Main

Although H5N1 viruses continue to cause outbreaks in poultry and there are cases of human infection in Indonesia, Vietnam, Egypt and elsewhere (http://www.who.int/influenza/human_animal_interface/H5N1_cumulative_table_archives/en/index.html), they have not acquired the ability to cause human-to-human transmission. Investment in H5N1 vaccines has therefore been questioned. However, because humans lack immunity to influenza viruses possessing an H5 HA, the emergence of a transmissible H5-HA-possessing virus would probably cause a pandemic. To prepare better for such a scenario, it is critical that we understand the molecular changes that may render H5-HA-possessing viruses transmissible in mammals. Such knowledge would allow us to monitor circulating or newly emerging variants for their pandemic potential, focus eradication efforts on viruses that already have acquired subsets of molecular changes critical for transmission in mammals, stockpile antiviral compounds in regions where such viruses circulate, and initiate vaccine generation and large-scale production before a pandemic. Therefore, we studied the molecular features that would render H5-HA-possessing viruses transmissible in mammals.

Previous studies suggested that HA has a major role in host-range restriction of influenza A viruses1,2,3. The HA of human isolates preferentially recognizes sialic acid linked to galactose by α2,6-linkages (Siaα2,6Gal), whereas the HA of avian isolates preferentially recognizes sialic acid linked to galactose by α2,3-linkages (Siaα2,3Gal)3. A small number of avian H5N1 viruses isolated from humans show limited binding to human-type receptors, a property conferred by several amino acid changes in HA4,5,6,7,8,9. None of the H5N1 viruses tested transmitted efficiently in a ferret model10,11,12,13, although, while our paper was under review, one study14 reported that a virus with a mutant H5 HA and a neuraminidase (NA) of a human virus in the H5N1 virus background caused respiratory droplet transmission in one of two contact ferrets.

To identify novel mutations in avian H5 HAs that confer human-type receptor-binding preference, we introduced random mutations into the globular head (amino acids 120–259 (H3 numbering), which includes the receptor-binding pocket) of A/Vietnam/1203/2004 (H5N1; VN1203) HA (Supplementary Fig. 1). Although this virus was isolated from a human, its HA retains avian-type receptor-binding properties6,15. We also replaced the multibasic HA cleavage sequence with a non-virulent-type cleavage sequence, allowing us to perform studies in biosafety level 2 containment (http://www.who.int/csr/resources/publications/influenza/influenzaRMD2003_5.pdf). The mutated polymerase chain reaction (PCR) products were cloned into RNA polymerase I plasmids16 containing the VN1203 HA complementary DNA, which resulted in Escherichia coli libraries representing the randomly generated HA variants. Sequence analysis of 48 randomly selected clones indicated an average of 1.0 amino acid changes per HA globular head (data not shown). To generate an H5N1 virus library, plasmids for the synthesis of the mutated HA gene and the unmodified NA gene of VN1203 were transfected into human embryonic kidney (293T) cells together with plasmids for the synthesis of the six remaining viral genes of A/Puerto Rico/8/34 (H1N1; PR8), a laboratory-adapted human influenza A virus.

Turkey red blood cells (TRBCs; which possess both Siaα2,6Gal and Siaα2,3Gal on their surface (data not shown)) were treated with Salmonella enterica serovar Typhimurium LT2 sialidase, which preferentially removes α2,3-linked sialic acid (that is, avian-type receptors), creating TRBCs that predominantly possess Siaα2,6Gal on the cell surface (Siaα2,6-TRBCs; Supplementary Fig. 2). The virus library was then adsorbed to Siaα2,6-TRBCs at 4 °C and extensively washed to remove nonspecifically or weakly bound viruses. Bound viruses were eluted by incubation at 37 °C for 30 min, and then diluted to approximately 0.5 viruses per well (on the basis of a pilot experiment that assessed the approximate number of eluted viruses). We screened one-third of the library (that is, 2.1 × 106 viruses) in three separate selection experiments (that is, 0.7 × 106 viruses per experiment) and isolated 370 viruses that bound to Siaα2,6-TRBCs (Supplementary Fig. 2). Individual viruses were then grown in Madin-Darby canine kidney (MDCK) cells modified to overexpress Siaα2,6Gal (AX4 cells17), and screened again for their ability to agglutinate Siaα2,6-TRBCs (Supplementary Fig. 2). The parental control virus (designated VN1203/PR8) with avian-type receptor-binding specificity agglutinated untreated TRBCs (which possess both human- and avian-type receptors on their surface), but not TRBCs possessing predominantly human-type receptors (Siaα2,6-TRBCs; Supplementary Table 1). By contrast, of the 370 viruses originally isolated, nine agglutinated Siaα2,6-TRBCs, albeit with different efficiencies (Supplementary Table 1). All nine viruses possessed mutations in the region targeted for random mutagenesis; one mutant also possessed an additional mutation (E119G) in an area that was not targeted for mutation. Most of the mutations clustered around the receptor-binding pocket (Fig. 1a). Several of the selected viruses possessed mutations known to increase binding to human-type receptors, including N186K (ref. 9), S227N (ref. 5) and Q226L (which confers human-type receptor binding together with G228S)15 (all shown in blue in Fig. 1a). The identification of known determinants of human-type receptor-binding specificity from a library of random mutants validates our approach. Notably, our screen also identified mutations not previously associated with receptor-binding specificity.

Figure 1: Localization of amino acid changes identified in this study on the three-dimensional structure of the monomer of VN1203 HA (Protein Data Bank accession 2FK0). 15
figure 1

a, Close-up view of the globular head of VN1203 HA. Mutations known to increase affinity to human-type receptors are shown in blue. Amino acid changes not previously known to affect receptor binding are shown in green. Additional mutations that occurred in the HA of H5 avian–human reassortant viruses during replication and/or transmission in ferrets are shown in red. b, The positions of four mutations in the HA of H5 transmissible reassortant mutant virus, HA(N158D/N224K/Q226L/T318I)/CA04, are highlighted in red. The fusion peptide of HA is shown in cyan. All mutations are shown with H3 numbering. Images were created with MacPymol (http://www.pymol.org/).

PowerPoint slide

Although viruses were diluted to 0.5 viruses per well for amplification in AX4 cells, we cannot exclude the possibility that some wells were infected with more than one virus, resulting in mixed populations. To confirm the significance of the identified mutations in HA for human-type receptor binding, the mutations were engineered into a VN1203/PR8 virus (possessing an avirulent HA cleavage site sequence, as described earlier). All nine mutants were generated; however, after two passages in MDCK cells, the S136N mutation reverted to the wild-type sequence. This mutant was excluded from further evaluation.

First, we confirmed the binding of the remaining eight variants to Siaα2,6-TRBCs (Supplementary Table 1). For comparison, we included a VN1203/PR8 virus with two changes in its HA (Q226L and G228S) previously shown to have increased binding to Siaα2,6Gal6,15. Indeed, compared to the wild-type VN1203/PR8 virus, the Q226L/G228S mutant displayed an increased ability to bind to human-type receptors. For the recreated variants, haemagglutination titres were higher and slightly different from the initial characterization, which we attribute to biological differences (the initial characterization was carried out with non-concentrated cell culture supernatant and potentially mixed virus populations, whereas the recreated viruses were concentrated and purified) and to experimental differences (that is, differences between the TRBC batches or the efficiency of α2,3-sialidase treatment, or both). Collectively, however, these experiments demonstrate that this random mutagenesis approach allows the identification of hitherto unrecognized amino acid substitutions that permit avian virus HAs to bind to human-type receptors.

To characterize further the receptor-binding properties of the selected variants, we used solid-phase binding assays in which sialylglycopolymers were absorbed to plates, which were then incubated with virus (Fig. 2a). A virus possessing the HA and NA genes of the seasonal human A/Kawasaki/173/2001 (H1N1; K173) virus and the remaining genes from PR8 (K173/PR8) served as a control virus with typical human-type receptor specificity. Indeed, K173/PR8 preferentially bound to Siaα2,6Gal. In contrast, VN1203/PR8 bound to only Siaα2,3Gal. As reported elsewhere6,15, the Q226L/G228S mutations led to increased binding to Siaα2,6Gal. Variants I202T/R220S, W153R/T160I, N169I/H184L/I217M and H130Q/K157E resembled VN1203/PR8 in their binding to glycans, despite the fact that these mutants weakly agglutinated Siaα2,6-TRBCs (see Supplementary Table 1). These viruses may have bound to glycans on TRBCs that were different from Siaα2,6Galβ1,4GlcNAc used in this study. However, variants N186K/M230I, S227N/G228A and Q226L/E231G showed an appreciable increase in binding to Siaα2,6Gal but also retained binding capacity for Siaα2,3Gal. Of all of the variants tested, only E119G/V152I/N224K/Q226L exhibited specificity for only Siaα2,6Gal. Thus, only one H5 HA variant with receptor-binding capability akin to that of seasonal influenza viruses was isolated from the library screen of 2.1 × 106 viruses. To identify the amino acid change(s) responsible for the conversion from Siaα2,3Gal to Siaα2,6Gal recognition in the E119G/V152I/N224K/Q226L virus HA, we tested the amino acid changes at positions 119, 152, 224 and 226 individually and in various combinations. Solid-phase binding assays demonstrated that the N224K/Q226L combination is critical for the shift from Siaα2,3Gal to Siaα2,6Gal recognition (Fig. 2b); Q226L in combination with V152I also conferred weak binding to α2,6-glycans.

Figure 2: Characterization of the receptor-binding properties of isolated viruses.
figure 2

a, Binding of VN1203 mutants to sialylglycopolymers in solid-phase binding assays. A human virus (K173/PR8), an avian virus (VN1203/PR8) and mutant VN1203/PR8 viruses were compared for their ability to bind to sialylglycopolymers containing either α2,3-linked (blue) or α2,6-linked (red) sialic acids. b, Identification of mutations that confer binding to human-type receptors. c, Binding of VN1203 mutant viruses to human respiratory tissues. K173/PR8, VN1203/PR8 and mutant VN1203/PR8 viruses were incubated with human tissue sections and then stained with either anti-K173 antiserum (green) or anti-VN1203 HA antibodies (green). All sections were subsequently incubated with labelled secondary antibodies and Hoechst dye (blue). d, Characterization of the receptor-binding properties of N158D/N224K/Q226L, N158D/N224K/Q226L/T318I and T318I viruses. The direct binding of virus to sialylglycopolymers containing either α2,3-linked (blue) or α2,6-linked (red) sialic acids was determined as described in panel a.

PowerPoint slide

To assess the effect of enhanced α2,6-glycan recognition on the attachment of viruses to human respiratory tracts, sections of tracheal and lung tissues were exposed to K173/PR8 (human-type receptor binder), VN1203/PR8 (avian-type receptor binder) and mutant VN1203/PR8 viruses (Fig. 2c). Because the N186K/M230I, S227N/G228A, Q226L/E231G, E119G/V152I/N224K/Q226L and N224K/Q226L mutants exhibited appreciable binding to Siaα2,6Gal (Fig. 2a, b), the attachment of these mutants was also tested. On tracheal sections, the K173/PR8 virus bound extensively to ciliated epithelial cells (Fig. 2c and Supplementary Fig. 3), whereas the VN1203/PR8 virus bound poorly. By contrast, on lung sections, both viruses bound extensively to the alveolar epithelial surface (both type I and II pneumocytes; Fig. 2c and Supplementary Fig. 4). The binding patterns of these viruses correlate with the distribution of Siaα2,3Gal (that is, avian-type receptors; present in lung epithelia) and Siaα2,6Gal (that is, human-type receptors; present in both trachea and lung epithelia) on the tissues, as observed with lectin staining18 (Supplementary Fig. 5). Like the human K173/PR8 virus, the E119G/V152I/N224K/Q226L and N224K/Q226L mutants exhibited strong binding to the ciliated epithelial cells of the trachea (Fig. 2c and Supplementary Fig. 3). By contrast, the N186K/M230I, S227N/G228A and Q226L/E231G mutants displayed little-to-no binding to tracheal epithelia (Fig. 2c), despite their binding to Siaα2,6Gal (Fig. 2a). A number of sialylated oligosaccharides with differing branching patterns and chain lengths are thought to be present on the cell surface19. We therefore speculate that the mutants can recognize a short glycan structure such as Siaα2,6Galβ1,4GlcNAc, but may not recognize longer, more complex glycan structures, which are possibly required for binding to human tracheal epithelium. On the other hand, all mutants bound to alveolar epithelial cells (both type I and II pneumocytes; Fig. 2c and Supplementary Fig. 4). When the tissue sections were pre-treated with Arthrobacter ureafaciens sialidase (which cleaves all non-reducing terminally branched and unbranched sialic acids), virus binding to the tissues was substantially reduced (Supplementary Fig. 6a–c), confirming the sialic acid binding specificity of the virus. These data indicate that alterations in the receptor specificity of the E119G/V152I/N224K/Q226L and N224K/Q226L mutants have profound effects on virus attachment to human respiratory epithelium.

In an avian H3 HA, the Q226L mutation changed the binding preference from avian- to human-type20. A previous study found that the Q226L mutation on an H5 HA does not confer efficient binding to α2,6-glycans in a glycan array15; however, when tested in combination with G228S, increased binding to human-type receptors, but not a complete switch from avian- to human-type receptor-binding specificity, was observed15. By contrast, here we found that Q226L in combination with N224K resulted in a switch from Siaα2,3Gal to Siaα2,6Gal binding in an H5 HA and allowed virus binding to human tracheal epithelia (Fig. 2c). The receptor-binding domain of HA is formed by the 190-helix at the top of HA, the 220-loop at the edge of the globular head, and the 130-loop at the other edge of the globular head (Fig. 1a). Crystal structure analysis revealed that the 220-loop of avian H5 HA is closer to the opposing 130-loop than in human H3 HA, indicating that a wider binding site for human H3 HA, compared to that of avian H5 HA, may be required to optimize contacts with the larger Siaα2,6-glycans21. N224 lies on the turn leading into the 220-loop, adjacent to position 226 (Fig. 1a). Replacement of N224 may alter the orientation of the 220-loop and thus optimize contacts between L226 and Siaα2,6Gal-containing receptors, thereby increasing the preference for α2,6 linkages.

Recent studies reported that 2009 pandemic H1N1 and H5N1 viruses show high genetic compatibility22,23. These two viruses have been isolated from pigs24,25,26,27,28, which have been considered as ‘mixing vessels’ for the reassortment of avian, swine and human strains. Thus, the coexistence of H5N1 and 2009 pandemic H1N1 viruses could provide an opportunity for the generation of transmissible H5 avian–human reassortants in mammals. Therefore, we generated reassortant viruses possessing the mutant VN1203 HAs generated above, and the seven remaining gene segments from a prototype 2009 pandemic H1N1 virus (A/California/04/2009, CA04). Experiments with viruses possessing the wild-type HA cleavage site were performed in enhanced biosafety level 3 (BSL3+) containment laboratories approved for such use by the Centers for Disease Control and Prevention (CDC) and the United States Department of Agriculture (USDA). Because efficient human-to-human transmission is a critical feature of pandemic influenza viruses, we examined the growth and transmissibility of reassortant viruses in ferrets, which are widely accepted as an animal model for influenza virus transmissibility and pathogenesis studies. Because the E119G/V152I/N224K/Q226L and N224K/Q226L variants bound extensively to human tracheal epithelia (Fig. 2c), we generated by reverse genetics (rg) three H5 reassortant viruses possessing the VN1203 HA or mutant HAs (all with the wild-type multibasic cleavage site) and the remaining genes from the CA04 virus. The VN1203 HA mutants tested included the one containing four mutations, E119G, V152I, N224K and Q226L (designated rg(E119G/V152I/N224K/Q226L)/CA04), and another containing two mutations, N224K and Q226L (designated rg(N224K/Q226L)/CA04).

To determine whether the introduced HA mutations affected the replication of the H5 reassortant viruses, six ferrets were inoculated intranasally with 106 plaque-forming units (p.f.u.) of virus. On day 3 after infection, a recombinant virus whose genes all came from CA04, rgCA04, replicated efficiently in the respiratory organs of infected animals, and was isolated from the colon, but not from any other organs tested (Fig. 3 and Supplementary Table 2). A virus possessing H5 VN1203 HA and the remaining genes from CA04 (designated rgVN1203/CA04) replicated to titres comparable to those of rgCA04 in nasal turbinates, but substantially less in the lungs. By contrast, the two H5 reassortant viruses with HA mutations (rg(E119G/V152I/N224K/Q226L)/CA04 and rg(N224K/Q226L)/CA04) were severely limited in their replicative ability in trachea. Although virus titres in nasal turbinates and lung were not statistically different between rg(N224K/Q226L)/CA04 and rgCA04, the virus titre in nasal turbinates was significantly lower in animals inoculated with rg(E119G/V152I/N224K/Q226L)/CA04 than in animals inoculated with rgCA04 (Dunnett’s test; P = 0.0002; Fig. 3). Notably, rgVN1203/CA04 (avian-type receptor binder) replicated efficiently in nasal turbinates of ferrets, which have a similar sialic acid receptor distribution pattern to that of the human respiratory tract29,30. The reason for this discrepancy is unclear; however, replication of avian H5N1 viruses in ferret nasal turbinates has been reported12,13.

Figure 3: Virus replication in respiratory organs.
figure 3

Ferrets were infected intranasally with 106 p.f.u. of virus. Three ferrets per group were killed on days 3 and 6 after infection for virus titration. Virus titres in nasal turbinates, trachea and lung were determined by use of a plaque assay on MDCK cells. Horizontal bars show the mean. Asterisks indicate virus titres significantly different from that of rgCA04 (Dunnett’s test; P < 0.05).

PowerPoint slide

Although virus titres in respiratory organs were generally lower on day 6 after infection than on day 3 after infection, rg(N224K/Q226L)/CA04 still showed high levels of replication at day 6 after infection; titres in nasal turbinates ranged from 104.6 to 108.1 p.f.u. g−1 (Fig. 3). Sequence analysis of viruses in nasal turbinates on day 6 after infection revealed that viruses in ferret 2 and ferret 3 possessed N158D and N158K mutations in their HA (in addition to the original two mutations), respectively, leading to the loss of the glycosylation site at position 158 (that is, 158N-S-T to 158D-S-T or 158K-S-T; Fig. 1a and Supplementary Table 3). In nasal turbinates on day 6 after infection, the titre of the virus with the N158D/N224K/Q226L mutations (108.1 p.f.u. g−1; see Fig. 3, ferret 2 of rg(N224K/Q226L)/CA04) was approximately four orders of magnitude higher than that of the original rg(N224K/Q226L)/CA04 (104.6 p.f.u. g−1; Fig. 3, ferret 1 of rg(N224K/Q226L)/CA04), whereas the virus with the N158K/N224K/Q226L mutations (105.6 p.f.u. g−1; Fig. 3, ferret 3 of rg(N224K/Q226L)/CA04) grew to one order of magnitude higher than the original mutant. These data indicate that the additional mutation N158D improved the replication of rg(N224K/Q226L)/CA04 in ferrets. To test the effect of this mutation on the replication of H5 reassortant viruses in ferrets, we examined the replicative ability of a virus with the triple N158D/N224K/Q226L HA substitutions in ferrets. This HA(N158D/N224K/Q226L)/CA04 virus replicated efficiently in infected animals, except in the trachea (Fig. 3 and Supplementary Table 2). On day 3 after infection, this virus was isolated from the brain of two of the three animals tested, although we did not observe neurological signs in these animals. These results indicate that the N158D mutation contributed to the efficient growth in the nasal turbinates of ferrets of an H5 reassortant virus with the N224K/Q226L mutations. Removal of the glycosylation site at position 158 has been reported to result in enhanced binding of H5N1 viruses to human-type receptors in combination with the Q226L/G228S mutations7. A previous study showed that H5N1 viruses lacking this glycosylation site transmit efficiently by direct contact among guinea-pigs31. By contrast, H5N1 viruses that acquire this glycosylation site lose the ability to transmit among guinea-pigs. Therefore, we speculated that the loss of the glycosylation site in HA(N158D/N224K/Q226L)/CA04 virus may affect its transmissibility in ferrets.

To assess the ability of H5 reassortant viruses with human-type receptor specificity to transmit between ferrets, we placed naive ferrets in wireframe cages next to ferrets inoculated with 106 p.f.u. of rgCA04, rgVN1203/CA04, rg(N224K/Q226L)/CA04, or HA(N158D/N224K/Q226L)/CA04 (Supplementary Fig. 7). Similar to previous experiments32, rgCA04 was efficiently transmitted via respiratory droplets to all three contact ferrets, as evidenced by the detection of virus in nasal washes and haemagglutination inhibition (HI) antibody in these animals (Table 1 and Fig. 4). By contrast, rgVN1203/CA04 and rg(N224K/Q226L)/CA04 were not transmitted; neither virus shedding nor seroconversion was detected in any contact animals, despite the binding of the latter to Siaα2,6Gal. This result was consistent with that of previous studies in which human-type receptor recognition was shown to be necessary but not sufficient for respiratory droplet transmission of an H5N1 virus in a ferret model12,14. In the HA(N158D/N224K/Q226L)/CA04-inoculated group, virus was recovered from two of the six contact ferrets (pairs 1 and 2) between days 5 and 7 after contact. Moreover, seroconversion was detected in five animals including those from which virus was recovered. No animals died in the course of these transmission experiments. This finding demonstrates the generation of an H5 HA that supports virus transmission by respiratory droplets among ferrets.

Table 1 Transmission in ferrets inoculated with H5 avian–human reassortant viruses
Figure 4: Respiratory droplet transmission of H5 avian–human reassortant viruses in ferrets.
figure 4

af, Groups of three, five, or six ferrets were inoculated intranasally with 106 p.f.u. of rgCA04 (a), rgVN1203/CA04 (b), rg(N224K/Q226L)/CA04 (c), HA(N158D/N224K/Q226L)/CA04 (d), HA(N158D/N224K/Q226L/T318I)/CA04 (e), or rgT318I/CA04 (f). One day after infection, three, five, or six naive ferrets were placed in adjacent cages. Nasal washes were collected every other day from both inoculated (left panel) and contact (right panel) animals for virus titration. Virus titres in organs were determined by using a plaque assay on MDCK cells. The lower limit of detection is indicated by the horizontal dashed line.

PowerPoint slide

To determine whether additional mutations occurred in the HA of HA(N158D/N224K/Q226L)/CA04 during transmission, viral RNA was analysed from nasal washes of inoculated and contact ferrets (Fig. 4 and Supplementary Table 4). On day 5 after infection, the A242S and T318I mutations in HA were present in five (pairs 1, 3, 4, 5 and 6) and one (pair 2) of the six inoculated animals, respectively. Viruses derived from the contact animals of pair 1 on day 7 after contact had two changes in HA (K193N and A242S) (Fig. 1a), whereas those derived from the contact animals of pair 2 contained a single change in HA (T318I) (Fig. 1b), indicating that additional changes in HA occurred during the infection of ferrets with HA(N158D/N224K/Q226L)/CA04. No mutations in the remaining genes were detected in any of these viruses from nasal washes compared with the CA04 virus sequences.

Because HA(N158D/N224K/Q226L)/CA04 was isolated from only one-third of the contact animals, we isolated a virus from the nasal wash of the contact ferret that shed a high titre (107.5 p.f.u. ml−1) of virus on day 7 after contact (pair 2) (Fig. 4d) to evaluate the replication and transmissibility of that virus in ferrets. This mutant virus, designated HA(N158D/N224K/Q226L/T318I)/CA04, replicated efficiently in the nasal turbinates and was isolated from brain tissue (Fig. 3 and Supplementary Table 2). In the transmission study, four of the six contact ferrets were positive for virus between days 3 and 7 after contact, and all contact animals were seropositive; no animals died in the course of the transmission experiments (Table 1; Fig. 4e and Supplementary Fig. 8). Notably, this transmission pattern is comparable to that of the 1918 pandemic H1N1 virus when tested under the same experimental conditions; the 1918 pandemic virus was recovered from the nasal wash of two of three contact animals (our own unpublished data). Sequence comparison of viruses from inoculated and contact animals identified mutations at positions 225 and 242 as well as a reversion at position 224 (Fig. 1a and Supplementary Table 5) (in addition to the original four mutations) although the 224 reversion was found only in viruses from inoculated ferrets. Collectively, these findings demonstrate that four amino acid substitutions (N158D/N224K/Q226L/T318I) in H5 HA confer efficient respiratory droplet transmission in ferrets to a virus possessing an H5 HA in a 2009 pandemic H1N1 backbone. We also confirmed that recombinant viruses possessing the three HA mutations N158D, N224K and Q226L, or the four HA mutations N158D, N224K, Q226L and T318I, and the NA of VN1203 in a PR8 backgrand (designated N158D/N224K/Q226L or N158D/N224K/Q226L/T318I, respectively) preferentially bind to Siaα2,6Gal and attach to human tracheal epithelia (Fig. 2c, d).

HA(N158D/N224K/Q226L/T318I)/CA04 transmitted by respiratory droplet more efficiently than HA(N158D/N224K/Q226L)/CA04, raising the possibility that the T318I mutation is involved in the efficient transmission of avian H5N1/pandemic H1N1 reassortants. To explore the functional role of this mutation in respiratory droplet transmission, we generated an H5 reassortant expressing the H5 HA with the T318I mutation and examined its receptor-binding specificity and transmissibility. This reassortant (designated rgT318I/CA04) bound to only Siaα2,3Gal and showed little binding to human tracheal epithelia (Fig. 2c, d). rgT318I/CA04 did not transmit via respiratory droplet among ferrets (Table 1 and Fig. 4f), although it replicated in nasal turbinates and trachea as efficiently as rgCA04 (Fig. 3 and Supplementary Table 2). These results indicate that the T318I mutation alone is not sufficient for H5 reassortant viruses to transmit efficiently among ferrets.

Influenza virus HA protein has membrane-fusion as well as receptor-binding activity. Notably, in the three-dimensional model of influenza A virus HA, residue 318 is located proximally to the fusion peptide (Fig. 1b), which has key roles in the membrane fusion process. To assess the effect of HA mutations on low-pH-induced membrane fusion activity, we examined the pH at which the fusion activity of wild-type and mutant HA was activated (Fig. 5). The wild-type HA had a threshold for membrane fusion of pH 5.7; the N224K/Q226L and N158D/N224K/Q226L mutations raised the threshold for fusion to >pH 5.9, whereas the T318I mutation reduced the threshold for fusion to pH 5.5. The N158D/N224K/Q226L/T318I mutations showed wild-type fusogenic properties (that is, a threshold at pH 5.7). The HA of influenza virus undergoes a low-pH-dependent conformational change, which is required for fusion of the viral envelope with the target membrane33. Such a conformational change to a fusion-active form can also lead to viral inactivation. Therefore, sustained and efficient human-to-human transmission of virus may require a certain level of stability of the HA protein in an acidic environment, as the pH of human nasal mucosa, where human influenza viruses replicate primarily, is approximately pH 5.5–6.5 (ref. 34). Our findings suggest that an increase in the pH threshold for fusion as a result of the N224K/Q226L mutations that shift the HA receptor recognition from avian-type to human-type may reduce HA protein stability; however, the T318I mutation decreases the pH threshold for fusion activity, resulting in a stable mutant HA.

Figure 5: Polykaryon formation by HeLa cells expressing wild-type or mutant HAs after acidification at low pH.
figure 5

a, The efficiency of polykaryon formation over a pH range of 5.4–6.0 was estimated from the number of nuclei in polykaryons divided by the total number of nuclei in the same field. The mean and standard deviations determined from five randomly chosen fields of cell culture are shown. Single asterisks indicate values significantly different between the wild-type HA and the N224K/Q226L or N158D/N224K/Q226L HA (Tukey test; P < 0.05). The double asterisk indicates values significantly different between the T318I HA and the N224K/Q226L or N158D/N224K/Q226L HA (Tukey test; P < 0.05). b, Representative fields of cells expressing the indicated HAs and exposed to pH 5.4, 5.6, or 5.8 are shown. Images were taken at ×10 magnification.

PowerPoint slide

Because heat treatment at neutral pH is also known to promote a fusogenic form of HA protein35,36 and serve as a surrogate assay for HA stability37, we next tested whether the HA mutations described above affect the heat stability of the HA protein. Wild-type and mutant HA viruses were incubated at 50 °C for various times, after which the loss of infectivity and haemagglutination activity were determined. The wild-type and N224K/Q226L viruses lost most of their infectivity by heating for 60 min (>5.5-log10 decrease in titre; Fig. 6a), whereas the N158D/N224K/Q226L and N158D/N224K/Q226L/T318I mutants exhibited considerable tolerance to high temperature (3.9- and 3.4-log10 decrease after a 60-min incubation, respectively) and the T318I mutant was most resistant (only a 1.4-log10 decrease under the same conditions). In haemagglutination assays, the N224K/Q226L mutant HA lost activity more rapidly than did the wild-type HA, and N158D/N224K/Q226L lost activity more rapidly than did the N158D/N224K/Q226L/T318I mutant (Fig. 6b). Thus, addition of the N158D mutation to the N224K/Q226L HA increased HA stability and subsequent addition of the fourth mutation, T318I, rendered the HA protein even more stable. Taken together, these results suggest that the addition of the T318I mutation to H5 HAs that preferentially recognize human-type receptors restores HA protein stability, thereby allowing a virus carrying the N158D/N224K/Q226L/T318I mutations in HA to transmit efficiently via respiratory droplet among ferrets. In conclusion, a fine balance of mutations affecting different functions in HA (such as receptor-binding specificity and HA stability) may be critical to confer transmissibility in ferrets.

Figure 6: Effect of heat treatment on the infectivity and haemagglutination activity of viruses.
figure 6

Aliquots of a virus stock containing 128 HA units were incubated for the times indicated at 50 °C. a, Virus titres in heat-treated samples were determined by plaque assays on MDCK cells. b, Haemagglutination titres in heat-treated samples were determined by using haemagglutination assays with 0.5% TRBCs. Each point represents the mean ± standard deviation from triplicate experiments.

PowerPoint slide

We next compared the pathogenicity in ferrets of H5 avian–human reassortants with that of the pandemic H1N1 virus CA04 (Fig. 7, Supplementary Information and Supplementary Figs 9–11). The control virus, rgCA04, caused substantial body weight loss (15.1%) (Table 1 and Supplementary Fig. 9). By contrast, the four reassortant viruses caused only modest weight loss (<10%) in most of the animals. However, no statistically significant differences in body weight loss were found between the reassortant viruses and rgCA04. Pathological examination revealed similar histological changes and levels of viral antigens in the nasal mucosa of rgCA04-, HA(N158D/N224K/Q226L)/CA04- and HA(N158D/N224K/Q226L/T318I)/CA04-infected ferrets (Fig. 7a, b). In the rgVN1203/CA04 and rg(N224K/Q226L)/CA04 groups, however, less tissue damage was found in the nasal mucosa compared with the rgCA04 group on day 3 after infection (Dunnett’s test; P = 0.0057 and 0.0175, respectively; Fig. 7b). In addition, all three viruses caused lung lesions (Supplementary Information and Supplementary Figs 10 and 11).

Figure 7: Pathological analyses of H5 avian–human reassortant viruses.
figure 7

a, Representative histological changes in nasal turbinates from influenza-virus-infected ferrets. Three ferrets per group were infected intranasally with 106 p.f.u. of virus, and tissues were collected on day 3 after infection for pathological examination. Uninfected ferret tissues served as negative controls (normal). Left panel, haematoxylin-and-eosin staining. Right panel, immunohistochemical staining for viral antigen detection (brown staining). Scale bars, 50 μm. b, Pathological severity scores in infected ferrets. To represent comprehensive histological changes, respiratory tissue slides were evaluated by scoring the pathological changes and viral antigen expression levels. The pathological scores were determined for each animal in each group (n = 3 per group on days 3 and 6 after infection) using the following scoring system: 0, no pathological change/antigen negative; 1, affected area (<30%) or only interstitial lesion/rare viral antigens; 2, affected area (<80%, ≥30%)/moderate viral antigens; 3, severe lesion (≥80%)/many viral antigens. Nasal, pathological changes in the nasal mucosa; nasal Ag, viral antigens in the nasal mucosa. Asterisks indicate virus pathological scores significantly different from that of rgCA04 (Dunnett’s test; P < 0.05). Error bars denote standard deviation.

PowerPoint slide

To assess whether current control measures may be effective against the H5 transmissible reassortant mutant virus, we examined the reactivity of sera from individuals vaccinated with an H5N1 prototype vaccine38 against a virus possessing the N158D/N224K/Q226L/T318I mutations in HA. We found that pooled human sera from individuals immunized with this vaccine reacted with the virus possessing the mutant H5 HA (N158D/N224K/Q226L/T318I) at a higher titre than with a wild-type H5 HA virus (VN1203/PR8; Supplementary Table 6), indicating that current H5N1 vaccines would be efficacious against the H5 transmissible reassortant mutant virus. In addition, the H5 transmissible reassortant mutant virus (HA(N158D/N224K/Q226L/T318I)/CA04) was highly susceptible to a licensed NA inhibitor, oseltamivir (Supplementary Table 7). These experiments show that appropriate control measures would be available to combat the transmissible virus described in this study.

Currently, we do not know whether the mutations that we identified in this study that allowed the HA(N158D/N224K/Q226L/T318I)/CA04 virus to be transmissible in ferrets would also support sustained human-to-human transmission. In particular, we wish to emphasize that the transmissible HA(N158D/N224K/Q226L/T318I)/CA04 virus possesses seven segments (all but the HA segment) from a human pandemic 2009 H1N1 virus. Human-virus-characteristic amino acids in these seven segments may have critically contributed to the respiratory droplet transmission of the HA(N158D/N224K/Q226L/T318I)/CA04 virus in ferrets. Examples include amino acids in the PB2 polymerase protein that confer efficient replication in mammalian, but not avian, cells39,40,41,42,43. As the PB2 gene of the HA(N158D/N224K/Q226L/T318I)/CA04 virus is of human virus origin, the virus possesses high replicative ability in mammalian cells. In contrast, most avian virus PB2 proteins lack these human-type amino acids, although one of these changes (a glutamic-acid-to-lysine mutation at position 627) is found in highly pathogenic avian H5N1 viruses circulating in the Middle East44. As a second example, the viral NA gene may contribute to viral transmissibility. The NA protein cleaves α-ketosidic linkages between a terminal sialic acid and an adjacent sugar residue, an activity that balances the sialic-acid-binding activity of HA. A recent study found that a human virus NA gene was critical to confer limited transmissibility to a mutant H5 avian-human reassortant virus14. In general, a human-type receptor recognizing H5 HA alone may not be sufficient to confer transmissibility in mammals, but may have to act together with other human-virus-characteristic traits (in PB2, NA, and/or other viral proteins). Therefore, at this point we cannot predict whether the four mutations in the H5 HA identified here would render a wholly avian H5N1 virus transmissible.

Three of the residues identified here (N224, Q226 and T318) have been strictly conserved among H5 HA proteins isolated since 2003. However, as H5N1 viruses continue to evolve and infect people, receptor-binding variants of H5N1 viruses, including avian–human reassortant viruses as tested here, may emerge. One of the four mutations we identified in our transmissible virus, the N158D mutation, results in loss of a glycosylation site. Many H5N1 viruses isolated in the Middle East, Africa, Asia and Europe do not have this glycosylation site. Therefore, only three nucleotide changes are needed for the HA of these viruses to support efficient transmission in ferrets. In addition, the H5N1 viruses circulating in these geographic areas also possess a glutamic-acid-to-lysine mutation at position 627 in the PB2 protein, which promotes viral replication in certain mammals, including humans40,45. Therefore, these viruses may be several steps closer to those capable of efficient transmission in humans and are of concern.

Our study highlights the pandemic potential of viruses possessing an H5 HA. Although current vaccines may protect against a virus similar to that tested here, the continued evolution of H5N1 viruses reinforces the need to prepare and update candidate vaccines to H5 viruses. The amino acid changes identified here will help individuals conducting surveillance in regions with circulating H5N1 viruses (for example, Egypt, Indonesia, Vietnam) to recognize key residues that predict the pandemic potential of isolates. Rapid responses in a potential pandemic situation are essential in order to generate appropriate vaccines and initiate other public health measures to control infection. Furthermore, our findings are of critical importance to those making public health and policy decisions.

Our research answers a fundamental question in influenza research: can H5-HA-possessing viruses support transmission in mammals? Moreover, our findings have suggested that different mechanisms (that is, receptor-binding specificity and HA stability) may act in concert for efficient transmissibility in mammals. This knowledge will facilitate the identification of additional mutations that affect viral transmissibility; the monitoring of this expanded set of changes in natural isolates may improve our ability to assess the pandemic potential of H5N1 viruses. Thus, although a pandemic H5N1 virus may not possess the amino acid changes identified in our study, the findings described here will advance our understanding of the mechanisms and evolutionary pathways that contribute to avian influenza virus transmission in mammals.

Methods Summary

Viruses

All recombinant viruses were generated by using reverse genetics essentially as described previously16. All experiments with the viruses possessing the wild-type HA cleavage site were performed in an enhanced biosafety level 3 (BSL3+) containment laboratory approved for such use by the CDC and the USDA.

Infection and transmission in ferrets

Six–ten-month-old female ferrets (Triple F Farms) were intramuscularly anaesthetized and intranasally inoculated with 106 p.f.u. (500 μl) of virus. On days 3 and 6 after infection, ferrets were killed for virological and pathological examinations. The virus titres in various organs were determined by use of plaque assays in MDCK cells.

For transmission studies in ferrets, animals were housed in adjacent transmission cages that prevented direct and indirect contact between animals but allowed spread of influenza virus through the air (Showa Science; Supplementary Fig. 7). Ferrets were intranasally inoculated with 106 p.f.u. (500 μl) of virus (inoculated ferrets). Twenty-four hours after infection, naive ferrets were each placed in a cage adjacent to an inoculated ferret (contact ferrets). To assess viral replication in the nasal turbinates, we determined viral titres in nasal washes collected from virus-inoculated and contact ferrets on day 1 after inoculation or co-housing, respectively, and then every other day. Animal studies were performed in accordance with Animal Care and Use Committee guidelines of the University of Wisconsin-Madison.

Biosafety and biosecurity

All recombinant DNA protocols were approved by the University of Wisconsin-Madison’s Institutional Biosafety Committee after risk assessments were conducted by the Office of Biological Safety, and by the University of Tokyo’s Subcommittee on Living Modified Organisms, and, when required, by the competent minister of Japan. In addition, the University of Wisconsin-Madison Biosecurity Task Force regularly reviews the research program and ongoing activities of the laboratory. The task force has a diverse skill set and provides support in the areas of biosafety, facilities, compliance, security and health. Members of the Biosecurity Task Force are in frequent contact with the principal investigator and laboratory personnel to provide oversight and assure biosecurity. Experiments with viruses possessing the wild-type HA cleavage site were performed in enhanced BSL3 containment laboratories approved for such use by the CDC and the USDA. Ferret transmission studies were conducted by three scientists with both DVM and PhD degrees who each had more than a minimum of 6 years of experience with highly pathogenic influenza viruses and animal studies with highly pathogenic viruses. Our staff wear powered air-purifying respirators that filter the air, and disposable coveralls; they shower out on exit from the facility. The containment facilities at University of Wisconsin-Madison were designed to exceed standards outlined in Biosafety in Microbiological and Biomedical Laboratories (5th edition; http://www.cdc.gov/biosafety/publications/bmbl5/BMBL.pdf). Features of the BSL3-enhanced suites include entry/exit through a shower change room, effluent decontamination, negative air-pressure laboratories, double-door autoclaves, double HEPA-filtered exhaust air, and gas decontamination ports. The BSL3-Agriculture suite features include all those listed for BSL3-enhanced plus HEPA-filtered supply and double-HEPA-filtered exhaust air, double-gasketed watertight and airtight seals, airtight dampers on all ductwork, and the structure was pressure-decay tested during commissioning. The University of Wisconsin-Madison facility has a dedicated alarm system that monitors all building controls and sends alarms (500 possible alerts). Redundancies and emergency resources are built-in to the facility including two air handlers, two compressors, two filters each place filters are needed, two effluent sterilization tanks, two power feeds to the building, an emergency generator in case of a power failure and other physical containment measures in the facility that operate without power. Biosecurity monitoring of the facility is ongoing. All personnel undergo Select Agent security risk assessment by the United States Criminal Justice Information Services Division and complete rigorous biosafety, BSL3 and Select Agent training before participating in BSL3-level experiments. Refresher training is scheduled on a regular basis. The principal investigator participates in training sessions and emphasizes compliance to maintain safe operations and a responsible research environment. The laboratory occupational health plan is in compliance with the University of Wisconsin-Madison Occupational Health Program. Select agent virus inventory is checked monthly and submitted to the University of Wisconsin-Madison Research Compliance Specialist. Virus inventory is submitted 1–2 times per year to the file holder in the Select Agent branch of the CDC. The research program, procedures, occupational health plan, documentation, security and facilities are reviewed annually by the University of Wisconsin-Madison Responsible Official and at regular intervals by the CDC and the Animal and Plant Health Inspection Service (APHIS) as part of the University of Wisconsin-Madison Select Agent Program.

Online Methods

Viruses

All recombinant viruses were generated by using reverse genetics essentially as described previously16. All experiments with the viruses possessing the wild-type HA cleavage site were performed in an enhanced biosafety level 3 (BSL3+) containment laboratory approved for such use by the CDC and the USDA.

Infection and transmission in ferrets

Six–ten-month-old female ferrets (Triple F Farms) were intramuscularly anaesthetized and intranasally inoculated with 106 p.f.u. (500 μl) of virus. On days 3 and 6 after infection, ferrets were killed for virological and pathological examinations. The virus titres in various organs were determined by use of plaque assays in MDCK cells.

For transmission studies in ferrets, animals were housed in adjacent transmission cages that prevented direct and indirect contact between animals but allowed spread of influenza virus through the air (Showa Science; Supplementary Fig. 7). Ferrets were intranasally inoculated with 106 p.f.u. (500 μl) of virus (inoculated ferrets). Twenty-four hours after infection, naive ferrets were each placed in a cage adjacent to an inoculated ferret (contact ferrets). To assess viral replication in the nasal turbinates, we determined viral titres in nasal washes collected from virus-inoculated and contact ferrets on day 1 after inoculation or co-housing, respectively, and then every other day. Animal studies were performed in accordance with Animal Care and Use Committee guidelines of the University of Wisconsin-Madison.

Biosafety and biosecurity

All recombinant DNA protocols were approved by the University of Wisconsin-Madison’s Institutional Biosafety Committee after risk assessments were conducted by the Office of Biological Safety, and by the University of Tokyo’s Subcommittee on Living Modified Organisms, and, when required, by the competent minister of Japan. In addition, the University of Wisconsin-Madison Biosecurity Task Force regularly reviews the research program and ongoing activities of the laboratory. The task force has a diverse skill set and provides support in the areas of biosafety, facilities, compliance, security and health. Members of the Biosecurity Task Force are in frequent contact with the principal investigator and laboratory personnel to provide oversight and assure biosecurity. Experiments with viruses possessing the wild-type HA cleavage site were performed in enhanced BSL3 containment laboratories approved for such use by the CDC and the USDA. Ferret transmission studies were conducted by three scientists with both DVM and PhD degrees who each had more than a minimum of 6 years of experience with highly pathogenic influenza viruses and animal studies with highly pathogenic viruses. Our staff wear powered air-purifying respirators that filter the air, and disposable coveralls; they shower out on exit from the facility. The containment facilities at University of Wisconsin-Madison were designed to exceed standards outlined in Biosafety in Microbiological and Biomedical Laboratories (5th edition; http://www.cdc.gov/biosafety/publications/bmbl5/BMBL.pdf). Features of the BSL3-enhanced suites include entry/exit through a shower change room, effluent decontamination, negative air-pressure laboratories, double-door autoclaves, double HEPA-filtered exhaust air, and gas decontamination ports. The BSL3-Agriculture suite features include all those listed for BSL3-enhanced plus HEPA-filtered supply and double-HEPA-filtered exhaust air, double-gasketed watertight and airtight seals, airtight dampers on all ductwork, and the structure was pressure-decay tested during commissioning. The University of Wisconsin-Madison facility has a dedicated alarm system that monitors all building controls and sends alarms (500 possible alerts). Redundancies and emergency resources are built-in to the facility including two air handlers, two compressors, two filters each place filters are needed, two effluent sterilization tanks, two power feeds to the building, an emergency generator in case of a power failure and other physical containment measures in the facility that operate without power. Biosecurity monitoring of the facility is ongoing. All personnel undergo Select Agent security risk assessment by the United States Criminal Justice Information Services Division and complete rigorous biosafety, BSL3 and Select Agent training before participating in BSL3-level experiments. Refresher training is scheduled on a regular basis. The principal investigator participates in training sessions and emphasizes compliance to maintain safe operations and a responsible research environment. The laboratory occupational health plan is in compliance with the University of Wisconsin-Madison Occupational Health Program. Select agent virus inventory is checked monthly and submitted to the University of Wisconsin-Madison Research Compliance Specialist. Virus inventory is submitted 1–2 times per year to the file holder in the Select Agent branch of the CDC. The research program, procedures, occupational health plan, documentation, security and facilities are reviewed annually by the University of Wisconsin-Madison Responsible Official and at regular intervals by the CDC and the Animal and Plant Health Inspection Service (APHIS) as part of the University of Wisconsin-Madison Select Agent Program.

Cells

Madin–Darby canine kidney (MDCK) cells and MDCK cells overexpressing Siaα2,6Gal (AX4 cells17) were maintained in Eagle’s minimal essential medium (MEM) containing 5% newborn calf serum. Human embryonic kidney 293T cells were cultured in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum (FBS). HeLa cells were maintained in MEM containing 10% FBS. All cells were maintained at 37 °C in 5% CO2.

Plasmid construction and reverse genetics

Plasmid constructs for viral RNA production (pPolI)—containing the genes of the A/Vietnam/1203/2004 (H5N1; VN1203), A/Puerto Rico/8/34 (H1N1; PR8), A/Kawasaki/173/2001 (H1N1; K173) and A/California/04/2009 (H1N1; CA04) viruses flanked by the human RNA polymerase I promoter and the mouse RNA polymerase I terminator—were constructed as described16. The multibasic amino acids at the haemagglutinin (HA) cleavage site (RERRRKKR↓G) of the reassortant viruses between VN1203 and PR8 were changed to RETR↓G by site-directed mutagenesis. All transfectant viruses were generated by using reverse genetics essentially as described previously16. Recombinant viruses were amplified in MDCK or AX417 cells and stored at −80 °C until use. The HA segment of all viruses was sequenced to ensure the absence of unwanted mutations. All experiments with the reassortant viruses between VN1203 and CA04 were performed in enhanced biosafety level 3 containment laboratories approved for such use by the CDC and the USDA.

To introduce random mutations into the globular head of the VN1203 HA protein, a 143-amino-acid region spanning residues 120–259 (H3 numbering) was selected. This region was subjected to PCR-based random mutagenesis by use of the GeneMorph II kit (Stratagene) following the manufacturer’s instructions. The targeted mutation rate (1–2 amino acid replacements per molecule) was achieved through optimization of the template quantity, and was confirmed by sequence analysis of 48 individual clones. By using a PCR-based cloning strategy, we inserted the mutagenized region into its respective vector containing the VN1203 HA gene between the human RNA polymerase I promoter and mouse RNA polymerase I terminator sequences. The composition of the plasmid library was confirmed by sequencing. The plasmid library was then used to generate an influenza virus library, essentially as described16. The size of the virus library was 7 × 106 p.f.u.

Preparation of sialidase-treated TRBCs

Turkey red blood cells (TRBCs) were washed three times with phosphate-buffered saline (PBS), and diluted to 20% (vol/vol) in PBS. TRBCs (1 ml) were incubated with 500 U of α2,3-sialidase from Salmonella enterica serovar Typhimurium LT2 (NEB) for 20–24 h at 37 °C, washed three times in PBS, and re-suspended in PBS or MEM containing 1% bovine serum albumin (BSA) (MEM/BSA).

Haemagglutination assay

Viruses (50 μl) were serially diluted with 50 μl of PBS in a microtitre plate. An equal volume (that is, 50 μl) of a 0.5% (vol/vol) TRBC suspension was added to each well. The plates were kept at room temperature and haemagglutination was assessed after a 1-h incubation.

Virus library screening

To select VN1203 HA variants that had acquired the ability to recognize human-type receptors, three parallel experiments were carried out, each with 0.7 × 106 viruses. The virus library was first incubated with 0.1 ml of 10% (vol/vol) α2,3-sialidase-treated TRBCs for 10 min at 4 °C. After this incubation, the TRBCs and bound viruses were pelleted at 1,000 r.p.m. for 1 min, and the pellets then washed ten times in MEM/BSA containing 313 mM NaCl. Bound viruses were eluted by incubation at 37 °C for 30 min and then diluted to approximately 0.5 virus per well (determined by virus titration in a pilot study). Individual viruses were then amplified in AX4 cells, which overexpress Siaα2,6Gal17. Individual viruses were re-screened by using haemagglutination assays with α2,3-sialidase-treated TRBCs.

Solid-phase binding assay

Viruses were grown in MDCK cells, clarified by low-speed centrifugation, laid over a cushion of 30% sucrose in PBS, and ultracentrifuged at 25,000 r.p.m. for 2 h at 4 °C. Virus stocks were aliquoted and stored at −80 °C. Virus concentrations were determined by using haemagglutination assays with 0.5% (vol/vol) TRBCs. The direct receptor-binding capacity of viruses was examined by use of a solid-phase binding assay as previously described9. Microtitre plates (Nunc) were incubated with the sodium salts of sialylglycopolymers (poly-l-glutamic acid backbones containing N-acetylneuraminic acid linked to galactose through either an α2,3 (Neu5Acα2,3Galβ1,4GlcNAcβ1-pAP) or an α2,6 (Neu5Acα2,6Galβ1,4GlcNAcβ1-pAP) bond) in PBS at 4 °C overnight. After the glycopolymer solution was removed, the plates were blocked with 0.15 ml of PBS containing 4% BSA at room temperature for 1 h. After four successive washes with ice-cold PBS, the plates were incubated in a solution containing influenza virus (8–32 HA units in PBS) at 4 °C overnight. After washing as described above, the plates were incubated for 2 h at 4 °C with rabbit polyclonal antiserum to either K173 or VN1203 virus. The plates were then washed again as before and incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG antiserum for 2 h at 4 °C. After washing, the plates were incubated with O-phenylenediamine (Sigma) in PBS containing 0.01% H2O2 for 10 min at room temperature, and the reaction was stopped with 0.05 ml of 1 M HCl. The optical density at 490 nm was determined in a plate reader (Infinite M1000; Tecan).

Virus binding to human airway tissues

Paraffin-embedded normal human trachea (US Biological) and lung (BioChain) tissue sections were deparaffinized and rehydrated. Sections were then blocked by using 4% BSA in PBS and covered with virus suspensions (64 HA units in PBS) at 4 °C overnight. After being washed four times in ice-cold PBS, the sections were incubated with primary antibodies for 3 h at 4 °C. The primary antibodies used were as follows: a pool of mouse anti-VN1203 HA monoclonal antibodies (15A3, 3G2, 7A11, 8A3, 14C5 and 18E1; Rockland); rabbit anti-K173 polyclonal antibody; rabbit anti-surfactant protein A polyclonal antibody (Millipore); and mouse anti-surfactant protein A monoclonal antibody (Abcam). Antibody binding was detected by using an IgG secondary antibody conjugated with Alexa Fluor 488 or Alexa Fluor 633 (Molecular Probes). Sections were also counterstained with Hoechst 33342, trihydrochloride, trihydrate (Molecular Probes). The samples were examined by using confocal laser scanning microscopy (model LSM 510; Carl Zeiss).

To confirm sialic-acid-specific virus binding, tissue sections were treated, before incubation with viruses, with Arthrobacter ureafaciens sialidase (Sigma) for 3 h at 37 °C. Viruses bound to tissue were detected as described above.

Experimental infection of ferrets

Animal studies were performed in accordance with the Animal Care and Use Committee guidelines of the University of Wisconsin-Madison. We used 6–10-month-old female ferrets (Triple F Farms) that were serologically negative by haemagglutination inhibition (HI) assay for currently circulating human influenza viruses. Six ferrets per group were anaesthetized intramuscularly with ketamine and xylazine (5–30 mg and 0.2–6 mg kg−1 of body weight, respectively) and inoculated intranasally with 106 p.f.u. (500 μl) of viruses. On days 3 and 6 after infection, three ferrets per group were killed for virological and pathological examinations. The virus titres in various organs were determined by use of plaque assays in MDCK cells.

Excised tissue samples of nasal turbinates, trachea, lungs, brain, liver, spleen, kidney and colon from euthanized ferrets were preserved in 10% phosphate-buffered formalin. Tissues were then trimmed and processed for paraffin embedding and cut into 5-μm-thick sections. One section from each tissue sample was stained by using a standard haematoxylin-and-eosin procedure, whereas another one was processed for immunohistological staining with a mixture of two anti-influenza virus rabbit antibodies (1:2,000; R309 and anti-VN1203; both prepared in our laboratory) that react with CA04 and VN1203, respectively. Specific antigen–antibody reactions were visualized by using an indirect two-step dextran-polymer technique (Dako EnVision system; Dako) and 3,3′ diaminobenzidine tetrahydrochloride staining (Dako).

Ferret transmission study

For transmission studies in ferrets, animals were housed in adjacent transmission cages that prevented direct and indirect contact between animals but allowed spread of influenza virus through the air (Showa Science; Supplementary Fig. 7). Three, five, or six ferrets were inoculated intranasally with 106 p.f.u. (500 μl) of virus (inoculated ferrets). Twenty-four hours after infection, three, five, or six naive ferrets were each placed in a cage adjacent to an inoculated ferret (contact ferrets). The ferrets were monitored for changes in body weight and the presence of clinical signs. To assess viral replication in nasal turbinates, we determined viral titres in nasal washes collected from virus-inoculated and contact ferrets on day 1 after inoculation or co-housing, respectively, and then every other day.

Serological tests

Serum samples were collected between days 14 and 20 after infection, treated with receptor-destroying enzyme, heat-inactivated at 56 °C for 30 min, and tested by use of an HI assay with 0.5% TRBCs (http://www.wpro.who.int/entity/emerging_diseases/documents/docs/manualonanimalaidiagnosisandsurveillance.pdf). Viruses bearing homologous HA were used as antigens for the HI tests.

Polykaryon formation representing membrane fusion activity

Monolayers of HeLa cells grown in 12-well plates were transfected with the protein expression vector pCAGGS46 encoding wild-type or mutant HA. At 24 h after transfection, cells transiently expressing HA protein were treated with trypsin (1 μg ml−1) in MEM containing 0.3% BSA for 30 min at 37 °C to cleave the HA into its HA1 and HA2 subunits. Polykaryon formation was induced by exposing the cells to low-pH buffer (145 mM NaCl, 20 mM sodium citrate (pH 6.0–5.4)) for 2 min at 37 °C. After this exposure, the low-pH buffer was replaced with MEM containing 10% FBS and the cells were incubated for 3 h at 37 °C. The cells were then fixed with methanol and stained with Giemsa’s solution and photographed with a digital camera mounted on an inverted microscope (Nikon, Eclipse Ti). For quantitative analyses, cell nuclei were counted in five randomly chosen fields of cell culture. Polykaryon formation activity was calculated from the number of nuclei in polykaryons divided by the total number of nuclei in the same field.

Thermostability

Viruses (128 HA units in PBS) were incubated for the times indicated at 50 °C. Subsequently, infectivity and haemagglutination activity were determined by use of plaque assays in MDCK cells and haemagglutination assays using 0.5% TRBCs, respectively.

Neuraminidase (NA) inhibition assay

To assess the sensitivity of viruses to the NA inhibitor oseltamivir, NA inhibition assays were performed as described previously32.

Statistical analysis

All statistical analyses were performed using JMP 9.0.0 (SAS Institute Inc.). The statistical significance of differences between rgCA04 and H5 avian/human reassortant viruses was determined by using a Dunnett’s test. Comparisons of polykaryon formation between wild-type and mutant HAs were done using Tukey’s test. P values of <0.05 were considered significant.