SUMMARY. Highly pathogenic avian influenza viruses (HPAIVs) (“bird flu”) devastate the poultry industry, threaten
wildlife, damage economies, and constitute a permanent pandemic threat. HPAIVs emerge from low pathogenic
avian influenza viruses (LPAIVs) upon transmission from wild waterfowl (e.g., ducks, geese, gulls), their main
reservoir, to terrestrial poultry (e.g., chickens, turkeys). The transition from LPAIV to HPAIV results from the
insertion of nucleotides coding for multiple basic amino acids in the protease cleavage site of the viral
hemagglutinin (HA) gene during replication of the viral genome by the influenza virus polymerase. This change
in HA leads to systemic virus dissemination characterized by an endotheliotropism in poultry with mortality rates
up to 100%. In contrast, systemic virus dissemination, severe disease and endotheliotropism upon HPAIV
infection are rare or absent in most species of duck, wild and domestic. Interestingly, the transition from LPAIV
to HPAIV has only been observed in influenza viruses of the H5 and H7 subtypes. Moreover, although LPAIVs
circulate extensively in wild waterfowl, there is no evidence that they can evolve into HPAIVs in these species.
HPAIV emergence is currently unpredictable because the mechanisms of initial emergence through nucleotide
insertion by the influenza virus polymerase, and subsequent process of natural selection in avian hosts remain
poorly understood. To understand the molecular mechanism of nucleotide insertion, we have recently predicted
subtype-specific RNA stem-loop structures at the HA cleavage site. Here, we hypothesize that the stem of the
stem-loop structure refolds during viral RNA replication leading to the template closing on itself, trapping the
polymerase in the loop and causing it to stutter and insert nucleotides. Additionally, we hypothesize that specific
RNA sequences present in H5 and H7 stem-loops determine why insertions only occur in these subtypes. To
test these hypotheses, we successfully developed in vitro polymerase assays, including single-molecule assays,
with which nucleotide insertions in HA RNA can be reliably detected with high throughput via circular
resequencing. Secondly, we hypothesize that intrinsic differences in the ability of HPAIV to spread systemically
in poultry versus waterfowl determine the process of natural selection of HPAIVs from LPAIVs and explain the
host species-specificity of HPAIV emergence. More specifically, we hypothesize that HPAIV are selected in
poultry and not in waterfowl because of their endotheliotropism in poultry supporting their systemic
dissemination. To test this hypothesis, we designed competition experiments between LPAIV and HPAIV to
study the natural selection at the host level in chickens (as a model for poultry) and ducks (as a model for
waterfowl) and at the cellular level using newly developed in vitro transwell co-culture models of primary
respiratory and intestinal epithelial and endothelial cells of chickens and ducks. Increased knowledge about
HPAIV emergence will fill crucial knowledge gaps on influenza and may provide a point of action to
predict – and thus possibly control - HPAIV emergence and subsequent outbreaks that are threatening
animal and human health.