Self-assembling nanoparticle engineered from the ferritinophagy complex as a rabies virus vaccine candidate

The study addresses persistent manufacturing and immunogenicity challenges in ferritin-based nanoparticle vaccines, including particle heterogeneity from chemical conjugation, folding/assembly interference in gene fusions, and unintended immunogenicity from bulky adaptors (e.g., Tag/Catcher systems). Leveraging the physiological interaction between FTH1 and NCOA4 in ferritinophagy, the authors hypothesize that a minimal peptide adapter engineered from the NCOA4 binding region can bind ferritin with high affinity and specificity, enabling efficient, homogeneous antigen display. As a proof-of-concept, they target rabies virus (RABV), a lethal and neglected disease, focusing on the conserved glycoprotein domain III (GDIII) that contains key neutralization epitopes. The goal is to develop a stable, immunogenic, and potentially dose-sparing ferritin nanoparticle vaccine capable of rapid and durable protection, and to demonstrate the platform’s broader applicability.

Ferritin nanoparticles are versatile, biocompatible carriers forming 24-mer protein cages (FTH/FTL) that present repetitive antigen arrays, enhancing B cell activation. Prior ferritin-based vaccines have used chemical conjugation, genetic fusion, or Tag/Catcher strategies, but face issues such as particle heterogeneity, misfolding and impaired assembly, and immunogenicity against non-essential tags. CD71/FTH1 interactions are extensive (~1900 Å2) and unsuitable for adapter design. NCOA4 is a selective ferritin cargo receptor; its C-terminal region binds FTH1, mediating ferritinophagy. Recent work (Frank et al.) reported a 2.9 Å cryo-EM structure of NCOA4/FTH1, confirming key residues. For rabies, extensive vaccine modalities exist (inactivated, live-attenuated, recombinant, nucleic acid, oral), but many require multiple doses. RABV-G is the sole surface glycoprotein and neutralization target; GDIII (PHD domain) is conserved and bound by potent neutralizing antibodies (e.g., RVC20), blocking pH-triggered fusion. There remains a need for safer, more stable, cost-effective vaccines with optimized immunogenicity and lower dosing regimens.

• Structural determination and peptide engineering: Prepared NCOA4 C-terminus (475–511) with human FTH1 and solved the complex by single-particle cryo-EM at 2.2 Å. Mapped the interaction (NCOA4 484–499 forms two short α-helices engaging FTH1 Helix A, Helix C, and BC loop; hydrophobic cores and salt bridges/hydrogen bonds defined). Designed 298 peptide variants with ProteinMPNN, evaluated structures by AlphaFold2 and Rosetta (ddg, contact_molecular_surface, score), filtered to 146, then selected 16 top sequences (PLDDT > 0.944) for testing. Peptides were displayed as GST fusions and screened by ELISA for FTH1 binding.
• High-affinity adapter development: Identified Peptide 10 (EC50 ~53.11 nM vs WT ~288.5 nM). Performed restricted co-design of peptide/ferritin to generate Peptide 10-1 with FTH1-1, achieving EC50 ~34.08 nM. Determined structures: Peptide 10/FTH1 by cryo-EM at 2.02 Å; Peptide 10-1/FTH1-1 by X-ray crystallography at 2.3 Å. Compared binding modes and defined the engineered Fagy tag (Peptide 10-1).
• Vaccine construction and biophysical characterization: Fused GDIII to Peptide WT, Peptide 10, or Fagy tag; measured SPR binding to ferritin (FTH1 or FTH1-1). Produced GDIII-Ferritin nanoparticles by mixing purified subunits for self-assembly. Characterized by analytical ultracentrifugation (AUC) to estimate antigen valency, DLS for size/stability (including 4 °C storage up to 20 days, freeze–thaw), negative-stain EM with 2D classification and 3D reconstruction (O symmetry), and thermal shift assays across pH and denaturants.
• Antigenicity assays: Measured binding of neutralizing antibodies Docaravimab, Rafivirumab, and RVC20 to GDIII antigens and nanoparticles by SPR.
• Immunogenicity in mice: Female BALB/c mice received subcutaneous vaccinations (Freund’s adjuvant) with GDIII-Ferritin, GDIII monomer, GDIII-Fc, or full-length G at equal mass (50 μg) in a prime–boost–booster (weeks 0, 3, 6). Assessed serum antigen-specific IgG (ELISA) at weeks 2, 5, 8, 11; neutralizing titers (FAVN using CVS-11). Evaluated plasma cell and memory B cell frequencies (flow cytometry), T helper (CD4+ IFN-γ/IL-4) and cytotoxic T cell (CD8+ IFN-γ) responses, and cytokines (IFN-γ, IFN-α, IL-4, Granzyme B) pre/post T-cell activation.
• Protection studies: Six groups (n=10/group) including GD-Ferritin single-dose (30 or 60 μg), GD-Ferritin two-dose (30 μg/dose, days 0 & 7), commercial inactivated rabies vaccine BRP (0.1 dose; single- or two-dose), and Mock. Challenged intracerebrally with 50× LD50 CVS-24 on day 14. Monitored body weight, clinical signs, survival for 21 days. Assessed brain viral antigen by DFA and viral load by qRT-PCR. Measured neutralizing titers at day 21 in survivors and durability up to 9 months post two-dose GD-Ferritin.

• Structural insights: NCOA4 (residues 484–499) binds FTH1 via two α-helices forming extensive hydrophobic cores and specific salt bridges/hydrogen bonds. Cryo-EM of NCOA4/FTH1 at 2.2 Å; Peptide 10/FTH1 at 2.02 Å; crystal structure of Peptide 10-1/FTH1-1 at 2.3 Å showed virtually conserved binding mode with enhanced interactions.
• Engineered adapter affinity: Peptide WT EC50 ~288.5 nM for FTH1; Peptide 10 EC50 ~53.11 nM (~5.5× improvement); co-designed Peptide 10-1 with FTH1-1 EC50 ~34.08 nM (~8–9× vs WT). SPR: GDIII–Fagy tag binding to FTH1-1 KD ~2.90×10^-8 M vs GDIII–Peptide WT to FTH1 KD ~1.913×10^-7 M (~7× tighter).
• Antigen display: AUC and geometric analysis indicated ~19 GDIII antigens displayed per 24-mer ferritin (~80% occupancy). Negative-stain EM confirmed monodispersity and uniform antigen density; 3D reconstruction mapped spike distribution on ferritin.
• Stability and antigenicity: GDIII-Ferritin maintained particle integrity over 15–20 days at 4 °C, resisted multiple freeze–thaw cycles, and was stable across pH 5–10 and in 1 M urea or 20% FBS. GDIII-Fc formed ~30 nm particles but showed heterogeneity. SPR showed strong binding of neutralizing mAbs (Docaravimab, Rafivirumab, RVC20) to GDIII-Ferritin with minimal epitope shielding, indicating effective neutralization epitope presentation.
• Humoral immunity: GDIII-Ferritin elicited rapid and sustained high IgG-binding titers against GDIII and G after priming and boosts, surpassing GDIII monomer, GDIII-Fc, and G. Even at ~50% reduced molar dose, GDIII-Ferritin maintained substantial titers, highlighting valency effects. Neutralizing titers (FAVN) exceeded WHO protective thresholds and were substantially higher than comparators; reported as several-fold greater than GDIII-Fc and far greater than G.
• B cell responses: GDIII-Ferritin significantly increased splenic CD138+ plasma cells and IgG+ memory B cells in germinal centers, supporting durable humoral immunity.
• Cellular immunity: Th1-biased responses—elevated CD4+ IFN-γ+ T helper cells without significant Th2 (IL-4) increase; increased CD8+ IFN-γ+ cytotoxic T cells; higher IFN-α and Granzyme B consistent with strong cellular activation and innate signaling.
• Protection efficacy: After 50× LD50 intracerebral CVS-24 challenge on day 14: 100% survival in GD-Ferritin single-dose 60 μg and GD-Ferritin two-dose 30 μg/dose groups; partial protection in GD-Ferritin single-dose 30 μg (5/10 survived); BRP two-dose protected most (8/10 survived); BRP single-dose showed low survival (2/10 survived); Mock 0/10. Survivors in GD-Ferritin high-dose/two-dose had rapid recovery and minimal clinical signs. DFA showed no detectable virus in brains of surviving GD-Ferritin groups; qRT-PCR revealed significantly lower brain viral loads in GD-Ferritin groups than BRP two-dose and far lower than BRP single-dose. Neutralizing titers in survivors remained high; two-dose GD-Ferritin induced neutralizing antibodies durable for >9 months.

The work demonstrates that a minimal, structure-guided adapter derived from the natural NCOA4–FTH1 ferritinophagy interaction can drive high-affinity, noncovalent antigen display on ferritin with improved homogeneity, stability, and immunogenicity compared with conventional approaches. Structural elucidation at atomic resolution enabled targeted mutations that expanded hydrophobic cores and optimized hydrogen-bond/salt-bridge networks, explaining affinity gains. When applied to GDIII from RABV-G, the Fagy-tag ferritin platform achieved high antigen valency (~19/24 sites), preserved neutralization epitopes, and produced robust Th1-skewed humoral and cellular responses. Functionally, single-dose (60 μg) or two-dose (30 μg/dose) immunization afforded complete protection against a stringent intracerebral challenge, with rapid viral clearance in the CNS and durable neutralizing antibodies. Compared with bulky conjugation systems (e.g., SpyCatcher), the 16-aa Fagy tag potentially reduces non-essential immunogenic components while maintaining strong binding and ease of assembly. The platform is compatible with terminal fusions and could support multi-antigen display. Overall, these findings support the Fagy-tag ferritin system as a versatile, cost-effective nanoparticle vaccine strategy that addresses prior limitations in ferritin-based vaccines and offers translational promise for diverse pathogens.

This study introduces the Fagy tag, a 16-residue peptide engineered from the NCOA4–FTH1 interface, enabling high-affinity, noncovalent antigen display on ferritin nanoparticles. Structural data (cryo-EM and X-ray) guided affinity optimization and validated conserved binding modes. Applied to rabies GDIII, the GDIII-Ferritin vaccine displayed high antigen valency, exhibited excellent physicochemical stability, preserved neutralization epitopes, elicited potent Th1-biased humoral and cellular responses, and conferred complete protection in mice after single or two-dose regimens, with neutralizing antibodies maintained for at least 9 months. The approach reduces reliance on bulky adaptors and complex chemistries and is compatible with potential multi-antigen presentation. Future work should systematically evaluate the immunogenicity of non-antigen components (ferritin, Fagy tag), benchmark against alternative conjugation/fusion platforms in multiple species, optimize multi-antigen orientations/spacing, and advance to translational studies.

• Immunogenicity of non-antigen components (ferritin scaffold and Fagy tag) was not comprehensively characterized; dedicated profiling is needed.
• Direct, head-to-head comparisons with alternative conjugation strategies (e.g., genetic fusion, Tag/Catcher) were not fully executed.
• Multivalent, multi-antigen display (1:1 or 1:1:1) was proposed but not validated in vivo; spacing/orientation optimization remains to be addressed.
• Efficacy and safety were evaluated in female BALB/c mice with intracerebral challenge; broader species, sexes, administration routes, and clinically relevant challenge models are required for generalization.
• Durability beyond 9 months and dose-sparing across diverse adjuvants/regimens were not fully explored.