An engineered influenza virus based on a haemagglutinin protein from H5N1 avian influenza, with just four mutations, can be transmitted between ferrets, emphasizing the potential for a human pandemic to emerge from birds. See Letter p.420
Influenza pandemics in humans arise from animal influenza viruses, yet the molecular changes required for an animal virus to be transmitted efficiently between humans are poorly understood. Highly pathogenic H5N1 avian flu viruses have circulated in poultry for more than 16 years, only rarely resulting in human infections. But when people do catch H5N1 bird flu, their disease is of unusual severity, raising concerns that a human H5N1 pandemic might have a catastrophic impact on public health. However, an H5N1 virus that can be efficiently transmitted from human to human has not yet emerged, leading some researchers to question whether these viruses are inherently incapable of acquiring this capacity. On page 420 of this issue, Imai et al.1 demonstrate that H5N1 viruses do have the potential to cause a human pandemic. The authors identify mutations in the avian virus that permit viral transmission between ferrets by means of respiratory droplets — the best available model for influenza transmission in humans.
Imai et al. focused on the haemagglutinin (HA) protein of a highly pathogenic H5N1 influenza virus, which is involved in binding and fusion of the virus to the cells that it infects; the HA type is used in the viral nomenclature, together with the other influenza surface glycoprotein involved, a neuraminidase (NA). The HA of H5N1 viruses preferentially binds sialic acids in receptors on the surface of avian cells (Siaα2,3), whereas cells of the human upper airway predominantly have another type of sialic acid (Siaα2,6), which is recognized by human influenza viruses. The researchers introduced random mutations into the globular head of the HA molecule, where the receptor-binding domain is located, and searched for mutated viruses that exhibited enhanced binding to Siaα2,6. Using reverse genetics, a technique that allows genetic manipulation of the virus genome, they then made a 'hybrid' H5N1 virus in which the gene encoding one of these mutated H5 HA proteins replaced the HA gene in the H1N1 virus that caused a human pandemic in 2009.
The researchers infected ferrets with this hybrid H5N1 virus and, following multiple rounds of infection, isolation of the virus from the upper respiratory tract of the infected animals, and reinfection in animals not previously exposed to the virus, they obtained a virus that can be transmitted efficiently between the animals. Four key amino-acid changes in the HA are associated with ferret respiratory-droplet transmissibility in the authors' virus (Fig. 1a). Three of the mutations (N158D, N224K and Q226L) contribute to Siaα2,6 specificity. The fourth (T318I) lowers the pH at which the protein undergoes a structural change that allows it to release its genetic material into the cytoplasm of an infected cell by fusion of the viral envelope with intracellular membranes.
There have been many previous attempts to determine whether H5N1 can acquire transmission capacity in ferrets. Two studies2,3 assessed H5N1 and H3N2 hybrid viruses, and another study4 introduced mutations into the H5N1 HA that are known to increase Siaα2,6 binding in H2 and H3 haemagglutinin — but neither approach conferred transmissibility by respiratory droplets. One study achieved partial transmissibility by introducing three HA mutations (Q196R, Q226L and G228S) into an H5N1 virus and combining this with the NA protein of a human seasonal H3N2 virus (Fig. 1b). It is interesting that the two H5 HAs that demonstrate transmissibility among ferrets1,5 (Fig. 1a,b) contain similar mutations: one at amino-acid residues 158–160, which removes an N-linked glycosylation site of the globular head6, and the other at residues 221–228, which alters the structure of the loop of the receptor-binding domain6.
Another research group previously produced a transmissible H9N2 hybrid virus7 using the HA and NA of a low-pathogenic avian virus and other genes from a human seasonal H3N2 virus. The HA of this hybrid contains a Q226L mutation, which confers human-like Siaα2,6 binding7 (Fig. 1c). During the 10 rounds of infection that were needed for this H9N2 virus to acquire transmissibility, it accumulated two additional HA mutations, one located at residue 189 of HA1 (close to the receptor-binding domain) and the other at residue 192 of HA2 (close to the membrane-fusion domain), as well as one NA mutation located in the transmembrane domain. Together, these studies demonstrate that mutations that increase Siaα2,6 binding1,5,7, as well as those that stabilize HA structure1,7, seem to be functionalities required for the HA of viruses that are transmissible among mammals.
Thus, Imai and colleagues' generation of a respiratory-droplet-transmissible H5N1 virus represents the culmination of a sustained effort by numerous groups to better understand the mechanisms that can confer mammalian transmissibility in avian influenza viruses. However, it is likely that different combinations of HA mutations can achieve the same effects, and further studies are needed to explore this possibility. Furthermore, although transmission of influenza A virus among humans is already known8,9 to involve the virus HA, NA and basic polymerase protein 2, it is possible that other viral proteins, not explored by Imai and colleagues, also contribute to mammalian transmissibility.
It is intriguing that, although the parent H5N1 virus strain (A/Vietnam/1203/2004) used by Imai et al. causes lethal disease in ferrets when they are directly infected, infection with the ferret-transmissible H5N1 virus does not kill the animals. It is possible that the change in receptor binding from Siaα2,3, which is present on alveolar epithelial cells in ferret and human lungs, to Siaα2,6, found on cells of the upper respiratory tract, may change the virus from one that causes alveolar infection, likely to result in more severe disease, to one targeting the upper airways, which is associated with milder symptoms. However, Imai and colleagues did not find that the two viruses differ markedly in their targeting of the upper or lower respiratory regions.
Imai and colleagues' H1N1 virus with an H5 HA is a laboratory creation, but it should not be considered an experimental artefact. Natural emergence of an H5N1–H1N1 hybrid virus is plausible. Some H1N1 and H5N1 viruses readily swap genes with one another in vitro10,11, generating hybrid viruses. Furthermore, pandemic H1N1 viruses are established in pigs in many parts of the world, and H5N1 viruses have been isolated from pigs12, suggesting that opportunities exist for the viruses to combine in these animals.
However, these findings do not only provide further indication that such a virus may arise naturally; they also pave the way for improved influenza surveillance and pandemic preparedness. For example, one of the four mutations reported by Imai et al., N158D, results in a loss of N-linked glycosylation, and loss of glycosylation at this residue is increasingly seen among naturally occurring H5N1 isolates. Although the other three mutations have only very rarely been seen in H5N1 isolates from the field, mutations in HA that change the virus from binding Siaα2,3 to binding both Siaα2,3 and Siaα2,6 have been reported in both birds and humans13. These findings reinforce the need for focusing even greater attention on H5N1 infections in humans and other mammals (including pigs). Influenza viruses exist as genetic variants termed quasi-species even within a single clinical specimen, but such genetic diversity may not be fully assessed by conventional sequencing methods. Investigation of mammalian clinical specimens using new deep sequencing methods for these mutations, and for other mutations that confer similar functionality, will allow us to evaluate the extent of H5N1 adaptation that is occurring in mammalian hosts. More broadly, understanding the mutations that confer mammalian transmission of avian influenza viruses will allow better risk assessment of the animal viruses that represent a pandemic threat, and help to select virus strains against which pre-pandemic vaccines should be generated.
Imai, M. et al. Nature http://dx.doi.org/10.1038/nature10831 (2012).
Jackson, S. et al. J. Virol. 83, 8131–8140 (2009).
Maines, T. R. et al. Proc. Natl Acad. Sci. USA 103, 12121–12126 (2006).
Maines, T. R. et al. Virology 413, 139–147 (2011).
Chen, L.-M. et al. Virology 422, 105–113 (2012).
Stevens, J. et al. J. Mol. Biol. 381, 1382–1934 (2008).
Sorrell, E. M., Wan, H., Araya, Y., Song, H. & Perez, D. R. Proc. Natl Acad. Sci. USA 106, 7565–7570 (2009).
Belser, J. A., Maines, T. R., Tumpey, T. M. & Katz, J. M. Expert Rev. Mol. Med. 12, e39 (2010).
Sorrell, E. et al. Curr. Opin. Virol. 1, 635–642 (2011).
Cline, T. D. et al. J. Virol. 85, 12262–12270 (2011).
Octaviani, C. P., Ozawa, M., Yamada, S., Goto, H. & Kawaoka, Y. J. Virol. 84, 10918–10922 (2010).
Nidom, C. A. et al. Emerg. Infect. Dis. 16, 1515–1523 (2010).
Yamada, S. et al. Nature 444, 378–382 (2006).
About this article
Interdisciplinary Sciences: Computational Life Sciences (2017)
Aerosol Disinfection Capacity of Slightly Acidic Hypochlorous Acid Water Towards Newcastle Disease Virus in the Air: AnIn VivoExperiment
Avian Diseases (2015)
PLoS ONE (2014)