A Novel Neuraminidase-Dependent Hemagglutinin Cleavage Mechanism Enables the Systemic Spread of an H7N6 Avian Influenza Virus

The identification of virulence markers in influenza viruses underpins risk assessment programs and the development of novel therapeutics. The cleavage of the influenza virus HA is a required step in the viral life cycle, and phenotypic differences in viruses can be caused by changes in this process. Here, we describe a novel mechanism for HA cleavage in an H7N6 influenza virus isolated from a mallard duck. The mechanism requires the N6 protein and full activity of thrombin-like proteases and allows the virus to cause systemic infection in chickens, ducks, and mice. The thrombin-mediated cleavage of HA is thus a novel virulence determinant of avian influenza viruses.

A vian influenza A viruses cause disease not only in birds, but also in humans and other mammals. The H5 and H7 strains of these viruses occasionally develop high pathogenicity (1-4), causing adverse economic effects in the poultry industry. Although influenza A virus can infect many hosts, wild aquatic birds (including ducks, geese, gulls, and shorebirds) are believed to constitute the principal natural reservoir for most virus subtypes (5,6).
Influenza A viruses possess 2 major glycoproteins, hemagglutinin (HA) and neuraminidase (NA), which protrude from the viral envelope. To infect host cells, the HA protein mediates binding to the cellular receptor, composed of ␣2,3or 2,6-sialylated glycans; exposure to low pH in the endosome then drives the virus to fuse with the cellular membrane following cleavage of the precursor HA molecule (HA 0 ) into HA 1 and HA 2 subunits by cellular proteases (7,8). Specific cleavage of the HA 0 precursor protein is required for viral infectivity and is an important determinant of viral tropism and pathogenicity in the infected host.
Typically, the HA proteins of highly pathogenic avian influenza (HPAI) H5 and H7 viruses contain additional basic amino acids (R_X_R/K_R/G) at the cleavage site that can be cleaved by an intracellular subtilisin-type enzyme. In contrast, most low-pathogenic avian influenza (LPAI) viruses have a single arginine at the cleavage site, and the HA of these viruses can be cleaved by trypsin-like proteases, such as type II transmembrane serine proteases (9). Because expression of the trypsin-like proteases is limited to the respiratory tract in mammalian hosts, the replication of LPAI viruses is confined to this target site. Similarly, the requirement for an exogenous trypsin-like protease to support efficient viral replication in cell culture is determined by the nature of the monobasic cleavage site in the HA protein. However, the subtilisin-like proteases that can cleave the HA in HPAI viruses are ubiquitously expressed in many tissues; consequently, these viruses spread systemically in susceptible hosts, and such systemic spread is considered the major determinant for increased virulence (10,11).
HPAI viruses can also replicate in cultured cells without the need for exogenous trypsin. Additionally, the 1918 pandemic influenza H1N1 virus, which does not harbor a multibasic HA cleavage site, can replicate to high titers in Madin-Darby canine kidney (MDCK) cells without the addition of trypsin; notably, NA is essential for this trypsinindependent growth (11). Similarly, the NA of a related strain, A/WSN/33 (H1N1), also facilitates efficient viral replication in the absence of trypsin by activating serum plasminogen, expanding viral tropism in vivo (11). Recently, Tse et al. demonstrated that the HA of an LPAI H9N2 virus contains an R-S-K-R cleavage site that can be cleaved in cells with high levels of furin expression and that a glycosylation site-eliminating mutation in HA allows it to be universally cleaved in cell culture (12). Other molecular mechanisms for trypsin-independent replication of LPAI viruses beyond NA-dependent plasminogen activation remain unclear.
In this study, we report a novel replication mechanism exhibited by a naturally occurring LPAI virus, A/Mallard duck/Korea/6L/2007 (A/Mdk/6L/07 [H7N6]). This virus was isolated from multiple tissues of a dead mallard duck. Despite the virus having many features in common with HPAI viruses, it has maintained a monobasic HA cleavage site. Using several in vitro and in vivo assays, we provide evidence for a thrombin-and neuraminidase-dependent HA activation mechanism that underpins many of the unique properties of this virus. Thrombin-mediated HA cleavage resulted in wider dissemination of the virus in infected chickens, ducks, and mice; thus, it is a novel virulence determinant of H7 influenza viruses.

RESULTS
Genetic characterization of the A/Mdk/6L/07 (H7N6) virus. During routine surveillance for avian influenza activity in migratory birds, in December 2007, we isolated A/Mdk/6L/07 from multiple organs of a dead mallard duck, including the trachea, lung, spleen, and brain. Whole-genome sequence analysis revealed that virus isolates from multiple tissues of the bird were homologous and of the H7N6 subtype. Therefore, we designated the lung isolate A/Mdk/6L/07. Despite the presence of virus in multiple organs of the duck, deduced amino acid sequence analysis of the HA cleavage site showed that the virus has characteristics associated with an LPAI virus (PEIPKGR/G) and that additional basic amino acids were not present (Table 1). Similarly, an examination of other well-characterized molecular markers did not explain the disseminated nature of the infection.
Sequence analysis of the wider receptor-binding site (RBS) revealed that A/Mdk/ 6L/07 possessed the highly conserved residues Tyr88, Trp143, His174, and Tyr186 (equivalent to Tyr98, Trp153, His183, and Tyr195 in H3 numbering, respectively), as seen in recent wild bird H7 viruses. No deletions were observed in the NA stalk region, where mutations occur upon virus adaptation in terrestrial birds. A functional PB1-F2 open reading frame was detected, which is typical of viruses of wild bird origin and affects  (13). Amino acid sequence analysis of the NS1 protein revealed that all isolates had a 5-residue deletion at positions 80 to 84, which has been observed in poultry isolates in Hong Kong as far back as 2001 (14). The NS1 C-terminal PDZ-binding motif had the ESEV sequence, which is typical for avian viruses; these motifs are known to confer a severe disease phenotype in mice (15,16) (Table 1). All 8 gene segments of the Mdk/Korea/6L/07 virus were related to those of viruses isolated in Asia (South Korea, Japan, and China), and the nucleotide (nt) similarity among these segments was 98.7 to 99.9% (see Table S1 in the supplemental material). Phylogenetic analysis of the HA gene of the Mdk/Korea/6L/07 virus showed that it belongs to the Eurasian lineage and is part of a cluster containing recent H7 avian influenza viruses isolated in South Korea.
As sequence analyses did not reveal any clues as to the underlying mechanism behind the disseminated nature of the observed infection, we also investigated receptor-binding preferences by performing a solid-phase direct binding assay, as previously described (10,17). The 2009 pandemic H1N1 virus (A/California/07/2009 [CA/07 {H1N1}]) was used as a positive control for binding to ␣2,6-linked sialic acids. Although the CA/07 (H1N1) virus exhibited strong affinity for ␣2,6-linked sialic acids, A/Mdk/6L/07 mainly bound the ␣2,3-linked sialic acid (Fig. S1). This result confirmed that the A/Mdk/6L/07 virus preferentially binds to avian influenza virus receptors, as suggested by sequence analysis.
A/Mdk/6L/07 replicated in infected chickens and reached a peak viral titer (measured using tracheal swabs) of 4.2 log 10 EID 50 /ml at 5 days postinfection (dpi), with virus shedding persisting up to 7 dpi (Table 2). Notably, A/Mdk/6L/07 was recovered from the lungs (4.2 log 10 EID 50 /g), kidneys (1.2 log 10 EID 50 /g), and heart (1.2 log 10 EID 50 /g), indicating an ability to replicate outside the respiratory and gastrointestinal tracts ( Table 2). In contrast, A/Ab/W44/05 was only detected in the respiratory and intestinal organs where typical LPAI viruses replicate. Although A/Mdk/6L/07 was isolated from a wild mallard duck, virus titers in experimentally infected in ducks were lower than those in chickens; the peak viral titer in tracheal swabs was 2.5 log 10 EID 50 /ml. However, as seen in the original mallard duck, the virus exhibited broad tissue tropism, as virus was found in the kidney (1.2 log 10 EID 50 /g), spleen (1.2 log 10 EID 50 /g), heart (1.5 log 10 EID 50 /g), and in the respiratory and intestinal organs of infected ducks. The virus, however, replicated to higher titers in the chicken lung than in the duck lung (Table 2). In contrast, A/Ab/W44/05 was only detected in the duck respiratory tract and cloaca. We evaluated the intravenous pathogenicity index (IVPI) of A/Mdk/6L/07 in chickens and found it to be 0.22. Further, depression with decreased food consumption was observed in A/Mdk/6L/07-infected chickens at early time points (1 to 3 dpi), although no death was observed during the test. Although the virus was isolated from fatal cases in migratory mallard ducks, we only observed mild clinical symptoms for 2 to 3 dpi in the experimental setting with SPF birds, suggesting that the Mdk/Korea/6L/07 strain is an LPAI virus. In mice, the 50% mouse lethal dose of A/Mdk/6L/07 is Ͼ8.7 log 10 EID 50 /ml, and this virus induces only mild pathological changes in the lung (Table S2 and Fig. S6). However, A/Mdk/6L/07 could replicate in mice without preadaptation and was detected in the lung, brain, spleen, and heart of infected mice, with peak titers of 4.2 log 10 EID 50 /ml in lung tissues, showing that its systemic spread was not limited to avian hosts (Table 2). A/Ab/W44/05 was detected only in the lungs (4.2 log 10 EID 50 /ml at 3 dpi).

Trypsin-independent growth of A/Mdk/6L/07 (H7N6) in vitro.
As discussed above, an additional characteristic of HPAI viruses is their ability to replicate in cultured cells without exogenous trypsin. Because we had observed systemic spread in infected animals, we wanted to determine whether A/Mdk/6L/07 replication in MDCK cells requires trypsin. For comparison, A/Ab/W44/05 (a negative control) and A/WSN/33 (a positive control) were included.
To prepare to study further the mechanism behind the trypsin-independent growth of A/Mdk/6L/07, we used reverse genetics to generate recombinant versions of A/Mdk/ 6L/07, A/Ab/W44/05, and A/WSN/33. These recombinant viruses exhibited growth patterns similar to those observed for their respective parental viruses, indicating that wild-type and recombinant viruses shared similar characteristics (Fig. 1B).

Role of the NA gene of A/Mdk/6L/07 in trypsin-independent viral growth.
Because the most obvious genetic difference between A/Mdk/6L/07 and A/Ab/W44/05 was the presence of a different NA (N6 and N3, respectively), we hypothesized that N6 was associated with the trypsin-independent growth of A/Mdk/6L/07. To test this hypothesis, recombinant A/Mdk/6L/07-like viruses differing only in their NA genes were generated (Table S3). Of the 9 recombinant viruses (H7N1 through H7N9), only H7N6 replicated to substantial levels in the absence of trypsin (greater than 3.5 log 10 TCID 50 /ml), indicating that the NA gene (specifically N6) was associated with trypsinindependent viral growth ( Fig. 2A).
To determine whether this trait was associated with a specific N6 protein, we generated 6 more A/Mdk/6L/07-like viruses (all H7N6) differing only in the source of their N6 gene (Table S2, various N6 gene panel). All tested viruses reached titers of 5.63 to 7.47 log 10 TCID 50 /ml with or without exogenous trypsin (Fig. 2B). Transferring the N6 gene from A/Mdk/6L/07 to A/Ab/W44/05 enabled growth in the absence of trypsin (Fig. 2C); A/Ab/W44/05 did not replicate well with its native N3 protein in the absence of trypsin (Fig. 1). Correspondingly, transferring the N3 gene from A/Ab/W44/05 to A/Mdk/6L/07 resulted in the loss of its cytopathic effect 72 h after infection (Fig. S2). These results indicate that N6 plays a critical role in the trypsin-independent viral growth of the tested H7 viruses. Genetic analysis of the H7 cleavage site residues suggests the presence of a potentially novel proteolytic motif. Although we were able to provide evidence that the trypsin-independent growth of A/Mdk/6L/07 is dependent on the N6 gene, the HA protein of this virus, H7, must still be cleaved for efficient growth. Therefore, we explored the genetic characteristics of the cleavage motifs in the HA proteins of H7 viruses by performing single amino acid polymorphism (SAP) analysis of sequences deposited in GenBank and the Influenza Virus Resource (Influenza Virus Resource; Information, Search and Analysis; https://www.ncbi.nlm.nih.gov/genomes/ FLU/Database/nph-select.cgi?goϭdatabase). SAP analysis revealed that the H7 HA protein contains an invariant arginine at position 339 (H3 numbering), which is the P1 position of the proteolytic cleavage site, and highly conserved proline, lysine, and glycine residues at positions 336, 337, and 338 (the P4, P3, and P2 positions), respectively (Fig. S3A).
The residue at the P2 position correlates with the geographic distribution of each virus (Fig. S3B). Glycine (the dominant residue) was common (39.1%) in Eurasian H7 lineages (found in Italy, Spain, Germany, England, Sweden, the Netherlands, Egypt, South Korea, China, Japan, Hong Kong, Thailand, and other countries), whereas proline (34%), tyrosine (18.9%), and other residues were observed only in American LPAI lineages (the United States, Canada, Chile, Mexico, and New York) or in HPAI H7 isolates (Fig. S3B). We also noted that glycine, arginine, and glycine in positions P2, P1, and P1=, respectively, form an optimal cleavage site for thrombin (Fig. S3C) (18), suggesting that thrombin-like proteases can recognize the HA cleavage motif of H7 viruses, especially those of Eurasian lineages (including A/Mdk/6L/07).
Effects of HA cleavage motifs on virus replication. To further investigate the effects of HA cleavage motifs on replication, we modified the native A/Mdk/6L/07 HA cleavage motif (H7 PKGRG ) using PCR-mediated site-directed mutagenesis. We generated recombinant A/Mdk/6L/07 viruses containing serine, alanine, aspartic acid, or asparagine substitutions at the P2 position and a mutant virus (HA PKG338_QS ) with glutamine and serine substitutions at the P3 and P2 positions, respectively. By doing this, we generated the cleavage motif of the A/PR/8/34 H1N1 virus, which exhibits trypsindependent viral growth. Although all of the mutant viruses exhibited similar replication properties in media containing trypsin, a significant reduction in replication was observed in the absence of trypsin (P Ͻ 0.005; Fig. 3A). The same effect was observed in multicycle replication studies conducted using an MOI of 0.0001 and an incubation time of 60 h (Fig. 3B).
We next replaced the HA cleavage site in A/PR/8/34 with the thrombin motifcontaining site from A/Mdk/6L/07 (creating the virus PR8 S338G ). We then made an additional virus by replacing the N1 gene of PR8 S338G with the N6 gene (creating the virus PR8 S338GϩN6 ) (Fig. 3C). The wild-type A/PR/8/34 (H1N1), PR8 S338G (H1N1), and PR8 S338GϩN6 (H1N6) viruses replicated efficiently in the presence of trypsin in MDCK cells, indicating that all recombinant PR8 viruses (H1N1 and H1N6) could replicate ( Fig. 3C and D). In the absence of trypsin, the PR8 virus replicated only minimally (3.05 log 10 TCID 50 /ml), PR8 S338G replicated to intermediate levels (4.63 log 10 TCID 50 /ml), and PR8 S338GϩN6 (H1N6) replicated to levels almost comparable to those of the parental virus in the presence of trypsin (7.38 log 10 TCID 50 /ml). These results demonstrate that the N6 NA and the HA thrombin cleavage sites are necessary and sufficient for trypsin-independent growth in MDCK cells.
To further investigate whether the thrombin cleavage motif was also responsible for systemic infection, as observed in A/Mdk/6L/07-infected mice (Table 2), we infected mice with our panel of recombinant viruses with altered cleavage sites and examined their tissue distribution. As expected, wild-type A/Mdk/6L/07 was recovered from multiple organs, including the brain (1.2 log 10 EID 50 /0.1 g), spleen (1.2 log 10 EID 50 /0.1 g), heart (2.2 log 10 EID 50 /0.1 g), and lungs (3.5 log 10 EID 50 /0.1 g). None of the recom-binant A/Mdk/6L/07 viruses with altered cleavage sites were recovered from any of these organs except the heart; instead, these viruses were primarily detected in the lungs of infected mice (Table 3). In contrast, the PR8 S338G and PR8 S338GϩN6 recombinant viruses displayed broad tissue tropism that was expanded (i.e., brain involvement) compared to that of the parental virus. Collectively, these results demonstrate that the  HA thrombin cleavage motif operates in concert with N6 to permit expanded tissue tropism in infected mice. The mouse 50% lethal dose (LD 50 ) for each wild-type or recombinant virus is reported in Table S2, and the body weight and histopathological results for each virus are shown in Fig. S5 and S6. Impact of the HA subtype on trypsin-independent growth. Because the cleavage site and NA were important contributors to the trypsin-independent growth of A/Mdk/ 6L/07, we next wanted to determine the impact of the thrombin cleavage site on the growth of viruses with different HA subtypes. To do so, we created a panel of recombinant A/Mdk/6L/07-based viruses carrying a variety of HA proteins (H1 to H12), each of which was altered to contain the thrombin motif and the N6 NA. We made the same combinations in the A/PR/8/34 virus backbone (Table S4). Of these, the H4, H5, H6, H10, H11, and H12 subtypes were successfully rescued on both backbones and further assayed for growth in MDCK cells with or without exogenous trypsin.
The 6 A/Mdk/6L/07-based recombinant viruses replicated to substantial levels in the absence of trypsin (greater than 6.13 log 10 TCID 50 /ml), but their parental viruses exhibited basal levels of replication (lower than 2.47 log 10 TCID 50 /ml) in the absence of trypsin ( Fig. 4 and Table S4). Similarly, all of the A/PR/8/34-based recombinant viruses replicated to titers greater than 3.63 log 10 TCID 50 /ml in the absence of trypsin, although  Table S3. (B) Wild-type or recombinant viruses that contain a variety of HA GRG and N6 genes were generated by using the A/Mdk/6L/07 (H7N6) and PR8 viruses as a backbone. Viral titers were determined in the presence of trypsin. The source of each gene is listed in Table S3. ϩ, treated; Ϫ, untreated. *, P Ͻ 0.05; **, P Ͻ 0.01 (Student's t test). most of their parental viruses did not reach 2.55 log 10 TCID 50 /ml in the absence of trypsin (Fig. 4A). As expected, all of the tested viruses exhibited growth patterns that were similar to those of their parental viruses in the presence of trypsin (Fig. 4B).
Impact of protease inhibitors on virus growth. Thrombin-like proteases were implicated in the trypsin-independent growth of A/Mdk/6L/07 based solely on the analysis of the HA cleavage site sequence. Therefore, to provide more direct evidence for the contribution of thrombin-like proteases, we assessed the impact of select protease inhibitors on viral growth and HA cleavage patterns. Because aprotinin (a broad-spectrum serine protease inhibitor that inhibits trypsin, plasmin, papain, and cathepsin B but not thrombin) efficiently blocks A/WSN/33 replication in the absence of trypsin (19), we tested whether aprotinin or leupeptin (an inhibitor of trypsin, plasmin, and several other serine proteases) could attenuate the replication of A/Mdk/ 6L/07 and inhibit the cleavage of HA in the absence of trypsin. We also tested the impact of argatroban (a thrombin-specific inhibitor).
The replication of A/Mdk/6L/07 was not attenuated in the presence of aprotinin or leupeptin and exhibited growth kinetics similar to those of A/Mdk/6L/07 in the absence of trypsin (Fig. 5A). Argatroban, the thrombin-specific protease inhibitor, significantly attenuated the growth of A/Mdk/6L/07 (Fig. 5A). In contrast, the HA PKG338_QS recombinant virus grew to levels below the limit of detection when serine protease inhibitors or argatroban were added to the medium, and this virus exhibited normal growth kinetics only in the presence of trypsin (Fig. 5B).
These results demonstrate that the HA cleavage mechanism used by A/Mdk/6L/07 is associated with a thrombin-like protease but differs from that used by the A/WSN/33 virus, the growth of which is affected by broad-spectrum serine protease inhibitors.
Thrombin activation following A/Mdk/6L/07 infection. While we were able to show that thrombin-like proteases are important for A/Mdk/6L/07 growth, it was not clear if the virus activated the protein from its inactive precursor form. To assess this, we looked for the activation of thrombin-like proteases during A/Mdk/6L/07 replication in MDCK by Western blotting.
Wild-type A/Mdk/6L/07 and recombinant A/Mdk/6L/07 viruses containing the N3 from A/Ab/W44/05 or the HA and NA from A/PR/8/34 were used to infect MDCK cells in the absence of trypsin. Cleavage of A/Mdk/6L/07 HA 0 into HA 1 and HA 2 was observed at 24 and 48 h but was not seen for the recombinant H7N3 and H1N1 viruses (Fig. 6A). Using the same cell lysates, we next sought evidence of cleavage of the prothrombin complex to the active thrombin based on Western blotting with a commercial prothrombin antibody. A 30-kDa band, the expected size of thrombin, was evident in A/Mdk/6L/07-infected cultures but not in cultures of the recombinant viruses (Fig. 6B), suggesting prothrombin activation by the H7N6 virus. To investigate the effect of N6 protein on prothrombin activation, prothrombin protein was incubated with sucrosepurified H7N3 or H7N6 viruses in vitro at 37°C for 30 min, and then Western blotting was conducted with prothrombin antibody. Although phosphate-buffered saline (PBS) treatment induced only a basal level of thrombin activation, in vitro incubation with H7N6 virus significantly enhanced thrombin activation compared with H7N3 virus (Fig. S4). This result suggests that the N6 protein of H7N6 viruses could directly activate the thrombin by cleavage of the prothrombin during the replication. Argatroban, a thrombin-specific inhibitor, protects mice from lethal A/Mdk/ 6L/07 challenge. Because we determined that addition of argatroban was sufficient to block the replication of Mdk/Korea/6L/07 in cultured cells, we next sought to determine its impact in vivo. Groups of mice (n ϭ 10) were infected with A/PR/8/34 (H1N1), Ab/Korea/Ma44/05 (H7N3) (Ma44), or A/Mdk/6L/07 (H7N6) (6L) virus, and argatroban was administered as indicated (Fig. 7). All PBS control-treated mice died within 10 days, regardless of challenge virus used (Fig. 7A). In the low-dose (4.5 mg/kg) argatroban treatment groups, 6L (H7N6)-infected mice (40%) survived (Fig. 7B). In the high-dose  , and 135 g (C) in 100 l was administered intraperitoneally every day for (9 mg/kg) argatroban treatment groups, 9 of 10 6L (H7N6)-infected mice survived with less than 10% of body weight loss. However, no A/PR8/34 (H1N1) virus-infected mice survived, and only 3 of 10 Ma44 (H7N3)-infected mice survived (Fig. 7C).

DISCUSSION
Here, we report the discovery of a novel replication mechanism used by a naturally occurring LPAI virus, A/Mdk/6L/07, which was isolated from a wild bird. The mechanism relies on a specific NA subtype (N6) and a specific cleavage motif in HA (GRG), and it culminates in altered viral growth in vitro and expanded tissue tropism and virulence in vivo. Although A/Mdk/6L/07 and A/Ab/W44/05 share a Eurasian H7 HA ancestry (Fig. S3), only A/Mdk/6L/07 replicated in MDCK cells in the absence of trypsin and was recovered from multiple organs of infected birds and mice.
Other than the prototypic HPAI H5 and H7 viruses, only 2 human H1N1 viruses, WSN (H1N1) and the 1918 Spanish influenza virus, have been shown to replicate independently of trypsin and to disseminate to nontarget organs in mice; in the case of the H1N1 viruses, plasminogen was shown to be a critical factor (9,20). Post et al. reported that certain LPAI viruses, including the H5N2, H7N1, H7N7, and H9N2 subtypes, could systemically infect multiple tissues in infected chickens via an unknown mechanism, whereas the tissue distribution of LPAI viruses in ducks is typically limited to the lungs and to peripheral blood mononuclear cells (21,22). However, although the molecular properties of A/Mdk/6L/07 were characteristic of low-pathogenicity viruses, the virus was originally isolated from several organs (trachea, lung, spleen, and brain) of a dead mallard duck.
Infection studies in domestic ducks and chickens have revealed that A/Mdk/6L/07 can spread to several organs, including the heart, spleen, and kidney, in both bird species (Table 2). A/Mdk/6L/07 was also recovered from the brains and spleens of infected mice, whereas LPAI A/Ab/W44/05, a H7N3 virus, was recovered only from the lungs of infected mice, as previously described. We demonstrated that N6 is a required component of trypsin-independent growth and systemic spread and confirmed that this NA-dependent replication mechanism is common to Eurasian N6 lineages rather than being specific to the N6 gene of A/Mdk/6L/07.
The HA proteins of most influenza viruses are cleaved by activated proteases within a membrane-proximal surface loop at an R or K residue adjacent to the G residue that constitutes the N terminus of the newly generated and highly conserved fusion peptide of the HA 2 subunit. In low-pathogenicity viruses, the cleavage site commonly consists of a monobasic R/K residue, which may be cleaved by trypsin-like serine proteases for activation. An SAP analysis of recently discovered Eurasian H7 genes, including the avian H7N9 viruses that recently infected humans in China, revealed that the H7 influenza viruses contain highly conserved proline, lysine, and glycine residues at the P4, P3, and P2 positions of the cleavage site, respectively, with a conserved arginine at the P1 position that can be recognized and potentially cleaved by thrombin (GRG) (9,16). Thrombin cleavage motifs that connect peptides in the HA genes of avian influenza species are not commonly found. A comparison with published sequences, including sequences in the GenBank database, showed that the thrombin cleavage motif in HA is highly conserved in, and exclusive to, Eurasian lineages of H7 viruses, including A/Mdk/6L/07 and A/Ab/W44/05 isolates and recent human infectious H7N9 viruses (e.g., A/Anhui/1/2013) (see Appendix and Fig. S3A to C in the supplemental material). Therefore, we hypothesized that the specific HA cleavage motif (GRG) present in A/Mdk/6L/07 is associated with the trypsin-independent replication of the virus and with the N6 neuraminidase. Our recombinant PR/8/34 viruses bearing the modified GRG HA cleavage motif without (PR8 S338G [H1N1]) or with PR8 S338G-N6 (H1N6), the N6 gene segment of A/Mdk/6L/07, were used to confirm that the H1N6 virus had acquired the ability to replicate even in the absence of trypsin and that the PR8 S338G (H1N1) strain, which contained a modified HA cleavage motif alone, could not grow as well as the wild-type virus in the absence of trypsin (Fig. 3C and D).
We demonstrated that the dual modification of HA cleavage sites (GRG) along with the N6 NA subtype in various avian influenza A subtypes (H4N6, H5N6, H6N6, H10N6, H11N6, and H12N6) could alter their growth properties to make them trypsin independent in MDCK cells. These results strongly support the idea that the novel replication mechanism used by the A/Mdk/6L/07 virus is not specifically limited to the H7 subtype but could be a general mechanism inherent in many of the avian influenza A viruses. The monobasic cleavage motifs in the HA proteins of the A/WSN/33 and 1918 pandemic H1N1 viruses were originally shown to be cleaved by plasmin after the activation of serum plasminogen in MDBK cells (23), and the virulence properties of these viruses were subsequently linked to the neuraminidase (NA) gene (23,24). Although the monobasic motifs of H7 subtypes (GRG) have only one amino acid difference from that of A/WSN/33 (SRG), aprotinin and leupeptin (broad-spectrum serine protease inhibitors that efficiently block the replication of A/WSN/33 in the absence of trypsin [14]) failed to affect the replication of the A/Mdk/6L/07 virus in media lacking trypsin (Fig. 5B). However, treatment with argatroban (a direct thrombin inhibitor) significantly attenuated the replication of A/Mdk/6L/07 in vitro and in vivo ( Fig. 5 and 7). These results clearly demonstrate that both the GRG motif in Eurasian H7 viruses and the N6 neuraminidase play crucial roles in the trypsin-independent growth of LPAI H7N6; however, this trypsin-independent growth differs from that of the WSN/33 virus, which was affected by broad-spectrum serine protease inhibitors. Further in vitro studies revealed that direct incubation of prothrombin with purified H7N6, but not H7N3, virus significantly enhanced prothrombin cleavage (Fig. S4). Therefore, we hypothesized that the prothrombin activation by the N6 neuraminidase during replication could facilitate thrombin-dependent cleavage of the GRG motif leading to trypsin-independent replication and multiorgan spread in vivo. However, further mechanistic studies are needed to elucidate the details of protease activation by the N6 neuraminidase and, additionally, to further explore the inherent enzymatic activity of this group of proteins.
Since 2013, human infections by novel avian H7N9 virus strains in China are of increasing concern. Although the viruses still replicate mainly in respiratory organs, both H7N9 and H7N6 viruses can be maintained in poultry or migratory bird species. Because the HA proteins of human infectious H7N9 viruses cluster with Eurasian avian lineages and share the cleavage motif of A/Mdk/6L/07 (H7N6), we cannot exclude the emergence of novel H7 reassortants in nature, especially novel H7N6 viruses, which have the HA and internal genes of human infectious H7N9 viruses and the N6 segment of a Eurasian lineage virus. Moreover, domestic ducks and chickens might serve as an interface between the natural gene pool of avian influenza viruses in wild birds and land-based poultry. Therefore, continuous monitoring and systematic surveillance should be instituted to contain these viruses and to reduce their opportunities for further genetic evolution.
In summary, we found that the amino acids at the P4, P3, and P2 positions of the HA cleavage site, along with the N6-specific NA subtype, were related to the efficiency of HA cleavage, replication, and systemic distribution of H7N6 influenza virus in several animal species. We propose that the identity of the amino acids at the GRG cleavage position and the presence of N6 may be indicators of viruses that could pose a public health risk. Therefore, these findings might aid in the prediction and prevention of pandemics emerging at the human-avian interface.