Receptor-Binding Profiles of H7 Subtype Influenza Viruses in Different Host Species

The major natural reservoirs of influenza A viruses are wild aquatic birds of the orders Anseriformes (ducks, geese, and swans) and Charadriiformes (gulls, terns, and waders), which harbor viruses of all 16 hemagglutinin (HA) and 9 neuraminidase (NA) antigenic subtypes currently known. Dabbling ducks (Anatinae), such as mallards and teals, carry almost all subtypes and show particularly high virus isolation rates, suggesting a unique role of these species in the persistence of influenza viruses in nature (for reviews, see references 14, 41, 46, and 64). Occasionally, influenza viruses of aquatic birds infect other birds and mammals, adapt to efficiently replicate and transmit in the new species, and continue to circulate, forming new stable host-specific virus lineages. All known lineages of influenza A viruses in land-based birds and mammals are believed to originate from the viruses of wild aquatic birds. Adaptation of influenza viruses to their sialic acid-containing receptors in a new host species is often required for successful interspecies transmission. Thus, avian influenza viruses bind to receptors containing terminal sialyl-galactosyl residues linked by an α2-3 linkage (Neu5Acα2-3Gal), whereas swine and human viruses bind to receptors which contain terminal α2-6-linked sialyl-galactosyl moieties (Neu5Acα2-6Gal) (10, 18, 27, 32, 35, 44, 47, 50), and a corresponding switch in the receptor specificity of the avian precursor is essential for the emergence of new stable virus lineages in humans and pigs (reviewed in references 1, 25, and 34).

Based on early data (10, 35), it was assumed that all avian influenza viruses have similar receptor-binding specificities and, therefore, that there is no significant receptor-mediated restriction on viral interspecies transmission in birds. The first evidence arguing against this concept was obtained in a study on H5N1 viruses from Hong Kong isolated in 1997, when virus isolates from poultry and humans were found to have a lower receptor binding affinity and a lower neuraminidase activity than closely related viruses of aquatic birds (33). In addition, analysis of the HA and NA sequences of H5 and H7 influenza viruses from various avian species revealed that poultry influenza viruses often differ from duck influenza viruses by additional N-linked glycans at the top of HA and by large deletions in the stalk of NA (2, 33). Furthermore, the H9N2 viruses widely circulating in poultry in Eurasia were found to differ significantly in their receptor specificity from H9N2 viruses of other evolutionary lineages. In particular, these Eurasian poultry influenza viruses displayed good binding to α2-6-linked sialic acids (37, 53). Studies on expression of sialic acids in intestinal and respiratory epithelia of different birds revealed substantial host-specific distinctions, among them, expression of both Neu5Acα2-3Gal- and Neu5Acα2-6Gal-terminated sialyloligosaccharides in chicken and quail, in contrast to ducks, which mainly contain Neu5Acα2-3Gal in their intestinal epithelium (16, 19, 23, 29, 49, 63). Taken together, these findings indicated that influenza viruses perpetuated in different birds can have different receptor specificities owing to distinctions in the sialic acid receptors in these avian species.

Subsequent detailed studies on viral receptor-binding specificity revealed that preferential binding to terminal Neu5Acα2-3Gal disaccharide is shared by the majority of avian viruses; however, viruses adapted to ducks, gulls, and land-based gallinaceous poultry differ in their ability to recognize the subterminal saccharides of Neu5Acα2-3Gal-terminated receptors (reviewed in references 34 and 43). Duck influenza viruses of various HA subtypes (H1 to H5, H9 to H11) preferentially bound to receptors with type 1 and type 3 oligosaccharide sequences, i.e., having the β1-3 linkage between the terminal Neu5Acα2-3Gal moiety and the penultimate sugar residue such as Neu5Acα2-3Galβ1-3GlcNAc (SLe^c) and Neu5Acα2-3Galβ1-3GalNAcα (STF). Sulfation at the 6-OH group of the subterminal GlcNAc had little effect on binding of duck influenza viruses, whereas fucosylation of this residue reduced the binding significantly (15, 17, 20, 21). In contrast to duck influenza viruses, the H4, H6, H13, and H14 subtype viruses isolated from gulls showed high-avidity binding to fucosylated sialyloligosaccharides Neu5Acα2-3Galβ1-4(Fucα1-3)GlcNAc (SLe^x) and Neu5Acα2-3Galβ1-3(Fucα1-4)GlcNAc (SLe^a) (17, 67). Influenza viruses with HA subtypes H5, H7, and H9 are commonly reported in terrestrial gallinaceous poultry (7, 9). The common properties shared by these poultry-adapted viruses are (i) preferential binding to receptors with type 2 sequences having a β1-4 bond between the Neu5Acα2-3Gal moiety and the next sugar residue, such as Neu5Acα2-3Galβ1-4GlcNAc (3′SLN) and (ii) particularly strong binding to the corresponding sulfated analogues Neu5Acα2-3Galβ1-4(6-O-HSO₃)GlcNAc (Su-3′SLN) and Neu5Acα2-3Galβ1-4(Fucα1-3)(6-O-HSO₃)GlcNAc (Su-SLe^x) (17, 20, 21).

In the case of H5 and H9 subtypes, virus strains with both duck-virus-like and poultry-virus-like receptor phenotypes were identified in corresponding avian species (17, 20, 21). In contrast to these relatively extensive analyses of H5 and H9 influenza viruses, previous studies on receptor specificity of H7 influenza viruses were limited to a few virus strains from North American and Eurasian poultry. All tested H7 influenza viruses displayed a poultry-virus-like receptor specificity, and some of them showed the ability to bind to human-type receptors (4, 20, 68). To fill the gap in current knowledge on the receptor specificity of H7 influenza viruses, especially those circulating in wild aquatic birds, here we analyzed a large collection of viruses from various host species.

We previously found that duck influenza viruses with various divergent HA subtypes (H1 to H5, H9 to H11) shared receptor binding specificity, suggesting that these viruses either evolved from a common duck ancestor or acquired duck-virus-like receptor specificity during their prolonged circulation in ducks (20, 34). The H5N1 and H9N2 stable virus lineages in gallinaceous poultry originated from viruses of aquatic birds (7, 9, 48) and have distinctive poultry-virus-like receptor specificity (17, 20). These findings indicate that the duck-to-poultry adaptation of H5 and H9 influenza viruses altered their receptor specificity.

The influenza viruses with H7 HA subtypes were isolated from a variety of aquatic and land-based birds. The viruses were associated with poultry outbreaks of avian influenza (“fowl plaque”) of low and high pathogenicity in both hemispheres and occasionally infected mammals, including humans (reviewed in references 5, 9, and 65). Interestingly, little genetic diversity has been observed between H7 influenza viruses isolated from wild aquatic and land-based domestic birds in the same geographical area (2, 8, 42), suggesting their relatively frequent and perhaps unrestricted interspecies transmission. Given the importance of H7 influenza viruses with respect to animal and human health and their apparent expanded host range, we characterized receptor specificity of H7 influenza viruses isolated from different host species and representing almost all currently known evolutionary lineages (Fig. 1 and Table 1).

The equine H7 influenza viruses differed from all other H7 influenza viruses by their clear preference for the N-glycolyl form of sialic acid, which is the major sialic acid species expressed in the horse tracheal epithelium (59). The ability of H3 and H7 influenza viruses to bind both Neu5Ac and Neu5Gc was noticed previously (17, 18, 59). The receptor-binding assay used in this study shows that H7 equine influenza viruses bind Neu5Gc significantly better than Neu5Ac, whereas H3 equine influenza viruses display the opposite binding preference. Because Neu5Gc is absent in birds and humans (54), one can assume that equine H7 influenza viruses were strongly adapted to their host and that their receptor specificity could have restricted their transmission to other species.

All non-equine H7 influenza viruses displayed relatively homogeneous receptor-binding properties. The key feature of these viruses was their enhanced avidity for sulfated receptor sequences with type 2 cores (Su-3′SLN and Su-SLe^x), most likely owing to ionic interactions of the negatively charged sulfate with a conserved positively charged amino acid in position 193 (Lys or Arg). In addition, most H7 influenza viruses tolerated fucosylation of the third sugar in both type 1 and type 2 sequences (SLe^a and SLe^x). These features, individually or in combination, are known hallmarks of poultry-adapted viruses with H5 and H9 HAs that differentiate these viruses from viruses in ducks (17, 20, 21). An independent acquisition by H5, H7, and H9 poultry influenza viruses of a strong binding to Su-SLe^x suggests that this property provides the viruses with a selective replicative advantage in poultry. In mammals, SLe^x and its sulfated forms, among them Su-SLe^x, are well-known naturally occurring sequences recognized by E-, P-, and L-selectins (12, 39, 56). SLe^x and related molecules are constitutively expressed glycan sequences typical for endothelial cells and leukocytes, and their levels can increase manyfold during the inflammation process (12). Given that both selectins and their ligands are evolutionary conserved molecules, one can predict expression of Su-SLe^x on avian endothelial cells. Interestingly, three previous reports described a strong HA-dependent tropism of H5 and H7 poultry influenza viruses to endothelial cells of chicken embryo cells (13) and to cultured human pulmonary microvascular endothelial cells (45, 69). Thus, enhanced binding of poultry influenza viruses to Su-SLe^x and other sulfated and fucosylated sialyloligosaccharide sequences could have emerged as an adaptive change required for viral endotheliotropism in birds, a well-known feature of highly pathogenic poultry influenza viruses with H5 and H7 HA (28).

In is believed that influenza A viruses in land-based birds and mammals originate from the viruses of wild aquatic birds. Remarkably, we found that none of the H7 influenza viruses isolated from North American and Eurasian ducks and shorebirds tested in this study showed a binding phenotype typical of the duck influenza viruses with non-H7 HAs. Because all H7 influenza viruses, irrespective of the host of origin, share the unique structure of the 190 loop and distinctive poultry-virus-like binding phenotype, one can speculate that structural and functional features of the RBS of contemporary H7 influenza viruses were formed after the introduction of their common precursor from ducks to a new host, presumably, gallinaceous poultry. The original H7 duck influenza virus seems to have become extinct; alternatively, the H7 HA could have evolved from a non-H7 HA subtype after its duck-to-poultry transmission.

Efficient circulation of H7 influenza viruses in ducks despite their poultry-virus-like receptor specificity could be explained by a relatively good binding of these viruses to the “duck-type” receptor sequence SLe^c (see Fig. 2; see also Table S1 in the supplemental material). In contrast, the H9 poultry-adapted viruses have almost completely lost binding to this sequence (20). The ability of the progenitor H7 subtype virus to replicate in both ducks and poultry without major changes in the receptor specificity could have contributed to its fast expansion.

Although distinctive receptor-binding properties of the H7 HA permit virus replication in various avian species without significant changes in the RBS and receptor specificity, adaptive changes in the neuraminidase seem to be essential for the duck-to-poultry adaptation. Thus, deletions in the stalk of the NA that affect its catalytic activity are commonly observed in H5 and H7 poultry influenza viruses but not in their putative duck precursors (2, 8, 31, 33). Emergence of the NA stalk deletions as the first essential change in the duck influenza viruses during their experimental adaptation to poultry was confirmed in several independent studies (22, 26, 30, 40). In addition to deletions in the NA, HA mutations leading to emergence of new glycosylation sites near the RBS of H5 and H7 influenza viruses correlate with virus isolation from poultry. These mutations do not seem to emerge immediately after the virus transmission to poultry (2, 33). They rather correlate with the acquisition of the multibasic cleavage site by the virus and could thus be a consequence of the extended tissue tropism of the virus such as the ability to replicate in endothelial cells (13). We have tested only a few HPAI H7 influenza viruses, and their receptor-binding phenotypes were similar to phylogenetically close low-pathogenic strains (Table 1). Further studies are needed to understand the role of NA deletions and enhanced HA glycosylation in virus replication and pathogenicity in poultry.

Similarly to most avian viruses, many H7 influenza viruses do not bind Neu5Acα2-6Gal-terminated receptors. The potential structural reasons underlying the lack of binding of the human-type receptors by the typical H7 influenza virus A/Turkey/Italy/02 were previously discussed (52). Since 1994, the H7 influenza viruses circulating in North American live bird markets have acquired an eight-amino-acid deletion in the RBS (positions 221 to 228) (see Fig. 1; see also Fig. S1 in the supplemental material). The viruses of this lineage maintained preferential binding to Neu5Acα2-3Gal-terminated sequences and acquired weak-to-moderate binding to 2-6-linked sialic acids (4, 20, 68). Structural analysis revealed that Arg220 and Arg229 compensate for the deletion and interact with receptor analogues (68). Importantly, we found in this study that not only H7 poultry influenza viruses of a lineage with the HA deletion but H7 aquatic bird and poultry influenza viruses from many other evolutionary lineages displayed detectable binding to human-type receptor analogue 6′SLN (see Fig. 1B). We speculate that the peculiar structure of the 185 to 189 loop located close to the 221 to 228 loop reduces HA conflicts with the Neu5Acα2-6Gal-terminated receptor sequences and that various additional substitutions within and in the vicinity of the RBS allow binding to Neu5Acα2-6Gal. It is not clear whether this property provides the virus with some selective advantage or merely reflects occasional HA variations. Whatever is the case, the frequently emerging ability of the H7 influenza viruses to bind to 2-6-linked sialic acids can facilitate transmission of these viruses to humans.

This work was supported by the European Union FP6 project FLUPATH. Parts of the study were funded by the Von Behring-Roentgen-Stiftung (M.N.M.), the Deutsche Forschungsgemeinschaft SFB 593 (H.-D.K.), NIAID-NIH contracts HHSN266200700010C (R.A.M.F.) and HHSN266200700005C (R.G.W.), NIAID-NIH Division of Intramural Research (V.J.M.), research grant “Molecular and Cell Biology” from the Presidium of the Russian Academy of Sciences (N.V.B.), and research grant 11-04-00517 from the Russian Foundation for Basic Research (A.S.G.).

We thank John Skehel (MRC National Institute for Medical Research, London, United Kingdom) and Rupert Russell (University of St Andrews, St. Andrews, United Kingdom) for atomic coordinates of the H7 HA complex with LSTa and for critical comments on the manuscript.