A parsimony approach was used to construct phylogenetic trees of the H1, H2
and H3 influenza hemagglutinin subtypes. The parsimony trees were then com
pared with randomly generated trees to identify regions of the proteins con
taining the most phylogenetic information, i.e. those regions making the pa
rsimony trees shorter. We reasoned that any areas of the hemagglutinin prot
ein that were phylogenetically 'information-rich' would be good candidates
for sites involved in virus-host interactions and their identification migh
t lead to a better understanding of the protein. Molecular modelling, based
upon the crystal structure of the H3 hemagglutinin, demonstrated that most
phylogenetically important regions of the H1 subtype were on the surface o
f the hemagglutinin trimer, primarily in the globular region. Many correspo
nded to known antigenic or receptor binding sites, while others appear to b
e novel and specific for H1. (C) 1999 Elsevier Science B.V. All rights rese
rved.