Substitution Arg140Gly in HA/H7 reduced the virulence highly pathogenic avian influenza virus H7N1

The H7 subtype of avian influenza viruses (AIV) stands out among other AIV. The H7 viruses circulate in ducks, poultry, equine and have repeatedly caused outbreaks of disease in humans. In order to study the pathogenicity factors of H7N1 viruses, several variants were obtained, starting with laboratory strain, with a history of 12 passages through chicken embryos. This strain, A/chicken/Rostock/R0p/1934(H7N1) (R0p) had only 3 substitution in HA relatively A/Chicken/Rostock/45/34(H7N1), substitution Arg140Gly among them. 10 variants of this strain was obtained and studied to ascertain its biological property, genome stability and factors of pathogenicity. Strain R0p had decreased virulence for chicken, comparing with described in literature virulence of A/FPV Rostock/34 and A/chicken/Rostock/34 viruses. After 10 passages through the chicken lungs variant was obtained much more pathogenic than the starting R0p. The study of intermediate passages through the chicken lungs showed that the jump in pathogenicity had occurred sharply between the fifth and sixth passage. By cloning these variants, a pair of strains (R5p and R6p) were obtained, and the complete genomes of these strains were sequenced. Single amino acid substitution was revealed, namely reversion Gly140Arg in HA1. This amino acid is located at the head part of the hemagglutinin, adjacent to the receptor-binding site. In addition to the increased pathogenicity for chicken and mice, R6p differs from R5p in the pattern of foci in cell culture and an increased affinity for a negatively charged receptor analogue, while maintaining a pattern of receptor binding specificity and the pH optimum of the HA conformational change.


Introduction
The main hosts of influenza viruses (IV) H1 -H6, H8 -H12 and H14 are wild ducks, while separate evolutionary branches have formed in other hosts [1]. The H13 and H16 subtypes are primarily gull viruses. The H7 subtypes and the phylogenetically close H15 also stand out among avian influenza viruses (AIV). Although they actively circulate in both ducks and poultry, evolutionary trees show that chickens are the primary hosts, and ducks are secondary hosts of H7 viruses [2]. H7 viruses in ducks are generally completely apathogenic and lack a polybasic cleavage site. However, they easily introduced into the poultry population and rapidly evolved with increasing pathogenicity [3]. Numerous outbreaks of the H7N2 virus occurred in 1996 in Pennsylvania. Despite the absence of a polybasic cleavage site, the virus caused economic losses of the order of several million USD. Later, viruses of this evolutionary branch caused outbreaks in Miami (2001) non-permissive temperatures and are incorrectly trimerized [16,17]. The Ile512Leu mutation in PB2 also led to temperature sensitivity and attenuation [18].
The H7 viruses differ from other AIVs not only in their epidemiology, but also in the structure of the receptor-binding site (RBS). In HA H7, the conserved for all subtypes Pro185 is replaced by Ser. The residues at positions 185-189 of H7 HA have small side chains (usually Ser-Gly-Ser-Thr-Thr), while other avian viruses besides Pro185 have at least one large amino acid residue at these positions. As a rule, Lys or Arg are located in 193 positions of the RBS in H7 HA, which is also typical for chicken H5N1 viruses. The positively charged amino acid residue 193 provides binding of the negatively charged sulfo group of the Neu5Ac2-3Gal1-4-(6-O-Su)GlcNAc (Su3'SLN) and Neu5Ac2-3Gal1-4(Fuc1-3)(6-Su)GlcNAc (Su-SLex) receptors [2]. Another characteristic property of H7 HA is a pair of arginines: 140 and 141. The Arg141 is strictly conserved for H7 HA, while the Arg140 in rare cases is replaced by Lys or a small amino acid.
The study of the causal relationship between the structural features of the RBS H7 and the biology of these viruses is the goal of this work. The origin virus, a laboratory strain A/chicken/Rostock/R0p/1934(H7N1) (R0p) had only 3 substitution in HA relatively A/Chicken/Rostock/45/34(H7N1), namely Arg140Gly, Asn157Asp and Ser197Asn (numbering according to H3). The R0p was passed through the lungs of chickens, and passage variants were compared with parent virus. The mechanism of increasing pathogenicity was investigated.

Reagents
Fetuin and horseradish peroxidase were from Serva, Switzerland. Antibodies against mouse and chicken immunoglobulins conjugated with horseradish peroxidase were from (Sigma, USA). Sialylglycopolymers were from GlycoNZ (Auckland, New Zealand).

Viruses
The laboratory strain A/chicken/Rostock/R0p/1934(H7N1) (R0p), derived from A/Chicken/Rostock/45/34(H7N1), was obtained from the collection of the Chumakov Federal Scientific Center. Non-pathogenic mallard virus A/mallard Sweden 91/02 (H7N9) was kindly provided by Dr. R. Fouchier (Department of Virology, Erasmus Medical Center, the Netherlands). Passage variants of A/chicken/Rostock/R0p/1934 were obtained in this study. Ten-day-old embryonated chicken eggs (CE) were inoculated with 10 2 EID of viruses, incubated at 36°C, monitored and cooled immediately after death or after 60 hour of incubation. Infectious allantoic fluids (IAF) were harvested and tested by hemagglutination assay. The virus amount was expressed in hemaglutinating units (HAU). 50% infective dose (EID50) for each virus stock was determined by titration in CE.

Animals
Chickens and embryonated chicken eggs were purchased from State poultry farm "Ptichnoe" (Moscow, Russia). All studies with HPAIV viruses were conducted in a biosafety level 3 containment facility. BALB/c mice (weight in the range from 8 to 10 g) were purchased from "Lesnoye" farm, Moscow, Russia.

Ethics Statement
Studies involving animals were performed in accordance with the (European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes, Strasbourg, 18 March 1986). All appropriate measures were taken to ameliorate animal suffering. 94 chickens were used in the study; 39 chickens survived and were subsequently kept in the bird housing facility for repeated detection of antibody levels. 55 chickens were humanely euthanized after they showed signs of severe disease. The study design was approved by the Ethics Committee of the Chumakov Federal scientific center, Moscow, Russia (Approval #4 from 2 December 2014).

Sequencing
Viral RNA was isolated from the allantoic fluid of infected chicken embryos with a commercial QIAamp Viral RNA mini kit (Qiagen, # 52904). Full-length viral genome segments were obtained by revers transcription and PCR with specific terminal primers [19], MMLV and Taq-polymerase (Alpha-Ferment Ltd., Moscow, Russia). The amplified fragments were separated by electrophoresis in 1-1.3% agarose gel and subsequently extracted from the gel with the Diatom DNA Elution kit (Isogene Laboratory Ltd., Moscow, Russia, # D1031). Sequencing reactions were performed with terminal or internal primers [20] with the BrightDye ™ Terminator Cycle Sequencing Kit v3.1 (Nimagen, the Netherlands), followed by analysis on an ABI PRISM 3100-Avant Genetic Analyzer (Applied Biosystems, Foster City, USA). The Lasergene software package (DNASTAR Inc., Madison, WI, USA) was used for assembly and analysis of nucleotide sequences. The complete genomes were sequenced for the original strain and its variant obtained after 10 passages in the chicken lungs, as well as cloned variants of R5p and R6p. Their GenBank Accession Numbers are MT914267-MT914274, MT916934-MT916941, MT916987-MT916994, MT917013-MT917020.

Infection of chickens
The 7-day-old or 56-day-old chickens were infected intranasal with undiluted allantoic fluid containing the tested viruses. All birds were assessed daily for body weight, clinical signs of disease and mortality. On day 3 post-infection, lungs of one chick from each group were taken for virus determination.

Passaging the virus through chicken lungs
For the first passage, four 7-day-old chickens were infected intranasal with undiluted virus-containing IAF (100 μL per chicken). On days 3 after infection, one chick were humanely euthanized; lungs were excised, homogenized, and centrifuged. The rest chicks were monitored for signs of disease. The lung supernatant was used for infection of the new group of chicken. In order to study the passage variants of the virus, the CE was infected with lung homogenates and IAF was used to analyze the viral genome, to determine receptor specificity, virulence, and for other purposes.

Assessment of pathogenicity index
The intravenous pathogenicity index (CPI) was measured out according to World Organ-ization for Animal Health recommendation with minor modifications [21]. Six weeks old specific pathogen-free chickens (SPF) with no previous history of vaccination against influenza virus were used. Chickens were injected intravenously into a wing vein, using 0.2 mL of an inoculum containing 1:10 dilution (using sterile PBS) of IAF. Birds were examined twice daily for up to 10 days. At each observation, each bird was scored 0 if normal, 1 if sick, 2 if severely sick, 3 if dead. If birds were too sick to eat or drink, they were humanely sacrificed and scored as dead at the next observation. Dead birds were assigned a score of 3 daily, up to the tenth day of the experiment. The scores for all birds in the group were summed daily, and then summed up over all 10 days of observation. The result was divided by the product of the number of birds in the group by the number of days. The maximum possible index value calculated using this method is 3.0.

Determination of the pH optimum of the HA conformational change (hemolysis test)
Allantoic fluid clarified by low-speed centrifugation was diluted with PBS to 128 HAU. To 250 μL of the obtained viral sample, 50 μL of 2.5% chicken erythrocytes diluted in the same buffer were added and incubated, periodically shaking, at +4°С for 1 h. The erythrocytes with the virus adsorbed on them were centrifuged at 2800 rpm /min for 1 min at + 4 ° C, the supernatant was removed, and 250 μL of 0.1 M MES buffer with a pH in the range from 4.5 to 6.0 was added. After that, samples were incubated with shaking for 1 h at 37 ° C. Untreated erythrocytes without virus served as a negative control, and erythrocytes with the addition of 0.5% Tween-20 served as a positive control. After incubation, the samples were centrifuged for 1 min at 2800 rpm, 170 μL of the supernatant was transferred to a flat-bottom 96-well plate for measuring optical density at a wavelength of 415 nm (major maximum absorption of hemoglobin) using an iMark Microplate Reader (Bio-Rad, USA). Based on the measurement results, a graph was constructed on which the value of the pH-dependent conformational change of HA was determined.

Determination of virus affinity with the receptor analogue -fetuin
The viruses' affinity for the peroxidase-labelled fetuin was determined in a direct solid phase binding assay [22]. In brief, 100 μL of IAFs were added to each well of the fetuin-coated microplates. After incubation at 4°C overnight, the plates were washed with ice-cold washing buffer (0.02% Tween 80 in PBS). Serial twofold dilutions of peroxidaselabelled fetuin in the reaction buffer (RB; 0.02% Tween 80, 0.02% bovine serum albumin, 1 μM oseltamivir carboxylate in PBS) were added into the wells (50 μL/well), and the plates were incubated at 4°C for 1 h. After washing, the peroxidase activity in the wells was assayed with tetramethylbenzidine substrate solution. The absorbencies at 450 nm were determined, the data were converted to Scatchard plots (A450/C versus A450), where C is the fetuin concentration expressed in μM sialic acid.

Determination of receptor specificity by competitive inhibition
The association constants of viral complexes with non-labeled sialylglycopolymers (SGPs) were determined in a solid-phase fetuin binding inhibition assay [23]. The viruses were adsorbed in the well of fetuin-coated plates as described above. Serial twofold dilutions of SGPs in solution of peroxidase-labelled fetuin in RB were added into the wells (50 μL/well), and the plates were incubated at 4°C for 1 h. After washing, the peroxidase activity in the wells was assayed as described above. The absorbencies at 450 nm were measured, transferred to a PC and processed using Microsoft Excel software.

Infection of mice
Six week-old BALB/c mice were used. Six group of six mice were formed for every virus tested one group for one dose of virus. Groups of mice were anesthetized and inoculated intranasally with placebo (PBS) or diluted IAF with 10 1 -10 5 EID50 of viruses. Survival and body weight following infection were monitored daily. On days 15 post-infection, serum samples were taken from survivor mice for antibody titration.

Measurement of antibodies against influenza viruses in mouse sera
The levels of antibody were assessed by ELISA assay with anti-mouse IgG. Plates (Nunc, MaxiSorp) were covered with fetuin. 100 μL of IAFs were added to each well of the fetuin-coated microplates, kept overnight at 4°C, then washed and blocked with 0.2% BSA solution in PBS, 1 h. The blocking solution was removed, 100 μL of buffer (0.1% Tween-20, 0.2% BSA on PBS) was added to the wells, on which the sera were titrated, starting from a dilution of 1:20. Wells without virus were used as blank. Sera from uninfected mice served as negative controls. Incubation was performed for 4 h at 4°C. After washing, peroxidase-labeled antibodies against mouse immunoglobulins (Sigma, USA) were added. It was incubated for 2 h; after washing, a color reaction with tetramethylbenzidine substrate solution was carried out.

Foci of virus infection on MDCK culture.
The MDCK cells were grown in 96-well plates. The cultures were washed, and 200 μL of the DMEM (Gibco) medium with 0.1% BSA were added to the wells. 50 μL of IAFs were added to the outermost wells and titrated by transferring 50 μL of the solution. After 16 h, glutaraldehyde solution (20 μL) was added to the wells to a final concentration of 0.02%, incubated for 30 min; the medium was poured out, and the wells were washed. Solution of chicken antibodies to FPV/Rostock/34 in PBS supplemented with 0.1% Tween-20, 0.2% BSA was added, the plates were incubated for 2 h at 4°C and washed. Solution of anti-chicken antibodies conjugated with horseradish peroxidase was added, the plates were incubated for 1 h at 4°C and washed. Infected cells were visualized by incubation with 0.1 mL of substrate solution per well (0.05% 3-amino-9-ethylcarbazole, 0.01% H2O2 in 0.05 M sodium acetate buffer, pH 5.5) for 30 min.

Statistical analysis
Statistical significance was determined with GraphPad Prism (Graphpad Software, Inc.) using the Student's t-test. Statistical significance was defined as P < 0.05.

Molecular models
Atomic coordinates of H7 HA (1ti8) were obtained from the Protein Data Bank [24]. The molecular model was generated with DS ViewerPro 5.0 software (Accelrys Inc.).

Emergence of mutations during passaging of the H7N1 virus through the chicken lungs
In order to study the pathogenicity factors of H7N1 viruses, several variants were obtained, starting with laboratory strain, with a history of 12 passages through chicken embryos. This strain, A/chicken/Rostock/R0p/1934(H7N1) (R0p) had substitution Arg140Gly in HA relatively A/Chicken/Rostock/45/34(H7N1) (MT914270 vs CY077420). Ten passages of the strain R0p through chickens lungs were carried out. Sequencing of the entire genome of the original and final variants revealed 9 amino acid (a.a.) substitutions between them, namely Val109Phe in PB2, Gln621Lys in PB1, Thr32Ala and Leu586Phe in PA Gly140Arg in HA1 and Ala101Thr in HA2 (numbering by H3), Ser82Arg in M2, Arg118Lys and Met124Arg in NS1 (Table 1). No differences were found in proteins NA, NP, M1 and NS2. Among these mutations, the Gly140Arg substitution in HA1 is a reversal, since Arg is located at position 140 of HA1 in all derivate strains of the A/FPV/Rostock/34 (H7N1) according to the GenBank data (V01105, MT914270, GU052946, CY077420, M24457). In all likelihood, substitution Arg140Gly appeared in the chicken/Rostock/R0p/1934 virus during the passages through chicken embryos, since the Gly140 variant is less pathogenic for embryos.

Location of R140 in the HA molecule
Arg140 is adjacent to strictly conserved Cys139, which forms a disulfide bridge with strictly conserved Cys97, which, in turn, is adjacent to the key amino acid of receptor binding site Tyr98. Arg141 in HA1 is a conserved amino acid in viruses of some subtypes. Arg140 is located at the edge of the receptor-binding site, at the apex of the hemagglutinin. Arg140 is unique to the H7 subtype, it is very conserved. The positively charged atoms of the Arg140 amino groups are directly adjacent to the edge of the receptor binding site and can influence binding to the negatively charged sialic acid, the terminal group of the receptor.  [24]. The protein surface is colored blue. The key amino acids of receptor binding site are colored red. The Arg140 colored by element.

Increasing the pathogenicity of R0p variants during passage through the chicken lungs
The original laboratory strain R0p with the Gly140 in HA1 was low pathogenic for chickens. Chicken embryos infected with this virus died within 24-48 hours, nevertheless, infected chicks, even one weeks of age, died rarely (Table 2). During the passage of this virus through the chicken lungs until the fifth passage, no changes in pathogenicity occurred, but starting from the sixth passage the virus became very pathogenic for chickens.
None of the birds infected with variants obtained after the fifth passage survived ( Table  2).  Since a sharp jump in the pathogenicity occurs between the fifth and sixth passage, we carried out a complete sequencing of these variants (R5p and R6p). In all genes of R5p and R6p, only one amino acid difference was found: replacement Gly140Arg in HA1 (tabl.1)

Pathogenicity index of strains R5p and R6p
Two groups of three 6-week-old chickens were infected intravenous with 10 times diluted IAFs of the R5p and the R6p strains. One bird threated with the R5p strain died on the 2nd day, the second showed signs of illness until the 10th day, and the third bird was sick until the 4th day and recovered. All birds infected with the R6p strain died on the day 2 after infection. The pathogenicity index for these strains was 1.7 and 2.8, respectively. According to the OIE instructions, avian influenza viruses with a pathogenicity index of 1.2 or more (maximum value 3.0) are considered highly virulent. Thus, the R6p is a typical highly virulent virus, while the R5p retains its highly virulent status, although its pathogenicity was reduced in comparison with the R6p.

Pathogenicity of R5p and R6p for mice, compared with highly pathogenic virus A/chicken/Kurgan/5/2005 (H5N1) and apathogenic virus A/mallard/Sweden/91/02(H7N9).
Mice were infected intranasal with varying doses (10 1 -10 5 EID/mouse) of R5p and R6p strains, as well as apathogenic virus A/mallard/Sweden/91/02 (H7N9) and highly pathogenic A/chicken/Kurgan/5/2005 (H5N1), which were taken for comparison. The mice were monitored and weighed daily. We recorded three variants of response to virus infection in mice: 1) immune response without signs of disease 2) illness with subsequent recovery and 3) death. Mice, which were reduced in weight within 5 days, and survived, were classified as diseased. On the 15th day, antibodies against the infection viruses were measured in all surviving mice. In the H5N1 virus, the lethal dose differs little from the immunogenic dose; only occasionally surviving mice with antibodies were detected, and sick but survivors mice were almost never detected. All these doses are close to one infectious unit EID50 of the virus. The mallard/Sweden/91/02 (H7N9) virus caused an asymptomatic infection in mice when challenged at the maximum dose, and elicited a strong immune response. The pathogenicity of R5p and R6p variants for mice are between the H7N9 and the highly pathogenic H5N1viruses. Both strains R5p and R6p killed mice when infected with high doses of the virus. The lethal dose of the R6p strain was an order of magnitude lower than that of the R5p strain. When infected with sublethal doses of both strains, a reliable picture of the disease in mice was observed. Lower doses, up to nearly 10 EID50, did not cause disease, but induced an immune response.
The pathogenicity of the studied viruses correlates with the structure of the cleavage site of HA, although in highly pathogenic H7 and H5N1 viruses it may be due to other factors associated with other genes [25]. Substitution Gly140Arg undoubtedly increases the pathogenicity of the virus for mice, since this is the only difference between the R5p and R6p variants. Fig. 2 shows a weight dynamics and survival of mice infected with 10 4 EID50 the three above-mentioned H7 viruses. 1/6 and 4/6 mice infected with R5p and R6p, respectively, died, while mice infected with the A/mallard/Sweden/91/02 virus suffered an asymptomatic infection. Antibodies to H7 HA were found in all surviving mice. One of the factors influencing the pathogenicity of influenza viruses is the value of pH-dependent conformational change of HA. In viruses of wild ducks, the conformational change occurs at pH close to five. A shift in the pH transition towards less acidic pH is characteristic of highly virulent strains of poultry influenza viruses. During attenuation of the highly pathogenic virus A/chicken/Kurgan/5/2005 (H5N1) a sharp drop in the transition pH, due to mutations in the stem part of HA, was shown [26]. It has been shown that a high pH transition increases the tropism of viruses to human endothelial cells [27]. We compared the transition pH of strains R5p and R6p to test the hypothesis that substitution increases the pathogenicity of the virus by increasing the pH of transition. However, this hypothesis was not confirmed. As can be seen from Figure 3, the curves of pHdependent erythrocyte hemolysis for strains R5p and R6p practically coincide, that is, the Gly140Arg substitution did not affect the value of the pH-dependent conformational change of the virus. Obviously, this is due to the surface arrangement of Arg140.  Fetuin -fetal bovine serum albumin -is a glycoprotein that contains double and quadruple branched N-glycans ending in Siaα2-3Galβ1-4GlcNAc and Siaα2-6Galβ1-4Glc-NAc motifs [28]. This makes it a good model receptor for most influenza viruses. We compared the affinity of R5p and R6p for fetuin labeled with horseradish peroxidase. The analysis of Fet-HRP binding to virions adsorbed on 96 well panels was performed in Scatchard coordinates (Fig. 4). The affinity of fetuin for the R6p virus is two times higher than for the R5p virus (Dissociation constants are 0.4 μM and 0.8 μM, respectively). This is probably due to the strong negative charge of fetuin and a sharp increase in the positive charge in the R6p strain compared to R5p due to the replacement of Gly by Arg.

Comparison receptor specificity patterns of viruses R5p and R6p.
Receptor specificity of the viruses was characterized in fetuin-binding inhibition assay using synthetic sialylglycosaccharides, attached to the polymer [29]. The structures and designations of the oligosaccharide moieties are shown below.

Foci of the infected cells produced by the viruses in MDCK monolayer.
The assay used was based on the method of cultivation influenza viruses in MDCK cells in 96-well microplates and detection of foci of infected cells by immunoperoxidase staining [30]. Infected cells were treated with chicken antibodies against R0p virus followed to peroxidase labeled anti-chicken antibodies and stained with aminoethylcarbazole -hydrogen peroxide solution. Cells infected with the virus acquired a red color.
Monolayer of MDCK cells was infected with R5p and R6p variants, incubated 16 h and stained as described in material and method section. The R5p produces large indistinct foci, which bear gaps of noninfected cells inside the focus. The foci of R6p are small, round-shaped and reveal no gaps of noninfected cells between the stained cells surrounding the center of the focus. These features of the foci suggest that the R6p virus progeny spreads directly from cell to cell, while R5p virus progeny released into the culture medium and infect the distance cells. The different shape of foci produced by the R5p and R6p viruses can be explained by the higher affinity of R6p for cells than affinity of R5p.

Discussion
The viruses with H7 HA subtypes were isolated from a wide range of hosts such as wild ducks, poultry, horses, seal and humans. They caused disease outbreaks in poultry ("fowl plaque") in both hemispheres. Little genetic diversity has been observed between H7 viruses isolated from wild birds and poultry birds in the same time and region suggesting their unrestricted interspecies transmission.
Such a wide range of hosts may be because that modern H7 AIVs are descendants of viruses, which long ago had transmitted from ducks to chickens, adapted to a new host, and then returned to wild ducks again, which allows the virus to persist and spread. This is confirmed by the topology of the evolutionary trees of H7 viruses, at the base of which not duck, but chicken viruses are grouped [2].
The receptor specificity of H7 AIVs also has a dual character. They recognize the chicken receptors -Su3'SLN and Su-SLe x as well as the main duck receptor SLe c . Many H7 AIVs also show some affinity for the "human" receptor, 6'SLN. Naturally, this broad receptor specificity contributes to the ability to infect a wide range of hosts. The structure of the receptor-binding site of hemagglutinin H7 indicates the rearrangement that occurred during adaptation to the new receptor in new hosts. All H7 AIVs have Lys193 in HA, similar to HPAIV H5N1. Lys193 provides increased binding of viruses to sulfated receptors Su-3'SLN and Su-SLe x due to electrostatic interaction of the sulfate residue of the receptor with the positively charged lysine side chain [23]. The set of amino acids 185-189 located at the bottom of the RBS is unique [2]. The pair Arg140 and Arg141 is another feature of H7 HA. Arg141 in HA1 is the strictly conserved amino acid in viruses of H7, H15, H10, H14, H3 and H4 subtypes. It plays an important role in maintaining the structure of the RBS, since the Arg141 residue is inside the molecule and it's positively charged groups contact with negatively charged oxygen Asp77, Gly72, and Phe147 [24]. Arg140 is unique to the H7 subtype; it is located at the edge of the receptor-binding site, at the apex of the hemagglutinin. Arg140 is conserved in H7 HA, although in rare cases it replaced by Lys or a small amino acid.
Reversion of Gly140Arg dramatically increases the surface positive charge of virions and promotes the binding of the virus to the negatively charged surface and molecules. R6p strain differs from R5p by an increased pathogenicity for chickens and mice, increased affinity for a negatively charged receptor analogue, an increased affinity for MDCK cells. In the same time, a pattern of receptor specificity and the pH of the hemagglutinin transition of the R6p do not change. The increased pathogenicity of R6p for chickens and mice is probably due to an increased affinity for negatively charged host cells and receptors.