Structures and Receptor Binding of Hemagglutinins from Human-Infecting H7N9 Influenza Viruses

See allHide authors and affiliations

Science  11 Oct 2013:
Vol. 342, Issue 6155, pp. 243-247
DOI: 10.1126/science.1242917

Two Viruses to Bind

Structural studies of two different H7N9 influenza viruses isolated from humans—A/Shanghai/1/2013 and A/Anhui/1/2013—which have different amino acid sequences in the receptor binding site, provide data indicating that the virus is in transition with respect to host adaptation. The Shanghai virus was one of the first isolated in humans that binds avian receptor glycans with high affinity, but binds poorly to human receptors. However, the later Anhui isolates can bind both avian and human receptors at high affinity. Shi et al. (p. 243, published online 5 September) show that four hydrophobic mutations contribute to acquisition of affinity for the human receptor by the virus hemagglutinin (HA) and confirm this effect in binding studies with virus particles. Further comparison of a mutant H7N9 A/Anhui/1/2013 HA with the bird flu H5N1 virus revealed the significance of some of the naturally occurring changes observed in circulating H7N9 viruses, which helps to explain how these viruses have been able to cause many severe human infections in a short time.


An avian-origin human-infecting influenza (H7N9) virus was recently identified in China. We have evaluated the viral hemagglutinin (HA) receptor–binding properties of two human H7N9 isolates, A/Shanghai/1/2013 (SH-H7N9) (containing the avian-signature residue Gln226) and A/Anhui/1/2013 (AH-H7N9) (containing the mammalian-signature residue Leu226). We found that SH-H7N9 HA preferentially binds the avian receptor analog, whereas AH-H7N9 HA binds both avian and human receptor analogs. Furthermore, an AH-H7N9 mutant HA (Leu226 → Gln) was found to exhibit dual receptor-binding property, indicating that other amino acid substitutions contribute to the receptor-binding switch. The structures of SH-H7N9 HA, AH-H7N9 HA, and its mutant in complex with either avian or human receptor analogs show how AH-H7N9 can bind human receptors while still retaining the avian receptor–binding property.

In February 2013, a novel reassortant influenza A (H7N9) virus was identified in eastern China, which by 30 April had spread to more than 11 provinces and municipalities. This virus is a low-pathogenicity avian influenza (LPAI) virus in domestic poultry (14). Until now, only sporadic cases of severe human infection with an LPAI virus have been reported (5, 6). Understanding the underlying mechanism of the avian-human host “jump” is crucial for the development of effective preventive and therapeutic measures (712).

The viral surface glycoprotein hemagglutinin (HA) is responsible for host receptor binding and is the major determinant of the virus host “jump” (13). Recent work has shown that the Gln226 → Leu (Q226L; H3 numbering used throughout) substitution in avian H5N1 HA confers human receptor (α-2,6-linked galactose) binding and simultaneously reduces avian receptor (α-2,3-linked galactose) binding (1417). The observation of a receptor shift in influenza viruses creates concern that a pandemic in human beings might begin this way (1417). It is noteworthy that H7N9 HA has a naturally occurring Q226L substitution observed in most of the isolates [e.g., AH-H7N9, A/Anhui/1/2013 (H7N9)], with the exception of an earlier Shanghai isolate that retained the glutamine at position 226 [SH-H7N9, A/Shanghai/1/2013 (H7N9)] (8). These findings have led to the assumption that the AH-H7N9 lineage virus might have acquired high-affinity human receptor–binding properties.

To characterize the receptor-binding properties of AH-H7N9 and SH-H7N9 at the virus level, we rescued the viruses with reverse genetics technology (18); the rescued viruses were named rAH-H7N9 and rSH-H7N9. We analyzed their receptor-binding properties through solid-phase binding assays using the 2009 pandemic influenza virus isolate [CA04-H1N1, A/California/04/2009 (H1N1)] and avian H5N1 influenza virus isolate [AH05-H5N1, A/Anhui/1/2005 (H5N1)] as control viruses that have typical human or avian receptor specificity, respectively. rAH-H7N9 binds both the human and avian receptor, whereas rSH-H7N9 preferentially binds the avian receptor (Fig. 1, A and B). In contrast, CA04-H1N1 specifically binds the human receptor (Fig. 1C), and AH05-H5N1 specifically binds the avian receptor (Fig. 1D).

Fig. 1 Receptor-binding properties at virus level.

Binding of virus to α-2,3-linked (3′SLNLN) or α-2,6-linked (6′SLNLN) sialylglycan receptors was determined by solid-phase binding assays. (A) rAH-H7N9 (reverse genetics–rescued A/Anhui/1/2013) virus; (B) rSH-H7N9 (reverse genetics–rescued A/Shanghai/1/2013) virus; (C) CA04-H1N1 (A/California/04/2009) virus; (D) AH05-H5N1 (A/Anhui/1/2005) virus. Blue, binding to 3′SLNLN; red, binding to 6′SLNLN. rAH-H7N9 binds to both 3′SLNLN and 6′SLNLN, whereas rSH-H7N9 binds preferentially to 3′SLNLN. As a control, CA04-H1N1 specifically binds 6′SLNLN, and AH05-H5N1 specifically binds 3′SLNLN.

To further evaluate the binding affinities of AH-H7N9 and SH-H7N9 to canonical avian-like and human-like receptor analogs (18), we prepared soluble HA proteins for both viruses and showed by surface plasmon resonance (SPR) experiments that, similar to the rescued viruses, AH-H7N9 HA binds both avian- and human-like receptors, whereas SH-H7N9 HA binds only the avian-like receptor (Fig. 2, A to F). AH-H7N9 HA bound both avian and human receptors, with high affinities of 0.10 μM and 0.33 μM, respectively (Fig. 2, A to C). SH-H7N9 HA also showed a binding preference for the avian receptor (with a similarly high affinity of 0.16 μM), but by contrast had extremely weak or no binding to the human receptor (>1 mM, beyond the SPR measurement range) (Fig. 2, D to F). The binding kinetics of SH-H7N9 HA are similar to the previously reported H7N7 HA from a highly pathogenic avian influenza (HPAI) A/H7N7 virus isolate (A/Netherlands/219/2003) (19). In contrast to the Q226L substitution of H5N1, which has endowed this virus with the ability to bind to human receptors but reduced its affinity for avian receptors (1417), AH-H7N9 HA has maintained dual receptor-binding properties despite having the same substitution.

Fig. 2 Receptor-binding properties at protein level.

(A, B, D, E, G, and H) BIAcore plots showing binding of AH-H7N9 HA to 3′SLNLN (A) and 6′SLNLN (B), binding of SH-H7N9 HA to 3′SLNLN (D) and 6′SLNLN (E), and binding of AH-H7N9 mutant HA to 3′SLNLN (G) and 6′SLNLN (H). (C, F, and I) Response units were plotted against protein concentrations. Blue, binding to 3′SLNLN; red, binding to 6′SLNLN. The binding affinity (KD) values were calculated using a steady-state affinity model produced with BIAcore 3000 analysis software (BIAevaluation Version 4.1).

In total, there are eight amino acid substitutions between the SH-H7N9 and AH-H7N9 HAs, of which four residues—S138A (Ser138 → Ala), G186V (Gly186 → Val), T221P (Thr221 → Pro), and Q226L—locate in the receptor-binding site (RBS) (fig. S1). There are likely several distinct mechanisms by which a shift in host receptor-binding preference can take place in different HA subtypes (20, 21). This is illustrated by differences between avian H1 and avian H2 and H3 viruses, in which the H2 and H3 viruses require Q226L and G228S (Gly228 → Ser) substitutions, whereas the H1 virus HA requires E190D (Glu190 → Asp) and G225D (Gly225 → Asp) substitutions in the RBS, retaining the residue Gln226 (2126). Thus, to investigate whether the Q226L substitution is a key determinant for obtaining the human receptor binding in H7, we introduced one L226Q substitution into AH-H7N9 HA, which binds the avian receptor analog with an affinity of 0.2 μM, similar to that of wild-type AH-H7N9 HA. Surprisingly, it retained its ability to bind to the human receptor analog, albeit with a reduced affinity of 1.2 μM (Fig. 2, G to I). Therefore, the other three amino acid substitutions are sufficient for human receptor binding for H7, without the Q226L substitution.

Using x-ray crystallography, we solved the structures of SH-H7N9 HA and of AH-H7N9 HA and its mutant, all in their free form or in complex with the two sialo-pentasaccharides 3′SLNLN and 6′SLNLN. These sialo-pentasaccharides are analogs of the avian and human receptors, respectively, and contain the three terminal saccharides sialic acid (Sia), galactoside (Gal), and N-acetylglucosamine (GlcNAc) (27). Given the resolution of the structures (2.6 Å, 2.6 Å, 2.8 Å, 2.5 Å, 3.0 Å, and 3.1 Å, respectively) (tables S1 to S3), there is unambiguous electron density for the ligands in the six complexes (fig. S2).

The RBS is at the membrane-distal end of each monomer. Conventionally, the RBS of H7 is divided into two parts: (i) the base, consisting of the conserved residues Tyr98, Trp153, His183, and Tyr195; and (ii) the side, consisting of the secondary elements 130-loop, 190-helix, and 220-loop. The structure of AH-H7N9 HA in complex with the avian receptor analog 3′SLNLN revealed that the analog bound in a cis conformation (Fig. 3A), similar to that seen in two recently reported H5 mutants in complex with the avian receptor analogs (16, 17, 28). However, there were clear differences of the molecular interactions with respect to the H5 mutants. In addition to the seven conserved hydrogen bond interactions between Sia-1 and RBS residues, the adjacent glycan rings (Gal-2 and GlcNAc-3) formed substantive van der Waals interactions with the 220-loop in the AH-H7N9 HA–avian receptor complex (Fig. 3A), which has not been observed in other HA–avian receptor complexes. Crucially, the main-chain carbonyl oxygen of Gly225 formed one hydrogen bond with Gal-2, and the side chain of Gln222 formed a hydrogen bond with GlcNAc-3, which stabilized the conformations of both rings. Moreover, the residue Leu226 created a hydrophobic environment for the nonpolar portion of Gal-2. Interestingly, in the presence of Gln226, SH-H7N9 HA also bound avian receptor analog in a cis conformation, in a manner similar to that of avian receptor analog bound to AH-H7N9 HA (Fig. 3B). In contrast, although exhibiting the same Gln226, the AH-H7N9 mutant HA bound the avian receptor analog in a trans conformation (Fig. 3C), which is usually observed in other avian HA–avian receptor complexes. This indicates that the amino acid at position 226 is not a key determinant for the conformation of avian receptor binding in the H7 subtype. In fact, both cis and trans configurations are in low-energy conformations for α-2,3 linkages, and both have been observed in studies of sialosides in solution and bound to other proteins (2934). We have captured two conformations of avian receptor binding in H7 HAs with a Gln226 in this study.

Fig. 3 Structural comparative analyses of the interactions of AH-H7N9 HA, SH-H7N9 HA, and AH-H7N9 mutant HA with either avian or human receptor analogs.

The three secondary structural elements of the binding site (i.e., the 130-loop, 190-helix, and 220-loop; H3 numbering) are labeled in ribbon representation, together with the selected residues in stick representation. Hydrogen bonds are shown as dashed lines. Green, AH-H7N9 HA; orange, SH-H7N9 HA; light blue, AH-H7N9 mutant HA; yellow, glycans. (A to C) AH-H7N9 HA (A), SH-H7N9 HA (B), and AH-H7N9 mutant HA (C) with the avian receptor analogs bound. The receptor analogs bind in different conformations in AH-H7N9 HA (cis), SH-H7N9 HA (cis), and AH-H7N9 mutant HA (trans). (D to F) AH-H7N9 HA (D), SH-H7N9 HA (E), and AH-H7N9 mutant HA (F) with the human receptor analogs bound. For AH-H7N9 HA, the receptor analog binds in a cis conformation. Only the sialic acid moiety is observed in the SH-H7N9 HA–human receptor complex. For AH-H7N9 mutant HA, the receptor analog binds in a cis conformation.

The structure of AH-H7N9 HA in complex with the human receptor analog 6′SLNLN showed that the analog was bound in a cis conformation (Fig. 3D). The residue Leu226 formed van der Waals interactions with Cα atoms of Gal-2 around the glycosidic linkage, and the main-chain carbonyl oxygen of residue Gly225 formed one hydrogen bond with the 3-OH of Gal-2 (Fig. 3D). In the SH-H7N9 HA–human receptor complex, only the sialic acid moiety was observed and the remaining four glycan rings of the pentasaccharide were not visible (Fig. 3E), implying a weak interaction as observed in the SPR experiment. The AH-H7N9 mutant HA bound the human receptor in a cis conformation, and the Gln226 formed two hydrogen bonds with sialic acid and one hydrogen bond with Gal-2 (Fig. 3F). Although the Leu226 was substituted as Gln226 in the AH-H7N9 mutant HA, the other three residues (Ala138, Val186, and Pro221) were able to create a hydrophobic region (fig. S3) in the RBS, allowing the human receptor to bind in a cis conformation.

Structural comparison revealed that a similar cis conformation occurred when SH-H7N9 HA and AH-H7N9 HA bound to an avian receptor analog, where the glycan moieties sat lower on the 220-loop (by ~2 Å) in SH-H7N9 HA than in AH-H7N9 HA (Fig. 4A). We assume that the four hydrophilic residues (Ser138, Gly186, Thr221, and Gln226) of SH-H7N9 HA are more compatible for the hydrophilic glycosidic oxygen of the avian receptor analog than the four hydrophobic residues (Ala138, Val186, Pro221, and Leu226) of AH-H7N9 HA (fig. S3). The trans conformation of avian receptor analog binding was observed in the AH-H7N9 mutant HA complex (Fig. 4B), and structural comparison revealed a ~180° rotation around the glycosidic linkage between the AH-H7N9 mutant HA complex and the AH-H7N9 or SH-H7N9 HA complex (Fig. 4, B and C). The residues Gly216 and Gln213 further stabilized the conformations of Gal-2 and GlcNAc-3, and dragged the two glycan rings toward the 220-loop by ~4 Å or ~6 Å, respectively (Fig. 3, A and C, and Fig. 4, B and C).

Fig. 4 Structural comparison of AH-H7N9 HA, SH-H7N9 HA, and AH-H7N9 mutant HA complexes.

(A) Comparison of RBSs of the AH-H7N9 HA–avian receptor (green) and SH-H7N9 HA–avian receptor (orange) complexes. A similar cis conformation for glycan binding is observed between these two complexes, but the glycan receptor sits lower (by ~2 Å) in the SH-H7N9 complex. (B) Comparison of RBSs of the AH-H7N9 HA–avian receptor (green) and AH-H7N9 mutant HA–avian receptor (light blue) complexes. The glycan receptor sits lower (by ~4 Å) in the AH-H7N9 HA complex. (C) Comparison of RBSs of the SH-H7N9 HA–avian receptor (orange) and AH-H7N9 mutant HA–avian receptor (light blue) complexes. The glycan receptor sits lower (by ~6 Å) in the SH-H7N9 HA complex. (D) Comparison of RBSs of the AH-H7N9 HA–human receptor (green) and AH-H7N9 mutant HA–human receptor (light blue) complexes. There is a 70° rotation around the Gal-2 C6-C5 bond. (E) Comparison of RBSs of AH-H7N9 HA (green), SH-H7N9 HA (orange), and AH-H7N9 mutant HA (light blue). The RBS of AH-H7N9 HA is wider than those of SH-H7N9 HA and AH-H7N9 mutant HA. (F) Comparison of RBSs of AH-H7N9 HA (green) and 57H2N2 HA (yellow). AH-H7N9 HA has a similarly wide RBS relative to 57H2N2 HA (PDB code 2WR7), but the 150-loop of AH-H7N9 HA is much closer to the RBS (by >6 Å) than that of 57H2N2 HA.

By contrast, the four hydrophilic residues (Ser138, Gly186, Thr221, and Gln226) in SH-H7N9 HA are expected to create an unfavorable environment for the nonpolar portion of the human receptor analog around the glycosidic linkage, which results in no binding to the human receptor. However, we found that when AH-H7N9 HA bound to the human receptor analog, the hydrophobic residues (Ala138, Val186, Pro221, and Leu226) stabilized the cis conformation of the receptor analog by creating a favorable platform on which the Gal-2 can sit to form a tighter interaction between the receptor and the ligand. In the AH-H7N9 mutant HA complex, the Leu226 was substituted by Gln226, but the receptor analog still bound in a cis conformation, indicating that the other three hydrophobic residues maintained the hydrophobic interaction even in the absence of hydrophobic Leu226. Analysis of the human receptor analog showed that there is a 70° rotation around the Gal-2 C6-C5 bond between the AH-H7N9 HA and AH-H7N9 mutant HA complexes (Fig. 4D). Depending on the orientation of the Gal-2, the remaining three glycan rings of the pentasaccharide may be oriented in different directions (Fig. 4D). For example, the glycan receptor in the AH-H7N9 HA complex would extend from the RBS, whereas the glycan receptor in AH-H7N9 mutant HA complex would extend toward the space between the 190-helix and the 220-loop.

Previous studies have shown that the RBSs of the human and swine influenza virus HAs are larger than those of the avian influenza virus HAs (35). We compared the HAs of AH-H7N9, SH-H7N9, and the 1957 Singapore human H2N2 (57H2N2) to see whether AH-H7N9 HA had acquired similar characteristics. The distances between the 130-loop and the 220-loop of the RBS are larger (by ~1.5 Å) in the AH-H7N9 HA structure than in the SH-H7N9 HA structure (Fig. 4E). By contrast, the distances are comparable between the AH-H7N9 mutant HA and SH-H7N9 HA (Fig. 4E). It is noteworthy that one hydrogen bond was formed between the residues Gln226 and Ser138 in SH-H7N9 HA and that the 220-loop and 130-loop were connected tightly (fig. S3). In the AH-H7N9 mutant HA, as a result of the S138A substitution, the residue Gln226 was displaced by ~2 Å (fig. S3). The different residue interactions might be responsible for the different binding properties of AH-H7N9 mutant HA and SH-H7N9 HA. The distances between the 130-loop and the other two sides of the RBS (190-helix and 220-loop) are comparable in the AH-H7N9 HA and 57H2N2 HA structures (Fig. 4F). The 150-loop moves closer toward the 190-helix of RBS by >6 Å in the AH-H7N9 HA structure relative to the 57H2N2 HA structure (Fig. 4F). Previous modeling suggests that longer α-2,6 linkage receptor analogs may clash with the 150-loop (36), and such loop interference probably causes multiple conformations of the glycan rings. This might explain why we obtained a poor electron density map for the glycan rings GlcNAc-3, Gal-4, and GlcNAc-5 of the human receptor analog 6′SLNLN in AH-H7N9 HA–human receptor complex.

Previous studies have shown that HA receptor binding of appropriate affinity and specificity is a requirement for efficient virus transmission between individuals and between species (16, 17, 37). The loss of affinity for the avian receptor appears to be an important factor for efficient human-to-human transmission; however, to date, limited human-to-human transmission has been observed for H7N9, which might be a result of retention of high affinity for the avian receptor. Maintenance of avian receptor binding can trap the virus in the human upper airways, which contain mucin molecules rich in α-2,3-linked galactose (38); this in turn leads to a requirement of high-dose virus to reach the susceptible cells, making human-to-human transmission more difficult. Anhui Leu226-containing virus binds human receptors with a greater affinity, and it dominates most of the later isolates. The earlier Shanghai Gln226-containing virus became less prominent, implying that the H7N9 virus is evolving. Furthermore, our mutagenesis study shows that, in contrast to H5N1 HA, the Q226L substitution is not solely responsible for the avian-to-human receptor-binding switch for H7 HA. Previous studies have also shown that many North American and Eurasian H7 influenza viruses display weak but detectable binding to the human-type receptor, highlighting the potential of H7 influenza viruses for avian-to-human transmission (39). We believe that surveillance of H7N9 virus isolates for detection of the new amino acid substitutions is essential for the future implementation of control strategies.

Supplementary Materials

Materials and Methods

Figs. S1 to S3

Tables S1 to S3

References (4050)

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: Supported by the China Ministry of Science and Technology National 973 Project (grant 2011CB504703), the National Natural Science Foundation of China (NSFC, grant 81290342), and Chinese Academy of Sciences intramural special grant KSZD-EW-Z-002 for influenza virus research. G.F.G. is a leading principal investigator of the NSFC Innovative Research Group (grant 81021003). We thank E. Dong for NSFC Medical Science Department director’s special grant 81341002; staff at the Shanghai Synchrotron Radiation Facility (beamline 17U), especially J. He and S. Huang, for assistance; the Genewiz Corporation for rapid synthesis of HA genes; and the Consortium for Functional Glycomics for providing the biotinylated SA analogs. Coordinates and structure factors are deposited in the Protein Data Bank (PDB) with the following codes: 4KOL (AH-H7N9 HA), 4KOM (AH-H7N9 HA–3′SLNLN), 4KON (AH-H7N9 HA–6′SLNLN), 4LCX (SH-H7N9 HA), 4LKG (SH-H7N9 HA–3′SLNLN), 4LKH (SH-H7N9 HA–6′SLNLN), 4LKI (AH-H7N9 mutant HA), 4LKJ (AH-H7N9 mutant HA–3′SLNLN), and 4LKK (AH-H7N9 mutant HA–6′SLNLN).
View Abstract

Navigate This Article