INTRODUCTION
Avian influenza viruses usually induce asymptomatic infection in their natural primary hosts, which include waterfowls of the Anseriformes and Charadriiformes orders. They sporadically infect other species of wild and domestic birds with a variable degree of clinical manifestations. The prolonged circulation of viruses of the H5 and H7 subtypes among chickens is accompanied by the accumulation of mutations that contribute to an increased virulence and the emergence of highly pathogenic strains [1, 2]. Since the sporadic infection of animals and humans with highly pathogenic avian viruses can lead to the emergence of new viruses with zoonotic and pandemic potential, these viruses attract much attention.
The major influenza epidemics and pandemics in the 20th–21st centuries were often induced by viruses that contain genome segments of avian origin [3–5]. One of the hypotheses of the origin of the H1N1 influenza virus, which caused the devastating Spanish flu pandemic in 1918, suggests that this virus consisted entirely of genes of avian origin [6, 7]. The alternative hypothesis assumes that the 1918 pandemic was caused by the virus strain that had emerged as a result of reassortment between avian, swine, and human influenza viruses circulating several years before the pandemic rather than by direct transmission of the virus from birds to humans [8]. In any case, some genes of the 1918 pandemic virus are still present in the influenza virus gene pool [9].
Highly pathogenic influenza viruses of the H5 subtype, which periodically induce devastating epizootics, are known to repeatedly evolve from low-virulent precursors asymptomatically circulating among poultry [1, 2]. The highly pathogenic avian viruses currently circulating on the Eurasian continent emerged there in the late 1990s and belong to a lineage originating from the H5N1 A/goose/Guangdong/1/1996 (GsGd) virus. Later on, they were taken to Africa and the North American continent by migratory birds. In some areas, they became endemic. The new variants of the H5Nx subtypes of the highly pathogenic influenza virus, which carry the gene coding for hemagglutinin (HA) of the GsGd line (H5N2, H5N3, H5N4, H5N5, H5N6, H5N8), have emerged as a result of the reassortment of highly pathogenic H5N1 influenza viruses with low pathogenic ones. Some of these viruses occasionally infect humans (H5N1, H5N8, H5N6) [10]. For example, in February 2021, the World Health Organization (WHO) reported the detection of the H5N8 influenza virus in the nasopharyngeal swabs of seven employees who had been working on a poultry farm in the Astrakhan region of the Russian Federation during an outbreak of avian influenza [11].
Many mallards live in the city of Moscow water reservoirs in close contact with people and animals. In the 2006–2019 period, influenza viruses of different subtypes, including H1 and H5, were isolated from mallards habituating this area [12]. The goal of the present study was to investigate the probability of transmission of the avian influenza viruses of H1 and H5 subtypes isolated from mallards to mammals. To this end, we adapted avian influenza viruses to the growth in mouse lungs, studied their phenotypic properties, and identified the genetic changes that arose during the adaptation.
MATERIALS AND METHODS
Viruses
The two prototype avian influenza viruses A/duck/Moscow/4970/2013 (H1N1) and A/duck/Moscow/4182/2010 (H5N3) were kindly provided by Dr. A. Gambaryan (Chumakov Federal Scientific Center for Research and Development of Immune and Biological Products, Russia). The virus A/duck/Moscow/4182/2010 (H5N3) was one-time cloned in chicken embryos by the limiting dilution method; the obtained virus was named A/duck/Moscow/4182-C/2010 (H5N3). The viruses A/duck/Moscow/4970/2013 (H1N1) (4970/H1), A/duck/Moscow/4182/2010 (H5N3) (4182/H5), and A/duck/Moscow/4182-C/2010 (H5N3) (4182C/H5) were deposited in the State collection of viruses (Gamaleya National Center of Epidemiology and Microbiology, Moscow, Russia) with the following assigned numbers 2848, 2849, and 2847, respectively.
Adaptation of viruses to the growth in the mouse lungs
Groups of white outbred mice were infected intranasally with the allantoic fluid of 4970/H1 or 4182C/H5 viruses. In 48 h, the mice were euthanized with an overdose of diethyl ether, the lungs were removed, pooled for 2 animals, and homogenized in 1.5 ml of lactalbumin hydrolysate medium without glutamine (PanEko, Russia). The pulmonary suspension was centrifuged for 10 min. at 850 g, and the supernatant was used for the next intranasal infection of the mice. The presence of the virus in the lungs was determined by hemagglutination assay. After the successive 9 passages of the 4970/H1 virus and 10 passages of the 4182C/H5 virus in the lungs of mice, both viruses were cloned in chicken embryos by the limiting dilution method.
All of the studies with the animals were carried out in strict compliance with the rules for carrying out work with the use of experimental animals (Order No. 266 of the Ministry of Health of the Russian Federation dated 6/19/2003).
Hemagglutination assay
Series of twofold dilutions of the influenza virus were performed in microtiter plates in a buffer solution (0.15 M NaCl, 0.01 M Tris-HCl, pH 7.2). Then, an equal volume of 1% chicken red blood cells was added to each well and incubated at 4°C for 20–30 min. The reciprocal of the last virus dilution that agglutinates erythrocytes was considered as a virus titer and was expressed in hemagglutinating units (HAU) [13].
Viral infectious titer in chicken embryos
In order to determine the viral infectious titer, 10-day-old chicken embryos were infected with 100 μl of tenfold serial dilutions of the virus into the allantoic cavity (5 embryos with every dilution). After 48 h of incubation at 37°C, the chicken embryos were kept overnight at 4°C, and then the presence of the virus in the allantoic fluid of each embryo was assessed by the hemagglutination assay with a 1% suspension of chicken erythrocytes. The viral titer was calculated as the 50% embryo infectious dose (EID50) by the Reed & Muench [14] and expressed as log10EID50.
Assessment of the mouse lethal dose
BALB/c mice were infected intranasally with consecutive dilutions of influenza virus (80 μl per mouse) under light ether anesthesia (six animals per dilution). The animals were observed for 14 days, and changes in body weight and their survival were assessed. The lethal dose (LD50) was calculated according to the method of Kerber in Ashmarin’s edition and expressed as log10EID50 in one unit of LD50 [15].
Nucleotide Sequencing by the Sanger method
Viral RNA was isolated from the allantoic fluid of infected embryos using a commercial QIAamp Viral RNA mini kit (Qiagen, Germany) according to the manufacturer’s instructions. Reverse transcription was performed at 42°C during 1 h in 25 μl of a reaction mixture containing 8 μl of RNA, 1 μl of uni12 primer with a concentration of 50 ng/μl (13.5 nM), 10 μl of water, 1 μl of 10 mM dNTP, 5 μl of 5x buffer, and 100 units of Moloney Murine Leukemia Virus (MMLV) reverse transcriptase (Alpha Ferment Ltd., Moscow). The resulting cDNA was used for PCR with specific terminal primers in order to synthesize full-length genome segments [16]. The amplified fragments were separated by electrophoresis in 1.0–1.3% agarose gel in the presence of ethidium bromide and eluted from the gel with a Diatom DNA Elution kit (Isogene Laboratory Ltd., Russia). Sequencing was performed with terminal or internal primers [17] using the BrightDye™ Terminator Cycle Sequencing Kit v3.1 (NimaGen, Netherlands) followed by analysis on an automatic DNA sequencer ABI PRISM 3100-Avant Genetic Analyzer (Applied Biosystems, USA). The DNASTAR Sequence Analysis Software Package (DNASTAR Inc., USA) was used for the assembling and analysis of nucleotide sequences.
The nucleotide sequences of the complete genomes of 4970/H1 (accession numbers MW897994-MW897999, MF969258, MF969259) and 4970MA/H1 viruses (accession numbers MW898002-MW898009) were deposited in the GenBank (https://www.ncbi.nlm.nih.gov/nuccore). For all the segments of the 4182C/H5 and 4182C-MA/H5 viruses, the numbering was assigned according to the wild strain 4182/H5. The sequences of 8 gene segments of 4182/H5 virus were deposited in the GenBank with the accession numbers KF885672-KF885679.
Influenza virus HA thermal stability
The virus-containing allantoic fluid was clarified by low-speed centrifugation and diluted with phosphate buffered saline (PBS) to 128 HAU. Then, 120 μl aliquots were placed into thin-walled PCR tubes (0.5 ml, SSI, USA), incubated in a Master-cycler Gradient 5331 thermal cycler (Eppendorf, Germany) at temperatures ranging from 50°C to 70°C for 40 min., and then immediately chilled on ice. The control sample was stored for 40 min. at 0°C. After incubation, the virus hemagglutination titer in each sample was determined.
HA receptor-binding activity
Determination of the receptor-binding activity of viruses was carried out using the synthetic analogs of cell receptors – sialo-oligosaccharides conjugated with high-molecular-weight polyacrylamide [18]. We used the following set of sialo-oligosaccharides: Neu5Acα2-3Galβ1-4Glcβ (3’SL), Neu5Acα2-3Galβ1-4GlcNAcβ (3’SLN), Neu5Acα2-3Galβ1-4(6-O-su)GlcNAcβ (6-su-3>SLN), Neu5Acα2-6Galβ1-4Glcβ (6’SL), Neu5Acα2-6Galβ1-4GlcNAcβ (6’SLN), and Neu5Acα2-6Galβ1-4(6-O-su)GlcNAcβ (6-su-6>SLN). The receptor-binding activity of viruses was assessed by the inhibition of hemagglutination by sialoglycoconjugates [19, 20]. Aliquots (25 μl) of virus-containing allantoic fluid diluted to 4 HAU with 0.01 M Tris-HCl buffer (pH 7.2) and containing 0.15 M NaCl and 10 μM of a neuraminidase inhibitor (2,3-didehydro-2,4-dideoxy-4-amino-N-acetyl-D-neuraminic acid, Sigma, USA) were added to the twofold serial dilutions of sialoglycoconjugates (25 μl) in 96-well microtiter plates. After the incubation at 4°C for 30 min., 50 μl of 0.5% suspension of chicken erythrocytes were added to each well and the plates were incubated at 4°C for another 45 min. The results were determined by the maximum dilution of sialoglycoconjugate, at which the inhibition of hemagglutination was observed and expressed as the concentration of sialic acid (μM) at this dilution.
Hemolysis assay
Virus containing allantoic fluid was clarified by low-speed centrifugation and diluted to 128 HAU with PBS. 50 μl aliquots of 2.5% chicken erythrocytes diluted with the same buffer were added to 250 μl samples of the obtained solution and incubated at 4°C for 1 h with periodic shaking. The erythrocytes with the adsorbed virus were centrifuged at 2800 rpm for 1 min. at 4°C, then the supernatants were carefully removed, and 250 μl aliquots of 0.1 M MES buffer with pH ranging from 5.0 to 7.0 were added to the pellets and incubated for 1 h at 37°C with periodic shaking. The pellet of erythrocytes without virus with the addition of 0.5% Tween-20 in PBS served as a positive control. The pellet of erythrocytes without virus with the addition of 250 μl of MES (pH 7.0) served as a negative control. After incubation, the samples were centrifuged for 1 min. at 2800 rpm and 170 μl aliquots of the supernatant were transferred to microtiter plates for optical density measurement at λ 415 nm using an iMark Microplate Reader (BioRad, USA). The pH value of the HA conformational transition was determined by the analysis of a graph based on the obtained data [21].
Molecular dynamics simulations of HA conformational change
For the analysis of HA conformational change by molecular dynamics simulation, we used data from the Protein Data Bank (PDB; https://www.rcsb.org/) for HA influenza virus isolated from mallard A/WDK/JX/12416/2005 (H1N1) (PDB ID: 3HTO). The calculations were performed using the GROMACS software package [22]. The study of the model protein acid-base properties was conducted using the PROPKA program, version 3.4.0 [23]. The AMBER model was used to estimate the interaction potentials [24].
Statistics
Statistical analysis of the obtained data was carried out using a parametric Student’s t-test (for pH of conformational transition), Friedman’s nonparametric test (ANOVA), and Mann-Whitney test (for calculations of virus titers and pathogenicity). The critical level of significance was p<0.05. The corresponding calculations were performed using MS Office Excel 2016 and Statistica 8.0 software. The obtained results are presented as the arithmetic mean ± standard deviation (mean±SD).
RESULTS
Virus adaptation to reproduction in the lungs of mice
The adaptation of avian influenza viruses 4970/H1 and 4182C/H5 to the reproduction in the lungs of mice was carried out by repeated sequential passages. This experimental approach enables the selection of the virus variant that is capable of growing more actively in the lung tissue of laboratory mice, and finally causes a lethal infection in mice. After 9 passages of the 4970/H1 virus and 10 passages of the 4182C/H5 virus followed by single cloning in chicken embryos the mouse-adapted (MA) variants A/duck/Moscow/4970-MA/2013 (H1N1) (4970MA/H1) and A/duck/Moscow/4182-C-MA/2010 (H5N3) (4182C-MA/H5) were obtained. Viruses were deposited in the State collection of viruses (Gamaleya National Center of Epidemiology and Microbiology, Moscow, Russia) with assigned numbers 2921 and 2922, respectively. Upon adaptation to mice, the virulence of both H1 and H5 viruses increased, and at the same time the 4970MA/H1 virus turned out to be more virulent than 4182C-MA/H5 (Table 1). The original virus 4182C/H5 grew in chicken embryos to the titer of 8.9 log10EID50 and was not pathogenic for mice. The signs of infection were observed in the lungs of mice at the 7th passage. After the 10th passage, infection of mice with virus in a dose of 5.3 log10EID50 resulted in the death of 50% of the animals. 10 50 The prototype virus 4970/H1 was also not lethal for mice. After 9 passages in the mouse lungs, the virus induced the death of 50% of animals upon intranasal infection at a dose of 3.9 log10EID50.
| Virus | Abbreviation | pH of HA transition a | Virulence b | Temperature of HA inactivation, °C |
|---|---|---|---|---|
| A/duck/Moscow/4970/2013 (H1N1) | 4970/H1 | 5.0±0.2 | > 7.5±0.2 | 57.5±0.2 |
| A/duck/Moscow/4970 MA/2013 (H1N1) | 4970MA/H1 | 5.5±0.2 | 3.9±0.2 | 57.5±0.2 |
| A/duck/Moscow/4182 C/2010 (H5N3) | 4182C/H5 | 5.5±0.2 | > 8.9±0.2 | 60.1±0.2 |
| A/duck/Moscow/4182 C–MA/2010 (H5N3) | 4182C–MA/H5 | 5.5±0.2 | 5.3±0.2 | 60.1±0.2 |
The pH values of the HA conformational transition are presented as mean ± SD from the results of three independent experiments.
Virulence in mice is presented as log10EID50 in one unit of LD50. A lower value corresponds to a higher virulence.
The pH of the HA conformational transition
Using the hemolysis of the erythrocytes assay, we have shown that the pH of the HA conformational transition of the adapted 4970MA/H1 virus increased by 0.5 units in comparison to that of the original 4970/H1 virus (5.5 ±0.2 versus 5.0 ± 0.2) that is typical for viruses with increased virulence for birds and mice [25]. For the virus 4182C-MA/H5, no change of the pH of the HA conformational transition (5.5 ± 0.2) was observed compared with the parent virus (Table 1).
Thermal stability of HA
The analysis of HA thermal stability revealed a difference in the temperature of HA inactivation between viruses of different subtypes: 57.5°C for the influenza viruses of the H1 subtype and 60.1°C for the viruses of the H5 subtype (Fig. 1). It is noteworthy that the differences in the thermal stability between the adapted variants and the corresponding prototype viruses were not statistically significant (Fig. 1).
Viral binding to sialoglycopolymers
The receptor specificity of viruses was assessed by the degree of interaction with the synthetic analogs of cellular sialylated carbohydrate chains – oligosaccharides terminated by Neu5Acα2-3Gal units (3`SL, 3`SLN, and 6-Su-3`SLN), typical for the avian influenza viruses – avian-like receptors, or Neu5Acα2-6Gal (6`SL, 6`SLN and 6-Su-6`SLN), known as human-like receptors. All of the investigated viruses showed similar high affinity for the analogs of avian-like receptors: 3`SL, 3`SLN, and 6-Su-3`SLN (Table 2). None of the original or mouse adapted viruses of the H5 and H1 subtypes interacted with the human-like receptor analogs such as 6’SL, 6’SLN, or 6-Su-6’SLN. Therefore, both avirulent for mice original 4970/H1 and 4182C/H5 influenza viruses as well as adapted virulent strains 4970MA/H1 and 4182C-MA/H5 showed an affinity only for avian-like receptor analogs – carbohydrate chains terminated by Neu5Acα2-3Gal-units (Table 2).
| Viruses | Sialoglycoconjugates ([Neu5Ac], μM) a | |||||
|---|---|---|---|---|---|---|
| Analogs of avian receptors | Analogs of human receptors | |||||
| 3’SL | 3’SLN | 6-Su-3’SLN | 6’SL | 6’SLN | 6-Su-6’SLN | |
| 4970/H1 | 0.25 | 0.125 | 0.25 | > 4 | > 4 | > 4 |
| 4970MA/H1 | 0.25 | 0.25 | 0.25 | > 4 | > 4 | > 4 |
| 4182C/H5 | 0.25 | 0.063 | 0.5 | > 4 | > 4 | > 4 |
| 4182C–MA/H5 | 0.25 | 0.125 | 0.25 | > 4 | > 4 | > 4 |
Concentration of sialoglycoconjugates calculated as the concentration of sialic acid (N-acetylneuraminic acid). A lower value corresponds to a higher affinity of the virus for sialoside. The data from one of three typical experiments are presented.
Genome sequencing of the original and adapted virus variants
In order to find out the causes of the phenotypic changes that occurred during the adaptation of mallards’ viruses to mice, we sequenced the complete genome of the original and adapted virus variants. The variant 4182C/H5 differed from the wild-type virus 4182/H5 only by two amino acid substitutions located in HA: Ala188/200Glu and Asp264/277Asn (H3 numbering / H5 numbering). These substitutions remained unchanged in the mouse adapted variant 4182C-MA/H5. In addition, in the adapted variant 4182C-MA/H5, mutations were found in six viral proteins: two in the surface glycoproteins – in HA (HA2 subunit, Lys52Arg) and in neuraminidase (NA, Ser59Ile), one in non-structural PB1-F2 (Ser82Pro) protein, as well as three in the internal proteins, PB2 (Glu627Lys), PB1 (Val113Ala), and nucleoprotein (NP, Arg65Lys).
During adaptation of the 4970/H1 influenza virus to mice, amino acid substitutions occurred in only two proteins: HA1 (Asp35/52Asn) (numbering according to H3/ H1; Table 3) and PB2 (Glu627Lys). Therefore, the change of the pH value of the HA conformational transition is caused by the substitution Asp35/52Asn in the HA1 subunit. In order to confirm this, we performed molecular dynamics simulations of HA conformational change.
| Virus subtype, Protein | Amino acid residue | Localization (function) | References | |
|---|---|---|---|---|
| Subtype H1N1 | 4970 а | 4970MA а | ||
| PB2 | 627Glu | 627Lys | Domain 627 (contact with host cellular factors; host specificity; determinant of pathogenicity) | [26] [27] [28] |
| HA1 | 35Asp b | 35Asn b | HA stem region (HA conformational change, virus penetration into the cell) | [29] |
| Subtype H5N3 | 4182C а | 4182C–MA а | ||
| PB2 | 627Glu | 627Lys | See above | [26–28] |
| PB1 | 113Val | 113Ala | Outside the active site (transcription and replication, the exact role of the 15-187 region is unknown) | [30, 31] |
| PB1-F2 | 82Ser | 82Pro | Non-structural protein (induces apoptosis depending on the cell type, modulates the activity of PB1 polymerase; strain specific, host specific, and cell specific) | [32] |
| НА2 | 51Lys | 51Arg | HA stem region (highly conserved region, HA conformational change, virus penetration into the cell) | [25, 33, 34] |
| NP | 65Arg | 65Lys | 1–77 – RNA binding region; 1–161 – PB2 interaction domain | [35] [36] |
| NA | 59Ser | 59Ile | Stem region | [37] |
The designations of viruses are given according to Table 1.
HA numbering by subtype H3: position 35 corresponds to position 52 in immature HA protein of H1 subtype.
Molecular dynamics simulations of HA conformational change
The conformational transition of HA is caused by changes in the spatial arrangement of electric charges depending on the pH of the medium [25, 34, 38]. Acidification of the medium leads to the protonation of the basic amino acid residues (Lys, Arg, His) and the neutralization of the negative charge of acidic amino acid residues (Asp, Glu) resulting in the change of protein conformation.
The mathematical assessment of the influence of the Asp35/52Asn mutation in the HA1 chain of the 4970MA/H1 virus on the pH of the HA conformational transition was performed using a three-dimensional HA model of the mallard’s influenza virus A/WDK/JX/12416/2005 (H1N1) (PDB ID: 3HTO), which is the closest in the primary structure to the HA of the studied virus. The HA molecules of these viruses differ in 12 amino acid residues that are located outside of the N-terminal region of the HA1 chain (positions 1-52/1-59). The Asp35/52Asn mutation was introduced into the model molecule (3HTO/52N) by replacing the oxygen atom with a nitrogen with subsequent correction of the C – N bond length. The model (PDB, 3HTO) lacks 62 amino acids at the C-end of the HA2 subunit corresponding to the transmembrane domain. It can be assumed that this fragment does not affect the protein’s ability to change the conformation at acidic pH, since it is inaccessible to the solvent and not prone to protonation.
The crystal structure (PDB, 3HTO) contains the structural model of one HA monomer consisting of the HA1 and HA2 subunits. The trimer model was obtained mathematically by the addition of two more structural copies of the monomer rotated by + 120° and –120° with respect to the trimer axis and, therefore, does not fully correspond to the structure of the trimer in solution. The models of the 3HTO trimer and mutant 3HTO/52Asn trimer were approximated to their most probable conformations using molecular dynamics simulations methods [22] by the minimization of their conformational energy in aqueous NaCl (0.15 M) at pH 5.0. Therefore, the conformations of HA models were optimized for the conditions close to physiological.
The charges of HA trimers at a different pH were calculated using the PROPKA program. The calculations (Table 4) showed that as a result of the Asp52Asn substitution, the positive charge of HA trimer at pH 5.0 became significantly higher than that of the original protein.
| pH | Molecule charge, e a | |
|---|---|---|
| HA/52Asp | HA/52Asn | |
| 0.00 | 140.69 | 179.67 |
| 1.00 | 138.63 | 177.75 |
| 2.00 | 128.96 | 167.82 |
| 3.00 | 99.74 | 134.90 |
| 4.00 | 55.98 | 87.96 |
| 5.00 | 24.51 | 49.93 |
| 6.00 | 13.32 | 27.80 |
| 7.00 | 9.33 | 14.87 |
| 8.00 | 3.96 | 7.02 |
| 9.00 | –6.12 | –3.05 |
| 10.00 | –36.84 | –33.86 |
| 11.00 | –86.00 | –83.64 |
| 12.00 | –128.94 | –126.40 |
| 13.00 | –159.93 | –157.64 |
| 14.00 | –179.60 | –177.16 |
e – elementary charge corresponding to the absolute value of the proton/electron charge
It can be assumed that the increased protonation of amino acid residues upon acidification of the medium induces the electrostatic repulsion and subsequent conformational transition of HA. According to the calculations, the protonation of the mutant HA/52Asn to a charge comparable to that or the original HA/52Asp occurs at a higher (about 6.0) pH (Table 4). That can explain the increase of the pH of HA conformational transition of the mouse adapted variant 4970MA/H1 to 5.5 from 5.0 known for the parental virus 4970/H1 in the hemolysis assay. Therefore, the Asp52Asn mutation in the HA of the 4970MA/H1 virus led to the pH shift of HA conformational transition by 0.5 units.
DISCUSSION
The two avian influenza viruses 4970/H1 and 4182C/H5 isolated from mallards, which are not pathogenic for mice, became pathogenic after 7–10 serial passages in the lungs of the mice. As a result of the adaptation of the virus 4970/H1, mutations emerged in the HA (HA1 subunit, Asp35/52Asn) and PB2 (Glu627Lys) proteins. In the 4182C/H5 virus mutations were detected in six viral proteins: HA (HA2 subunit, Lys52Arg), NA (Ser59Ile), PB2 (Glu627Lys), PB1 (Val113Ala), PB1-F2 (Ser82Pro), and NP (Arg65Lys). Any mutation affecting a functionally important site of the proteins’ interaction with each other, with viral RNA, or with the host cell components can lead to the change in virulence.
Influenza virus surface glycoprotein HA is responsible for the binding of the viral particle to the cellular receptors and for the penetration of the viral genome into the cell cytoplasm. The penetration of the virus into the cell begins in the endosome upon acidification of the surrounding media to the pH that triggers an irreversible conformational change (pH of transition) of HA, that was previously cleaved by the cellular protease into the HA1 and HA2 subunits. That in turn leads to the fusion of the viral and cellular membranes followed by the destruction of the endosome and release of the viral genome into the cytoplasm. The pH of the HA transition varies depending on the virus host and subtype [25]. The main determinant of the pathogenicity of the H5 influenza viruses is the composition of the HA cleavage site (activation site), located at the border of the HA1 and HA2 subunits [33]. No changes in the HA cleavage site were revealed upon the adaptation of 4970/H1 and 4182/H5 viruses to mice. Both adapted variants retained the avian-like receptor specificity of original strains (Table 1). The virus 4182C-MA/H5 acquired the mutation Lys51Arg in the HA2 subunit of HA, which is located near the fusion peptide adjacent to the N-terminus of the proteolytic cleavage site. The HA amino acid Lys51 is invariant in viruses of different subtypes. Amino acid residues near the fusion peptide may be involved in the initiation of the HA conformational transition [25, 33, 34]. The substitution Lys51Arg found in the 4182C-MA/H5 virus did not affect the pH of the HA conformational transition, probably because the positively charged Lys was replaced by the positively charged Arg.
The virus 4970MA/H1 acquired the mutation Asp35/52Asn in the HA1 subunit, which is located in the upper part of the HA stem region at the border with the HA globular domain. The Asp35/52Asn replacement led to a change in the amino acid residue charge (eliminating the negative charge) in this position. Since this is the only substitution in HA, it can be assumed that the change in the pH of the HA conformational transition of the 4970MA/H1 virus compared to the parental virus is due to this very mutation. The polymorphism in this position is observed according to the IRD (Influenza Research Database, http://www.fludb.org). The Asp residue prevails (95.3%) in the avian influenza viruses, while the Asn is present in 3.5% and some isolates contain Gly or Ser. In human and mammalian influenza viruses, Asp also predominates (88.5– 99.5%), while Asn accounts for 0.4–11.3%, and Glu, Gly, His, Ser, Thr, Tyr, or Val are present very rarely. It should be noted that the proportion of viruses with Asn35 in the pool of human influenza viruses decreased to 0.3% in the period from 2009 to the present. It seems unlikely that the amino acid in position 35 is host specific.
The second surface glycoprotein of influenza virus is NA, the enzyme, which cleaves neuraminic acid residues from carbohydrate chains on the cell surface. It is believed that, by destroying the receptors involved in virus binding on the cell surface, NA facilitates the budding of virions. Virus 4182C-MA/H5 acquired the mutation Ser59Ile in the position located in the stem region of the NA, connecting the transmembrane domain with the head of NA where the catalytic site of the enzyme is located. The 3D structure of the transmembrane and stem domains of NA has not yet been determined [37]. The deletions from 15 to 22 aa occur in this very region of the NA of N1 subtype of highly pathogenic H5N1 influenza viruses that contribute to their increased virulence [39]. Ser predominates at position 59 (94.2%), while Asn (1.9%), Gly, and Thr (0.8% and 0.4%, respectively) are found to be extremely rare. The Ser59Ile mutation, identified in the NA of 4182C-MA/H5 variant, is unique.
The NP protein covers all 8 RNA segments of the viral genome, forming ribonucleoproteins (RNP), which also include the polymerase complex proteins PB2, PB1, and PA, which are necessary for viral RNA replication. The Arg65Lys mutation found in the NP of the 4182C-MA/ H5 virus is located at the N-terminal end of the molecule, between the two domains that form the groove for RNA binding. The Arg65 is one of the positively charged residues that line up on the surface of the groove. These residues are highly conserved in influenza viruses of A, B, and C types [35]. As it was previously shown by Li et al. [36], the replacement of the positively charged amino acid residue by a neutral one Arg65Ala in the NP molecule did not significantly affect the enzymatic activity of the polymerase complex. The substitution Arg65Lys (both positively charged amino acids) found in 4182C-MA/H5 most likely does not affect the NP-RNA interaction. The IRD database screening for the amino acid residues at position 65 of the NP protein showed that, in all avian influenza viruses, with rare exceptions, this position is occupied by Arg. In human influenza viruses of H1 and H2 subtypes as well as in mammalian viruses of different subtypes, Arg is also present at this position of the NP protein. It is noteworthy, that the human influenza viruses of the H3N2 subtype, which appeared in 1968, had only NP with Arg65 until 1992. The replacement Arg65Lys appeared in 1993 and began to prevail by 1997, then Arg was completely replaced by Lys in this position. About 20,000 influenza virus isolates contain Lys65, including human influenza viruses exclusively of the H3N2 subtype, several H3N2 pigs’ and dogs’ isolates, a small group of H1N1 and H1N2 viruses isolated from humans and dogs as well as 5 H7N7 strains isolated from horses in 1956-1966. The presence of the Arg65Lys mutation in human H3N2 viruses may be associated with their adaptation to the host or with a certain combination of proteins in viruses of this subtype. Therefore, the Arg65Lys mutation in the NP protein of the influenza virus can be conditionally considered as adaptive to mammals.
The PB1, PB2, and PA proteins form a polymerase complex of the influenza virus, which is responsible for the transcription and replication of viral RNA in the nucleus of the host cell [26, 30]. The PB1 protein contains highly conserved motifs and forms the catalytic center for RNA polymerization. The role of the homologous substitution Val113Ala in PB1 of the 4182C-MA/H5 variant is not well understood, while the amino acid at position 113 is not involved in the conservative motifs. On the threedimensional structure (PDB: 4WSB), it is located in the middle of the heterotrimer, in a small cavity, screened by the protruding parts of the polymerase complex. According to the IRD data, polymorphism is observed in this position with a predominance of three amino acid residues, including Val, Ile, Ala, and extremely rare Phe, Thr, Leu, Ser, and Met. The ratio of dominant amino acids (Val, Ile, Ala) in influenza viruses isolated from birds, humans, and pigs over the period from 1918 to 2021 differs significantly depending on the host species. In avian viruses, Val prevails (72.4%), while Ile (27.2%) and Ala (0.1%) are less common. In human viruses, two amino acids are present equally, of which Ala (50.5%) slightly prevails over Val (47.6%); in pig viruses, Ile prevails (65.9%) over Val (32.3%). Viruses with Ala113 account for only 0.4%. It is possible that the Val113Ala mutation in the 4182C-MA/H5 virus adapted to mice is associated with the adaptation to the cellular factors of the new host, which is necessary at certain stages of the virus life cycle.
The non-structural protein PB1-F2 is encoded by the same segment as the PB1 protein, as a result of a reading frame shift by +1 nucleotide. PB1-F2 has pleiotropic effects: it can (1) induce apoptosis depending on the cell type, (2) stimulate inflammation, and (3) affect the activity of viral polymerase by interacting with the PB1 subunit. Some of the effects of PB1-F2 are strain, host and cell type specific [32]. However, no clear understanding of the role of PB1-F2 protein for virus pathogenicity exists so far. In this regard, the role of the Ser82Pro mutation in PB1-F2 of the 4182C-MA/H5 virus is unclear.
The PB2 protein is also a part of the viral polymerase heterotrimer (PB2, PB1, and PA). The amino acid residue at position 627 depends on the host. Avian influenza viruses have Glu at this position, while viruses of humans and other mammals have Lys [27]. The mutation Glu-627Lys was found by Shi et al. [40] during the adaptation of avian viruses to mice and ferrets. The Lys residue at position 627 was also found in highly pathogenic H5N1 and H7N9 avian influenza viruses isolated from humans [40, 41]. The replacement Glu627Lys is host adaptive. Position 627 is located in the C-terminal part of the PB2 protein – in the domain-627 (538–693 aa) and is in close contact with the neighboring nuclear localization signal domain (NLS-domain, 694-741 aa), which interacts with the host cell importin in order to transport viral RNP from the cytoplasm to the nucleus [26]. So far, there is no solid data on the host factors associated with domain-627, nor on the mechanism of their interaction. Presumably, these could be the importin family proteins [26], transcription regulator ANP32A [42, 43], or others [44].
Previously, Yamayoshi et al. [28] found that the replacement Glu627Lys in the PB2 protein of avian influenza viruses led to an increase in polymerase activity, replication efficiency, and an increase in virulence during the cultivation of viruses in mammalian cells as well as in experiments in mice. We have shown that the adaptation of the wild avian influenza viruses 4970/H1 and 4182/H5 to mice leads to the emergence of the Glu627Lys mutation in PB2 and to an increase in virulence. It is likely that this amino acid substitution plays the major role in the virulence changes of the influenza H1 and H5 viruses in our study. The contribution of mutations found in other proteins is most likely less significant to the increased virulence of the viruses.
CONCLUSION
Summing up, the influenza viruses of the H1 and H5 subtypes isolated from wild ducks were adapted to growth in the lungs of mice in 7–10 passages. The viruses initially apathogenic for mice changed their phenotype to pathogenic thereby inducing a lethal infection. The mouse-adapted 4970MA/H1 variant was found to be more virulent than 4182C-MA/H5, despite the fact that only two mutations occurred in two viral proteins (Glu627Lys in PB2 and Asp35/52Asn in HA1), which apparently led to an increase in virulence. In the influenza virus of the H5 subtype 4182C/H5 mutations were found in six proteins. Three mutations turned out to be unique: Lys51Arg in HA2, Ser59Ile in NA, and Ser82Pro in PB1-F2. Other mutations were host specific: Glu627Lys in PB2, Arg65Lys in NP, and Val113Ala in PB1. The Glu627Lys substitution in the PB2 protein, which has long been recognized as the determinant of pathogenicity, is common for the H1 and H5 mouse-adapted influenza viruses.