#18,897
For more than 20 years repeated migratory bird die offs at Qinghai Lake, China - a major migratory bird stopover site - have heralded significant changes in the HPAI H5Nx virus's evolution, and behavior.
The first, in the spring of 2005 (see H5N1 Influenza Continues To Circulate and Change 2006 by Webster et. al.) saw the emergence of a new clade (2.2) which was particularly pathogenic in some species, while some migratory birds were better able to carry it over long distances.
This sparked the first major diaspora of the virus, with H5N1 turning up in Europe, then the Middle East, and Africa, within a matter of months (see H5N1 Virus Evolution in Europe—An Updated Overview).
While these pivotal evolutionary milestones may have first occurred elsewhere, they were often first detected at Qinghai Lake. And the sharing of viruses there has undoubtedly aided and abetted their spread.
All of which brings us to a new Letter to the Editor of the Journal of Infection from researchers affiliated with various Chinese national research institutes, that describes the detection of the PB2-E627K mutation in a migratory bird at Qinghai lake.
PB2-627K is often associated with enhanced replication and pathogenicity in mammals, and is one of the most important mutations that H5Nx is thought to need in order to spread more efficiently in mammals (see A rapid review of the avian influenza PB2 E627K mutation in human infection studies).
Luckily, these mammalian adaptations tend to exact a heavy `fitness penalty' in avian hosts, making detection (and spread) in birds a rare event. An exception, however, is a close cousin - an emerging PB2-627V - which appears to be making inroads in Chinese poultry.
First the letter, after which I'll have some closing thoughts on its significance, and the 3-year delay in releasing this information.
Zoonotic Threat: Emergence of Mammalian-Adapted H5N1 Virus in Migratory Birds at Qinghai Lake
Xiang Li 1, Xinru Lv 1, Yi Li, Mengdan Fei, Songbai Liu, Hongliang Chai
https://doi.org/10.1016/j.jinf.2025.106623
Under a Creative Commons license
Open access
In October 2020, a novel clade 2.3.4.4b H5N1 highly pathogenic avian influenza virus (HPAIV) was first detected in wild ducks in the Netherlands and rapidly disseminated globally, causing unprecedented outbreaks among wild birds, poultry, and an expanding range of mammalian hosts.1, 2, 3 By June 2025, over 1,000 confirmed H5N1 infections had been reported in dairy cattle across 17 U.S. states, with epidemiological evidence pointing to interstate transmission via asymptomatic or pre-symptomatic animals.4, 5 More than 70 human infections have also been reported in the U.S., including one fatality, and genomic data revealed close similarity between viruses isolated from humans and infected cattle.4, 6 The widespread host range and repeated spillover events suggest increasing mammalian adaptation of the virus, with profound implications for wildlife ecology, food security, and global public health.7
In July–August 2022, a cluster of wild waterfowl deaths occurred at Qinghai Lake, a key breeding site for migratory birds in China. We detected H5N1 viruses (Q-H5N1, n = 7) directly from seven freshly dead brown-headed gulls (Larus brunnicephalus Jerdon) (Table S1). All Q-H5N1 isolates carried the multibasic HA cleavage site (REKRRKR↓GLF), confirming high pathogenicity, and belonged to clade 2.3.4.4b.
Genomic analysis showed 99.2%–100% nucleotide identity across all eight segments. Molecular clock analysis dated their most recent common ancestor to early 2022 and revealed close phylogenetic links to strains circulating in wild birds and poultry along the East Asian–Australasian and Central Asian flyways in 2021–2022 (Figure S1 and Table S2). These findings underscore the continuing role of migratory birds in viral evolution and long-distance spread.
Notably, one virus, A/brown-headed gull/Qinghai/Q12/2022(H5N1) (QZ12), carried the PB2-E627K mutation, a key marker of mammalian adaptation (Table S3). This mutation is known to increase replication and pathogenicity in mammals and to enable airborne transmission in experimental ferret models.8 Its detection in a wild bird in China highlights the risk of avian-to-mammal transmission and the possible public health impact.
Globally, 277 H5Nx HPAIVs with the PB2-627K mutation were reported between 2021 and early 2025, spanning five continents and showing distinct spatiotemporal patterns over time (Table S4). Europe was an early hotspot, peaking in 2023 (n = 72), mostly in wild carnivores. In Asia, detections rose in 2022, driven by emu outbreaks in Japan (n = 13), and surged again in 2024 when infections were found in large-billed crows (n = 9). North America recorded its first cases in 2022 and a sharp increase in early 2025 linked to H5N1 infections in U.S. dairy cattle (n = 18). The virus with the PB2-627K mutation even reached Antarctica in 2023, where it was isolated from southern elephant seals.
Although H5N1 remained dominant, sporadic detections of H5N8 (notably in 2021), H5N5 (peaking in 2024), and H5N6 (mainly in humans) were also reported.
The growing detection of PB2-627K across species and regions raises concern that H5 viruses could establish sustained mammal-to-mammal transmission, a critical step toward human-to-human spread. In early 2025, the mutation was identified for the first time in H5N1 from U.S. dairy cattle, along with a human case carrying the same change. These events show the virus has crossed into the livestock sector, increasing the risk of human exposure through direct contact or contaminated dairy products.9
Our detection of PB2-627K in clade 2.3.4.4b H5N1 HPAIVs from Qinghai Lake gulls is the first such report in wild birds in China between January 2020 and February 2025. Qinghai Lake, a key breeding and stopover site along the Central and East Asian flyways, has long been recognized as a hotspot for the emergence and dispersal of HPAIVs. Since 2005, multiple distinct outbreaks have been recorded here, including clade 2.2 H5N1 in 2005, clade 2.3.2 in 2009, clade 2.3.2.1c in 2015, and clade 2.3.4.4 H5N8 in 2016. This history underscores the lake’s role as a viral convergence node where intercontinental flyways intersect and facilitate reassortment and dissemination.
The detection of a mammalian-adaptation marker in wild birds at such a site suggests that H5N1 is continuing to acquire traits that could expand its host range. PB2-627K is of particular concern because it has repeatedly emerged in mammalian infections worldwide, from wild carnivores in Europe to dairy cattle in the United States.
The independent appearance of this substitution in migratory birds adds a new dimension, implying that adaptive variants may circulate undetected in natural reservoirs before spilling over into domestic animals or humans. This raises the possibility that Qinghai Lake and similar high-risk wetlands function not only as mixing vessels for avian viruses but also as staging points for mutations with pandemic potential.
These findings highlight the need for strengthened surveillance at high-risk sites such as Qinghai Lake, integrating genomic analysis, early detection, and monitoring of mammalian spillover. Critical questions remain regarding the selective pressures and fitness costs of PB2-627K, as well as its interaction with other polymerase or HA mutations. Addressing these gaps will require coordinated efforts linking viral genomics, infection experiments, and ecological studies of bird migration and cross-species transmission.
(Continue . . . )
The detection of a mammalian-adapted H5N1 strain in migratory birds is admittedly a worrying sign, although one incident does not a paradigm shift make.
The problem is, surveillance is very sporadic, and the release of information (in this case taking 3 years) is often glacial-paced, making it difficult to know what the state of the virus's evolution truly is today.
A three-year delay in reporting is not all that unusual out of China (see recent examples here and here), but as we saw last March (see Nature: Lengthy Delays in H5N1 Genome Submissions to GISAID), belated sharing of data is not just a Chinese problem.
While still relatively rare, we've seen a large increase in the number of PB2 -627K in mammals around the globe, and the assumption has been these occur post-infection in the mammalian host.
And that may still very well hold true.
But with this detection of PB2-627K in a migratory bird, the possibility that some of those cases might have come from avian hosts already carrying a mammalian-adapted virus has to be at least considered.
Once again (see here, here, here, here, and here) we see calls for better surveillance, more rigorous testing, and the timely sharing of data.
As basic and as sensible as that might sound, over the past few years we've seen a serious decline in both surveillance programs, and the sharing of data.
The world, it seems, prefers to treat the spread of HPAI as more of an economic or political concern, than as a potential public health threat.
A strategy that admittedly works well, until suddenly it doesn't anymore.