Bat-derived influenza-like viruses H17N10 and H18N11

Highlights

Abstract

Influenza A and influenza A-like viruses

Influenza virus is a member of the Orthomyxoviridae family. There are three types of influenza virus based on its internal proteins of nucleoprotein and matrix protein, namely A, B, and C [1]. Among these, influenza A virus is the most prevalent pathogen for both humans and animals, causing the so-called seasonal flu. Influenza A virus was also the causative agent of four major pandemics, in other words the 1918 Spanish flu, 1957 Asian flu, 1968 Hong Kong flu, and the 2009 swine-origin pandemic flu (2009 pH1N1), as well as of the relatively milder pandemic, the 1977 Russian flu 2, 3, 4. For the origin, genesis and ecology of influenza viruses, migratory birds (shorebirds and waterfowls) are regarded as reservoir hosts, providing a large pool of virus gene segments that can contribute to novel reassortant viruses 5, 6, 7. Meanwhile, swine are believed to be ‘mixing vessels’ or at least intermediate hosts that influenza A viruses can utilize to ‘jump’ from poultry to humans [5].

Influenza A virus is an enveloped negative-stranded RNA virus, with a segmented genome of 8 pieces, which encode a total of 14 proteins [8]. There are three major envelope proteins embedded in the virus surface: hemagglutinin (HA), neuraminidase (NA), and matrix protein 2 (M2) [5]. HA is responsible for virus binding to susceptible cells, which harbor sialylated proteins (as virus receptors), eventually resulting in fusion to and entry into the cells. NA is a sialidase that enables mature viruses to be released from infected cells [5]. To date 16 HA subtypes and nine NA subtypes have been identified (Figure 1 ), and the different reassorted combinations are used to name the viruses, for example, the common seasonal flu viruses H1N1, H2N2, H3N2, and the sporadic human infections with H5N1 and the recent H7N9 9, 10, 11.

Open in a separate window

Figure 1

Phylogenetic trees of the hemagglutinin (HA) and neuraminidase (NA) genes of all known influenza A virus subtypes including H17/18 and NA-like N10/11. (A) The categorized HA molecules can be grouped into two groups, group 1 and group 2. The bat-derived H17 and H18 (each marked with a star) belong to group 1, displaying a typical sequence feature of group 1 HA molecules. (B) All the known NAs could also be grouped into two groups: group 1 and group 2. The bat-derived N10 and N11 (each marked with a star) do not belong to either group. They are NA homologs or could be called NA-like molecules. Here we propose that they be categorized as influenza A-like group 3.

In 2012, an astonishing story published in Proc. Natl. Acad. Sci. USA [12] caused some concerns because a new influenza virus genome (H17N10) was isolated from bats by next-generation sequencing (NGS). This raised a serious scientific and public health question as to the origin and evolution of influenza virus. If this genome can produce a real influenza virus that can cause human or animal infections, then the situation for influenza virus-caused diseases would be more complicated because bats are known to harbor many viruses and are regarded as a reservoir host for many human- and animal-infecting viruses, including the SARS (severe acute respiratory syndrome) coronavirus [13]. If this is the case, then our understanding of the influenza virus ecology will need to be rewritten. Therefore research on this NGS-identified novel genome is urgently needed for the sake of public health. Studies on the two surface envelope proteins, HA and NA (NA-like) 14, 15, 16, 17, demonstrate that neither protein has the corresponding canonical influenza virus functions or structures [18]. Therefore the new genome does not represent a ‘true’ influenza virus, and it should be renamed influenza-like virus, at most. Whether or not the genomic segments will reassort with canonical influenza A virus genomes should be vigorously tested in the near future. Recently a similar virus genome, H18N11, was again identified by NGS, and neither HA nor NA have canonical structures or functions [19].

In this review we summarize recent work on the functions and structures of both HA and NA derived from H17N10 and H18N11. In addition, all influenza proteins are examined using bioinformatics, and proteins from the six internal genes of the H17N10/H18N11 genomes are shown to conserve known structural and functional modules or domains. We also further discuss our recent work on the N-terminal domain of the polymerase subunit PA. We believe that the functions and structures of all the encoded proteins should be examined in detail to understand these influenza-like viruses better. Moreover, their reassortment potential should be assessed by reverse genetics experiments.

HAs H17 and H18 do not bind to the canonical sialic acid receptors

HA is the receptor binding protein of the influenza A virus, and is responsible for virus entry into host cells. Before the discovery of the H17 and H18 genes there were 16 subtypes of HA described, H1–H16. Based on their primary sequences, these HA molecules can be categorized into two groups (Figure 1): group 1 (H1, H2, H5, H6, H8, H9, H11, H12, H13, H16, H17, and H18) and group 2 (H3, H4, H7, H10, H14, and H15) 12, 19, 20, 21, 22. Bat-derived H17 and H18 should be placed into group 1 based on their primary sequences (Figure 1) 12, 16, 17, 19.

As expected from their primary sequences, the H17 and H18 structures resemble the structures of group 1 HAs rather than those of group 2. Previously solved HA structures demonstrate that there are group-specific features at sites where extensive conformational changes occur for HA activation, including the conformation of the interhelix loop and the rigid body orientation of the globular domain [22]. Taking H17 as an example, it displays a similar interhelix loop conformation to the HAs from group 1, and this is consistent with the phylogenetic analysis. Superimposition with other solved HA structures by means of the long central α-helices of HA2 revealed that the globular domains fall into three subgroups 16, 23: subgroup 1, including H1, H2, H5, and H9; subgroup 2, including H3, H7, and H14; and subgroup 3, consisting of H13, H16, H17 and H18. These differences may result from subtle variation in the interhelix loops among different HA subtypes and could signify different mechanisms during HA activation 16, 23.

Influenza A virus enters susceptible cells through endocytosis after binding to cell surface receptors 24, 25, 26. The receptors for influenza A viruses are sialic acids (SA) linked to cell surface glycolipids or glycoproteins 24, 25, 26. After entering the cells, the virus fuses with the endosomal membrane under low pH and subsequently the genetic material is released into the cell 22, 27, 28, 29, 30. Avian- and human-adapted influenza A viruses harbor distinct HA molecules that have different capacities to bind specifically linked SA-moieties. In Figure 2 , the configurations of the α-2,3-linked SA and α-2,6-linked SA are shown. The molecular basis of the interaction of these specific receptors with the virus-derived HA has been extensively studied 22, 27, 28, 29, 30. In the trimeric structure of the HA, resolved as early as in 1981 in an outstanding collaborative work by Wilson, Wiley, and Skehel [31], the HA head domain of each monomer was shown to be responsible for SA binding. We now know that there are three secondary elements and one base element involved directly in SA binding. The three secondary elements (the 130-loop, the 190-helix, and the 220-loop, in H3 numbering) form the edge portion, and four conserved residues (Y98, W153, H183, and Y195) form the base portion. These two portions usually form a shallow cavity to accommodate sialylated glycans. Typically, the SA moiety of the sialylated glycan forms several conserved hydrogen bonds with the 130-loop and the base residue Y98, and the remaining glycan moieties interact with the 220-loop or 190-helix.

Open in a separate window

Figure 2

Configurations of avian and human receptor analogs. (A) The avian receptor analog. The Sia-1 (sialic acid, SA) is linked to Gal-2 (galactose) via an α2,3-glycosidic bond. The glycans extend forward and the hydrophilic glycosidic oxygen atom is exposed to the receptor binding site. (B) The human receptor analog. The Sia-1 is linked to Gal-2 via an α2,6-glycosidic bond. The glycans fold back and the hydrophobic C6 atom is exposed to the receptor binding site. In both (A) and (B) N-acetylglucosamine is abbreviated as GlcNAc-3; black arrows indicate the orientations of the glycosidic linkage.

Substitution of the residues in the three secondary elements is important for the HA protein to obtain avian or human SA receptor preference. Different HA subtypes use different substitutions to achieve this goal. For H2 and H3 HAs, Q226L and G228S substitutions in the 220-loop are responsible for the switch between avian and human receptor binding specificities 30, 32 whereas, for H1 HA, different combinations of substitutions at residues 190 and 225 are important for the SA binding preference 33, 34. For H5 HA, a single Q226L substitution is enough to change the receptor binding preference 35, 36. By contrast, for H7 HAs, Q226L substitution is not solely responsible for the acquisition of human receptor binding, and other amino acid substitutions also contribute to the receptor binding switch, especially G186 V substitution 37, 38.

Multiple lines of evidence, including surface plasmon resonance (SPR) experiments, MDCK cell binding assays, and glycan microarray analysis, revealed that the bat-derived H17 and H18 do not bind to canonical human or avian receptors 16, 17, 19. This lack of canonical receptor binding is likely due to specific structural features in the putative receptor binding sites of H17 and H18 HA. In the H17 and H18 structures (Figure 3 ) there is no obvious cavity to accommodate the sialylated glycans, due to strong interactions among three secondary elements (130-loop, 190-helix, and 220-loop) through a hydrogen bond and salt bridge network formed by residues D136, Q190, H226, and D228 16, 19. Furthermore, the negatively charged D136 in the 130-loop (all canonical influenza HAs have an uncharged threonine or serine at this position) could result in a charge conflict with the negatively charged carboxyl group of SA, which is unfavorable for SA receptor binding 16, 19. Moreover, residue 98 (usually a conserved tyrosine) in the base of the receptor binding site is a phenylalanine in H17 and H18, and this could also affect SA receptor binding capacity. Thus, these five key residues likely contribute to the lack of SA receptor binding by H17 and H18 and make the putative binding cavity a much smaller, pseudo-binding site (a ‘closed’ site) (Figure 3).

Open in a separate window

Figure 3

Receptor-binding ‘head’ domains of all the hemagglutinins (HAs) with known structures (including influenza A and B) showing the SA binding sites. This figure clearly shows that H17 and H18 have a smaller SA-binding cavity – with two acidic amino acids (labeled in red) in comparison to the non-charged amino acids in all the other HA molecules (labeled in yellow). Note, the base of the receptor binding groove is labeled in green; both H17 and H18 show a ‘closed’ conformation, and that of H18 is even smaller.

The possibility remains that we may have not detected the binding of H17 and H18 to canonical human or avian receptors using the soluble protein in vitro because it is plausible that a stronger interaction may occur through receptor clustering in vivo. It is also possible that H17 and H18 may bind to canonical influenza receptors very weakly, below the level of our detection (which is a limitation of the available methods). However, together with the extensive amino acid changes in the receptor binding site of H17 and H18 protein, it is likely that the putative bat influenza virus has acquired a different receptor, possibly protein-based. There are many examples of closely related viruses that switch between protein and SA receptors, for example, paramyxoviruses. The most common type of the paramyxovirus attachment protein recognizes SA receptors. It is called hemagglutinin–neuraminidase (HN) and is found on viruses such as Newcastle disease virus (NDV) and human parainfluenza virus 3 (hPIV3), and recognizes SA receptors. The structures of the globular heads of HN proteins display a conserved β-sheet propeller motif, which was identified originally in influenza virus NA, and an SA binding site located in the central cavity of the proteins 39, 40. Unlike HN, the hemagglutinin (H) of measles virus (MV), which also belongs to the Paramyxoviridae family, possesses an inactivated SA receptor binding site and recognizes specific proteins, such as signal lymphocyte-activating molecule (SLAM), CD46, and nectin-4 41, 42, 43. The structure of the globular heads of H protein still reveals a conserved β-sheet propeller motif, and the specific protein receptors bind to the side part of H protein with different orientations [44]. In particular, the immunoglobulin (Ig)-like SLAM molecule binds to the H protein mainly through interactions between two β-sheets. Interestingly, the bat-derived influenza-like virus H17/H18 and N10/N11 proteins have similar Ig-like fold elements 14, 15, 16, 17, 19, which possibly provide the β-sheets that could interact with putative protein-based receptors. If the H17 and H18 protein should lose its trimerization state and expose its Ig-like fold element, it might be able to bind a specific protein receptor. In this case, the bat-derived influenza-like virus would abrogate the need for an active N10 and N11. This hypothesis or notion needs further experimental work to be confirmed or rejected. Thus, the bat-derived H17 and H18 are unique among the characterized HAs and might use a different entry mechanism.

NA-like molecules N10 and N11 are not sialidases

NA is the second virus-surface envelope protein and is a sialidase, responsible for cleavage of SA from glycans on the host cell surface to release the emerging progeny virus, prevent virus aggregation, and help virus migration 45, 46. Therefore it has an opposite function from HA, SA-binding versus SA-releasing. If bat-derived HA does not bind SA, consequently it would have no need for a sialidase to help release the virus particles. There are nine subtypes (N1–N9) of NA identified so far, before the discovery of the NA-like N10 or N11 genes from bat. Similarly to HA, NA has also been classified into two groups, group 1 and group 2, based on primary sequences [47] (Figure 1). N1, N4, N5, and N8 are group 1 members and N2, N3, N6, N7, and N6 belong to group 2. It is clearly shown from the phylogenetic tree (Figure 1) that N10 and N11 are genetically far away from NA molecules found on canonical influenza A viruses (N1–N9), thereby forming a unique cluster, which we propose to name group 3.

The structure of the NA head domain is a homotetramer, each monomer of which is composed of six topologically identical β-sheets arranged in a propeller formation [48]. With the exception of N7, the structures of all NAs (including the newly discovered bat-derived NA-like molecules, N10 and N11) have been solved 49, 50, 51, 52, 53 (Figure 2). Analysis of the active site of N1-N9 shows that eight residues (R118, D151, R152, R224, E276, R292, R371, and Y406, N2 numbering) which directly interact with the substrate, and 11 framework residues (E119, R156, W178, S179, D198, I222, E227, H274, E277, N294, and E425) which support the catalytic residues, are all highly conserved [14].

With the exception of the 2009 pandemic H1N1 virus N1, all members of group 1 NA have an additional cavity (150-cavity) adjacent to the active site [54]. Crystal structures reveal that the 150-cavity results from the open form of the 150-loop (formed by amino acids 147–152, N2 numbering). By contrast, the 150-loop of group 2 NAs display closed conformations which result in 150-cavity deficiency. The open form of the 150-loop has been considered as a new opportunity to develop antiviral compounds [49]. Moreover, D151 on the 150-loop was confirmed to be related to enzymatic activity. The D151G mutant appears to mediate the weak hydrolytic and enhanced SA binding activities [55].

Thanks to fast technology development, the crystal structures of NA-like molecules N10 and N11 have been immediately solved after the genome sequencing, and they maintain an overall tetrameric NA structure but with distinct features, especially in the deficiency of the enzymatic active site or 150-cavity 14, 15, 19 (Figure 4 ). In both structures the active sites are a more opened form with no clue as to how the SA is accommodated. In fact, when scrutinized the primary sequences of both N10 and N11 show diversified amino acid usages in both the active and the framework sites. Only three of eight active site (R118, R224, and E276) and three (four for N11) of 11 framework (S179, D198, E425, and W178-N11) amino acids are conserved. Therefore the structures do not support a neuraminidase function. Indeed, no sialidase enzymatic activities were detected in either N10 or N11 proteins when tested in vitro 14, 15, 19. Therefore they are not canonical NA proteins, and instead are NA-like molecules. N10 and N11 might function as a receptor binding protein via their Ig-like folds, as discussed in the previous section. Thus, alterations of the conserved NA active site architecture observed in both bat-derived N10 and N11, combined with evidence for a lack of sialidase activity, indicates that they may have an unknown function distinct from canonical influenza NAs.

Open in a separate window

Figure 4

Enzymatic active sites and the adjacent 150-loop structures of all the neuraminidases (NAs) with known structures (including influenza B virus) and NA-like N10 and N11. All the NA structures have been crystallographically resolved except for N7. Amino acids of the 150-loop are labeled in yellow and the substrate-binding groove is labeled in blue. Members from group 1 have a 150-cavity (except for N1 of 2009 pH1N1), which group 2 members otherwise lack. N10 and N11 also lack a 150-cavity, and have a more open structure which does not support SA binding. Under some circumstances, 09N1 can open (I223R mutation) and N2 can half-open (oseltamivir carboxylate-induced). After binding to the inhibitors, group 1 members usually close their 150-cavity (N5 is shown with oseltamivir bound).

Conserved functional modules in other putative encoded proteins

Until now there has been no evidence that these virus genomes can assemble into a real virus, or at least it might not be a viable virus in non-bat cells because attempts to propagate the virus in vitro have so far failed [12]. Owing to the distinct functional and structural properties of both H17/18 and N10/11, a comprehensive analysis of the functional modules was performed in the other encoded proteins in addition to the HA and NA. Figure 5 shows that the functional modules in other proteins are conserved (explicitly described in detail in the legend of Figure 5) but their real roles need to be experimentally tested in the future. Recent work on the N-terminal domain of the polymerase subunit PA paved the road for such a research [56].

Open in a separate window

Figure 5

Schematic of the 14 influenza A virus-encoded proteins showing the functional and structural modules or domains. All the known modules are indicated and are highly conserved in the genomes of both H17N10 and H18N11. Abbreviations: ‘?’, unknown functions; CT, cytoplasmic tail; MT, mitochondrial targeting sequence; NES, nuclear export signal; NLS, nuclear localization signal; PBS, promoter binding site; RdRp, RNA-dependent RNA polymerase; RNP, ribonucleoprotein; TM, transmembrane anchor. Amino acid numbering is based on all the protein sequences of the 1918 pandemic H1N1 virus [Genbank: {"type":"entrez-protein","attrs":{"text":"ABA55038.1","term_id":"76786705","term_text":"ABA55038.1"}}ABA55038.1(PB2), {"type":"entrez-protein","attrs":{"text":"ABA55039.1","term_id":"76786707","term_text":"ABA55039.1"}}ABA55039.1(PB1/PB1N40), {"type":"entrez-protein","attrs":{"text":"ABW36320.1","term_id":"158344925","term_text":"ABW36320.1"}}ABW36320.1(PB1-F2), {"type":"entrez-protein","attrs":{"text":"ABA55040.1","term_id":"76786709","term_text":"ABA55040.1"}}ABA55040.1(PA), {"type":"entrez-protein","attrs":{"text":"AGG82783.1","term_id":"459681224","term_text":"AGG82783.1"}}AGG82783.1(PA-X), {"type":"entrez-protein","attrs":{"text":"AAD17229.1","term_id":"4325039","term_text":"AAD17229.1"}}AAD17229.1(HA), {"type":"entrez-protein","attrs":{"text":"AAV48837.1","term_id":"55273941","term_text":"AAV48837.1"}}AAV48837.1(NP), {"type":"entrez-protein","attrs":{"text":"AAF77036.1","term_id":"8572169","term_text":"AAF77036.1"}}AAF77036.1(NA), {"type":"entrez-protein","attrs":{"text":"AAN06597.1","term_id":"23986295","term_text":"AAN06597.1"}}AAN06597.1(M1), {"type":"entrez-protein","attrs":{"text":"AAN06598.1","term_id":"23986296","term_text":"AAN06598.1"}}AAN06598.1(M2/M42), {"type":"entrez-protein","attrs":{"text":"AAK14368.1","term_id":"13173348","term_text":"AAK14368.1"}}AAK14368.1(NS1), {"type":"entrez-protein","attrs":{"text":"AAK14369.1","term_id":"13173349","term_text":"AAK14369.1"}}AAK14369.1(NS2)].

In influenza virus there are six further genes, in addition to HA and NA, which encode 12 proteins, with some newly identified protein products 8, 56, 57, 58, 59. These proteins include PA, PB1, PB2, PB1-N40, PB1-F2, PA-X, NP, M1, M2, M42, NS1, and NS2 (Figure 5). Two additional proteins of N-truncated PAs, PA-N155, and PA-N182, have also been reported [60]. M1 is the third virus-surface envelope protein and NP is responsible for viral RNA binding to form ribonucleoprotein (vRNP), which then combines with PB1, PB2, and PA to form the vRNP complex, a core apparatus responsible for proper virus replication [5]. Previous studies on influenza virus have shown that NP binds to viral RNA destined for transcription by PB1, that harbors a canonical RNA-dependent RNA polymerase domain, in conjunction with PB2, with cap-binding capacity, that together form the polymerase complex [61]. PB1 also binds directly to the C-terminal domain of PA 62, 63. The N-terminal domain of PA (PAn) functions as an endonuclease to generate small RNA primers that are essential for initiation of viral gene transcription 64, 65. This cleavage of host-derived RNA occurs in a sequence-specific manner 66, 67. Tong and coworkers demonstrated that the combination of NP, PB1, PB2, and PA derived from bat H17N10 is able to transcribe RNA in an influenza A mini-genome reporter assay, and this activity was abrogated when PB1 was removed from the complex [12].

In studying the N-terminal domain of PA from H17N10 [56], the relative distance was examined between H17N10 PAn and other PAn molecules of which the crystal structures have been solved {A/goose/Guangdong/1/96 (avian H5N1) [65], A/Victoria/3/1975 (human H3N2) [64], A/Vietnam/1203/2004 (avian H5N1) [68], and A/California/04/2009 (human pH1N1) [14]}. The sequence identity between PAn from the four influenza A viruses analyzed is 93–98%, whereas the PAn sequence identity of H17N10 with the four influenza A strains is only 71–72%. The sequence identity between PAn from influenza B and PAn from the four influenza A viruses analyzed is much lower (31–34%), and is in the same range as the identity with H17N10 PAn (35%). PAn from influenza C is at a greater distance, sharing 22–28% identity with the PAn from the four influenza A viruses and 21% with H17N10 PAn. Clearly, based on this comparison, H17N10 PAn is evolutionarily distant from the PAn domains of other viruses used in this comparison.

The crystal structure of H17N10 PAn was solved and was shown to be highly similar to PAn domains of other influenza viruses, folding into five α-helices and seven β-strands [56]. The catalytically conserved residues, including H41, E80, D108, E119, and K134, are ordered in close proximity such that they can accommodate a manganese ion in the center. As expected from the structure, H17N10 PAn possesses manganese-dependent endonuclease activity on both single-stranded DNA and RNA. When the endonuclease activity of H17N10 PAn was compared to that of avian (H5N1) and human (pH1N1) origins, no obvious differences were observed. Similarly to H5N1 PAn [65], the activity of H17N10 PAn is lost by substituting the key active histidine residue at position 41 for an alanine. These observations demonstrate the functional competence of H17N10 PAn as an influenza viral endonuclease, indicating that H17N10 possesses at least one protein domain that suggests an influenza-like life cycle. More experimental work on the other protein molecules or their functional modules should be done to reveal the complete life cycle.

Concluding remarks

Bat-derived influenza A viruses, both H17N10 and H18N11, may not follow the canonical rule for their life cycle as other influenza A viruses, at least based on the recent studies of HA and NA proteins. So far we have not isolated the virus from bats, and we suspect that this genome or virus should be, at most, called influenza-like virus or even a completely new name for the sake of public health concerns in view of the wide public knowledge about potential influenza pandemics. However, it is possible that the determined genomic sequences do not correspond to a functional virus because they may contain mutations originating from the non-functional viruses present in the bats. More work therefore needs to be done, such as the isolation of the virus or more sequences to be determined from the bats. Otherwise, the reassortment ability of the six internal gene segments of either the H17N10 or the H18N11 genome should be examined soon. If reassortment ever occurred, though doubtful, our understanding of influenza virus ecology would be changed and the range of reservoir and ‘mixing vessel’ hosts for influenza virus would need to be expanded. Furthermore, we are also expecting more HxNy viruses to be discovered from bats in the future (Box 1 ).

Acknowledgments

References