Genotype A3 influenza A(H5N1) isolated from fur seals shows high virulence in mammals, but not airborne transmission Authors: Anastasia S Panova, Andrey S Gudymo, Natalia P Kolosova, Alexey V Danilenko, Kiunnei N Shadrinova, Natalia V Danilchenko, Olga N Perfilieva, Anastasia A Moiseeva, Elena I Danilenko, Galina S Onkhonova, Natalia I Goncharova, Svetlana V Svyatchenko, Natalia N Vasiltsova, Marina L Egorova, Vasiliy Yu Marchenko Published: 2025 ABSTRACT AbstractThe global spread of highly pathogenic avian influenza A(H5N1) clade 2.3.4.4b viruses has recently extended to include diverse mammalian species, raising new concerns about pandemic risk. In 2023, this clade was first detected in Russian marine mammals during a mass mortality event among northern fur seals in the Far East. Genetic analyses revealed the causative viruses to belong to genotype A3 of European origin, which is known to have circulated in wild birds across the Far East since 2022. Notably, these isolates harbor the mammalian-adaptive substitutions PB2-K482R and NP-N319K—mutations previously linked to enhanced virulence in non-H5 avian influenza viruses, but whose impact on A(H5N1) clade 2.3.4.4b viruses remained to be characterized. The heightened virulence of A3 genotype viruses is confirmed by data obtained via a mouse model. However, despite these adaptive changes, ferret transmission models showed no evidence of airborne transmission of the fur seal-derived virus. Our findings indicate that while PB2-K482R and NP-N319K may contribute to increased mammalian pathogenicity, they do not significantly increase the efficiency of respiratory transmission—a key prerequisite for human pandemic potential. Although suggesting a limited immediate pandemic threat from this A3 genotype, these results underscore the critical need for continued surveillance and functional assessment of emerging mammalian-adaptive mutations in circulating A(H5N1) viruses.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-28032-3.Subject terms: Influenza virus, Viral transmission Introduction In recent years, there has been an increase in mammalian infections caused by clade 2.3.4.4b A(H5N1) highly pathogenic avian influenza viruses (HPAIV). Highly pathogenic avian influenza (HPAI) outbreaks have been registered both in terrestrial carnivores and in marine mammals1. The adaptation of HPAIV to mammals may occur through reassortment events leading to novel genotypes or via the accumulation of adaptive mutations2. During its global spread, the 2.3.4.4b A(H5N1) virus has undergone extensive genetic evolution through reassortment with low-pathogenic avian influenza viruses, resulting in the emergence of multiple distinct genotypes. The importance of genotype-level analysis for clade 2.3.4.4b A(H5N1) influenza viruses has grown in recent years due to the emergence and global spread of numerous reassortant variants. A previous study identified 13 distinct genotypes circulating in Europe between 2020 and 20223. Among these, genotype EA-2020-C – also designated as G13—was first detected in the Netherlands in October 2020 and subsequently became one of the dominant genotypes during the 2021–2022 epizootic wave in Europe4. Given its absence from global surveillance data prior to this detection, EA-2020-C is thought to have emerged in Europe through reassortment between a European A(H5N8) virus and a subtype N1 low-pathogenic avian influenza (LPAI) virus, leading to the introduction into 2.3.4.4b H5 of five new internal segments (PB2, PB1, PA, NP and NS) derived from LPAIs circulating in Eurasia among wild birds since 20194,5. Since 2021, this genotype has spread to Africa, Asia, and North America3. Genotype A3 is a part of EA-2020-C and has been primarily detected in Japan6 and the USA7,8, as well as in wild and domestic birds in Russia’s Far East since 2022. While variations in virulence among different influenza A(H5N1) genotypes remain poorly understood, differences in pathogenicity have been consistently observed in mammalian models 2,3,9. For example, the introduction of Eurasian clade 2.3.4.4b A(H5N1) viruses into North America in late 2021 and their subsequent reassortment with locally circulating North American avian influenza viruses generated novel genotypes associated with enhanced virulence and more severe disease in mammals, including prominent neurological manifestations2. The virulence of highly pathogenic A(H5Nx) viruses in mice is often determined not by a single genetic marker but by complex interactions among multiple viral genes10–12. Limited data on contemporary clade 2.3.4.4b A(H5N1) viruses indicates moderate to high virulence in animal models2,13. Notably, increased virulence has been associated with the acquisition of internal gene segments of North American lineage and, in some cases, with the presence of previously described mammalian adaptation markers found in other influenza subtypes and genotypes. While some mammalian adaptation-associated mutations have been functionally characterized, the biological significance of many others remains poorly understood. Among clade 2.3.4.4b A(H5N1) viruses, the best-studied pathogenicity markers are substitutions in the PB2 protein. Notably, up to 50% of mammalian-derived 2.3.4.4b isolates harbor PB2 mutations associated with enhanced viral replication in mammalian cells, including E627K, D701N, Q591K, T271A, M631L, and K482R12,14–19. Several of these substitutions have been detected in genotype A3 viruses. Mammalian adaptation mutations have also been identified in other gene segments across various HPAIV subtypes, including H5, where they influence virulence, pathogenicity, receptor binding preference, and transmission12. However, due to limited information regarding such mutations in the 2.3.4.4b clade, the functional significance of many observed genetic changes has yet to be evaluated using virological approaches. Preliminary data suggest that previously under-characterized genetic features may play a role in virus adaptation. For instance, a study on A(H7N7) demonstrated that mammalian adaptation was associated with the accumulation of the NP-N319K amino acid substitution (AAS) alongside a PB2 mutation, resulting in increased virulence20. The NP-N319K substitution was also present in human A(H5N1) isolates from the 1997 Hong Kong outbreak, which was characterized by high case fatality21,22. Despite its recurrent detection in mammalian A(H5N1) cases, the role of NP-N319K in the context of contemporary 2.3.4.4b viruses remains unexplored. Genotype A3 viruses containing the NP-N319K mutation have been identified in mammals in the United States7,8, Canada23, Japan6, and some European countries24, including Russia25. In addition to the identified mutations in the NP gene that may facilitate mammalian adaptation, other genetic features within this gene could also contribute to cross-species transmission. For example, Nguyen et al. proposed that the NP gene acquired from wild birds through reassortment may have contributed to the infection of cattle in the United States26. The presence of mammalian-specific markers may be associated not only with increased virulence but also lead to the emergence of airborne transmission13. To date, limited airborne transmission has only been identified for 2.3.4.4b A(H5N1) viruses containing pathogenicity markers in PB2. Specifically, airborne transmission has been demonstrated in ferret transmission models for: (i) a virus with PB2-T271A isolated during a mink farm outbreak in Spain18; (ii) a Russian avian isolate harboring PB2-E627K13; and (iii) bovine A(H5N1) viruses from U.S. dairy farm workers carrying either PB2-E627K or PB2-M631L27–29. As emphasized by the WHO Tool for Influenza Pandemic Risk Assessment, enhanced transmission is a critical determinant in pandemic risk assessment30. Although human infections with clade 2.3.4.4b A(H5N1) remain sporadic and are linked to direct animal contact in most cases—with no confirmed sustained human-to-human transmission to date31—the extensive genetic diversity generated through reassortment and rapid mutation accumulation raises concerns about the emergence of a virus with increased pandemic potential. Given their widespread circulation, repeated spillovers into mammals, and acquisition of putative mammalian-adaptive mutations, A3 genotype 2.3.4.4b A(H5N1) viruses warrant close attention in pandemic preparedness efforts. In the present study, A3 genotype viruses exhibited high virulence in mammalian models but ferret models demonstrated no transmission via the airborne route. These findings indicate that current A3-associated mutations, including PB2-K482R and NP-N319K, do not increase the efficiency of respiratory airborne transmission. Nevertheless, they underscore the need for continued surveillance and functional characterization of genotype-specific virological traits to accurately assess the evolving pandemic risk posed by circulating 2.3.4.4b A(H5N1) viruses. Results Genetic analysis In this study, we compared the virological properties of the clade 2.3.4.4b A(H5N1) influenza virus A/fur seal/Sakhalin/399-1V/2023 with a subset of genetically related 2.3.4.4b A(H5N1) viruses isolated from avian hosts in Russia between 2021 and 2023. The seal-derived strain was assigned to genotype A3 in our previously study25. For comparative analysis, we selected three avian isolates with defined genotypes isolated in Russia. The strain A/chicken/Magadan/14-7V/2022, also classified as genotype A3, differs from the A/fur seal/Sakhalin/399-1V/2023 strain by only 8 AASs across the genome. In contrast, A/turkey/Tyumen/81-96V/2021 belongs to genotype G10, which, despite being genetically closely related to A/fur seal/Sakhalin/399-1V/2023 in the HA and NP segments, differs from A3 primarily in the PB2 gene segment (Figs. 1, 2, 3, Tables S4, S5). Additionally, we included A/gull/Mari El/345-5V/2023, a genotype BB virus collected during a large-scale outbreak among gulls in the European part of Russia in 202325. Phylogenetic tree of the HA gene of A(H5Nx) viruses. Studied strains A(H5N1) are indicated in blue. Reference strains used in the virulence analysis are indicated by black squares. HA sequences from reference strains used in the analysis were obtained from the EpiFlu database of GISAID. Phylogenetic tree of the NP gene of A(H5Nx) viruses. Studied strains A(H5N1) are indicated in blue. Reference strains used in the virulence analysis are indicated by black squares. NP sequences from reference strains used in the analysis were obtained from the EpiFlu database of GISAID. Phylogenetic tree of the PB2 gene of A(H5Nx) viruses. Studied strains A(H5N1) are indicated in blue. Reference strains used in the virulence analysis are indicated by black squares. PB2 sequences from reference strains used in the analysis were obtained from the EpiFlu database of GISAID. Genetic analysis showed that both A3 genotype viruses in our panel carried two notable substitutions, PB2-K482R and NP-N319K, which had been previously associated with increased virulence in mammalian models20,32 and were absent in the analyzed G10 and BB genotypes (Tables S1, S5). Furthermore, we identified virulence-associated changes in the NS1 and NS2/NEP proteins: NS1-K217E was present in studied BB genotype virus; NS1-N205S in combination with NS2-T48A was present in all studied viruses; NS2-E67G was detected specifically in A3 and G10 genotypes (Tables S9, S10). These findings highlight genotype-specific molecular signatures that may influence the pathobiological potential of circulating 2.3.4.4b A(H5N1) viruses. Analysis of sequencing data from the investigated viruses did not reveal any SNPs with a frequency exceeding 5% that resulted in AASs associated with previously described functional changes12. Virulence All tested A(H5N1) influenza viruses exhibited high virulence in 10-day-old embryonated chicken eggs (ECEs), resulting in embryo mortality within 48 h post-inoculation. Virus titers in allantoic fluid ranged from 8.7 to 10.7 log10 EID50/mL (Table 1). Table 1Virological properties of the studied influenza A(H5N1) viruses.StrainSubtypePB2-K482R and NP-N319K mutationsTiter in chick embryos, log10 EID50/mLBALB/c intranasal ID50, log10 EID50/mLBALB/c intranasal LD50, log10 EID50/mLReference virusesA/wigeon/Sakha/1/2014*H5N8No6.2 ± 0.52.9 ± 0.44.1 ± 0.4A/rook/Chany/32/2015*H5N1No8.7 ± 0.41.9 ± 0.42.9 ± 0.8A/great crested grebe/Tyva/34/2016*H5N8No9.3 ± 0.31.7 ± 0.52.3 ± 0.5A/chicken/Sergiyev Posad/38/2017*H5N8No9.2 ± 0.61.9 ± 0.42.9 ± 0.6A/chicken/Kostroma/1718/2017*H5N2No8.9 ± 0.41.6 ± 0.61.8 ± 0.6A/common gull/Saratov/1676/2018**H5N6No8.9 ± 0.62.4 ± 0.5 > 5.0A/Astrakhan/3212/2020**H5N8No9.2 ± 0.31.2 ± 0.22.4 ± 0.4Test virusesA/fur seal/Sakhalin/399-1V/2023 (genotype A3)H5N1Yes8.7 ± 0.81.0 ± 0.41.3 ± 0.6A/chicken/Magadan/14-7V/2022 (genotype A3)H5N1Yes9.7 ± 0.41.1 ± 0.32.0 ± 0.4A/turkey/Tyumen/81-96V/2021 (genotype G10)H5N1No10.7 ± 0.63.6 ± 0.65.4 ± 0.6A/gull/Mari El/345-5V/2023(genotype BB)H5N1No9.7 ± 0.43.6 ± 0.65.5 ± 0.4* Marchenko et al.,33; ** Marchenko et al.,13. Virological properties of the studied influenza A(H5N1) viruses. A/gull/Mari El/345-5V/2023 (genotype BB) * Marchenko et al.,33; ** Marchenko et al.,13. The 50% infectious dose (ID50) and 50% lethal dose (LD50) were determined for intranasal infection of Balb/c mice (Table 1). These values were compared with data for reference strains described in our previous studies13,33. The A3 genotype A(H5N1) strains demonstrated high virulence level in mice; this was in contrast to the genotypes G10 and BB A(H5N1) viruses, for which medium virulence level was shown, as was the case with H5Nx reference strains A/wigeon/Sakha/1/2014 and A/common gull/Saratov/1676/2018 (Table 1, Fig. 4). Survival rates of BALB/c mice following intranasal infection with influenza A(H5N1) clade 2.3.4.4b virus at varying doses. Groups of mice (n = 6 per group) were intranasally inoculated with increasing doses of clade 2.3.4.4b A(H5N1) influenza virus. Survival was monitored daily over a 14-day observation period. The data is presented for four viruses: (A) A/fur seal/Sakhalin/399-1V/2023 A(H5N1), (B) A/chicken/Magadan/14-7V/2022, (C) A/turkey/Tyumen/81-96V/2021, (D) A/gull/Mari El/345-7V/2023. In addition, the A/fur seal/Sakhalin/399-1V/2023 virus exhibited high virulence in ferrets following intranasal inoculation, demonstrating an ID50 of 1.9 ± 0.8 log10 EID50/mL and an LD50 of 2.6 ± 1.0 log10 EID50/mL. Transmission Airborne transmission of the A(H5N1) virus A/fur seal/Sakhalin/399-1V/2023 was evaluated in ferrets using an experimental setup illustrated schematically in Fig. 5. All intranasally inoculated donor ferrets exhibited clinical signs of disease. Ferrets infected with challenge doses of 2.7 and 3.7 log10 EID50/mL displayed symptoms including reduced appetite, dyspnea, hind limb paresis, ataxia, torticollis, and decreased activity (mean relative inactivity index: 1.5 for both groups; Table S11). In addition to these symptoms, ferrets receiving higher doses (4.7 and 5.7 log10 EID50/mL) developed sneezing and wheezing, with mean relative inactivity indices of 2.0 and 1.8, respectively (Table S11). Of the 12 inoculated donor ferrets, 10 succumbed to infection between 7 and 11th days post-inoculation (dpi), or were humanely euthanized when they reached 20% loss of body weight or exhibited such clinical signs as hind limb paresis, ataxia, or torticollis. Surviving donor animals showed a maximum weight loss of 12% by 14 dpi (Fig. 6). Airborne transmission experiment in ferrets infected with the clade 2.3.4.4b influenza A(H5N1) virus A/fur seal/Sakhalin/399-1V/2023. Donor ferrets (n = 3 per dose group) were intranasally inoculated with 5.7, 4.7, 3.7, or 2.7 log10 EID50/mL of virus. On day 1 post-inoculation (dpi), naïve recipient ferrets (n = 3 per group) were placed in separate cages within a TIEGEL ELC 04–60 dynamic aerobiology chamber, with unidirectional airflow from donor to recipient cages. Cages were arranged at a distance sufficient to prevent direct contact while allowing for airborne transmission. Animals were co-housed in this setup for 5 h daily over 5 consecutive days (dpi 1–5). Clinical signs were monitored throughout the study, and nasal wash samples were collected at defined time points from both inoculated and recipient ferrets. Blood samples were collected at 21 dpi from surviving donor ferrets and at 21 days post-contact (dpc) from all recipient ferrets to assess seroconversion. Results of airborne transmission experiment in ferrets. (A) Survival rates of donor ferrets following intranasal inoculation with clade 2.3.4.4b influenza A(H5N1) virus A/fur seal/Sakhalin/399-1V/2023 in the airborne transmission experiment. Four groups of three ferrets were inoculated with increasing doses of virus (2.7 – 5.7 log10 EID50/mL) and monitored for survival over the course of the study. (B) Infectious viral titers in nasal washes from donor ferrets following intranasal inoculation with varying doses of virus. Four groups of ferrets (n = 3 per group) were inoculated with 2.7, 3.7, 4.7, or 5.7 log10 EID50/mL, and nasal washes were collected daily to quantify viral shedding. Infectious virus titers are expressed as log10 FFU/mL (mean ± SD for each group of ferrets). The limit of detection is indicated by the dashed line. (C) Weight loss of donor ferrets in the airborne transmission experiment following infection with 2.7, 3.7, 4.7, 5.7 log10 EID50/mL, respectively (mean ± SD for each ferret group). Triangles indicate animal deaths. (D) Weight loss of recipient ferrets in the airborne transmission experiment following contact with donor ferrets. (E) Change in body temperature of donor ferrets in the airborne transmission experiment following infection with 2.7, 3.7, 4.7, 5.7 log10 EID50/mL, respectively (mean ± SD for each ferret group). Triangles indicate animal deaths. (F) Change in body temperature of recipient ferrets in the airborne transmission experiment. Fever responses varied with dose. Mean increase in body temperature registered over the course of infection in deceased donor ferrets from experimental groups inoculated with 5.7, 4.7, 3.7 and 2.7 log10 EID50/mL was 2.3 °C, 2.2 °C, 1.2 °C and 1.8 °C, respectively (Fig. 6). Survived donor ferrets exhibited milder fever responses, with maximum temperature increases of 1.4 °C and 1.6 °C following infection with 3.7 and 2.7 log10 EID50/mL, respectively (Fig. 6). High viral loads were detected in nasal washes from all donor ferrets throughout the observation period (from 1 to 5 dpi), as determined by RT-PCR (Table S12). Mean infectious virus titers, calculated across all sampling days per group, were 4.3, 4.7, 3.9, and 4.0 log10 FFU/mL in groups infected with 5.7, 4.7, 3.7, and 2.7 log10 EID50/mL, respectively (Fig. 6). Serological analysis of surviving donor ferrets at 21 dpi revealed hemagglutination inhibition (HAI) antibody titers of 1:80 and ≥ 1:160 following infection with 3.7 and 2.7 log10 EID50/mL, respectively. Microneutralization assay confirmed the presence of virus-specific neutralizing antibodies at titers > 1:640 in both groups (Table S11). No clinical signs of infection were observed in recipient ferrets during the entire observation period. Relative inactivity indices remained at 1.0 for all recipient groups from 1 to 7 days post-contact (dpc) with donor animals (Table S11). Recipient ferrets did not exhibit significant weight loss (Fig. 6). Mean increase in body temperature was 1.8 °C (Fig. 6). RT-PCR and viral titration on MDCK cells yielded negative results for all nasal wash samples collected from recipient ferrets between 2 and 12 dpc (Table S12). Furthermore, no virus-specific antibodies were detected in sera obtained from recipient ferrets at 21 dpc, based on HAI and microneutralization assays (Table S11). Discussion The global expansion of highly pathogenic avian influenza A(H5N1) clade 2.3.4.4b viruses into diverse mammalian hosts since 2022 has raised significant concerns regarding zoonotic potential and pandemic risk34. Although such spillover events are frequently accompanied by the acquisition of mammalian-adaptive mutations, the contribution of specific genetic changes—particularly in the context of emerging genotypes—remains incompletely understood. Reassortment among clade 2.3.4.4b viruses has generated multiple genotypes having varying phenotypic properties2. Virulence and transmission may be critically influenced by the interplay between adaptive substitutions and genomic background. In 2023, A(H5N1) clade 2.3.4.4b viruses of genotype A3 were isolated from northern fur seals (Callorhinus ursinus) in the Russian Far East during a mass mortality event. Genetically similar A3 viruses were concurrently detected in wild birds on Sakhalin Island, strongly implicating avian species as the likely source of infection. This genotype joins others—such as A2, B1.2, B3.2, and D1.1—identified in marine mammals across the Americas during 2022–2023 (GISAID), underscoring the widespread circulation of reassorted A(H5N1) variants in non-avian hosts. We used a murine model to demonstrate that A3 genotype viruses exhibit enhanced virulence compared to both a closely related genotype G10 virus (A/turkey/Tyumen/81-96V/2021) and genotype BB viruses, which have caused extensive outbreaks in gulls and other birds across Europe and western Russia. The heightened pathogenicity of A3 viruses correlates with the presence of two key mammalian-adaptive substitutions: PB2-K482R and NP-N319K. PB2-K482R has been shown to increase virulence in A(H7N9)32 and mouse-adapted seasonal influenza viruses35,36, while NP-N319K enhances viral polymerase activity in mammalian cells by strengthening NP binding to importin-α1 and -α7, thereby facilitating nuclear import—particularly when co-occurring with PB2 mammalian-adapting mutations37,38. Structural studies indicate that NP residue 319 lies within a domain that interacts directly with PB2, supporting the potential for synergistic effects between these proteins39,40. Notably, a previous study associated an A(H5N1) isolate from Russia with the same seal outbreak41 that harbored the PB2-E627K substitution, along with PB2-K482R and NP-N319K AASs characteristic of the A3 genotype. This further underscores the propensity of marine mammal-derived influenza viruses to accumulate multiple adaptive markers associated with mammalian replication. Additionally, the A3 viruses in our study carried substitutions in NS1 and NEP/NS2 previously linked to increased virulence, suggesting that genome-wide synergistic interactions may further modulate pathogenicity. Despite the potentially important role of NP-N319K and PB2-K482R mutations in mammalian virulence, they remain understudied in contemporary A(H5N1) 2.3.4.4b viruses, and their functional roles need to be investigated through reverse genetics or other functional characterization approaches. In light of previous studies, our findings underscore the capacity of A(H5N1) clade 2.3.4.4b viruses to rapidly acquire combinations of mutations—through reassortment or selection of adaptive variants during mammalian infection—that enhance virulence in mammalian hosts. Prior studies have shown that specific PB2 mutations associated with increased virulence may also enhance the transmission of avian influenza viruses, including clade 2.3.4.4b A(H5N1), in mammalian models13,18,27–29. However, enhanced virulence does not necessarily confer efficient airborne transmission. To evaluate this critical parameter, we assessed the airborne transmission of the fur seal-derived A3 virus (A/fur seal/Sakhalin/399-1V/2023) in a ferret model using a validated dynamic aerobiology chamber (TIEGEL ELC 04–60). Despite high and sustained viral loads in nasal washes of donor ferrets over five days post-infection, no airborne transmission to naïve contact animals was observed under controlled environmental conditions (20 °C, 30% RH). Under identical experimental conditions and using the same transmission system in our laboratory, efficient airborne transmission was previously demonstrated for pandemic A(H1N1)pdm09 and swine-origin A(H3N2) viruses, limited airborne transmission for 2.3.4.4b A(H5N1) carrying PB2-E627K, and no transmission for investigated H5 clades 2.3.4.4b and 2.3.2.1c viruses13,42. The lack of transmission for the A(H5N1) virus in this study is likely multifactorial. First, the A3 virus retains strong preference for avian-type α2,3-linked sialic acid receptors, whereas the upper respiratory tract of ferrets predominantly expresses human-type α2,6-linked receptors—a key barrier to efficient airborne spread. Second, although viral replication in donors was robust, titers and infectivity were lower than those achieved by a PB2-E627K—bearing A(H5N1) strain that itself showed only limited transmission under identical conditions13. Third, the absence of other known adaptation markers—such as HA mutations that lower the pH of fusion or enhance binding to α2,6 receptors—may further restrict transmission potential43. Our study has certain limitations. The donor-recipient exposure period was limited to 5 h per day due to technical constraints of the chamber, though this duration has proven sufficient to detect transmission in positive-control experiments. Only female ferrets were used; while sex-based differences in influenza immunity have been observed44, their impact on airborne transmission remains uncertain. Nevertheless, the experimental conditions used in our study closely mirror those used in benchmark transmission studies to enhance comparability. From a pandemic risk perspective, the World Health Organization’s Tool for Influenza Pandemic Risk Assessment (TIPRA) identifies mammalian adaptation, disease severity, and transmission as key determinants of pandemic potential30. Although the A3 genotype harbors mammalian-adaptive mutations PB2-K482R and NP-N319K and exhibits high virulence, it shows no detectable airborne transmission in the ferret model. These findings indicate that, despite enhanced pathogenicity due to its genetic features, the virus lacks the capacity for efficient respiratory transmission in mammals, thereby representing low pandemic risk. To date, no human infections with A3 genotype viruses have been reported (GISAID); however, the ongoing circulation of this genotype in wild and domestic birds across the Russian Far East and geographically neighboring regions—including other countries—poses a persistent risk of spillover into mammals, including humans. In conclusion, our data indicate that currently circulating A3 genotype A(H5N1) viruses, despite exhibiting enhanced virulence in mammals, pose a limited immediate pandemic threat owing to their inability to transmit efficiently via the airborne route. Nevertheless, their capacity to acquire mammalian-adaptive mutations during spillover events into marine mammals underscores the dynamic evolutionary potential of clade 2.3.4.4b viruses. Sustained, integrated surveillance—complemented by functional characterization of emerging variants—is therefore essential for monitoring the accumulation of mammalian-adaptive genetic features and enabling the timely detection of avian influenza virus variants with heightened pandemic potential. Methods Conclusion of the bioethics committee All animal procedures were conducted in accordance with international, national, and institutional guidelines for the care and use of laboratory animals. This study complies with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines (https://arriveguidelines.org/). Experimental protocols were approved by the BioEthics Committee at the SRC VB Vector under the protocols №1 (28 June 2022) and №9 (10 November 2023). All studies were conducted under Animal Biosafety Level 3 (ABSL-3) conditions. Avian influenza strains Avian influenza strains, used in the experiment: A/fur seal/Sakhalin/399-1V/2023 (H5N1) clade 2.3.4.4b (GISAID ID: EPI_ISL_18462515); A/chicken/Magadan/14-7V/2022 (H5N1) clade 2.3.4.4b (GISAID ID: EPI_ISL_16618968); A/turkey/Tyumen/81-96V/2021 (H5N1) clade 2.3.4.4b (GISAID ID: EPI_ISL_8769033); A/gull/Mari El/345-5V/2023 (H5N1) clade 2.3.4.4b (GISAID ID: EPI_ISL_18525123). Cell culture, eggs and animal models Cell culture The Madin-Darby Canine Kidney (MDCK) cell line used in the study was obtained from the Cell Culture Collection of the Federal Budget Research Institution State Research Center of Virology and Biotechnology “Vector” (FBRI SRC VB “Vector”), Rospotrebnadzor, Russia. The MDCK cell line was initially acquired by the collection from the O.G. Andzhaparidze Research Institute of Viral Preparations, Russian Academy of Medical Sciences. This cell line has been routinely maintained under standard culture conditions and was used in accordance with institutional biosafety and cell line authentication guidelines. Eggs For influenza virus propagation, 10-day-old clean embryonated chicken eggs (ECEs) were used. These eggs were obtained from farms free from infectious diseases (CJSC Poultry Farm “Novo-Baryshevskaya”). Animals Female ferrets (6–8 months old) and Balb/c mice (18–20 g) were sourced from the animal breeding facility at the SRC VB “Vector”. Animals were housed in individually ventilated cages and maintained on a standard diet. Pre-experimental serological screening confirmed the absence of pre-existing influenza-specific antibodies. Humane endpoints for experimental animals were applied based on predefined ethical criteria, including weight loss exceeding 20%. Euthanasia was performed using a two-stage CO2 exposure system (Euthanizer, Russia), with 100% CO2 gradually introduced at a displacement rate of 30% to 70% of the chamber volume per minute, in accordance with the AVMA 2020 euthanasia guidelines. Influenza virus propagation and titration Virus isolation of A/fur seal/Sakhalin/399-1V/2023 and A/gull/Mari El/345-5V/2023 have been previously described25. Virus isolation of A/chicken/Magadan/14-7V/2022 and A/turkey/Tyumen/81-96V/2021 was performed as follows: Influenza A(H5N1) viruses were isolated from internal organs of dead birds by inoculation into the allantoic cavities of 10-day-old ECEs, followed by incubation at 37 °C for 48 h. Viral titers were determined by serial tenfold dilution of viral suspensions and inoculation of 0.1 mL per egg into additional ECEs. The 50% egg infectious dose (EID50) was calculated using the method of Reed and Muench45. RNA extraction and influenza virus typing Viral RNA was extracted from 0.1 mL allantoic fluid using either the AmpliSens RIBO-prep kit (Central Research Institute of Epidemiology, Rospotrebnadzor, Russia) or MagnoPrime Uni kit (NextBio Ltd., Russia), according to manufacturer protocols. Reverse transcription was performed using the Reverta-L kit (Central Research Institute of Epidemiology, Russia). Virus typing and subtyping were performed using commercial two-step real-time RT-PCR kits: “AmpliSens Influenza virus A/B-FL”, “AmpliSens Influenza virus A H5N1-FL”, and “AmpliSens Influenza virus A-type-H5, H7, H9-FL” (Central Research Institute of Epidemiology, Russia), following the manufacturers’ instructions. The quantitative RT-PCR program contained 45 cycles (10 dark cycles followed by 35 cycles with detection). The limit of detection was defined as a cycle threshold (Ct) value of 33, in accordance with the manufacturers’ instructions; samples with Ct > 33 were considered negative or below the limit of detection. Sequence analysis of influenza viruses Viral RNA was extracted from samples using the AmpliSens RIBO-prep kit (Central Research Institute of Epidemiology, Rospotrebnadzor, Russia) and MagnoPrime Uni (NextBio Ltd., Russia), according to the manufacturers’ instructions. Reverse transcription was performed with a primer mix consisting of Uni12, Uni12.4, and Uni1346 using the RT-M-MuLV-RH kit (LLC BIOLABMIX, Russia). PCR amplification was performed using the BioMaster LR HS-Taq PCR kit (2 ×) (LLC BIOLABMIX, Russia). Library preparation and next-generation sequencing (NGS) were carried out on an Illumina MiSeq platform using the MiSeq Reagent Kit v3 (Illumina, San Diego, CA, USA). Full-length viral genomes were assembled by aligning the sequencing reads to reference genomes using BWA v0.7.1547. The resulting nucleotide sequences have been deposited in the Global Initiative on Sharing All Influenza Data (GISAID) database and NCBI repository. Maximum likelihood phylogenetic analysis was performed in IQ-TREE v2.0.748 applying the best-fitted nucleotide substitution model selected by ModelFinder49 with 500 bootstrap replicates and visualized in FigTree v1.4.450. Reference strains genome sequences were used from GISAID. Amino acid substitutions have been identified using Treesub51. The genotype was assigned using methodologies as previously described25. Virulence To assess the virulence of avian influenza viruses, female BALB/c mice and female ferrets were intranasally inoculated. Mice were divided into 12 groups of six animals each to determine the 50% infectious dose (ID50) and 50% lethal dose (LD50). Prior to infection, animals were anesthetized with a combination of Zoletil 100 (Delpharm Tours, France) and Xyla (Interchemie, Estonia) at a dose calculated as follows: tiletamine/zolazepam 20 mg/kg + xylazine 10 mg/kg (intramuscular), as described previously52. Then, the animals were administered 0.05 mL of serial dilutions of the second-passage virus intranasally. Six groups were humanely euthanized three days post-inoculation, and lung tissues were collected and tested for viral presence using 10-day-old ECEs, as described previously. The remaining six groups were monitored daily for clinical signs and mortality over a 14-day period. ID50 and LD50 values were calculated using the method of Reed and Muench45. Virus virulence levels in mice were categorized as low (MLD50 > 6.5 log10 EID50), medium (3 log10 EID50 < MLD50 ≤ 6.5 log10 EID50), or high (MLD50 ≤ 3 log10 EID50) pathogenicity, as described previously53. Ferrets were divided into 8 groups of four animals each to determine the 50% infectious dose (ID50) and 50% lethal dose (LD50). Prior to infection, animals were anesthetized with a combination of Zoletil 100 (Delpharm Tours, France) and Xyla (Interchemie, Estonia) at a dose calculated as follows: tiletamine/zolazepam 3 mg/kg + xylazine 3 mg/kg (intramuscular), as described previously54. Then, the animals were administered 0.5 mL of serial dilutions of the second-passage virus intranasally. The remaining 8 groups were monitored daily for clinical signs, relative inactivity indices, mortality over a 14-day period. Nasal wash samples were collected to confirm infection by virus isolation in MDCK cell culture. ID50 and LD50 values were calculated using the method of Reed and Muench45. Transmission Transmission was evaluated using the TIEGEL ELC 04–60 aerobiology chamber (TIEGEL GmbH, Radeberg, Germany), as previously described42. Eight groups of ferrets were used: four donor groups (n = 3 per group) and four recipient groups (n = 3 per group). Donors were intranasally inoculated with 5.7, 4.7, 3.7, or 2.7 log10 EID50/mL of A/fur seal/Sakhalin/399-1V/2023 (H5N1) virus under anesthesia. On day 1 post-inoculation (dpi), naïve recipient ferrets (n = 3 per group) were placed in the same chamber in separate cages within a TIEGEL ELC 04–60 dynamic aerobiology chamber (cages were separated by a distance of 10 cm to prevent direct contact while allowing airborne transmission), with unidirectional airflow from donor to recipient cages (flow rate 3 cm/sec). A relative humidity of 30%, an air temperature of 20 °C, and a reduced air pressure of − 200 Pa were maintained in the cabinet chamber during airborne transmission experiments. Animals were co-housed in this setup for 5 h daily over 5 consecutive days (dpi 1–5). During the experiment the monitored characteristics included weight loss, body temperature, relative inactivity indices, respiratory and neuropathologic symptoms. Weight and temperature measurements were expressed as mean ± SD for each group of ferrets. Nasal wash samples were collected at defined time points from both inoculated and recipient ferrets. Blood samples were collected at 21 dpi from surviving donor ferrets and at 21 days post-contact (dpc) from all recipient ferrets to assess seroconversion. Determination of viral load in biological samples in MDCK cells using focus forming unit (FFU) assay MDCK cell monolayers were cultured in 96-well plates. Nasal washings from experimental animals were serially diluted tenfold. Maintenance medium was used for cell washing, sample dilution, and incubation (DMEM supplemented with 100 units/mL penicillin, 100 μg/mL streptomycin, 0.25 μg/mL amphotericin B, 25 mM HEPES, 0.2% bovine serum albumin, and 2 μg/mL TPCK-treated trypsin). Growth medium was removed from wells containing confluent monolayers and cells were washed twice with maintenance medium, as described previously33. Serial dilutions of each sample (from 10–1 to 10–4) were added in triplicate at 100 μL per well. Plates were incubated for 1 h at 37 °C with 5% CO2. Following incubation, inoculum was removed, and cells were rinsed once with maintenance medium. A volume of 150 μL of fresh maintenance medium was added to each well. After 18–20 h of incubation at 37 °C and 5% CO2, the medium was removed and cells were fixed with 200 μL of ice-cold 80% acetone (− 20 °C) for 20–30 min. Plates were air-dried for 3–5 min. After then 50 μL of mouse monoclonal anti-influenza A nucleoprotein antibody (Abcam, Cat. No. ab128193; 1:1000 dilution in DPBS) was added to each well. The plate was incubated at 37 °C for 1 h, followed by three washes with phosphate buffered saline (PBS). Wash volumes of buffer were 200 μL per well. Wash buffer was removed from wells immediately. Rabbit anti-mouse IgG secondary antibodies conjugated to Alexa Fluor 488 (Abcam, Cat. No. ab150113; 1:1000 dilution in DPBS) were then added, and the plate was incubated for an additional hour at 37 °C. After three final washes with PBS, fluorescent foci indicative of viral replication were visualized and quantified using an automated imaging system. Virus titers were expressed as focus-forming units per milliliter (FFU/mL) of nasal washings (mean log10 FFU/mL ± SD for each group of ferrets). Determination of virus RNA load in biological samples by RT-PCR RNA was extracted from nasal washes using the RIBO-sorb kit (InterLabService Ltd., Moscow, Russia). Reverse transcription was carried out with the Reverta-L kit (InterLabService Ltd.). Real-time RT-PCR was performed using the two-step protocol of the commercial “AmpliSens Influenza virus A H5N1-FL” kit (Central Research Institute of Epidemiology, Russia) on a RotorGene 6000 thermal cycler (Qiagen). Quantification was based on comparison with control standards run in parallel. The quantitative RT-PCR program contained 45 cycles (10 dark cycles followed by 35 cycles with detection). The limit of detection was defined as a cycle threshold (Ct) value of 33, in accordance with the manufacturers’ instructions; samples with Ct > 33 were considered negative or below the limit of detection. Evaluation of seroconversion In order to assess seroconversion in ferrets over the course of the airborne transmission experiment, serum samples obtained from donor and recipient ferrets were assessed using hemagglutination inhibition assay (HAI) and microneutralization assay. Serum samples were collected 3 days before and 21 days after infection/exposure. Receptor-destroying enzyme (RDE) was added to the serum at a ratio of one part serum to three parts RDE. The mixture was incubated at 37 °C for 18 h, followed by heat inactivation in a water bath at 56 °C for 30 min. Then six parts of PBS were added. The processed sera were aliquoted and stored at − 20 °C until used in HAI and microneutralization assays. The remaining serum samples were stored at − 80 °C in an ultra-low-temperature freezer. HAI was performed as previously described55. Serial twofold dilutions of serum samples were incubated with 4 hemagglutination units of the studied virus A/fur seal/Sakhalin/399-1 V/2023 for 30 min at room temperature. Then, a standard dose of 0.5% turkey red blood cells suspension was added to reaction. After erythrocytes completely settled, HAI titer was registered as the highest serum dilution which inhibited hemagglutination. Microneutralization assay was performed as previously described55. In a 96-well immunological plate, serial twofold dilutions of the test sera (1/10, 1/20, 1/40, 1/80, 1/160, etc.) are prepared in 100 µL of maintenance medium. Note that the final dilution will be halved after the addition of 100 µL of viral suspension. To the diluted sera, 100 µL of titrated virus with a calculated concentration of 200–300 FFU per well is added. The mixture is then incubated for 1 h at 37 °C in a CO2 atmosphere. Following incubation, the serum-virus mixture is transferred to a 96-well plate containing a confluent monolayer of cells (90–100% confluence). It is further incubated for 1 h at 37 °C in a CO2 atmosphere. After incubation, the contents of the wells are removed and 150 µL of maintenance medium is added to each well. The plate is then incubated for 18 h at 37 °C in a CO2 atmosphere. After 18–20 h of incubation at 37 °C and 5% CO2, the medium was removed and cells were fixed with 200 μL of ice-cold 80% acetone (− 20 °C) for 20–30 min. Plates were air-dried for 3–5 min. After then 50 μL of mouse monoclonal anti-influenza A nucleoprotein antibody (Abcam, Cat. No. ab128193; 1:1000 dilution in DPBS) was added to each well. The plate was incubated at 37 °C for 1 h, followed by three washes with phosphate buffered saline (PBS). Wash volumes of buffer were 200 μL per well. Wash buffer was removed from wells immediately. Rabbit anti-mouse IgG secondary antibodies conjugated to Alexa Fluor 488 (Abcam, Cat. No. ab150113; 1:1000 dilution in DPBS) were then added, and the plate was incubated for an additional hour at 37 °C. After three final washes with PBS, fluorescent foci indicative of viral replication were visualized and quantified using an automated imaging system. The titer of neutralizing serum antibodies is calculated as the highest serum dilution at which the number of FFU is reduced by 50% or more compared to the average FFU value in the control wells containing only the working dilution of the virus. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Author contributions P.A.: data interpretation, original draft preparation, writing-reviewing and editing. G.A.: animal experiments, data interpretation, original draft preparation. K.N., G.N.: PCR analysis, N.G.S. experiments, bioinformatics analysis and data interpretation, writing-reviewing and editing. D.A., S.K.: PCR analysis, N.G.S. experiments, bioinformatics analysis and data interpretation. D.N., P.O.: animal experiments. M.A., D.E.: determination of viral load in biological samples in MDCK cells. S.S.: data interpretation, writing-reviewing and editing. O.G., V.N., E.M.: influenza virus propagation and titration, data interpretation. M.V.: formulation of overarching research goals and aims, original draft preparation, supervision. Funding This research was funded by the state assignment of FBRI SRC VB “Vector”, Rospotrebnadzor, Russia. Data availability The datasets generated and/or analysed during the current study are available in the NCBI repository, accession numbers PV998154-PV998169, PV998177-PV998192. Declarations Competing Interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Anastasia S. Panova and Andrey S. Gudymo have equally contributed to the work. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets generated and/or analysed during the current study are available in the NCBI repository, accession numbers PV998154-PV998169, PV998177-PV998192.