Computationally designed Spike antigens induce neutralising responses against the breadth of SARS-COV-2 variants

Since late 2019, SARS-CoV-2 has accumulated mutations that alter host interactions, tropism and transmission, with notable immune escape as human population immunity increased via infection and vaccination. Variants of concern (VOCs) such as Alpha, Beta, Gamma, Delta and the Omicron lineage evolved primarily through changes in the Spike protein, reducing neutralisation by antibodies elicited by first-generation Wu-Hu-1–based vaccines. With multiple co-circulating strains across geographies, reactive updates to match a single dominant strain (including bivalent formulations) have shown limited breadth and may be impacted by immune imprinting. The authors hypothesised that vaccine candidates expressing diverse, immunodominant epitopes from multiple VOCs would be more effective than monovalent or bivalent wild-type strain-matched vaccines. They designed three computationally engineered Spike antigens—T2_32 (pre-Delta design), and T2_35 and T2_36 (post-Gamma/Omicron-era designs)—to test whether multiepitope antigens can induce broader neutralising responses against past, contemporary and future VOCs.

The paper situates its approach within evidence that: (1) successive SARS-CoV-2 VOCs show increased immune escape and transmission; (2) neutralising activity elicited by historical Wu-Hu-1 vaccines wanes against later VOCs, especially Delta and Omicron sublineages; (3) bivalent boosters including ancestral + Omicron antigens yielded inferior breadth versus monovalent Omicron boosters, likely due to immune imprinting; and (4) rapid variant evolution and geographic diversity complicate strain-matching, echoing challenges in seasonal influenza vaccine updates. The authors also reference broader approaches to pan-sarbecovirus or variant-proof vaccines (e.g., multivalent RBD nanoparticles, consensus-based SPAN) and contrast these with their DIOSynVax strategy, which assembles specific immunodominant mutations from VOCs into novel Spike constructs to broaden responses.

In-silico antigen design: The Spike protein was partitioned into NTD, RBD (plus S1-CTD), and S2 regions. Using MAFFT-based multiple sequence alignments and IEDB epitope data, mutations in immunodominant epitopes from Alpha, Beta, and Gamma (Feb. 2021) were mapped onto a Wu-Hu-1 scaffold to create T2_32, incorporating key mutations (e.g., K417N/T, E484K, N501Y; selected NTD changes from Alpha and Gamma; S1-CTD P681H; D614G; Q498R), stabilising substitutions K986P/V987P, and deletion of 19 C-terminal residues (dER). For Omicron-era designs (Dec. 2021), two Delta–Omicron recombinatorial strategies were used: T2_35 with Delta-like mutations in NTD/S2 and Omicron BA.1-like mutations in RBD/S1-CTD, plus Q677H, I834V, K986P/V987P, GSAS furin site substitution, and dER; and T2_36 with Omicron BA.1-like mutations in NTD/S2 and Delta-like RBD/S1-CTD (including K417N), plus K986P/V987P, GSAS furin site substitution, and dER. Structural integrity was checked via Modeller. A furin-site–modified T2_32 (T2_32_mFur) was also generated. Vaccine constructs and production: DNA vaccine plasmids (pEVAC) encoding T2_32 or Wu-Hu-1 dER were codon-optimised and produced in E. coli. Recombinant MVA-CR19 expressing T2_32 was generated via homologous recombination in CR.PIX cells and plaque-purified. For mRNA studies, partially modified, capped, polyadenylated IVT mRNAs encoding T2_35, T2_36, T2_32_mFur, Wu-Hu-1 dER, or Omicron BA.1 dER were LNP-formulated and quality-controlled (particle size ≤100 nm, PDI ≤0.20, encapsulation ≥90%, integrity ≥90%). Animal immunisations: Guinea pigs (n=4/group) received intradermal DNA immunisations at 0, 14, 28, and day 70 (200 µg each; Wu-Hu-1 dER or T2_32), then a heterologous IM boost at day 112 with MVA-T2_32 (2.0E PFU/dose). Serial bleeds were collected; neutralisation was assessed after the third DNA dose (bleed 4) and post-MVA (bleed 6). BALB/c mice (female, 8–10 weeks; n=6/group) received two IM doses of LNP-mRNA (10 µg) 21 days apart; bleeds at days 42, 63, and terminal. Neutralisation assays: Lentiviral pseudotypes bearing Wu-Hu-1 or VOC spikes (Alpha, Beta, Gamma, Delta, Omicron sublineages including BA.1, BA.2, BA.2.75.2, BQ.1.1, XBB, XBB.1.5, BA.2.86) were produced in HEK293T/17 cells. Pseudotype microneutralisation assays incubated serially diluted sera with pseudoviruses prior to infection of ACE2/TMPRSS2-expressing HEK293T/17 cells. Luminescence readouts yielded IC50 values via non-linear regression. Statistics used two-tailed Mann–Whitney U tests for pairwise comparisons.

- T2_32 (DNA prime–boost) in guinea pigs elicited broader neutralisation than Wu-Hu-1 dER after three DNA doses, with ≈1-log higher titres against most VOCs except Wu-Hu-1 and Delta where titres were comparable. Higher titres to BA.1 and BA.2 were observed only with T2_32. - Heterologous MVA-T2_32 boost raised neutralisation by at least an order of magnitude across the VOC panel in both groups and significantly enhanced titres against BA.1, BA.2, and XBB, with notably higher titres to XBB.1.5 in T2_32-primed animals (Mann–Whitney U; p ≤ 0.05 to ≤ 0.001 where indicated). - In mice (mRNA): T2_35 and T2_36 induced broad neutralisation across pre-Omicron and Omicron variants. T2_35 showed higher titres from BA.1 onward (later Omicron sublineages), whereas T2_36 showed higher titres to pre-Omicron VOCs (Alpha–Delta), consistent with RBD immunodominance and the designs’ mutation placements. - T2_35 achieved titres comparable to BA.1 antigen against Omicron variants while significantly exceeding BA.1 against Delta. T2_36, while lower than BA.1 against Omicron sublineages, induced significantly higher titres than BA.1 against non-Omicron VOCs. - T2_32_mFur (mRNA) showed superior titres to Wu-Hu-1 across VOCs and comparable titres to BA.1 against several Omicron sublineages, likely driven by shared immunodominant-region mutations (e.g., HV69-70 deletion, Y144 deletion, K417N, Q496R, N501Y, P681H). - Neutralising titres were lower against highly divergent recent sublineages (e.g., BQ.1.1, XBB.1.5) than against earlier variants (e.g., Delta, BA.2), indicating the need for periodic design updates. - Overall, the DIOSynVax multiepitope designs produced broader neutralisation profiles, including against variants detected >2 years after the initial designs.

The study addresses the challenge that strain-matched, historical Spike vaccines provide limited protection against newly emerging immune-evasive variants and can be constrained by immune imprinting. By incorporating key immunodominant mutations from multiple VOCs into single Spike constructs, the DIOSynVax designs elicited broader neutralising responses than wild-type Wu-Hu-1 or Omicron BA.1 controls in small animal models. The distinct neutralisation patterns of T2_35 (enhanced against Omicron sublineages) versus T2_36 (enhanced against pre-Omicron VOCs) support the centrality of RBD-focused immunodominance and rational placement of mutations. Importantly, heterologous boosting with T2_32 expanded breadth even in Wu-Hu-1–primed animals, suggesting these non-wild-type boosters can overcome imprinting and better future-proof immune responses. While titres diminished against very recent, highly divergent Omicron sublineages (e.g., BQ.1.1, XBB.1.5), the approach remained broadly effective across >2 years of variant evolution, highlighting a proactive antigen design strategy superior to reactive strain-matching.

Computationally engineered multiepitope Spike antigens (T2_32, T2_35, T2_36) induced broad neutralising responses across historical and contemporary SARS-CoV-2 VOCs in guinea pigs and mice, outperforming or matching wild-type Wu-Hu-1 and Omicron BA.1 controls. Heterologous boosting with T2_32 significantly broadened responses in Wu-Hu-1–primed animals, underscoring utility as next-generation booster candidates. The DIOSynVax approach, distinct from consensus-based strategies, leverages specific immunodominant mutations to proactively target future variants. Future work should update designs to address highly divergent sublineages (e.g., XBB/BQ lineages), validate protection in challenge models and humans, and further dissect immunodominant-region contributions to breadth and durability.

- Neutralisation against highly divergent, recent Omicron sublineages (e.g., BQ.1.1, XBB.1.5) was reduced compared with earlier variants, indicating potential need for iterative antigen updates as variants evolve. - Findings are from small animal models (guinea pigs n=4/group; mice n=6/group) using pseudotype-based microneutralisation assays; protective efficacy against live virus and in humans was not assessed. - The study focuses on humoral neutralisation; breadth and durability of T-cell responses were not reported. - Antigen performance may be influenced by immune imprinting histories not fully recapitulated in naïve or limited-prime animal models.