Vaccine design via antigen reorientation

Rapid viral evolution in influenza and SARS-CoV-2 drives recurrent outbreaks and undermines vaccine efficacy. Seasonal influenza vaccines primarily elicit antibodies to the variable HA head, necessitating frequent updates and offering limited protection against pandemic strains. Focusing responses on conserved epitopes in the HA stem is a promising strategy, supported by human broadly neutralizing antibodies (bnAbs) that target HA-stem and correlate with protection. Existing immunofocusing strategies often require complex regimens or compromise immunogenicity. This study tests the hypothesis that reorienting intact HA on alum via genetically encoded aspartate-rich tags can sterically occlude the HA head and increase exposure and immunogenicity of the HA stem, thereby redirecting antibody responses toward conserved, broadly protective epitopes.

Prior approaches to elicit stem-focused immunity include sequential or mosaic immunizations using heterologous HAs or chimeric HAs sharing a common stem but variable heads, which can be complex and may not overcome HA-head immunodominance. Additional strategies such as glycan masking or PEGylation of the head can reduce overall immunogenicity in vivo, while structure-based HA-stem-only immunogens require laborious design to preserve conformational epitopes. These methods have achieved group-restricted breadth but have struggled to elicit responses that consistently cross-react with both group 1 and group 2 HAs using a single immunogen. Enhancing antigen binding to alum using phosphoserine tags improves humoral responses but involves peptide synthesis and conjugation steps. There remains a need for a simple, generalizable method to control antigen presentation and orientation to focus responses on conserved epitopes.

Design and antigen engineering: Short oligomers of aspartate (oligoD; 2D, 4D, 8D, 12D) were genetically inserted at defined positions on viral glycoprotein antigens to mediate binding to alum via electrostatic interactions. Test antigens included Ebola virus GP ectodomain (with mucin-like domain deleted), SARS-CoV-2 spike ectodomain (Wuhan-Hu-1, 2P-stabilized trimer context referenced by PDB 6VXX), and influenza HAs (H1 A/New Caledonia/20/99; H2 A/Japan/305/1957; additional HAs for cross-reactivity). For HA reorientation, 12D inserts were placed in the HA head of H2 after residue S156 to generate reoH2HA, enabling trivalent alum anchoring of each HA trimer through three head-inserted oligoD motifs. Control constructs included C-terminal 12D insertions (upright orientation) and biotinylated variants for orientation assays. Protein expression and characterization: Antigens and mAbs were expressed in Expi293F cells and purified via Ni-NTA (antigens) or Protein A (antibodies). Quality control included size-exclusion chromatography (SEC), SEC–MALS for oligomeric state/mass, differential scanning fluorimetry for thermal stability with and without alum, and bio-layer interferometry (BLI) for binding to antigen-specific mAbs or ACE2-Fc (spike) to confirm epitope preservation. Permissive sites for oligoD insertion within GP loops (R200, T294, A309) were screened. Alum binding assays: Antigen–alum mixtures (protein:alum = 1:10 w/w) were incubated 1 h at room temperature then 24 h at 37 °C (10% naive mouse serum where indicated). Bound antigen in alum pellets was assessed by Western blot; unbound antigen in supernatants quantified by ELISA. Immunogold labeling and TEM visualized antigen on alum. Binding to aluminum phosphate (Adju-Phos) was tested to assess charge dependence. Orientation assays: Streptavidin- vs alum-coated ELISAs compared accessibility of head- vs stem-directed epitopes on reoH2HA. Biotinylated H2 HA-12D or reoH2HA was captured on streptavidin plates; alum plates were built by capturing ZsGreen-12D, incubating with alum, then adding H2 HA-12D or reoH2HA. Binding profiles of head-directed (8F8, 8M2) and stem-directed (MEDI8852, FI6v3) mAbs were measured. Mouse immunizations: Female BALB/c mice received subcutaneous injections of 5 µg antigen with adjuvants. Regimens: GP or GP-12D single dose with alum (150 µg); spike or spike-12D three doses (days 0, 21, 49) with alum; H1 HA or H1 HA-12D prime-boost (days 0, 56) with alum; H2 HA or reoH2HA three doses (days 0, 21, 70) with alum (150 µg) plus CpG ODN 1826 at 1 µg or 5 µg. Serum was collected longitudinally. Immunogenicity assays: ELISAs measured antigen-specific IgG and IgG subclasses; stem-directed binding quantified using H1-stabilized stem (H1-SS). Competition ELISAs pre-incubated plates with bnAbs (MEDI8852, CR9114, FI6v3) or anchor-directed 222-1C06, or trimer-interface FluA-20 to assess epitope targeting. Cross-reactivity ELISAs used biotinylated HAs from group 1 (H1 NC/99, H1 CA/09, H5 VT/04) and group 2 (H3 VC/75, H7 NT/27, H7 SH/13). Neutralization assays: Pseudovirus neutralization employed EBOV GP- or SARS-CoV-2 spike-pseudotyped lentiviruses with luciferase/ZsGreen reporters in HEK293T or HeLa-ACE2/TMPRSS2 cells; NT50 calculated from technical replicates. Authentic influenza microneutralization in MDCK.2 cells quantified infection via nucleoprotein staining; validated with stem bnAbs (MEDI8852, CR9114, FI6v3). IgG purified from pooled sera (H2 HA vs reoH2HA) was tested against A/California/7/2009 (H1N1). Germinal center (GC) analysis: Flow cytometry quantified GC B cells (CD19+ CD95+ CD38−), IgG+ GC B cells, and T follicular helper cells (CD3− CD4+ PD1+ CXCR5+) in draining lymph nodes at days 7, 14, 21 after GP or GP-12D immunization. Polyclonal epitope mapping: nsEMPEM analyzed Fab-polyclonal immune complexes with H2, H7, and H3 HAs from pooled antisera at weeks 7 and 12. Negative-stain EM reconstructions and 2D class averages identified targeted epitope regions (head, stem, anchor, apex; ambiguous sites).

• OligoD length controls alum binding without perturbing antigen structure: Insertion of 2–12 aspartates at the GP C-terminus progressively increased alum binding, with 12D achieving ~100% alum binding; thermal melting temperature (~58 °C) and BLI mAb binding profiles were unchanged vs wild-type. • Generalizable enhancement of immunogenicity: GP-12D with alum elicited ~10-fold higher GP-specific IgG titers by week 6 and 3–5-fold stronger neutralizing responses vs GP. All GP-12D mice developed neutralizing responses by week 6, while only ~50% of GP mice did; one GP mouse remained non-neutralizing at week 12. • Robust GC responses: GP-12D increased GC B and Tfh responses vs GP (e.g., day 14 GC B and Tfh percentages elevated; figure values indicate GC B 7.52 vs 3.04; Tfh 0.23 vs 0.073), supporting improved germinal center activity. • SARS-CoV-2 spike-12D: Maintained ACE2 and mAb binding; in a 3-dose alum regimen, spike-12D increased spike- and RBD-specific IgG titers and NT50 against D614G pseudovirus by endpoint (week 9), and improved neutralization against variants (Beta, Delta, Omicron BA.1/BA.2). • H1 HA-12D: Preserved conformational epitope recognition; prime-boost with alum yielded 5–10-fold higher H1-specific IgG titers and higher NT50 against authentic H1N1 A/NC/20/99 vs H1 HA. Competition ELISA showed stronger competition with head-directed mAbs (H2897, 6649) and negligible competition with stem-directed bnAbs (MEDI8852, CR9114), indicating boosting of head-directed responses with C-terminal 12D. • Reorientation of H2 HA (reoH2HA): Insertion of 12D after S156 in HA-head enabled ‘upside-down’ anchoring on alum. On streptavidin plates, head- and stem-directed mAbs bound similarly to H2 HA and reoH2HA; on alum plates, only stem-directed mAbs bound reoH2HA, confirming head occlusion. • Stem-focused response and breadth: With alum/CpG, H2 HA and reoH2HA elicited similar H2-specific IgG, but reoH2HA induced ~10-fold higher stem-specific titers (H1-SS binding). Cross-reactive ELISAs showed significantly higher titers from reoH2HA against group 1 (H1 NC/99, H1 CA/09, H5 VT/04) and group 2 (H3 VC/75, H7 NT/27, H7 SH/13) HAs after two and three doses (multiple comparisons P ≤ 0.0021 to <0.0001 at week 7, depending on antigen). • Epitope targeting: Antisera from reoH2HA competed more strongly with stem bnAb MEDI8852 and with anchor-directed 222-1C06 than antisera from H2 HA; neither group competed with trimer-interface antibody FluA-20. • nsEMPEM: Both antigens elicited head, stem, and anchor responses on H2 HA; reoH2HA additionally elicited apex-directed antibodies (likely near the oligoD region). Critically, reoH2HA induced cross-reactive stem-directed antibodies to group 2 HAs (H7 HA; also low-level H3 HA-stem), whereas H2 HA did not show these stem responses in group 2 contexts. • Functional cross-neutralization: Purified IgG from reoH2HA-immunized mice showed modest neutralization of heterosubtypic A/California/7/2009 (H1N1), while IgG from H2 HA-immunized mice did not show significant cross-neutralization. • Adjuvanting and Th bias: Increasing CpG from 1 µg to 5 µg with high-dose alum (150 µg) maintained a Th2-biased response (IgG1 >> IgG2), though reoH2HA still enhanced stem-directed titers and breadth.

This study demonstrates that genetically programmed antigen reorientation on alum via oligo-aspartate insertion redirects the immune response from dominant but variable epitopes to conserved neutralization targets. By anchoring the HA head to alum, reoH2HA sterically occluded head epitopes and increased accessibility of the stem within an intact trimer, preserving native stem conformation. The approach was validated across multiple antigens (Ebola GP, SARS-CoV-2 spike, influenza HAs), where oligoD enhanced alum binding and immunogenicity without disrupting structural integrity or key epitope presentation. In the influenza model, reoH2HA consistently elicited higher stem-directed antibody titers and superior cross-reactivity across both group 1 and group 2 HAs, addressing a key challenge of broadening responses with a single immunogen. Competition ELISAs and nsEMPEM confirmed targeting of stem epitopes, including cross-reactive responses to group 2 HA stems. Although cross-neutralization was modest, the breadth and stem focus suggest potential for protection via Fc-mediated mechanisms and for further optimization with adjuvants/formulations favoring balanced Th1/Th2 responses. The simplicity of cloning-based oligoD insertion and the potential for multivalent alum anchoring make antigen reorientation a practical, scalable, and generalizable immunofocusing strategy.

Antigen reorientation using site-specific oligo-aspartate insertion enables controlled orientation on alum to modulate epitope accessibility and focus humoral immunity on conserved targets. This generalizable method increased alum binding and enhanced immunogenicity for Ebola GP, SARS-CoV-2 spike, and influenza HAs. An H2 HA engineered with head-inserted oligoD (reoH2HA) adopted an ‘upside-down’ configuration on alum, eliciting robust stem-directed responses with cross-reactivity spanning both group 1 and group 2 influenza A HAs and modest heterosubtypic neutralization. The approach preserves native antigen structure while mitigating head immunodominance. Future work should optimize adjuvant formulations to enhance Th1-associated effector functions, assess protection in challenge models, refine insertion sites and charge motifs, and explore compatibility with alternative adjuvants or delivery platforms to translate reorientation-based immunofocusing to other pathogens.

• Cross-neutralization achieved by reoH2HA-induced antibodies was modest; protective efficacy against heterologous challenge remains to be established and may rely on Fc-mediated effector functions. • High alum dosing with CpG produced a Th2-biased response (IgG1-dominant); further adjuvant optimization is needed to achieve balanced Th1/Th2 and potentially stronger effector functions. • The reorientation strategy depends on alum binding; compatibility with non-aluminum adjuvants (e.g., emulsions) is uncertain and may require alternative tags or strategies. • Identifying permissive insertion sites is protein-specific; some head insertions reduced HA expression, necessitating screening to preserve expression and structure. • Potential neo-epitopes near insertion sites (e.g., apex signals in nsEMPEM) warrant evaluation to avoid off-target responses.