More H5N1 Receptor Binding Domain Roulette
Recombinomics Commentary 14:35
April 13, 2008
The recent reports on human to human H5N1 transmission in Pakistan and China have again focused attention of the receptor binding domain. The recent report on a new combination of changes that converts an affinity for an avian receptor to a human receptor has also focused attention on such changes.
Previously, attention has been focused on positions 226 (Q226L) and 228 (G228S), but the latest report indicates that in clade 1 H5N1 the change at 228 can be traded for changes at 158 (N158S) and 248 (N248D) to change H5N1 binding specificity to a human receptor. These data highlight the fact that H5 is somewhat different that H1 and H3 and different combinations can lead to the same end point. Similarly, the results suggest that such a change is not necessarily “all or none”, and there may be incremental changes.
Such incremental changes were reported previously. A screening of H5N1 isolates identified two from Hong Kong in 2003 which had a change at position 227 (S227N). This change reduced affinity for avian receptors and increased affinity for human, but the affinity for human receptors was still below the levels in seasonal flu. This observation for S227N led to the prediction of human infections by the Qinghai strain of H5N1 in the Middle East in late 2005 / early 2006 because donor sequences were present in Middle Eastern H9N2, which was endemic to the region.
The first confirmation of Qinghai H5N1 in humans was confirmed in early 2006 in Turkey, and the sequence has S227N. The index case was linked to a cluster of four siblings. H5N1 was confirmed in the three fatal cases, but H5N1 from the sister of the index case did not have S227N. However, the receptor binding domain change would concentrate the H5N1 in the upper respiratory tract, and would be selected against if grown in chicken eggs. These differences may have contributed to the failure to find S227N in the sister. Two additional sequences from human cases in Turkey have been released, and one of the two also has S227N, suggesting the change was widespread in human cases in Turkey, the vast majority of which were in clusters. Thus, although changes at 226 and 228 were not present, the change at 227 was detected in clusters, pointing toward a role in more efficient infections of humans by H5N1.
Similar results were found in Egypt, where the H5N1 sequences from a Gharbiya cluster had changes at positions 223 (V223I) and 230 (M230I). The M230I change is found in seasonal flu (H1N1, H3N2, and influenza B) raising additional concerns that changes in this HA region could contribute to combinations that increase transmission to humans.
These concerns are raised because the loss of a glycosylation site at 158 has already been fixed in clade 2.2, which is the sub-clade in the Middle East with changes at positions 223, 227, and 230. Moreover, this sub-clade also has PB2 E627K, which is another human polymorphism that creates higher levels of H5N1 in the upper respiratory tract, where the temperature is close to the optimal 33 C.
These data raise concerns that the combination of PB2 E627K, the lack of a glycosylation site at HA 158, the receptor binding domain changes at positions 223, 227, 230 plus one more change may be sufficient to produce a catastrophic pandemic strain that combines the high case fatality rate seen in a variety of versions of H5N1, with the transmissibility seen in seasonal flu. The concentrated receptor binding domain changes are in addition to other changes at positions 186 (N186S and N186K) or 196 (Q196R).
Recent poultry sequences in Egypt have V223I and M230I. Other isolates, which have been found in vaccinated stocks, have M230V. The changes create a game of receptor binding domain roulette which may be deadly.
Media Links
Recombinomics Presentations
Recombinomics Publications
Recombinomics Paper at Nature Precedings