Rapid evolution of A(H5N1) influenza viruses after intercontinental spread to North America

The Goose/Guangdong-lineage A(H5N1) viruses detected in 1996 have diversified into multiple hemagglutinin clades through repeated expansions and contractions, causing major losses in poultry. Clade 2.3.4.4 viruses, often reassorting to carry different neuraminidase subtypes (N1, N6, N8), expanded markedly from 2020 onward. In late 2021, clade 2.3.4.4b viruses were detected in North America, indicating intercontinental spread from Eurasia. Shortly thereafter, A(H5N6) viruses closely related to European 2.3.4.4b strains were found in eastern Canada and then in U.S. wild birds. The study’s purpose was to define the genetic and phenotypic evolution of these 2.3.4.4b viruses after arrival in North America, including the extent of reassortment with North American (NAM) wild-bird influenza lineages and the impact on pathogenicity and transmission in avian hosts and mammalian models (ferrets, mice). The authors hypothesized that introduction into North America led to reassortment producing novel ribonucleoprotein constellations that could alter virulence and tissue tropism in mammals.

Background literature highlights: (1) Clade 2.3.4.4 viruses frequently reassort, yielding multiple H5Nx constellations that have circulated widely in Asia, Europe, and Africa. (2) Prior intercontinental spread to the Americas occurred in 2014 via the Pacific flyway, with limited mammalian adaptation noted then. (3) Human infections have been reported with H5N6 (notably in China) and sporadically with H5N1, underscoring zoonotic potential. (4) Known determinants of mammalian adaptation include receptor specificity (α2,6 sialic acid binding), hemagglutinin activation pH, and polymerase complex activity. (5) Surveillance reports up to early 2022 documented widespread detection of clade 2.3.4.4b in birds across dozens of countries, prompting concern for further spread and cross-species transmission.

Overview: The study combined in vivo pathogenesis/transmission experiments in chickens and ferrets, mouse virulence assays, virological phenotyping (receptor binding, HA activation/inactivation pH, polymerase activity, replication kinetics in human airway cells), antiviral susceptibility testing, and comprehensive genotypic/phylogenetic analyses of North American H5Nx viruses. Animal studies and transmission: - Chickens: Specific-pathogen-free white leghorns (n = 3 per virus) were inoculated with 10^6 EID50 of Wigeon/SC/21 (H5N1) or Eagle/FL/22 (H5N1) via multiple routes (0.2 mL intranasal, 0.1 mL intraconjunctival, 0.1 mL intraperitoneal). Contact transmission was evaluated by co-housing naïve chickens at 1 hpi. Oropharyngeal/cloacal samples at 2, 4, 6 dpi were titrated by EID50. Chicken-to-ferret exposure was assessed by housing naïve ferrets in adjacent cages (~12 inches) within the same bioisolator; ferret nasal washes were collected at 2, 4, 6, 10 dpi. - Ferrets: Multiple experiments were performed. Designs included (i) inoculation with 100 EID50 (n = 9 per virus; serial euthanasia at 3 and 5 dpi for tissue titers) and (ii) pathogenesis panels with 10^6 EID50 (n = 9 per virus; euthanasia at 5 dpi for titers and pathology). A separate methods section describes ferret inoculation with 10^4 EID50 for virulence/transmission monitoring (clinical signs, weights, temperatures, respiratory/neurological signs), with brain and respiratory tract sampling at defined time points. Survival was analyzed by log-rank tests. - Mice: Female BALB/c (6–8 weeks) were inoculated intranasally with serial dilutions (10^4–10^6 EID50) to determine LD50 over 14 days. For replication studies, mice (n = 5/group) received 10^4 EID50; lungs and brains were harvested at 5 dpi for titration. Viruses and genotyping: - Virus panel included early North American detections (e.g., A/Wigeon/SC21/2021, A/Bald eagle/Florida/W22-134/2022) and additional genotypically diverse A(H5N1) isolates (e.g., Ck/NL/21; Hawk/NC/22; Scaup/GA/22; Eagle/NC/22). In total, 58 A(H5N) viruses were sequenced for genotypic analysis. - RNA extraction, multi-segment RT-PCR, Nextera XT library prep, and Illumina MiSeq sequencing were performed. 423 gene segment sequences from 53 isolates were deposited (GenBank/GISAID; Supplementary Tables). - Phylogenetics: Sequences from public databases were aligned (BioEdit), analyzed by neighbor-joining (500 bootstraps) and maximum likelihood (MEGA7). Time-resolved trees for HA and PB2 were generated with Augur/TreeTime; visualization with ggtree. Redundant PB2 sequences were reduced using CD-HIT (97% identity threshold). Phenotypic assays: - Receptor binding: Solid-phase binding of whole virions to biotinylated sialylglycopolymers representing α2,3-linked (3'-SLN) and α2,6-linked (6'-SLN) sialic acids; detection via HRP-streptavidin ELISA. - HA activation pH: Syncytium assay in Vero cells; defining the highest pH yielding cell–cell fusion (resolution 0.1 pH units). pH inactivation: incubation of standardized virus at defined pH buffers (37 °C, 1 h), neutralization, and TCID50 titration (resolution 0.2 pH units). - Polymerase activity: Minireplicon assay in HEK293T with PB2/PB1/PA/NP plasmids and luciferase reporter normalized to β-gal. - Replication kinetics: Calu-3 cells (MOI 0.001) and primary differentiated human bronchial epithelial (HBE) cultures (MOI 0.005) incubated at 37 °C; supernatants/apical washes collected up to 72 hpi; titers by TCID50 in MDCK. Antiviral susceptibility and serology: - Neuraminidase inhibitor susceptibility (oseltamivir, zanamivir) measured by fluorometric MUNANA assay; IC50 values via nonlinear regression. - Baloxavir susceptibility assessed via plaque reduction/EC50 relative to reference A(H1N1)pdm09 viruses with PA I38 genotypes. - M2 inhibitor resistance inferred genotypically from M2 mutations at residues 26, 27, 30, 31, 34. - NA-directed antibodies assessed by ELLA using reverse-genetics H3N6 antigens carrying NA from test viruses. HAI performed for seroconversion where indicated. A 48-sample human donor serum panel (ages 18–46) was used to assess cross-reactive H5 HA neutralization and NA-binding antibodies. Statistics and ethics: - Analyses included unpaired t-tests, two-way ANOVA with Tukey’s post hoc test, and log-rank survival analysis (GraphPad Prism v9). Animal studies were IACUC-approved (St. Jude; protocol 428).

- Rapid reassortment after intercontinental introduction: North American wild-bird lineage gene segments were acquired by clade 2.3.4.4b H5Nx viruses soon after arrival. Wigeon/SC21 matched early Eurasian-like genotypes, whereas Eagle/FL22 had reassorted, acquiring NAM-lineage PB2, PB1, and NP. - Four genotypes among 58 sequenced A(H5N) viruses: All retained Eurasian-origin HA, NA, M, and NS, but differed in polymerase complex and NP combinations (EA vs NAM lineage). PB2 displayed two distinct phylogenetic lineages. The patterns suggested a limited number of reassortment events followed by geographic spread. - Antigenic homogeneity: Despite genotypic diversity, isolates remained antigenically homogeneous in HAI assays. - Enhanced mammalian virulence linked to NAM RNP acquisition: In ferrets, greater numbers of North American-lineage RNP segments (PB2/PB1/PA/NP) were associated with increased disease severity, weight loss, systemic spread, and mortality. Peak mean nasal wash titers rose across genotypes: 2.9 (Ck/NL/21), 3.8 (Eagle/NC/22), 5.4 (Hawk/NC/22), 6.0 log10 TCID50/mL (Scaup/GA/22). - Early virus comparison: Eagle/FL22 caused more severe disease than Wigeon/SC21 in ferrets. Mean nasal wash titers at 1, 3, 5 dpi were higher for Eagle/FL22 (4.2–5.2 log10 TCID50/mL) than Wigeon/SC21 (2.1–3.4 log10 TCID50/mL) (p < 0.0001). Viral loads in trachea/lung were higher for Eagle/FL22; virus was detected in brain tissue with mean titer ~6.2 log10 TCID50/g. Eagle/FL/22 infection led to humane endpoints with 0% survival by 10 dpi versus 100% for Wigeon/SC/21 in the depicted experiment. - Avian host outcome: Directly inoculated chickens (n = 3 per virus) died within 48 hpi; transmission to naïve chickens was inefficient under study conditions. - Neurotropism and pathology: The most virulent virus (Scaup/GA/22) produced severe necrotizing lesions in URT/LRT with abundant antigen staining; widespread antigen was observed in the nervous system. Less severe pathology was seen with Hawk/NC/22 and Eagle/NC/22; minimal lesions for Ck/NL/21. - Mouse virulence: Viruses that were highly lethal in ferrets had low LD50s in mice. Reported LD50s (log10 EID50/mL): Eagle/FL/22 ~2.2; Scaup/GA/22 ~3.8; Hawk/NC/22 ~4.6 (lower values indicate higher virulence). - Receptor specificity and HA stability: Wigeon/SC/21 and Eagle/FL/22 bound strongly to α2,3-linked sialic acids (3'-SLN) and poorly to α2,6 (6'-SLN), indicating avian-like receptor preference. HA activation pH was relatively high (5.8–5.9), consistent with avian adaptation; inactivation pH (~4.4) was unusually lower than activation pH and lower than control CA/04 (H1N1)pdm09 (inactivation pH ~5.4). - Replication in human airway models: In Calu-3 and primary differentiated HBE cultures, Scaup/GA/22 and Eagle/FL/22 achieved higher titers at 72 hpi than a seasonal H3N2 comparator (P ≤ 0.001), whereas Wigeon/SC/21 trended lower. - Human antibody landscape: In a 48-sample human serum panel, neutralizing antibodies to H5 HA were generally absent (≤1:20), but NA-directed antibody GMTs to H5N1 NAs (Wigeon/SC/21, Eagle/FL/22) were comparable to those against A(H1N1)pdm09 NA, reflecting ~89.6% amino acid identity and conserved antigenic sites. - Antiviral susceptibility: Six H5N1 viruses tested (reassortant and non-reassortant) had IC50/EC50 values to oseltamivir, zanamivir, and baloxavir comparable to drug-susceptible reference A(H1N1)pdm09 viruses, indicating retained susceptibility. M2 resistance was assessed genotypically (common resistance-associated substitutions tracked).

The study addressed how clade 2.3.4.4b H5Nx viruses evolved after arriving in North America and how reassortment affected mammalian pathogenicity. Genomic analyses showed rapid incorporation of North American wild-bird lineage RNP gene segments into Eurasian-origin H5 backbones, producing multiple genotypes that spread widely. Functionally, increasing numbers of NAM-lineage RNP segments correlated with higher virulence, systemic dissemination, and marked neurotropism in ferrets and lower LD50s in mice. Despite this enhanced virulence, viruses retained avian-like receptor specificity (α2,3 preference) and relatively high HA activation pH values, characteristics that are usually suboptimal for efficient human-to-human transmission. Thus, while the reassortants display worrisome pathogenicity and CNS involvement in mammals, additional adaptive changes (e.g., shifts toward α2,6 receptor binding and more acid-stable HA) would likely be required for sustained human transmission. From a public health standpoint, the rapid reassortment and phenotypic diversification underscore the need for vigilant surveillance across wild birds and mammals, preparedness for zoonotic spillovers, and assessment of vaccine/antiviral effectiveness. The observation of antigenic homogeneity across genotypes and preserved susceptibility to NA inhibitors and baloxavir is encouraging for countermeasure utility, but the broad and expanding host range, including sporadic mammalian infections, increases the risk of further adaptation.

This work demonstrates that clade 2.3.4.4b A(H5N1) viruses rapidly reassorted with North American wild-bird influenza lineages after intercontinental spread, generating multiple genotypes differing in polymerase/NP segments. Certain reassortants exhibited enhanced virulence with pronounced neurotropism in mammalian models while retaining avian-like receptor binding and HA activation properties. The viruses remained antigenically homogeneous and susceptible to approved antivirals in vitro. Key contributions include linking NAM-lineage RNP acquisition to increased mammalian pathogenicity and documenting conserved phenotypes relevant to zoonotic risk assessment. Future directions: (1) longitudinal genomic surveillance to track ongoing reassortment and emergence of mutations associated with human adaptation (α2,6 binding, HA acid stability, polymerase markers); (2) expanded pathogenicity and transmission studies across mammalian hosts, including ferret respiratory droplet models; (3) evaluation of vaccine antigen match and breadth of NA-directed immunity; (4) continued antiviral susceptibility monitoring and resistance risk assessment; (5) mechanistic studies on CNS invasion and neurovirulence determinants.

- Translation from animal models to human disease is uncertain; virulence and tissue tropism observed in ferrets and mice may not fully predict human outcomes. - The study assessed a subset of circulating genotypes; additional diversity may exist beyond the sampled 58 viruses. - While reassortment patterns were defined, specific causal mutations within RNP segments driving virulence were not pinpointed. - Efficient mammal-to-mammal respiratory transmission was not demonstrated; experiments primarily focused on pathogenesis and limited contact exposure designs. - Some reported parameters varied by experiment (e.g., ferret inoculum doses), which may affect direct comparison across cohorts. - Receptor binding and HA pH phenotypes were evaluated for select representatives (e.g., Wigeon/SC/21, Eagle/FL/22) and may not generalize to all genotypes.