Antigenic sites of H1N1 influenza virus hemagglutinin revealed by natural isolates and inhibition assays

Abstract

The antigenic sites of hemagglutinin (HA) are crucial for understanding antigenic drift and vaccine strain selection for influenza viruses. In 1982, 32 epitope residues (called laboratory epitope residues) were proposed for antigenic sites of H1N1 HA based on the monoclonal antibody-selected variants. Interestingly, these laboratory epitope residues only cover 28% (23/83) mutation positions for 9 H1N1 vaccine strain comparisons (from 1977 to 2009). Here, we propose the entropy and likelihood ratio to model amino acid diversity and antigenic variant score for inferring 41 H1N1 HA epitope residues (called natural epitope residues) with statistically significant scores according to 1572 HA sequences and 197 pairs of HA sequences with hemagglutination inhibition (HI) assays of natural isolates. By combining both natural and laboratory epitope residues, we identified 62 (11 overlapped) residues clustered into five antigenic sites (i.e., A-E) which are highly correlated to the antigenic sites of H3N2 HA. Our method recognizes sites A, B and C as critical sites for escaping from neutralizing antibodies in H1N1 virus. Experimental results show that the accuracies of our models are 81.2% and 82.2% using 41 and 62 epitope residues, respectively, for predicting antigenic variants on 197 paring HA sequences. In addition, our model can detect the emergence of epidemic strains and reflect the genetic diversity and antigenic variant between the vaccine and circulating strains. Finally, our model is theoretically consistent with the evolution rates of H3N2 and H1N1 viruses and is often consistent to WHO vaccine strain selections. We believe that our models and the inferred antigenic sites of HA are useful for understanding the antigenic drift and evolution of influenza A H1N1 virus.