Avian influenza virus neuraminidase stalk length and haemagglutinin glycosylation patterns reveal molecularly directed reassortment promoting the emergence of highly pathogenic clade 2.3.4.4b A (H5N1) viruses

The recently emerged panzootic clade 2.3.4.4b H5N1 influenza viruses show unprecedented extensive spread in wild birds and have been transmitted to several mammals, including humans. The virologic factors that have driven the success of the 2.3.4.4b H5N1 viruses, which has not been achieved by previous H5N1 clades, is unclear. We show that the 2.3.4.4b H5 haemagglutinin (HA) paired exclusively with full length (long stalk) N1 neuraminidase (NA) in birds and mammals, unlike previous clades of H5 viruses, which preferentially paired with N1 proteins with stalk deletions (short stalk). We found that the emergence of a 2.3.4.4b H5 HA with seven glycosylation sites was critical in driving its pairing with long stalk N1. The earlier H5 clades that paired with short stalk N1s showed a pattern of eight or more glycosylation sites. A prior shift in a glycosylation site from position 103 to 171 in the receptor binding domain of H5 HA and the subsequent S173A mutation that removed it triggered the emergence of 2.3.4.4b clade H5N1 viruses. Thus, the evolution of novel variations in the H5 HA and their preference for long stalk N1 pairing led to increased fitness and pathogenicity. These observations led us to establish and validate the hypothesis that the pairing of avian virus HA and NA subtypes is not stochastic but is rather molecularly programmed by HA glycosylation and NA stalk length, modulating fitness and emergence of novel avian influenza viruses. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=182 SRC="FIGDIR/small/595329v1_ufig1.gif" ALT="Figure 1"> View larger version (29K): org.highwire.dtl.DTLVardef@1659b26org.highwire.dtl.DTLVardef@1ace4b6org.highwire.dtl.DTLVardef@1db7c0corg.highwire.dtl.DTLVardef@ad3654_HPS_FORMAT_FIGEXP M_FIG C_FIG