key: cord-1046190-z1ftfsu0
authors: Song, Hao; Qi, Jianxun; Khedri, Zahra; Diaz, Sandra; Yu, Hai; Chen, Xi; Varki, Ajit; Shi, Yi; Gao, George F.
title: An Open Receptor-Binding Cavity of Hemagglutinin-Esterase-Fusion Glycoprotein from Newly-Identified Influenza D Virus: Basis for Its Broad Cell Tropism
date: 2016-01-27
journal: PLoS Pathog
DOI: 10.1371/journal.ppat.1005411
sha: 9800de42f4f105fbd335179bef0fd6a57036d5ed
doc_id: 1046190
cord_uid: z1ftfsu0

Influenza viruses cause seasonal flu each year and pandemics or epidemic sporadically, posing a major threat to public health. Recently, a new influenza D virus (IDV) was isolated from pigs and cattle. Here, we reveal that the IDV utilizes 9-O-acetylated sialic acids as its receptor for virus entry. Then, we determined the crystal structures of hemagglutinin-esterase-fusion glycoprotein (HEF) of IDV both in its free form and in complex with the receptor and enzymatic substrate analogs. The IDV HEF shows an extremely similar structural fold as the human-infecting influenza C virus (ICV) HEF. However, IDV HEF has an open receptor-binding cavity to accommodate diverse extended glycan moieties. This structural difference provides an explanation for the phenomenon that the IDV has a broad cell tropism. As IDV HEF is structurally and functionally similar to ICV HEF, our findings highlight the potential threat of the virus to public health.

Influenza viruses are enveloped, segmented, single-stranded, negative-sense RNA viruses and belong to the family Orthomyxoviridae [1] . The genomes of influenza A virus (IAV) and influenza B virus (IBV) consist of eight RNA segments, whereas influenza C viruses (ICV) only have seven segments. Both IAV and IBV contain two major surface glycoproteins: the hemagglutinin (HA), which binds to sialylated host cell receptors and mediates membrane fusion; and the neuraminidase (NA), which destroys the receptor by cleaving sialic acid from host cell membranes, thereby releasing newly assembled virus particles [1] , and likely assisting initial invasion by destroying sialylated mucin decoys [2] . ICV, however, has only one major surface glycoprotein, the hemagglutinin-esterase-fusion (HEF) protein, which possesses all-in-one of receptor binding, receptor destroying and membrane fusion activities [3, 4] . While IAV infects avian, human, swine, and many other mammalian species including dogs, horses, tigers and seals, IBV and ICV are found principally in humans and rarely infect other species [3] .

ICV usually causes mild upper respiratory tract infections in children with cough, rhinitis and rhinorrhea as clinical symptoms [5, 6] . The virus only occasionally spreads to the lower respiratory tract and causes bronchitis, bronchiectasie and broncho-pneumonia [7] . Encephalopathy has also been occasionally reported [8] . Seroepidemiological studies have revealed that ICV is widely distributed globally and that the majority of humans acquire antibodies against the virus early in life [9, 10] . Aside from humans, there is evidence that ICV possesses the ability to infect animals [3] . Serological studies showed that antibodies against ICV are present in pigs [11] [12] [13] . In 1981, fifteen strains of ICV were isolated from domestic pigs in China [14] , which showed characters highly related to viruses isolated from humans in Japan [15, 16] . Furthermore, pigs have been shown to be susceptible to experimental infection with both pig and human ICVs, and the virus is able to be transmitted from the infected to uninfected contact pigs [14] , suggesting that interspecies transmission of ICV between humans and pigs might occur in nature. Dogs may also serve as a natural reservoir for human ICV due to the presence of viral replication and clinical symptoms in experimental infections and the prevalence of antibody to ICV among dogs [12, [17] [18] [19] .

In 2011, an influenza C-like virus was isolated from swine in Oklahoma (D/swine/Oklahoma/1334/ 2011 [D/OK]) exhibiting influenza-like symptoms [20] . The genome of this virus also contains seven segments, and sequence analysis showed approximately 50% overall amino acid homology to either human or previous swine ICVs. D/OK did not cross-react with antibodies against human ICVs and, importantly, was unable to reassort with human ICVs or generate viable progeny [20] [21] [22] . However, the low seroprevalence rate observed in both swine and humans to D/OK (9.5% and 1.3%, respectively) suggested that swine and humans are not likely to be a major reservoir of this novel virus [20] . Subsequent serological studies have showed that antibodies against D/OK are almost ubiquitously present in cattle, and several novel D/OK-like virus strains have been isolated from cattle with respiratory disease which could be divided into two distinct lineages represented by D/OK and D/bovine/Oklahoma/660/2013 (D/660) [21, 23] . These two genetic and antigenic distinct clades have been shown to reassort with each other [23] . In addition, D/OK has a broader cell tropism than human ICV and is capable of infecting ferrets, pigs and guinea pigs and transmit to naive animals by direct contact [20, 24] . Based on these differences to ICV it was suggested that this virus warrants classification as a new genus of influenza virus, named influenza D virus (IDV) with cattle as the potential reservoir [21] . Subsequently, more IDVs or viral genomic segments were identified from China and France in cattle, suggesting the wide geographic distribution of IDV [25, 26] . Interestingly, IDV is common in clinical samples of bovine respiratory disease complex (BRDC), which is the leading cause of morbidity and mortality in feedlot cattle [23, 27] . BRDC is a challenging multi-factorial disease caused by viral, bacterial pathogens and environmental factors, leading to severe clinical signs and deaths [28] . IDV was detected in clinical BRDC samples, co-infected with bovine coronavirus (BCV), bovine viral diarrhea virus (BVDV), bovine respiratory syncytial virus (BRSV), bovine herpesvirus 1 (BHV-1), and Pasteurella multocida, Mannheimia haemolytica, Histophilus somni et al, suggesting IDV has the pathogenic potential in BRDC [23] . In addition, a latest serological study showed that antibodies to IDV were present in sheep and goats in United States, suggesting that small ruminants are also susceptible to IDV infection [29] .

To further evaluate the infectivity and transmissibility of IDV, we expressed and purified the ectodomain of D/OK HEF, and determined that it also uses 9-O-Acetyl-Sia as its receptor by glycan microarray. We also solved the crystal structure of D/OK HEF, both in its native state (resolution of 2.4 Å) and in complex with its receptor analogue (resolution of 3.1 Å), and the structure of the enzymatically inactive HEF (resolution of 2.4 Å) alone and in complex with two receptor analogs respectively (both resolutions of 2.2 Å). Indeed, our results show that IDV HEF is functionally and structurally similar to ICV HEF, but with some distinct characteristics.

The ectodomain of HEF from D/OK strain was cloned and expressed using a baculovirus expression system based on a previously reported method [30] [31] [32] [33] with slight modifications. To avoid the enzymatic cleavage of the receptor substrates, we generated the catalytic mutant for the binding experiments and for the structures of the relevant complexes. Previous studies have demonstrated that the residues S57, D356 and H359 create a catalytic triad in ICV HEF esterase, which has been proved by site-directed mutagenesis and structural analysis [34, 35] , and the sequence alignment between the ICV and IDV HEF proteins reveals that a highly conserved catalytic triad is also observed in IDV HEF protein (S2 Fig) . Thus, we designed the enzymatically inactive HEF protein (HEF-mut) containing S57A, D356A and H359A substitutions, and expressed it using the same method as wild type protein. Soluble protein was purified by metal affinity chromatography followed by ion-exchange and gel filtration chromatography. The proteins were tested in a series of assays to determine their biological functions. Largescale glycan microarray analysis with 610 different glycans was used to investigate the receptor binding properties of the HEF-mut protein. The result revealed that the HEF-mut protein only binds robustly to 9-O-Ac-Sia glycan derivatives, with different relative fluorescence units (RFU) (Fig 1A, S1 Table) . The structural formulas of the top four binding glycans are shown in Fig 1A. In order to further characterize the binding properties of IDV HEF, we chose to use a more extended array with broader and paired (Neu5Ac-or Neu5Gc-based, and α2-3 or α2-6-linked) 9-O-Ac and non-O-Ac-sialoglycans [36] . The result, summarized in Fig 1B and S2 Table, indicate that IDV HEF-mut bind both α2-3 and α2-6-linked 9-O-Ac-Sias. Another interesting finding was the IDV HEF-mut can tolerate differentially modifications at C5, not only bind to 5-N-Ac-Sias, but also 5-N-Gc-Sias ( Fig 1B) . We also tried to perform the glycan array analysis of ICV HEF-mut protein, and to our surprise, the ICV HEF-mut protein did not bind to the glycan array with synthetic short glycans at all, which might be due to the lower binding affinity, and/or relative instability of the mutated protein.

In addition, both the IDV and ICV HEF-mut proteins, but not HEF, display the hemagglutination activities (Fig 2) and specifically binding activities to the Madin-Darby canine kidney (MDCK) cells and bovine submaxillary mucins (BSM) which are enriched in 9-Ac Sias (Fig 3A  and 3B ). More importantly, the IDV HEF-mut displays much stronger binding capacity than the ICV HEF-mut, which is compatible with our glycan array analysis.

IDV HEF has esterase enzyme activity HEF protein was further tested for its enzymatic activity using two substrates, BSM and pnitrophenyl acetate (pNPA). Firstly, we found that the receptors in BSM could be destroyed by treatment with HEF protein (Fig 4A) . Secondly, we performed an enzyme kinetics assay using pNPA as a substrate at three different temperatures (37°C, 25°C and 4°C). Both IDV and ICV HEF displayed obvious esterase activities while HEF-mut proteins exhibited no enzymatic displayed a good avidity to Neu5,9Ac 2 ligands. The structure formulas of the top representative are shown. Error bars represent the standard deviation (SD) of the mean value. Percent coefficient of variation (%CV) = 100×standard deviation/mean. A %CV of less than 50% indicates data reliability. (B) Glycan microarray analysis of IDV HEF-mut proteins using a more extended paired array. The library of glycans tested was designed to test the influence on protein binding of (1) glycosidic linkage of 9-O-Ac-Sia (α2-3 versus α2-6) and (2) modifications at C5 (N-acetylation, 5-N-Ac, versus N-glycosylation, 5-N-Gc). Data presented as mean ± SD. doi:10.1371/journal.ppat.1005411.g001 Hemagglutination assay of the IDV and ICV HEF proteins at 25°C or 4°C. Receptor binding activity of HEF protein was assessed by hemagglutination assay with chicken erythrocytes. Serial dilutions of purified HEF proteins (100 μg/mL to 0.1 μg/mL per well) were mixed with washed chicken erythrocytes and incubated to analyze the receptor binding and cross-linking of chicken erythrocytes at two different temperatures 25°C (A) or 4°C (B). The H5 HA protein was used as positive control. Positive hemagglutination results formed a uniform reddish color across the well, whereas negative results appeared as dots in the center of roundbottomed plates due to erythrocytes sedimentation. activities (Fig 4B-4D ). With the decrease of temperature, the esterase activities of both IDV and ICV HEF decreased as well. However, both IDV and ICV HEF still retain noticeable esterase activities even at 4°C (Fig 4D) . This may explain why wild type HEF proteins do not show either hemagglutination activity or receptor binding activities clearly (Figs 2 and 3) , as the receptors are being destroyed by their esterase activities, even at 4°C.

To examine the host tissue tropisms of IDV and ICV, we used soluble recombinant HEF-mut proteins of these two viruses to stain paraffinized human, swine or bovine trachea sections. Interestingly, the apical surfaces of the human, swine and bovine trachea showed positive staining with ICV HEF-mut and IDV HEF-mut (Fig 5) , indicating the apical surfaces of the trachea of these three species all exhibit the receptor 9-O-Ac-Sia. Moreover, the tracheas of swine and bovine display brighter staining than that of the human trachea ( Fig 5) .

In order to study the molecular basis of IDV HEF and its receptor / substrate binding mode we solved the native X-ray crystallography structures of wild type and mutant HEFs both at a resolution of 2.4 Å ( Table 1) . A Dali search within the Protein Data Bank (PDB) revealed that the IDV HEF structure most resembles the ICV HEF structure which has two subunits HEF1 and HEF2 (PDB code, 1FLC; Z score, 53.5 for HEF1 and 9.0 for HEF2) [35] . Following the initial domain nomenclature used in the description of the structure of ICV HEF, we divided the IDV HEF structure into three domains: receptor binding domain (R), esterase domain (consisting of E1, E' and E2 subdomains) and fusion domain (consisting of F1, F2 and F3 subdomains). Both the overall structure ( Fig 6A) and the structural folds of individual subdomains (Fig 6B) of IDV HEF and ICV HEF are remarkably similar. The E domains are the most conserved regions and display root mean square differences (RMSDs) on main chain C atoms at 0.396 Å, 0.390 Å and 0.435 Å for E1, E' and E2, respectively, with corresponding sequence identities of 66.7%, 68.8% and 56.6%, respectively. A comparison of R domains reveals a RMSD of 0.642 Å and a sequence identity of 46.3%. The F1 and F2 subdomains share 42.1% and 41.2% sequence identities and display RMSDs of 0.567 Å and 0.817 Å, respectively. Notably, the F3 subdomain shares 56.8% sequence identity but the RMSD reaches 1.445 Å. The F3 subdomain contains the fusion peptide, which is important for the viral membrane fusion.

To understand the molecular basis of IDV HEF receptor binding, the HEF-mut crystals] were soaked with glycans 9-O-Ac-3'Sle C (Tr323, Neu5,9Ac 2 α2-3Galβ1-3GlcNAcβ-Sp0) and 9-O-Ac-3'SLN (Tr322, Neu5,9Ac 2 α2-3Gal β1-4GlcNAcβ-Sp0) [(kindly provided by the Consortium for Functional Glycomics (CFG, Scripps Research Institutes, Department of Molecular Biology, La Jolla, CA) to determine the structures of the respective receptor complexes, which were resolved at a resolution of 2.2 Å (Table 1 ). Similar to ICV HEF, the receptor-binding site of IDV HEF is located near the top of the HEF1 globular head in a shallow cavity, surrounded by residues from four secondary structure elements: the 170-loop, 190-loop, 230-helix and For the IDV HEF-mut/receptor complex structures, interpretable electron density is observed for all the three glycan rings of 9-O-Ac-3'Sle C , including Neu5,9Ac 2 (9-O-Ac-Sia-1), galactose-2 (Gal-2), and N-acetylglucosamine-3 (GlcNAc-3); while for 9-O-Ac-3'SLN, electron density for the first two sugars is well defined (Figs 7 and S1). The glycan rings of 9-O-Ac-3'Sle C go through the open channel between the 230-helix and 270-loop ( Fig 7F) . The open channel in the receptor-binding cavity of IDV could provide a structural basis for accommodating different receptors with diverse glycan rings in different cell types, which should be tested in the future.

Compared with the ICV HEF receptor complex, the 9-O-Ac-Sia-1 of the receptor bound to IDV HEF displays a very similar orientation (Figs 7 and 8) [35] . The 9-O-acetyl group of the receptor docks into a nonpolar, hydrophobic pocket, formed by V275, F229, F240 and F297 (Fig 8) . The acetyl carbonyl oxygen forms a hydrogen bond with the hydroxyl group of Y231 (ICV HEF: Y227). Y231 is highly conserved both in ICV and IDV HEF (See alignment in S2  Fig) . Importantly, the conserved amino acid Y127 (Y98 in IAV HA, H3 numbering) both in ICV HEF and IAV HA has been changed to F127 in IDV HEF, which is the same as IBV HA [37] . The absence of two hydrogen-bonding interactions between hydroxyl group of Y127 and the 8-hydroxyl and 9-amide of the ligand, pushes the 9-O-acetyl moiety of 9-O-Ac-sia-1 to the other side by~1.1 Å. In exchange, two more hydrogen-bonding interactions between the carbonyl group of C4 and T171 mediated by water stabilize the sugar ring conformation and prevent its excessive shift and rotation. The 5-N-acetyl group of the ligand fits into a hydrophobic pocket formed mainly by W185, and forms a hydrogen-bonding interaction with A172. In addition, the carboxyl group of C1 formed two hydrogen-bonding interactions with S173.

Both 9-O-Ac-3'Sle C and 9-O-Ac-3'SLN bind in a cis conformation (Figs 7 and 8) , similar to the IAV avian SH-H7N9 HA binding with avian-like (α2-3) receptor structures [38] . But the majority of the interactions between ligands and IDV HEF are made by the 9-O-Ac-Sia-1 moiety, whereas the glycan portion of the ligands do not make significant contacts with IDV HEF, except for a hydrogen bond between the Gal-2 of 9-O-Ac-3'SLN and S173. 

The E domain of IDV HEF, harboring the receptor-destroying enzyme (RDE) activity, has a hydrolase fold that is highly similar to that of ICV HEF. The active-site architecture of the HEF sialate-9-O-acetylesterase is fully conserved in ICV and IDV HEF with S57, D356 and H359 creating a catalytic triad and with the side chain of N117 and the NH groups of S57 and G85 forming an oxyanion hole (Fig 9A) [35] . The pocket is extremely conserved not only in IDV HEF and ICV HEF, but also in some nidovirus hemagglutinin-esterase (HE) proteins such as bovine coronavirus (BCoV) HE, porcine torovirus (PToV) HE and bovine torovirus (BToV) HE [39] . Phylogenetic analysis also shows the relationship between HEF and HE to be much closer than the relationship between HEF and HA or hemagglutinin-neuraminidase (HN) proteins (Fig 10) , implying common ancestral origins.

Serine 57 is the active-site amino acid of the acetylesterase and is modified by arsenic in the structure Due to the presence of cacodylate buffer in the crystallization conditions, S57 in our IDV HEF structure is covalently modified by the addition of a dimethylarsenic group. The covalent modification of active site serine by arsenic was first reported by Ian Wilson's group in 2003 [41] , when they solved the structure of an acetyl esterase, HerE, and its complex with the inhibitor dimethylarsinic acid, and illustrated the mechanism of the broad scope inhibition of serine hydrolases by As(V)-containing organic compounds [41] . The electron density of the covalent modification of S57 is shown in Fig 9B. This modification is only observed in the wild type HEF structure, while in HEF-mut structure, there is no such electron density due to a S57A mutation. Another serine esterase inhibitor diisopropyl fluorophosphate (DFP) is known to bind covalently to the serine in the active site of serine esterases. DFP-treated ICV bound specifically and irreversibly to cells expressing 9-O-Ac-Sias. This provided a probe for detecting 9-O-Ac-Sias [4] and S57 was demonstrated as the active-site serine in the acetylesterase of ICV [42] . DFP has also been used to determine the active site of enzymes by solving the complex crystal structure [43, 44] . In our IDV HEF complex structure with its receptor analogue 9-N-Ac-Sia, we found that the 9-N-acetyl group of the substrate was tilted up against the oxyanion hole as the covalent modification of S57 by arsenic blocked the insertion of the 9-N-acetyl group inserting into enzymatic site (Fig 9D) . An exposed fusion peptide

In our IDV HEF structure, IDV HEF is proteolytically cleaved into the subunits HEF1 and HEF2. For ICV, the HEF protein contain a monobasic cleavage site located in the stem region of the trimeric spike and can be cleaved by protease hydrolyses between R432 and I433 in to HEF1 and HEF2 [35, 45] . This monobasic cleavage site was very conserved both in ICV HEF (R432) and IDV HEF (R439), and the first eight residues of fusion loop were conserved between ICV and IDV HEF sequences too (S2 Fig). To confirm the proteolytic processing, we isolated IDV HEF crystals and checked for HEF2 band both by SDS-PAGE and western blot with an antibody recognizing the hexa-His tag engineered at the C terminus (S3A Fig) . Furthermore, N-terminal amino acid sequencing of the HEF2 band revealed that the first five amino acids of HEF2 were IFGID (S3B-S3F Fig) , the same with ICV HEF. In all solved cleaved influenza A and B HA structures, the N-terminal HA2 fusion peptide inserts into an electronegative cavity composed of different HA protomers, except in the bat-derived H17 and H18, which display an exposed fusion peptide [46, 47] . The exposed fusion peptide has also been observed previously in the HEF protein of ICV [35] (Fig 11A) and in the cleavage site of IDV HEF structure (Fig 11B) . Unambiguous electron density was seen from the ninth residue (F9) of the fusion peptide (S4 Fig). Although we cannot observe the first eight residues in the structure, different orientations of F9 in the HEF structure and other known cleaved HA structures helped to confirm that the fusion peptide does not insert into the cavity.

Since the first ICV strain (C/Taylor/1233/47) was isolated in 1947 during an epidemic of respiratory illness, hundreds of viruses have been isolated in clinical specimens [48] [49] [50] , due to its inconspicuous or mild symptoms and lack of suitable cell lines for virus isolation [51] .

Recently, a dozen of novel IDV strains have been isolated from both pigs and cattle, strains that are distantly related to human ICV [20, 21, 23, 25, 26, 52] . Ferrets, guinea pigs and small ruminants (sheep and goats) have also been found to be susceptible to the virus [20, 24, 29] . Clearly, the viral traits necessary for host switching are different between ICV and IDV.

Here, we solved the structure of IDV HEF alone and in complex with its receptor analogs. We found that the overall structure of IDV is extremely similar to that of ICV, despite sharing a relatively low sequence identity of only 53% [20] . This finding is similar to the structural similarity observed between HAs of IAV and IBV [37] . The IDV HEF structure was modeled by Modeller 9.10 using the ICV HEF as template in a previous study [20] , and they had an overall similar structure with ICV HEF. However, the details of interactions between the ligands and receptor-binding site and substrate binding site were not predicted precisely. Notably, we found that the receptor-binding site of IDV HEF occurs in an open channel between 230-helix and 270-loop. By contrast, in the receptor-binding site of ICV HEF the K235 of 230-helix and D269 of 270-loop form a salt bridge interaction, pulling the 270-loop up to connect with 230-helix and close the channel. The open channel in the receptor-binding site of IDV HEF could provide the space to accommodate an array of glycan linkages found in diverse host receptors which might explain why IDV has a broad cell tropism. Although the receptor specificity and adaptability of viral surface protein is one crucial determinant for host jump, other viral proteins may also be restrictive barriers to viral host range. PA, PB1 and PB2, which comprise the RNA-dependent RNA polymerase complex have long been implicated in playing a crucial role in determining host tropism [53, 54] . Furthermore, innate immune responses, intracellular factors and cross-species contacts may also affect host adaptation [55] . Therefore, a better understanding of the ecological, evolutionary and molecular mechanisms of IDV is essential in order to explain the broader host and cellular range of IDV and accurately assess the risk of transmission to other host species.

We determined that IDV HEF uses the glycan derivatives, 9-O-Ac-Sia as its receptor. We show that both ICV and IDV HEF proteins can bind to the trachea of human, swine and bovine. Considering the ability of IDV to transmit in ferrets and guinea pigs, and its pathogenicity in pigs and cattle, its public health threat for transmission to human must be monitored. A number of studies have examined occupational risk factors for zoonotic influenza virus infections, including open bird market workers, swine workers, meat processing workers, veterinarians and poultry workers, concluding that these populations are indeed at greater risk of infection with zoonotic IAV [56, 57] . Therefore, the surveillance of IDV infection in animal farm workers with influenza-like illness must be performed. Moreover, given the major economic importance of cattle and swine, further research into the pathobiology of IDV in these hosts, especially putative role in BRDC, needs to be conducted.

Another important phenomenon observed earlier is that IDV is unable to reassort with the ICV [20, 21] . As the only glycoprotein on the virus surface, it was suggested that the HEF protein may affect virus reassortment by incompatible protein functions in ICV and IDV [20, 21] . However, herein we have described the structure of IDV HEF and found it to be highly structurally and functionally similar to ICV HEF. Previous work has shown that there exist two discrepancies at the extremely conserved non-coding regions of seven RNA segments (the first twelve nucleotides at each 3' end as well as the last eleven nucleotides at each 5' end) between ICV and IDV genomes [20] . In addition, the non-conserved non-coding regions adjacent to each coding region are significantly variable [20] . Therefore, we propose that the packaging signal compatibility may be the important factor for the heterotypic incompatibilities between ICV and IDV.

The esterase pocket of IDV HEF is highly conserved among ICV and nidovirus HE proteins and could represent a potential drug target for developing broad-spectrum inhibitors. Therefore, rational modification of the substrate analog based on our structures may provide a potential route for the development of novel therapeutics against both orthomyxovirus HEFs and nidovirus HEs.

In conclusion, our functional and structural approach for the IDV surface protein HEF clearly reveals the virus' similarities and distinctions in comparison to the ICV and that the virus could be a potential concern for public health as the viral HEF can bind human trachea epithelia.

Paraffin-embedded normal human tracheal tissue sections were purchased from Auragene Bioscience (China). Formalin-fixed normal swine (5 months old domestic pig) and bovine (8 months old cattle) tracheal tissues were obtained from Zhongmu institutes of China animal husbandry industry with approval.

Madin-Darby canine kidney (MDCK) cells (NBL2, obtained from cell resource center of Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences) were cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco) supplemented with 10% fetal bovine serum in a humidified chamber containing 5% CO 2 at 37°C.

The gene of the HEF from D/OK strain (NCBI accession no. JQ922308) encoding the ectodomains (amino acid residues 3-605 after deletion of the signal peptide) was cloned into the baculovirus transfer vector pFastBac1 (Invitrogen) in-frame with an N-terminal gp67 signal peptide for secretion and a His 6 -tag at the C terminus for purification [31, 32] . To allow expression of an enzymatically inactive HEF protein (HEF-mut), the codons for the esterase catalytic residues S57, D356 and H359 were all substituted by Ala by site directed mutagenesis using the overlap extension PCR method and inserted to pFastBac1 in the same way. Recombinant pFastBac1 plasmid was used to transform DH10Bac Escherichia coli (Invitrogen). Transfection and virus amplification were performed according to the Bac-to-Bac baculovirus expression system manual (Invitrogen) [30, 33, 58] . HEF proteins were produced by infecting suspension cultures of Hi5 cells (Invitrogen) for 2 days. Soluble HEFs were recovered from cell supernatants by metal affinity chromatography using a HisTrap HP 5 ml column (GE Healthcare), then purified by ion-exchange chromatography using a RESOURCE Q 6 ml column (GE Healthcare). For crystallization, the proteins were further purified by gel filtration chromatography using a Superdex 200 10/300 GL column (GE Healthcare) with a running buffer of 20 mM Tris-HCl and 150 mM NaCl (pH 8.0), and the collected protein fractions were concentrated to 10 mg/mL using a membrane concentrator with a molecular weight cutoff of 10 kDa (Millipore). Both the wild type and enzymatically inactive (with S57A, D356A and H359A mutations) ICV HEF (C/Johannesburg/1/66) protein (NCBI accession no.AM410041, amino acid residues 1-597 after deletion of the signal peptide) were expressed and purified in the same as that of HEF-D/OK. HA of A/Anhui/1/2005 (H5N1) was prepared as described in our previous report [38] .

The initial screening trials were set up with commercial crystallization kits (Molecular Dimensions) using the sitting drop vapor diffusion method. Normally, 1 μL protein was mixed with 1 μL reservoir solution. The resultant drop was then sealed, equilibrating against 100 μL reservoir solution at 4 or 18°C. After optimization and seeding, diffractable crystals were obtained in a reservoir solution of 0.1 M PCTP (Propionic acid, Cacodylate, Bis-tris propane system) buffer pH 8.5, 22.5% w/v PEG 1500 for both HEF and HEF-mut protein at 4°C. For receptor complexes, HEF-mut crystals were soaked in a reservoir solution containing 10 mM 9-O-Ac-3'Sle C or 9-O-Ac-3'SLN at 4°C for 5 hr. For receptor analog complexes, HEF crystals were soaked in a reservoir solution containing 8 mM N -Acetyl-9-(acetylamino)-9-deoxyneuraminic Acid (or 9-N-Ac-Sia, TRC, Canada) at 4°C for 5 hr. All crystals were flash-cooled in liquid nitrogen after a brief soaking in reservoir solution with the addition of 17% w/v PEG 1500. The X-ray diffraction data were collected at Shanghai Synchrotron Radiation Facility (SSRF) beamline 17U, with a wavelength of 1.000 angstrom, at a temperature of 100K. All data were processed with HKL2000 software [59] .

The HEF structures were solved by the molecular replacement (MR) method using Phaser [60] from the CCP4 program suite [61] , with the structure of human ICV HEF (PDB: 1FLC) as the search model. Model building and refinement were performed using the COOT [62] and REFMAC5 [63] programs, respectively. The HEF receptor analog complexes were subsequently determined using the refined HEF structure as the input model. The receptor analogs were manually built using COOT based on the simulated anealing omit Fo-Fc maps and were further refined by PHENIX [64] . The stereochemical quality of the final models was assessed with the program PROCHECK [65] . Final statistics for data collection and structure refinement is represented in Table 1 .

The microarray analysis was first performed by applying the IDV HEF-mut protein to the array at 200 μg/mL and detecting with a His antibody labeled with Alexa488. The experiments were performed in replicates of six at CFG using a version 5.1 CFG array consisting of 610 glycans. The highest and lowest points from each set of six replicates were removed, so the average is of four values rather than six. This eliminates some of the false hits that contain a single very high or low point.

Then, we chose to use a more specific array with broader and paired (Neu5Ac-or Neu5Gcbased, and α2-3 or α2-6-linked) 9-O-Ac and non-O-Ac-sialoglycans to further characterize the binding properties of IDV and ICV HEF [36] . Glycan microarrays were fabricated using epoxide-derivatized slides as previously described [36] . Printed glycan microarray slides were blocked by ethanolamine, washed and dried. Slides were then fitted in a multi-well microarray hybridization cassette (AHC4X8S, Arrayit, Sunnyvale, CA, USA) to divide into 8 subarrays. The subarrays were blocked with Ovalbumin (1% w/v) in PBS (pH 7.4) for 1 hr at room temperature (RT), with gentle shaking. Subsequently, the blocking solution was removed and diluted IDV-HEF-mut and ICV-HEF-mut protein samples with 160 μg/mL were added to each subarray. After incubating the samples for 2 hr at RT with gentle shaking, the slides were washed. Diluted anti-His-HiLyte Flour 555 antibody (LifeSpan BioSciences) in PBS was added to the subarrays, incubated for 1 h at RT, washed and dried. The microarray slides were scanned by Genepix 4000B microarray scanner (Molecular Devices Corp., Union City, CA, USA). Data analysis was performed using Genepix Pro 7.0 analysis software (Molecular Devices Corp., Union City, CA). Heat map was generated according to the method previously described [36] . Ranked binding of IDV-HEF-mut and ICV-HEF-mut on the array. Binding was ranked as (glycan RFU/ maximum glycan RFU) Ã 100. Blue and white represent the maximum and minimum, respectively.

The cell binding assays were performed in 96-well plates as previously described [46] . When the density of the MDCK cells in the wells reached 90% coverage, the plate was washed with PBS twice and fixed with ice-cold 100% methanol for 20 min. After PBST buffer (PBS with 0.05% Tween-20) washing for three times, the wells were blocked with blocking buffer [PBS, 0.05% Tween-20, 4% bovine serum albumin (BSA)]. His-tagged HEF or H5 HA protein (10 μg/mL, 20 μg/mL, 40 μg/mL, 60 μg/mL, 80 μg/mL, 100 μg/mL) was then added to each well and each concentration was performed in triplicates wells. After incubation at 37°C for 1 hr, the plate was washed three times with PBST buffer. Mouse anti-His antibody (MBL, Japan) was added to each well at a 1:1000 dilution and the plate was incubated for 1 hr. Then, the plate was washed and incubated at 37°C for 45 min with HRP-conjugated goat anti-mouse antibody (Santa Cruz, USA) at a dilution of 1:2000. Peroxidase activity was detected using TMB and the reaction was stopped by adding 2M H 2 SO 4 . Absorbance was measured at an optical density of 450 nm.

Hemagglutination assay was performed in U-bottom 96-well microtest plates (Becton Dickinson, USA) according to the method previously described [66] [67] [68] . Briefly, two-fold serial dilutions in 25 μl PBS of purified HEF or HEF-mut protein (100 μg/mL to 0.1 μg/mL per well) mixed with 25 μl of a 0.5% chicken erythrocytes suspension and incubated for 1hr at 25°C or 2 hr at 4°C. Then the hemagglutination effects were observed and the plates were screened by CTL-ImmunoSpot S5 Versa Analyzer (Cellular Technology, USA).

SLBA was performed as previously described [69] . Briefly, Corning 96 well EIA/RIA plates were coated for 16 hr at 4°C with BSM (60 μg/mL in PBS; Abnova) at 100 μL per well. The wells were washed with PBST and treated with blocking buffer for 1 hr at RT. Twofold serial dilutions of proteins containing a C-terminal His 6 tag were prepared in blocking buffer (starting concentration 100 μg/mL) and 100 μL samples of these dilutions were added to the glycoconjugate-coated wells. Incubation was continued for 1 hr after which unbound protein was removed by washing five times. Then the wells were incubated with mouse anti-His antibody (1:1000), washed five times with washing buffer, incubated with HRP-conjugated goat antimouse IgG antibody (1:2000) , and washed five times. Finally, the bound proteins were detected using TMB, and the reaction was stopped with 2M H 2 SO 4 . The absorbance of the resulting yellow color was read at 450 nm. To assess the enzymatic activities of IDV HEF protein towards 9-O-Ac-Sias, BSM coated plates were treated with samples from two-fold serial dilutions of IDV HEF protein (starting at 1 μg/mL in PBS, 100 μL/well) for 1 hr at 37°C. The destruction of 9-O-Ac-Sia receptor determinants was determined by SLBA with IDV HEF-mut protein (50 μg/mL in blocking buffer) as described above.

The activities of purified HEF and HEF-mut were tested using p-nitrophenyl acetate (pNPA, Sigma-Aldrich) as a substrate [70] . Proteins (50 μL) were diluted to 12.5 ng/mL using PBS buffer in each well of a 96-well plate, after which the plate was incubated at different temperatures (37°C, 25°C and 4°C). Twofold serial dilutions (0-8 mM) of preheated pNPA (50 μL) were then added at corresponding temperatures. The absorbance at 405 nm was measured immediately in a spectrophotometer every 30 seconds for 1 hr at corresponding temperatures on a microplate reader (SpectraMax M5; Molecular Devices). All assays were performed in triplicate, and the Km and Vm value for HEF were calculated using GraphPad Prism.

Immunofluorescence assays were performed as described previously with slight modifications [71, 72] . Briefly, paraffinized human, swine or bovine trachea tissue sections were deparaffinized, rehydrated and incubated with 2% BSA in PBS for 30 min at RT to prevent nonspecific binding. Purified HEF protein was precomplexed with primary antibody (mouse anti-His-tag, MBL) and secondary antibody (Alexa Fluor 488 goat anti-mouse IgG, Invitrogen) in a molar ratio of 4:2:1, respectively, for 20 min on ice. The tissue binding was performed using precomplexed stock HEF (50 μg/ mL) in 1% BSA-PBS. Tissue sections were then incubated with the HEF-antibody complexes for 3 hr at RT. Sections were counterstained with 4', 6-diamidino-2-phenylindole (DAPI) (Beyotime; 1:2,000 in PBS) for nuclei for 20 min at RT. After thorough washing, the tissue sections were mounted and then examined by using Leica TCS SP8 laser scanning confocal microscopy.

The IDV HEF crystal samples were applied to SDS-PAGE and subsequently transferred to polyvinylidene fluoride (PVDF) membranes at 50 V for 1 hr. For western blot, the proteins were identified with a mouse monoclonal antibody of Anti-His-tag-HRP-DirecT (MBL, Japan) and a Super Signal West Pico Chemiluminescent Substrate (Thermo, USA). For N-terminal sequencing, the PVDF blot membrane was stained for 30s-50s in coomassie brilliant blue (CBB) R250 staining solution (0.1% CBB R250, 1% acetic acid, 40% methanol in Milli-Q water) and destained with destaining solution (50% methanol in Milli-Q water) under visual control until protein bands were well visible. The PVDF membrane was dried and bands of interest were cut for the N-terminal sequencing with the Edman degradation method using PROCISE491 (America Applied Biosystems).

Atomic coordinates and structure factors have been deposited in the Protein Data Bank under accession codes 5E64 for IDV HEF in native state and 5E66 in complex with 9-N-Ac-Sia, and 5E5W, 5E65, 5E62 for IDV HEF-mut and complexes with 9-O-Ac-3'SLN and 9-O-Ac-3'Sle C , respectively. [73] , and are labeled using our IDV HEF structure. The sequence logos were generated after the total 223 sequences of ICV HEF alignment or total 14 sequences of IDV HEF alignment to visualize the sequence conservation by Geneious [74] . All the sequences were obtained from the NIAID Influenza Research Database (IRD) online through the web site at http://www.fludb.org. Table. Glycan microarray analysis of IDV and ICV HEF protein using a more extended glycan array. (XLSX)

The biology of influenza viruses

Influenza A penetrates host mucus by cleaving sialic acids with neuraminidase

The molecular virology and reverse genetics of influenza C virus. Japanese journal of infectious diseases

Selective inactivation of influenza C esterase: a probe for detecting 9-O-acetylated sialic acids

Influenza C virus infection in military recruits-symptoms and clinical manifestation

A nationwide epidemic of influenza C virus infection in Japan in 2004

Study of influenza C virus infection in France

Acute encephalopathy associated with influenza C virus infection. The Pediatric infectious disease journal

Clinical features of influenza C virus infection in children. The Journal of infectious diseases

Influenza C virus high seroprevalence rates observed in 3 different population groups

Serological studies of influenza viruses in pigs in Great Britain 1991-2. Epidemiology and infection

Distribution of the antibody to influenza C virus in dogs and pigs in Yamagata Prefecture

Prevalence of antibody to influenza C virus among pigs in Hyogo Prefecture

Isolation of influenza C virus from pigs and experimental infection of pigs with influenza C virus. The Journal of general virology

Interspecies transmission of influenza C virus between humans and pigs

Genome analysis of influenza C viruses isolated in 1981/82 from pigs in China. The Journal of general virology

Natural infection of dogs by influenza C virus. Research in virology

Natural infection of dogs by influenza C virus: a serological survey in Spain. The new microbiologica

Serological evidence of infection of dogs with human influenza viruses in Japan. The Veterinary record

Isolation of a novel swine influenza virus from Oklahoma in 2011 which is distantly related to human influenza C viruses

Characterization of a novel influenza virus in cattle and Swine: proposal for a new genus in the Orthomyxoviridae family

Genomic and evolutionary characterization of a novel influenza-C-like virus from swine

Cocirculation of two distinct genetic and antigenic lineages of proposed influenza D virus in cattle

Replication and transmission of novel bovine influenza D virus in guinea pig model

Identification of a potential novel type of influenza virus in Bovine in China

Emerging infectious diseases

The epidemiology of bovine respiratory disease: what is the evidence for preventive measures? The Canadian veterinary journal La revue veterinaire canadienne

Structured literature review of responses of cattle to viral and bacterial pathogens causing bovine respiratory disease complex

Serological evidence for the presence of influenza D virus in small ruminants. Veterinary microbiology

Structure and receptor binding specificity of hemagglutinin H13 from avian influenza A virus H13N6

Structure and receptor-binding properties of an airborne transmissible avian influenza A virus hemagglutinin H5 (VN1203mut)

Molecular basis of the receptor binding specificity switch of the hemagglutinins from both the 1918 and 2009 pandemic influenza A viruses by a D225G substitution

An airborne transmissible avian influenza H5 hemagglutinin seen at the atomic level

The Catalytic Triad of the Influenza-C Virus Glycoprotein Hef Esterase-Characterization by Site-Directed Mutagenesis and Functional-Analysis

Structure of the haemagglutinin-esterase-fusion glycoprotein of influenza C virus

Cross-comparison of protein recognition of sialic acid diversity on two novel sialoglycan microarrays. The Journal of biological chemistry

Structural basis for receptor specificity of influenza B virus hemagglutinin

Structures and receptor binding of hemagglutinins from human-infecting H7N9 influenza viruses

Structural basis for ligand and substrate recognition by torovirus hemagglutinin esterases

MEGA6: Molecular Evolutionary Genetics Analysis Version 6.0

Observation of an arsenic adduct in an acetyl esterase crystal structure

Serine-71 of the Glycoprotein Hef Is Located at the Active-Site of the Acetylesterase of Influenza-C Virus

HtrA proteases have a conserved activation mechanism that can be triggered by distinct molecular cues

Large lateral movement of transmembrane helix S5 is not required for substrate access to the active site of rhomboid intramembrane protease

Hemagglutinin-esterase-fusion (HEF) protein of influenza C virus

Bat-derived influenza hemagglutinin H17 does not bind canonical avian or human receptors and most likely uses a unique entry mechanism

New world bats harbor diverse influenza A viruses

Studies on Survival of Influenza Virus Between Epidemics and Antigenic Variants of the Virus

Identification of Another Epidemic Respiratory Disease

Further Observations on the Jj and 1233 Influenza Viruses

Rescue of influenza c virus from recombinant DNA

Influenza D virus infection in Mississippi beef cattle

Adaptation of Avian Influenza A Virus Polymerase in Mammals To Overcome the Host Species Barrier

Influenza virus proteins as factors involved in interspecies transmission

Host species barriers to influenza virus infections

Influenza and other respiratory viruses

Evidence of Previous Avian Influenza Infection among US Turkey Workers

Insights into avian influenza virus pathogenicity: the hemagglutinin precursor HA0 of subtype H16 has an alpha-helix structure in its cleavage site with inefficient HA1/HA2 cleavage

Processing of X-ray diffraction data collected in oscillation mode

Pushing the boundaries of molecular replacement with maximum likelihood

The Ccp4 Suite-Programs for Protein Crystallography

Coot: model-building tools for molecular graphics

Refinement of macromolecular structures by the maximum-likelihood method

PHENIX: a comprehensive Python-based system for macromolecular structure solution

Stereochemical quality of protein-structure coordinates

Hemagglutination Assay for Influenza Virus

The Murine Coronavirus Hemagglutinin-esterase Receptor-binding Site: A Major Shift in Ligand Specificity through Modest Changes in Architecture

9-O-acetylated sialic acids have widespread but selective expression: analysis using a chimeric dual-function probe derived from influenza C hemagglutininesterase

Crystal structure of mouse coronavirus receptor-binding domain complexed with its murine receptor

Recombinant influenza C hemagglutinin-esterase as a probe for sialic acid 9-O-acetylation

Glycan topology determines human adaptation of avian H5N1 virus hemagglutinin

Structural basis for preferential avian receptor binding by the human-infecting H10N8 avian influenza virus

Deciphering key features in protein structures with the new ENDscript server

Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data

We acknowledge the Consortium for Functional Glycomics for glycan array analysis and synthetic glycans used for crystallization. We thank Dr. Peter Rosenthal from the Division of Physical Biochemistry at the MRC National Institute for Medical Research, London, for providing human C/Johannesburg/1/66 HEF and 9-N-Ac-Sia complex structure file. We thank Jin Xiao from Zhongmu institutes of China animal husbandry industry for providing normal swine and cattle tracheal tissues. We thank Prof. Jinghua Yan, Dr. Wei Zhang, Yanfang Zhang, Joel Haywood and Lifeng Fu from Institute of microbiology CAS for assistance. We also thank the staff at Shanghai Synchrotron Radiation Facility (SSRF) beamline 17U for their kind help with data collection.