Background
Mammalian serum and respiratory fluids contain a complex mixture of proteins, some of which can inhibit hemagglutination activity or neutralize the infectivity of influenza viruses. Three classes of such inhibitors have been reported. The α and γ inhibitors are sialylated glycoproteins that act as receptor analogues, binding to the receptor-binding site of influenza virus hemagglutinin (HA) to block access to cellular receptors. The β inhibitors are not receptor analogues, do not contain sialic acid and act via a mechanism distinct to that of α and γ inhibitors.
Studies by Anders
et al. demonstrated that the β inhibitors in bovine and mouse serum were mannose-binding lectins of the collectin family [
1]. Collectins are large multimeric proteins that bind to glycoconjugates rich in D-mannose and N-acetylglucosamine in a Ca
2+-dependent manner and play an important role in innate host defence against a range of microbial pathogens (reviewed by [
2,
3]). Members of the collectin family include the serum mannose-binding lectin (MBL), bovine serum proteins conglutinin and collectin-43 (CL-43) and lung surfactant proteins A (SP-A) and D (SP-D). For influenza viruses of the H3 subtype, the oligosaccharide side-chain at the tip of the HA spike was shown to be critical in determining the sensitivity of the virus to the antiviral activities of collectins in mouse and bovine serum [
1]. Mutant viruses selected in the presence of bovine serum (a rich source of conglutinin) were shown to have lost this glycosylation site and were resistant to hemagglutination inhibition by β inhibitors [
1].
Since their identification as β inhibitors, the role of collectins in innate host defence against influenza viruses has become an area of intense interest. MBL, conglutinin, CL-43 and SP-D all act as classic β inhibitors, binding in a Ca
2+-dependent manner to oligosaccharides expressed on the viral HA and NA glycoproteins. This mediates hemagglutination inhibition, neutralization, virus aggregation and opsonization of virus to promote neutrophil responsiveness to the virus (reviewed by [
2,
4,
5]). In contrast, the collectin SP-A is a sialylated glycoprotein and therefore acts as a γ inhibitor to mediate a similar range of antiviral activities against influenza viruses [
6,
7]. Of particular interest, both SP-A and SP-D are present in respiratory secretions, although current evidence suggests that the high avidity interaction between SP-D and carbohydrates on the viral HA is a major factor contributing to the neutralizing capacity of bronchoalvolar lavage fluids [
8‐
10].
Since their appearance in the human population in 1968, H3N2 subtype viruses have shown a progressive increase in N-linked glycosylation in and around the globular head of the HA molecule, while glycosylation sites located in the stem region of HA tend to be highly conserved [
11,
12]. Using a mouse model of influenza infection, we have demonstrated that for viruses of the H3 subtype (1968-1992), the level of glycosylation on the globular head of HA of a particular virus strain inversely correlates with its ability to replicate
in vivo [
8]. Virus strains bearing high levels of glycosylation (1977-1992) were more sensitive to neutralization by murine collectins, and this in turn correlated with a poor ability to replicate in mouse lung. In initial studies we were surprised to find that one virus strain, A/Beijing/353/89 (Beij/89), did not fit this trend and grew well in mouse lung despite the presence of 4 potential sites of N-linked glycosylation on the globular head of HA. Studies were therefore undertaken to determine the mechanisms underlying the enhanced virulence of this particular mutant for mice.
Methods
Viruses
A seed stock of wild-type (non-reassortant) A/Beijing/353/89 (Beij/89) from the WHO Collaborating Centre for Reference and Research on Influenza, Melbourne, Australia and was propagated once at a 10-4 dilution in the allantoic cavity of 10-day embryonated eggs to generate an uncloned stock of Beij/89. When this stock was plaqued on MDCK cell monolayers in the presence of trypsin [
8], two morphologically distinct plaque types were observed; a predominant round plaque type approximately 1 mm in diameter (large plaque phenotype), and a minor subpopulation (<5%) of small, star-shaped plaques (small plaque phenotype).
Plaque purification (PP) of virus was performed on MDCK cells and was monitored by the distinctive plaque morphology of the large and small plaque viruses. Well-separated plaques were picked, resuspended in PBS and inoculated into 10-day embryonated hens' eggs. Allantoic fluid was harvested and the PP procedure repeated. Stocks of allantoic fluid generated from the second PP were plaqued to ensure appropriate morphology and frozen at -70°C. Purified virus stocks were prepared using discontinuous sucrose gradients as described [
1].
Infection and treatment of mice
C57BL/6 mice were bred and maintained in the animal facility of this department. Adult mice (6-8 weeks) were used in all experiments. All research complied with the University of Melbourne's Animal Experimentation Ethics guidelines and policies. Mice were anaesthetized and infected intranasally (i.n.) with 10
5 PFU of influenza virus (unless otherwise stated) in 50 μl of PBS. Each day, mice were weighed individually and monitored for signs of illness. To determine viral titres, mice were euthanized and lungs and nasal tissues were removed and homogenates were clarified by centrifugation. The samples were assayed for infectious virus by plaque assay on MDCK monolayers [
8].
Differential leukocyte counts in bronchoalveolar lavage (BAL) fluids
For collection of BAL cells, mice were killed and the lungs flushed three times with 1 ml of PBS through a blunted 23-guage needle inserted into the trachea. Cells were treated with Tris-NH4Cl (0.14 M NH4Cl in 17 mM Tris, adjusted to pH 7.2) to lyse erythrocytes, washed in RPMI 1640 medium supplemented with 10% FCS and cell viability was determined via trypan blue exclusion. For differential counts, aliquots of approximately 5 × 104 BAL cells were cyto-centrifuged onto glass microscope slides, dried and stained with Diff Quick (Lab Aids, Australia). Slides were examined using a light microscope and a minimum of 100 cells in 4-8 random fields was counted (×1000 magnification). Macrophages, lymphocytes and neutrophils were identified by their distinct nuclear morphologies.
Sera, mAbs and SP-D
Mouse serum was collected from blood that had clotted at 4°C overnight followed by storage at -70°C. Recombinant rat SP-D was a gift from Prof. Erika C. Crouch, Department of Pathology, Washington University School of Medicine, St. Louis, Missouri, USA. The anti-HA mAbs CY3/3, PA1/1 and C1/1 raised against BJx109 (A/Beijing/353/89 × A/PR/8/34) were prepared by Dr. Georgia Kapakalis-Deliyannis, Department of Microbiology and Immunology, University of Melbourne. mAb D7/1, raised against A/Philippines/2/82 (Phil/82), was prepared by Dr. E. M. Anders, Department of Microbiology and Immunology, University of Melbourne.
Virus Neutralization assays
Neutralization of virus infectivity was measured by fluorescent-focus reduction in monolayers of MDCK cells cultured in 96-well plates (Nunc, Golstrup, Denmark) as described [
8]. Briefly, dilutions of mouse sera or recombinant rat SP-D were mixed with a constant dilution of virus, and after incubation for 30 mins at 37°C, added to MDCK cell monolayers. After adsorption of virus for 45 min at 37°C, the inoculum was removed and cells were incubated a further 7-8 hrs to allow for infection of MDCK cells. Cell monolayers were then fixed in 80% acetone and stained for fluorescent foci by incubation with mAb A-3, specific for the nucleoprotein (NP) of type A influenza viruses, followed by fluorescein-conjugated rabbit anti-mouse immunoglobulins (Silenus, Melbourne, Australia).
Hemagglutination and Hemagglutination Inhibition (HI) assays
Hemagglutination titrations and HI tests were performed by standard procedures using 1% (vol/vol) chicken erythrocytes in Tris-buffered saline (TBS; 0.05 M Tris-HCl, 0.15 M NaCl, pH 7.2) containing 0.1% NaN3 (TBSN3).
Sequencing of HA gene
Influenza virus RNA was extracted directly from allantoic fluid. Virus was digested with proteinase K and 0.5% sodium dodecylsulfate (SDS) and heated to 55°C for 5 min. RNA was extracted using hot phenol, followed by phenol-chloroform extraction and ethanol precipitation. Full length HA cDNA was prepared from viral RNA using AMV reverse transcriptase (Promega, U.S.A.). Two segments were then amplified from the HA gene PCR for direct sequencing. Sequences were determined using a PRISM Ready Reaction Dyedeoxy terminator cycle sequencing kit (Perkin Elmer, Applied Biosystems Division, Foster City, CA, USA). The complete sequence of HA for the L phenotype virus has been deposited in GeneBank (U97740).
SDS-PAGE and immunoblot for HA
Proteins from purified preparations of influenza virus were resolved by SDS-PAGE (5-12.5% gradient gels) under non-reducing conditions, transferred to nitrocellulose and probed with 1/500 dilution of ascitic fluid of mAbCY3/3 in TBS containing 2.5 mg/ml BSA. After washing, bound antibody was detected with 1/400 dilution of HRP-conjugated rabbit anti-mouse immunoglobulins (Dako, Glostrup, Denmark). SeeBlue pre-stained standards (Novex, San Diego, California) were used to estimate molecular weights.
Discussion
Members of the collectin family can function as β inhibitors of influenza virus and have been implicated in playing an important role in innate host defence during influenza infections. For H3 subtype influenza viruses, sensitivity to β inhibitors has been shown to correlate with the presence of a mannose-rich oligosaccharide at residue 165, located at the tip of the HA spike. In this study we have identified a second glycosylation site at residue 246 on the globular head of H3 subtype viruses that enhances sensitivity to collectins and modulates virulence for mice. Loss of glycosylation site 246 was not associated with classical resistance to β inhibitors when assessed by HI assay, however mutant viruses lacking this site were shown to be more resistant to neutralization by murine MBL and SP-D in vitro. Mutant viruses showed no difference in their ability to replicate in respiratory epithelial cells in vitro, but were markedly more virulent following intranasal inoculation of mice. These studies provide further evidence that multiple glycosylation sites on the HA of H3 subtype influenza viruses are involved in determining sensitivity to the antiviral activities of collectins both in vitro and in vivo.
In previous studies we have reported a strong correlation between the degree of glycosylation of the influenza virus HA, sensitivity of particular virus strains to neutralization by collectins and their virulence in mice [
8]. Virus strains of the H3 subtype isolated after 1982 were particularly sensitive to neutralization by murine SP-D and MBL
in vitro and were very poor in their ability to replicate in mouse lung. Compared to earlier virus strains, Phil/82, Beij/89 and Beij/92 had acquired an additional glycosylation site at residue 246 of HA
1 and this site appeared to be associated with the enhanced sensitivity of these viruses to neutralization by collectins. Virus mutants selected for resistance to β inhibitors lack oligosaccharide attachments at residue 165 (H3 subtype) or 104 (H1 subtype)[
1,
13], yet recent evidence suggests additional glycans on the viral HA also contribute to SP-D-mediated antiviral activity. The use of reverse genetics to engineer additional sites of glycosylation into the globular head of the HA of A/Hong Kong/1/68 (H3N2) resulted in viruses that were more sensitive to inhibition by human SP-D and showed attenuated disease in mice [
18]. Sequential addition of glycosylation to sites 63, 126 and 246 was associated with step-wise increases in sensitivity to SP-D. Glycosylation site 165 has been shown to carry high-mannose glycans in early H3 virus strains such as A/Aichi/2/68 and Memphis/102/72 [
19,
20], however Beij89 (L virus) carries additional sites of potential N-linked
glycosylation at residues 63, 126, and 246 on the head of HA and, if glycosylated, the nature of specific glycans remains to be defined. The enhanced resistance to SP-D associated with loss of glycan 246 from Beij/89 is consistent with the expression of mannose-rich oligosaccharides at this site although the sugar specificity of SP-D does not exclude interactions with complex or hybrid-type glycans. Of interest, recent studies comparing H3N2 virus strains implicated additional glycan attachments at positions 122, 133 and 144 in contributing to the enhanced sensitivity of post-1995 H3N2 subtype viruses to human SP-D [
21].
Inhibition of the hemagglutinating activity of influenza viruses represents a simple and convenient assay for determining sensitivity to β inhibitors by assessing the ability of the inhibitor to block access of viral HA to sialylated erythrocyte receptors. For H3 subtype viruses, the presence of a mannose-containing oligosaccharide at residue 165 has been shown to be critical for determining sensitivity to HI by β inhibitors in bovine and mouse serum [
1], as well as by purified SP-D [
22]. The current study indicates that additional glycosylation sites on HA, such as that at residue 246 on the HA of Beij/89, also play an important role in other antiviral activities mediated by collectins, such as neutralization of virus infectivity. Due to the multimeric nature of collectins, additional
glycosylation sites on the head of HA presumably facilitates an enhanced degree of binding, thereby strengthening the overall affinity with which a collectin can bind to and inactivate virus. Previously, a mutant H3N2 virus selected for resistance to rabbit serum was found to have lost a potential glycosylation site at residue 246 on HA
1 [
23] and subsequent studies demonstrated that the inhibitor in rabbit sera was a mannose-binding lectin [
24]. Our study has extended these findings to demonstrate the critical role of this glycosylation site in determining sensitivity to murine SP-D, which is found in airway secretions, and in modulating virulence in a mouse model of influenza infection.
The presence of the small plaque mutant in the seed stock of Beij/89 suggests that loss of the glycosylation site at residue 246 on the HA
1 of Beij/89 confers a growth advantage upon cultivation of virus in eggs. Egg adaptation has been associated with a number of sequence changes that alter potential N-linked glycosylation sites on the HA, including the site at residue 246 of H3 subtype viruses [
25‐
27] although loss of this site from particular H3 field strains was not associated with adaptation to growth in eggs [
16]. In general terms, amino acid substitutions associated with egg-adapted variants cluster around the receptor-binding site on the HA molecule [
28] and are likely to increase binding to Sia(α2,3)Gal moieties on cells of the chicken embryo chorio-allantoic membrane, the site of influenza virus replication in chicken eggs. N-linked glycans in the vicinity of the receptor-binding site have been proposed to sterically interfere with binding to Sia(α2,3)Gal-containing macromolecules and so may be lost during adaptation to growth in eggs [
29].
In studies not presented here, L, S and Mo viruses showed no differences in their ability to agglutinate chicken erythrocytes that had been treated with increasing concentrations of periodate or
Vibrio cholerae neuraminidase to oxidize or remove sialic acid residues, nor in their ability to agglutinate erythrocytes from a variety of avian or mammalian species (chicken, turkey, monkey, guinea pig, mouse, rabbit, cow and human erythrocytes), and on this basis, displayed no evidence of modified receptor specificity. Subtle differences in the fine specificity of receptor binding between the Beij/89 variants that went undetected in these assays could, however, be of particular relevance during growth of viruses in eggs or in mouse lung. It is well established that Sia(α2,3)Gal moieties are expressed throughout the murine respiratory tract [
30], however our findings that L, S and Mo viruses are similar in their ability to infect and replicate in murine respiratory epithelial cells (Fig.
6) suggests that changes in S and Mo mutant viruses has not altered the intrinsic ability of these viruses to replicate in the appropriate target cells. Instead, their enhanced growth in mouse lung likely represents evasion of a particular host defence mechanism.
Glycosylation site 246 is located on the globular head of HA
1 and alterations in this site could presumably influence other functions mediated by the viral HA. While receptor binding by the viral HA is required for binding and entry of virus into target cells, efficient fusion of viral envelope with endosomal membranes in the target cell is required to initiate infection. Previous studies implicated residue 246 on the HA of A/Philippines/2/82 (H3N2) as a critical residue influencing stability of HA at low pH [
13], a factor critical in inducing the conformational change of the HA to expose the fusion peptides in the acidic environment of the endosome [
31,
32]. However, L, S and Mo viruses were found to be similar in their ability to mediate hemolysis of human erythrocytes over a range of pH using the fusion assay described ([
33], unpublished observations). Sequencing of the NA gene of multiple clones of S or Mo viruses confirmed that there were no amino acid substitutions in the NA, consistent with our findings that its enzymatic activity was not significantly different between L, S and Mo viruses when assessed by ELISA [
8] (unpublished observations).
Particular stains of influenza virus appear to be intrinsically more virulent than others, due largely to their ability to cause viral pneumonia and/or predispose the host to bacterial super-infection. Factors determining the virulence of particular virus strains are not fully understood although a number of lines of evidence suggest that in humans, as in mice, glycosylation may modulate viral virulence. Neither the virulent 1918 Spanish influenza pandemic virus (H1N1) [
34] nor the mouse-adapted A/PR/8/34 (H1N1) [
35] virus carry potential sites for N-linked glycosylation on the head of HA and human H5N1 isolates were resistant to HI by human SP-D [
21], consistent with the notion that virulent strains carry less glycosylation to evade collectin-mediated defences. Following its appearance in the human population, the H3N2 strain of 1968 contained only 2 sites on the globular head of HA, and despite the presence of oligomannose at residue 165 [
20], early H3 strains were partially resistant to SP-D [
8,
21]. In the process of antigenic drift, the accumulation of glycosylation on the head of HA may shield the virus from pre-existing antibody in the human population [
12,
36] but increase sensitivity to collectin-mediated antiviral activities. While evidence indicates the maintenance of extensive glycosylation on the head of recent H3 viruses [
37,
38], loss of glycosylation from the HA of particular human isolates [
16,
39,
40] argues for the existence of a fine balance between protection afforded by antibody and increased sensitivity to SP-D.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PR carried out the majority of experiments described in the study, analyzed and interpreted data and wrote the manuscript. DP, MT, PW and EJ performed some of the experiments described in this study. AB contributed to interpretation of data and the writing of the manuscript. All authors read and approved the final manuscript.