Clade 2.3.4.4b highly pathogenic avian influenza H5N1 viruses: knowns, unknowns, and challenges

Abstract

Since 2020, the clade 2.3.4.4b highly pathogenic avian influenza (HPAI) H5N1 viruses have caused unprecedented outbreaks in wild birds and domestic poultry globally, resulting in significant ecological damage and economic losses due to the disease and enforced stamp-out control. In addition to the avian hosts, the H5N1 viruses have expanded their host range to infect many mammalian species, potentially increasing the zoonotic risk. Here, we review the current knowns and unknowns of clade 2.3.4.4b HPAI H5N1 viruses, and we highlight common challenges in prevention. By integrating our knowledge of viral evolution and ecology, we aim to identify discrepancies and knowledge gaps for a more comprehensive understanding of the virus. Ultimately, this review will serve as a theoretical foundation for researchers involved in related avian influenza virus studies, aiding in improved control and prevention of H5N1 viruses.

Keywords: Clade 2.3.4.4b; H5N1; evolution; highly pathogenic avian influenza; pathogenicity.

Fig 1

Timeline of HPAI H5N1 evolution. This timeline shows significant events and important clades of HPAI H5N1.

Fig 2

Genetic evolution of HPAI H5N1 HA gene from 1997 to 2023. All available H5 subtype AIV HA gene sequences were retrieved from GISAID (https://www.gisaid.org/). Multiple sequence alignment was conducted using the MAFFT program. All sequences were subjected to data cleaning using Cluster Database at High Identity with Tolerance (CD-HIT) version 4.8.1, where sequences were removed if they met the following criteria: (i) containing multiple ambiguous bases and (ii) having a similarity of 99.5% or higher. The best-fit model under the Bayesian framework was determined using ModelFinder. We utilized BEAST v.1.10.4 to run the phylogeny tree, with the best-fit model being GTR + F + I + G4. Divergence times and evolutionary rates were estimated using an uncorrelated relaxed clock model and Bayesian Skygrid model. Bayesian Markov chain Monte Carlo was set to 200 million total iterations, with sampling every 20,000 iterations.

Fig 3

Cases of human infections with highly pathogenic H5N1 AIV. (A) From 2010 to 2024, statistics on human infections with HPAI H5N1 from 2010 to 2024, by every 5 years. (B) Statistics on human infections with HPAI H5N1 from 2020 to 2024, on an annual basis. All data are from the WHO and were plotted using GraphPad 8.3 software.

Fig 4

Host factors influencing the cross-species transmission of clade 2.3.4.4b H5N1 virus. Domestic poultry, such as chickens and waterfowl, is naturally susceptible to H5N1 due to low host barriers and the binding of HA to α−2,3 SA receptors. However, for human and most mammalian infections, H5N1 must undergo mutations to bind to α−2,6 SA receptors (e.g., HA-Q226L) and overcome high host barriers (e.g., PB2-E627K). Swine, as mixing vessels for avian and human influenza, possess both α−2,3 and α−2,6 SA receptors in the upper respiratory tract. Green arrows in the figure represent natural infection; black arrows indicate resistance to infection; yellow arrows indicate infection after mutation; and red arrows indicate the potential for new recombinant viruses to emerge after infection.