Through continuous monitoring of markets and breeding farms in several provinces, we successively isolated 23 H7N9 viruses. Using the sequences of those viruses and a reference sequence from the GISAID database (12), we conducted a phylogenetic analysis to study the evolution of H7N9 virus over the past decade (Figure 1).

We rooted the maximum-likelihood phylogenetic tree with A/Anhui/1/2013 and identified the branches as Group.y.0. During 2013–2017, the 5 low-pathogenicity avian influenza H7N9 virus waves formed Group.y.0–Group.y.2 branch. The first wave was mainly prevalent in the Yangtze River Delta. In 2014, the second wave spread to the Pearl River Delta and gradually expanded to all parts of the country in the subsequent 3 waves. Around 2017, or even as early as mid-2016, researchers successfully isolated an HPAI variant of H7N9 virus (6). That variant was found to contain alkaline amino acids inserted into the cleavage site of the hemagglutinin protein (6,13). The discovery of that variant in live poultry markets in Guangdong Province indicated an increased pathogenicity to poultry and potentially posed a greater threat to human health. We found that those HPAI H7N9 virus variants clustered within the Group.y.2.1 branch and its subordinate branch.

After implementation of a universal immunization program in 2017, the H7N9 virus outbreak was effectively controlled. However, some H7N9 viruses evolved to evade the vaccine. Those viruses continued to evolve and formed a new branch, Group.y.2.2, which is mainly found in northern China (Figure 1). Further investigation revealed that H7N9 virus is prevalent in the Bohai Rim region. Of note, we found no great differences in geographic, temporal, or host distribution between the 2 newly differentiated branches, Group.y.2.3 (A/Chicken/Hebei/1009/2020-like) and Group.y.2.4 (A/Chicken/Yunnan/1001/2021-like) (Figure 1). Those branches showed an average of 2.41% pairwise nucleotide distances between 2021 and 2023. That finding suggests that the evolutionary differences between those clades might not be influenced by geographic isolation, period, or host species, but rather by the adaptation of a new virus to natural selection. Positive selection pressure, which encourages mutations that contribute to the virus’ adaptation to the environment, can play a role in viral evolution. Our analysis confirmed that an increase in positive selection pressure in H7N9 virus occurred at some sites after 2017 (Appendix Table 1).

To examine whether mutation and evolution of H7N9 viruses are a result of antigenic drift and discontinuous variation, we used serologic methods to assess the antigenicity of the more evolved viruses from different clades (Appendix Table 2). Using the H7-Re4 and rHN7901 vaccine viruses for comparison, we found a weak cross-reaction titer between the vaccine viruses and the epidemic viruses in Group.y.2.3 (Table). The antigenic map demonstrated that the Group.y.2.3 viruses were distantly located from the vaccine serum (Figure 2), implying a consistent antigenic drift and greater antigenic divergence from Group.y.2.4 viruses. However, the distance between the vaccine virus and certain Group.y.2.4 viruses was relatively close (Figure 2), suggesting minimal differences. Furthermore, some Group.y.2.4 viruses, including A/Chicken/BJ/732-1/2022 and A/Quail/HeN/621/2022, both originating from northern China, also exhibited antigenic drift.

Changes in antigenicity often are caused by accumulation of amino acid mutations in antigenic sites. Therefore, we compared virus sequences and observed that the cleavage sites were KRKRTAR↓GLF or KRKRIAR↓GLF, both of which exhibited the characteristics of HPAI viruses. However, we noted no substantial differences between Group.y.2.3 and Group.y.2.4 at positions 86, 129, 134, 141, 145, 148, 151, 159, 208, 284, and 319 of H7 (Appendix Table 3). Those findings demonstrated the high genetic diversity of the H7N9 virus. Except for position 208 in H7, all sites were antigenic sites, and positions 141, 145, and 148 were both antigenic sites and receptor-binding sites. For the Group.y.2.4 branch, we compared the hemagglutinin 1 peptide of the vaccine viruses against antigenically distant viruses A/Chicken/SD/1301-2/2022, A/Chicken/HeB/B363-4/2022, A/Chicken/BJ/B732-1/2022, and A/Quail/HeN/621/2022. We observed different amino acids that could affect H7N9 virus antigenicity (Appendix Table 4). Among the analyzed viruses, A/Quail/HeN/621/2022 exhibited the highest number of mutations compared with the vaccine viruses, followed by A/Chicken/BJ/B732-1/2022 and A/Chicken/HeB/B363-4/2022; A/Chicken/SD/1401-2/2022 displayed the fewest mutations. Moreover, the previously reported Q226L and G228S sites of H3 viruses (Appendix Table 5), which have the potential to enhance mammalian adaptation, remained unchanged in all H7N9 viruses. Those sites still showed a preference for avian receptors, except A/Quail/HeN/621/2022, which mutated to P at position 160. All V125T H3 sites were replaced, indicating that the receptor-binding capacity and immune escape of the virus might be affected, making the virus more compatible with avian receptors (14,15).

This study explored the evolution and antigenic differentiation characteristics of H7N9 virus over the past decade through continuous monitoring and selection of representative sequences from all publicly available H7N9 virus sequences. However, our research still had certain limitations, and further investigation is needed to understand the relationship between the evolution of viruses under positive selection pressure and the underlying cause of antigenic variation.

In summary, influenza A viruses are highly prone to mutation and evolution, making the H7N9 virus epidemic more complex and challenging to control. This study offers vital insights into the genetic evolutionary branches and recent antigenic drift, providing crucial clues for updating the H7N9 vaccine seed virus and for disease prevention and control.