with negative control in both tenfold and 500-fold dilution treatments (Fig. 3E). Syncytia formation in RNH1-S1139 group is between the two other groups (Fig. 3E). Our findings reveal that RNH1-S1139 immune serum blocks viral infection by inhibiting cell fusion.

We further tested the protection ability of RNH1-S1139 as a vaccine candidate immunogen in vivo. Strong luminescence signals were observed in K18-hACE2 mice that received placebo injection (PB), and mice immunized with S trimer showed the blockage of pseudovirus infection. For the RNH1-S1139 immunization group, weaker signals were observed compared with the PB group (Fig. S7). Although the decreased value was not statistically significant, it provided strong evidence that RNH1-S1139 could protect organisms from SARS-CoV-2 infection to a certain extent.

RNH1-S1139 elicits epitope-focused antibody responses, and is suitable to serve as a heterologous booster vaccine

The diversity and frequency of B cell responses play a crucial role during fighting against SARS-CoV-2 variants infection. To investigate whether RNH1-S1139 elicits effective antibody responses as a candidate immunogen and to compare the immune repertoire profiling of RNH1-S1139 and S Trimer, mice were immunized with S trimer or RNH1-S1139. Single-cell VDJ sequencing (scVDJ-seq) was performed using B cells enriched from spleens at 60 days post first immunization. B cells without productive IGH or IGL were filtered out. No less than 76% cells with productive V-J spanning pair were recovered in each group (13869/17731 in PB, 17408/22867 in S Trimer, and 14422/18828 in RNH1-S1139). Totally, 13303 and 11338 full-length heavy and light chains were identified in RNH1-S1139 immunization group, among which top three CDR3 frequent paired heavy and light chain clonotypes are IGHV2-2/IGKV12-46, IGHV1-64/IGKV6-15 and IGHV1-15/IGKV6-23 (Fig. 4A–C). CDR3 heavy chain diversity of the RNH1-S1139 immunization group is significantly lower compared with S Trimer, but upregulation of transcriptional regulators Bcl6 and IRF8 indicates successful activation of B cells in germinal center (Fig. S9) [47]. This finding supports the hypothesis that RNH1-S1139 triggers a more epitope-focused B cell repertoire and thus gives rise to less class-switched memory B cells. CDR3 region of the light chain did not show significant differences among three tested groups (Fig. 4D).

Fig. 4

Antibody responses elicited by RNH1-S1139 were directed against the S1139 epitope, and is suitable to serve as a heterologous booster vaccine to combat immune imprinting. A Clonal overlap between PB, S Trimer and RNH1-S1139 groups was visualized by Venn charts. B Heatmap showing the percentage of B cell clones with a particular VH-VL pairing. Enriched VH and VL genes with a frequency greater than 1.5% were shown in each group. C Circos diagram representation of the IGHV and IGKV frequency in RNH1-S1139 and S Trimer immunization groups and their respective pairing. Top 3 pairs were highlighted in bold. D Summary of IGH and IGL CDR3 diversity. E Binding ability against S1139 of selected monoclonal antibodies from S trimer or RNH1-S1139 immunization groups determined by ELISA. OD value equal to or higher than 1.5-fold negative control value (0.51) was defined as positive binding. F Binding antibody titers in human sera against S trimer and S1139 epitope elicited by two doses of S trimer immunization followed by Omicron breakthrough infection. G Upper, immunization and serum collection timeline of heterologous prime-boost regimen. Lower, S1139-specific IgG titer from different immunized groups. N = 3 mice per group in A-D and 4–6 mice per group in G. Data are presented as mean values with SD in G. p values were determined by One-way ANOVA followed by Dunnett’s multiple comparison test. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001

Monoclonal antibody sequences with top nine mapped reads were retrieved from each group and their binding abilities to the S1139 epitope were tested. For the RNH1-S1139 immunized group, six out of nine antibodies show strong binding tendency to the S1139 epitope, while for the S trimer immunized group, selected antibodies show relatively low binding ability to the S1139 epitope (Fig. 4E). Isolation of high-affinity antibodies against the target epitope from enriched B cell samples indicates that the antigen we designed can effectively stimulate germinal center B cell responses. Additionally, we determined that six antibodies with high binding abilities to S1139 epitope isolated from RNH1-S1139 immunization group have competitive binding activities against mAb B6 (Fig. S10).

To explore whether antibodies in human sera elicited by vaccination plus viral infection target the S proteins and S1139 epitope, the serum of 20 participants who received two doses of wild-type S Trimer COVID-19 vaccination followed by SARS-CoV-2 Omicron infection were collected and tested against S trimer, Omicron BA.4/5 variant S trimer, and S1139. Binding was observed for both spikes tested, with lower titer against SARS-CoV-2 Omicron BA.4/5, while both spikes showed significantly higher binding titers compared with S1139 epitope (Fig. 4F). Despite being a possible vaccine target, the conserved S1139 epitope is infrequently targeted following vaccination or by natural infection due to the presence of an immune dominant region. The facts that lower Omicron BA.4/5-binding antibody titer was elicited in human serum and only a small fraction of S1139-directed humoral responses elicited emphasize that vaccination strategies to counter immune imprinting are critically needed [48, 49].

Heterologous prime-boost regimen can better stimulate unique immune responses to improve immunogenicity than homologous regimen [50,51,52]. To restore the general process of heterologous booster immunization, after completing two shots of prime immunization, mice were boosted with full-length S Trimer (W.T. or Omicron BA.4/5) or RNH1-S1139 at 6-week intervals and S1139-specific IgG titers were tested (Fig. 4G). Significantly higher S1139-specific IgG titer was detected after the RNH1-S1139 booster vaccination and distinct types of high-frequency IGHV and IGKV were stimulated after RNH1-S1139 and S Trimer immunization, which indicated that RNH1-S1139 could be used not only as a heterologous booster vaccine to provide broader protection but also as a strategy that would be better adapted to the immunological imprinting observed in SARS-CoV-2 infection (Fig. 4C, G). These findings revealed that S1139-specific antibodies are rarely elicited by vaccination or natural infection. Thus, the value of RNH1-S1139 as heterologous booster vaccine to provide broader protection against SARS-CoV-2 variants should be evaluated.

RNH1 is a versatile skeleton to display helical epitopes from various antigens

Due to the leucine-rich repeats that form the solenoid protein domain of RNH1 and the success in displaying the S1139 epitope, it is possible that other helical epitopes could be presented on the outer layer of RNH1. Hence, we extend the usage of RNH1 as a skeleton to display helical epitopes from RNH1-S1139 to other antigens.

Fusion peptide, part of the S2 subunit, is buried in the prefusion state of SARS-CoV-2. Cleavage of S by transmembrane protease serine 2 (TMPRSS2) results in exposure of the fusion peptide, which is necessary to initiate the formation of the fusion pore that ultimately mediates fusion between viral and cellular membranes, allowing the viral genome to enter the cell [45]. Antibody 76E1 targeting the fusion peptide has been reported to have broad-spectrum neutralizing activity among multiple alpha- and beta-coronaviruses, which might afford protection against emerging variants (Fig. 5A) [31].

Fig. 5

RNH1 serves as a versatile scaffold to display various epitopes. A, B Crystal structure of FP peptide in complex with 76E1 (PDB 7X9E), and Fab D25-RSV fusion glycoprotein (PDB 4JHW). Targeted peptides were highlighted in red and green. Structure prediction of RNH1-FP and RNH1-RSV-F were performed using AlphaFold2 v.1.5.3. C, D Binding affinity of designed antigens and indicated mAb determined by ELISA (C) and BLI (D). E Epitope-specific antibody titers induced by immunization with designed antigens or controls determined by ELISA. N = 6 mice per group in E. Data are presented as mean values with SD in E. p values were determined by One-way ANOVA followed by Dunnett’s multiple comparison test. ^****p < 0.0001

RSV infects ciliated epithelial cells in the lungs and respiratory tract and is a major cause of respiratory illness in children. Childhood RSV infection might lead to chronic diseases, such as asthma [53]. RSV membrane envelope glycoproteins G (for attachment) and F (for fusion) are the main targets in vaccine development and drug therapeutics. We screened and selected neutralizing antibodies that bind to linear epitopes of RSV-F, among which, mAb D25 locks F in its prefusion state by binding to linear epitope located at the trimer apex (Fig. 5B) [35].

Plasmids containing modified sequences of epitopes and RNH1 were constructed, and proteins were purified from procaryotic cells, namely RNH1-FP (₈₁₆SFIEDLLFNKV₈₂₆), and RNH1-RSV-F (₁₉₃LDLKNYIDKQLLPIVNKQSCS₂₁₃) (Additional file 1; Table S1 & Fig. S2). Structures of RNH1 displaying selected epitopes were predicted by AlphaFold2 (Fig. 5A, B). Next, the binding ability of the antigen to respective monoclonal antibodies was tested using both ELISA and BLI. Both assays showed that two antigen designs maintained high binding affinity to their monoclonal antibodies (Fig. 5C, D and Fig. S4). Next, engineered antigens containing the epitopes of interest were used to vaccinate mice. Serum was collected and antibody titers against RNH1-FP and RNH1-RSV-F epitopes were measured. RNH1-FP and RNH1-RSV-F elicited significantly higher epitope-specific antibody titers compared with epitopes presented on origin protein, 56 days after first immunization (Fig. 5E). These epitopes were successfully transplanted to RNH1-based backbone, providing indications of RNH1 scaffold versatility, and further supporting that our approach can be used to present alpha helix motifs of diverse antigens.

Discussion

To minimize viral transmission after vaccination and/or infection, variant-specific SARS-CoV-2 booster shots were promoted [54, 55]. However, the efficacy of single heterologous booster vaccine is limited under the influence of wildtype S-induced immune imprinting since anti-S responses preferentially target the immunodominant, antigenically hypervariable region [48]. Development of vaccine immunogens using conserved and immunerecessive regions that are resilient to viral evolution is therefore necessary. Epitope-based vaccine that is engineered to induce desired immune responses is becoming an increasingly popular approach to vaccine development.

Mimicking the natural epitope structure and enhancement of epitope immunogenicity are two major obstacles for epitope-based vaccine immunogen design. Recently established strategies, including integration of the target epitopes into other scaffolds by side chain grafting, have been evaluated in RSV and SARS-CoV-2, showing the feasibility of scaffold-based vaccine design, but they rely on other techniques to achieve the goal of epitope multiply display [56,57,58]. Using the structural feature of horseshoe-shaped protein RNH1, we present a new concept of epitope multiply presentation. The vaccine immunogens of SARS-CoV-2 and RSV we designed can display the target epitopes regularly and repeatedly, and, at the same time, improve epitope antigenicity, which is crucial for effective immune response stimulation. LRRs have been brought into focus in structure-based design because of its helically twisted surface that is suitable for protein–protein interaction [59]. More than two thousand LRR proteins have been identified from viruses, prokaryotes and eukaryotes; geometrical features like length, curvature, and helical twist could be optimized during antigen engineering to better present epitopes (https://www.ebi.ac.uk/interpro/entry/pfam/PF13516/structure/PDB/#table) [26, 59, 60]. The arc or horseshoe shape of the LRR proteins offered favorable natural advantages in protein–protein or protein–ligand interaction. The LRR proteins had the potential to be explored as protein–protein interaction inhibitors using the inner and the outer face after modification according to a hydrophilic or hydrophobic environment [61, 62]. The inner face functions as a “scarf” to embrace the essential interaction region. The outer face functioned like a traditional peptide binder. In addition, the LRR proteins also achieved the goal of targeted drug delivery [63]. Thus, the LRR proteins represent highly adaptable and evolvable motifs for protein–ligand interactions, which are ubiquitous in a vast array of proteins exhibiting diverse functionalities.

Generation of high-affinity protective antibodies requires class-switch recombination and somatic hypermutation of B cell in the geminal center. Single-cell VDJ sequencing results reveal that S trimer elicited a broader B cell specific response, while RNH1-S1139 promoted a less diverse B cell response targeting a neutralizing epitope. The 16-fold epitope presentation enables more efficient S1139 recognition by the BCR and initiates subsequent somatic hypermutation and affinity maturation in the germinal center. Similar findings were reported on HIV. Display of polymer HIV antigen in a nanoparticle drives an expanded and earlier germinal center reaction compared with oligomer [64].

We subsequently proved that RNH1-S1139 inhibited viral infection by blocking S-mediated cell fusion. Furthermore, antibody Fc domain engages a wide range of receptors that contribute to in vivo protection, promoting viral clearance and antiviral immune responses via Fc receptor activation, Ab-dependent cell-mediated cytotoxicity or Ab-dependent cellular phagocytosis [65]. These findings suggest that antibodies might induce protective adaptive immunity by selective activation of the dendritic cell and T cell pathway, even if lacking neutralizing abilities. Additional research is needed to test the protective breadth of our RNH1-S1139 vaccine immunogen and to study the underlying protective mechanisms elicited by RNH1-S1139.

In addition, we report that generation of S1139-targeting antibodies by vaccination, combined with natural infection, is limited in human serum samples emphasizing the potential value of RNH1-S1139 as booster vaccine to provide broader protection. RNH1-S1139 elicits higher S1139-specific IgG titer than both SARS-CoV-2 WT and Omicron BA.5 booster vaccinations. The RNH1-S1139 antigen we constructed elicited intense and durable immune responses targeting the conserved but less-exposed S1139 epitope. The present study reports an immunogen design strategy and is a proof of concept for epitope multiply display using natural protein scaffold.

Competing interests

The authors declare no competing interests.