Abstract
BACKGROUND: The H1N1 influenza A virus evades host immunity through continuous antigenic drift, posing a significant challenge to broad-spectrum neutralizing antibody therapies. This study aims to systematically evaluate the neutralizing capacity of the broad-spectrum antibody C12H5 against H1N1 strains from different eras and identify key immune escape mutation sites. METHODS: Three representative H1N1 virus strains from 2009, 2018, and 2023 were selected. An antigen-antibody binding prediction model based on the ESM-2 large language model was constructed by integrating 48,762 GISAID sequence data and deep mutation scanning data from the Bloom laboratory. Candidate escape sites were screened using SHAP (SHapley Additive exPlanations) value analysis. Mutant viruses were constructed via reverse genetics, and their neutralizing capacity and replication fitness were validated through hemagglutination inhibition assays, microneutralization assays, and viral growth kinetics analysis. RESULTS: Machine learning scoring identified five potential escape sites, with K147 exhibiting the highest overall score (0.92). SHAP analysis revealed that the K147 site within the HA protein's 130-loop region received the highest importance score (0.28), significantly surpassing other candidate sites. Experimental validation revealed that the K147N mutation reduced neutralizing potency against C12H5 by 8-fold (from 1:1024 to 1:128) and approximately 6-fold in microneutralization assays (from 8.3 log(2) to 5.7 log(2)), while exhibiting a replication advantage in MDCK cells. Microneutralization assays further confirmed an approximately 6-fold reduction in neutralization sensitivity. Structural analysis indicated that K147 is located at the periphery of the HA receptor-binding domain, immediately adjacent to the receptor-binding site. CONCLUSIONS: K147N is identified as the critical mutation mediating C12H5 immune escape, and this mutation has emerged in 2023 circulating strains. This study provides important molecular targets and early warning mechanisms for broad-spectrum antibody optimization and influenza vaccine updates.