New computational pipeline improves effectiveness of pan-Ebola neutralizing antibodies

A team led by Professor Wei Yang at the National Institute of Pathogen Biology, Chinese Academy of Medical Sciences & Peking Union Medical College, has developed a computational-experimental pipeline that synergistically optimizes antibody framework regions (FR) and complementarity-determining regions (CDR). Using this approach, they engineered two pan-Ebola antibodies with markedly improved neutralization breadth and potency. The study, titled "Computational Design of Broad-Spectrum Ebola Antibodies through Framework and Complementarity-Determining Regions Synergistic Optimization" is published in Research.

Background:
Ebola virus, a member of the Filoviridae family, caused recurrent regional outbreaks between 2013 and 2022 and remains a major global public health threat. Among the six known species, Zaire (EBOV), Sudan (SUDV), and Bundibugyo (BDBV) are highly pathogenic. Viral entry is mediated by the envelope glycoprotein (GP), the primary target of neutralizing antibodies. Current approved antibody therapies are largely effective only against EBOV and have limited cross-species activity, often requiring cocktail formulations. Moreover, viral evolution under immune pressure drives immune escape, underscoring the need for both potent and broadly neutralizing anti-Ebola antibodies, as well as rapidly adaptable antibody optimization technologies.

Research progress:
Closed-loop computational-experimental pipeline. The team established a multi-parameter in silico prediction platform integrated with targeted wet‑lab validation. Using deep mutational scanning tools (mCSM‑AB, mmCSM‑AB), they prioritized mutations based on predicted binding free energy changes (ΔΔG_bind) and relative solvent accessibility (RSA%). The iterative "predict‑test‑refine" strategy markedly accelerated antibody engineering.

FR-CDR co‑engineering of ADI‑15878. Through Foldseek‑based framework grafting followed by CDR single‑point mutagenesis, the variant W32G‑LC was identified. Without compromising structural integrity, this variant improved neutralization against EBOV, BDBV, and SUDV by approximately 17‑fold, 7‑fold, and 2‑fold, respectively, and partially restored binding to an escape mutant (G528E) of EBOV GP.

Docking‑guided optimization of ADI‑15946. For ADI‑15946, which initially showed weak SUDV neutralization, the researchers used ZDOCK to model antibody‑antigen complexes and applied a multi‑template consensus strategy for single‑point mutation screening. mmCSM‑AB guided construction of multi‑point mutants. Two variants, H27Q/S52Y/G66R‑LC and A50Y/S52Y/L54R‑LC, enhanced SUDV neutralization by >40‑fold and >100‑fold, respectively, while retaining full activity against EBOV and BDBV.

Structural and affinity validation. AlphaFold3 predictions and surface plasmon resonance measurements revealed that W32G‑LC expands buried surface area and establishes a new interaction with GP glycosylation site N204. The ADI‑15946 triple mutants introduced tyrosine and arginine residues, forming cation‑π and π‑π stacking networks, which increased binding affinity to SUDV GP from 2.241 nM to 0.248-0.349 nM and enlarged the contact area with key epitopes.

Future prospects:
This work establishes a computational and generalizable optimization strategy that integrates FR‑CDR co‑engineering, docking‑guided multi‑template consensus screening, and iterative experimental feedback. The approach overcomes the limitations of single‑parameter affinity‑driven design, enabling balanced improvement of potency across multiple viral strains. In the future, this pipeline can be extended to other highly variable pathogens such as influenza and SARS‑CoV‑2, and combined with AI‑predicted antigen structures to counter emerging escape variants. It also holds promise for optimizing antibody‑based detection probes, offering an efficient and versatile engineering tool for rapid responses to sudden viral outbreaks.