iBio (IBIO) Status update summary

Integration of AI in antibody drug discovery

• Deep integration of AI with experimental validation pipelines enables pre-validation of molecules, reducing risk and increasing efficiency in drug discovery campaigns.

• The platform accelerates timelines, achieving clinic approval for lead candidates like IBIO-600 in just two years.

• AI-driven engineered epitope designs allow for targeted antibody discovery, overcoming immune tolerance and enabling efficient identification of functional antibodies.

• Generative AI creates compact, structurally accurate antigen mimics, facilitating immunization even for challenging or insoluble targets.

• The approach enables the discovery of antibodies with cross-species and subtype reactivity, expanding the range of druggable targets.

Case studies and experimental validation

• Engineered epitopes successfully broke immune tolerance in targets like latent TGF-beta 1 and Activin E, leading to functional antibody discovery where traditional methods failed.

• Nanoparticle display systems presenting multiple epitopes enhanced immune responses, especially for highly conserved proteins.

• AI-designed soluble GPCR surrogates, such as for GIP receptor and GPR75, demonstrated specific ligand binding and structural fidelity, validated by binding assays and electron microscopy.

• The platform enabled the selection of antibodies with desired specificity and cross-reactivity profiles using multi-dimensional mammalian display sorting.

• Functional assays confirmed the generation of selective amylin receptor agonists, a significant advance over existing peptide therapeutics.

Key learnings and future directions

• Diffusion models for protein design now achieve high success rates, with single designs per epitope often sufficient for efficacy.

• Compact engineered epitopes are synergistic with immunization strategies and may make previously undruggable targets accessible.

• The next innovation wave is expected at the intersection of de novo antibody and antigen engineering, enabling simultaneous optimization of both sides for drug discovery.

• Main failure mode in generative models is insufficient structural constraint, leading to non-compact or unrealistic designs, highlighting the need for careful model guidance.

• Soluble surrogates have potential utility beyond antibody discovery, including peptide drug discovery.