Background
H5N1 is an avian influenza virus which originally circulated in aquatic birds without causing major disease. However, a rapidly spreading variant(s) of H5N1 is highly pathogenic to avian species, causing a major economic loss due to culling of millions of potentially infected birds [
1‐
5]. Alarmingly, this virus has crossed the species barrier to cause numerous human (and other animal) fatalities in certain regions of Asia, Europe and Africa[
6,
7]. The unprecedented spread and the high mortality rate of this virus have raised a major concern for a potential global pandemic. The lack of effective vaccines for humans and the emergence of oseltamivir-resistant H5N1 strains[
8,
9] underscore the urgent need in developing novel prophylactic and therapeutic treatments against this virus.
Influenza virus is an enveloped, negative-stranded, and segmented RNA virus. Two viral glycoproteins, hemagglutinin (HA) and neuraminidase (NA), on the viral surface, determine antigenic subtypes. Although the role of NA in the influenza life cycle is not clear, one of its functions is to release the progeny viral particles from the cell surface during budding. In contrast, the roles of the prototypic HA in viral entry have been well characterized by molecular, biochemical, biophysical, and structural techniques. HA is synthesized as a precursor, HA
0, that forms trimers in the endoplasmic reticulum (ER). This precursor is cleaved into two subunits, HA
1 and HA
2, which are linked by a disulfide bond[
10]. Many HAs contain a consensus sequence R-X-R/K-R as the cleavage site recognized by host furin-like proteases[
11]. The presence of this polybasic motif in HA has been shown to correlate with the high pathogenicity of influenza viruses[
3,
10,
12‐
14].
The two subunits of HA perform distinct functions in viral entry. HA
2 mediates membrane fusion and viral entry, while HA
1 is involved in binding to the sialic acid (SA) receptors on the target cells[
10]. Comparison of the HA
1 and HA
2 sequences among influenza virus subtypes reveals that the HA
2 sequence is well conserved, suggesting a highly conserved membrane fusion mechanism. However, the HA
1 sequence is much more divergent, suggesting differences in affinity of receptor-binding and in antigenicity. Although HA
1 binds to SA with a low affinity, it is believed that interaction of multiple HA molecules on the viral surface with the SA-containing glycoproteins or glycolipids on the cell surface increases the avidity of influenza virus to the target cells[
15], and this interaction facilitates viral infection through endocytosis in a pH-dependent manner.
Although bird flu H5N1 is a highly contagious pathogen in avian species, its transmission to humans, or more rarely human-to-human transmission, has been very limited thus far[
6]. A more transmissible and sustained variant(s) of H5N1 in human populations may arise through accumulating mutations in multiple viral proteins of H5N1 and/or genomic reassortment between H5N1 and other influenza viruses[
12]. An outbreak of a highly pathogenic H5N1 influenza virus in migratory birds of several species in Qinghai Lake, China was reported recently[
16,
17]. It was feared then and realized now that the H5N1-infected migratory birds could carry and transmit the virus to avian and non-avian animals including humans in densely populated areas on different continents[
18]. In this study, we characterized several aspects of the host cell tropism of H5N1, and our results implicate an unidentified host factor in H5N1 entry.
Discussion
Different pseudotyping systems have been developed in dissecting the roles of the glycoproteins of highly pathogenic enveloped viruses such as Ebola virus in viral entry and host tropism. To date, such systems have not been used often in entry studies of influenza viruses, although several groups have previously shown the feasibility of this technique in applications such as vaccine development and gene targeting using various viral vectors [
32‐
37]. In this report, we have developed an efficient HIV-based pseudotyping system for avian influenza H5N1 to circumvent the strict requirements in handling this highly pathogenic virus. We anticipate that this system will be invaluable for screening and developing potential entry inhibitors against H5N1. In addition, this pseudotyping system can be easily adapted for entry studies of 1918 Spanish flu and other highly pathogenic influenza viruses with alleviated safety concerns.
Multiple viral factors including HA, NA, the polymerase complex, and NS1 have been implicated in determining the cell and host tropism of influenza viruses [
12,
13,
38‐
42]. Among them, HA is known to be one major determinant which dictates the host restriction [
12,
43,
44]. One of the applications of viral pseudotyping systems such as the one developed for H5N1 in this report is to distinguish the host restriction at the entry step determined by HA from those restrictions at the post-entry steps determined by other viral proteins. Here, we demonstrate that several salient features of HA have been recapitulated by the HIV-based pseudotyping system in this report. First, efficient HA-mediated transduction is dependent on neuraminidase treatment on the producer cells (see Results and Fig.
1B), consistent with the notion that NA is involved in cleaving the sialic acid (SA) and thus releasing the budding viral progeny from the cell surface [
45]. Furthermore, neuraminidase treatment can be replaced by cotransfection of a human N1 NA gene with the HA gene and HIV vector. It is likely that the HA-mediated transduction efficiency reported in this study, even though very efficient already, can be further optimized by co-expressing M2 in 293T producer cells since others have shown a synergy of M2 and NA in viral particle release [
37]. Second, it is well-documented that influenza viruses enter the host cells via receptor-mediated endocytosis in a pH-dependent manner [
26]. The sensitivity of HA(QH)-mediated transduction to both lysosomotropic agents demonstrated the strict requirement of low pH for the pseudoviral infection to the target cells (see Fig.
2). Together these results give compelling arguments for using HA/HIV pseudotyping as a simple and reliable surrogate system in elucidating the host tropism of H5N1 and other highly pathogenic influenza viruses at the entry step.
In this report, we have examined the cell tropism of a highly pathogenic avian influenza virus H5N1, which was originally isolated from the infected migratory birds in Qinghai Lake of Western China [
16], at the entry step. The surprising finding is that the most susceptible cell lines for the HA(QH)-mediated transduction are those derived from humans, 293T (kidney) and Huh 8 (liver), and all of the human lung cell lines tested, A549 and NCI-H661 consistent with the recent reports that H5N1 virus can attach to the lower respiratory tract and lung of humans [
46,
47]. In stark contrast, two avian cell lines, QT6 (quail fibroblasts) and DF-1 (spontaneously immortalized chicken embryo fibroblasts), were transduced by HA(QH)/HIV virions at much lower frequencies (see Tables
1 and
2). These results indicate that this highly pathogenic H5N1 virus can enter numerous human cells including those derived from human lungs more efficiently than that in the two avian cell lines, suggesting that the HA protein of H5N1 can effectively interact with the cognate cellular receptor(s) on human cells to initiate viral infection. Furthermore it appears that other viral and/or human determinants, rather than HA, restrict efficient transmission of H5N1 to humans in a sustained manner. Therefore we believe that while the H5N1 virus still needs to acquire the ability, either through mutations or genomic re-assortment or both, to produce sustained infection in humans, it appears that this final step for the virus to overcome is at the post-entry level.
Another important implication of the current study is that either an unidentified host co-factor or an alternate linkage of sialic acid may be necessary for mediating H5N1 entry and infection. It is well documented that avian influenza viruses preferentially bind to sialic acid (SA) receptors on avian cells where the SA is predominantly joined to the sugar chain through an α2–3 linkage, while human-adapted influenza viruses have an increased affinity to the SA receptors of the α2–6 linkage [
10], a major form in the human respiratory tract. It is believed that a switch in preferential receptor-binding from the α2–3 linked SA to the α2–6 linked SA by HA protein is a prerequisite for an avian influenza virus to emerge as a pandemic threat in human populations [
12]. Recently it was suggested that in addition to the SA linkages, glycan topology as well as sulfation and fucosylation can dictate human adaptation of avian H5N1 virus HA [
48]. The question is why the HA protein of H5N1, an avian influenza virus, can transduce human cells more efficiently than non-human cells including avian cells. It is possible that accumulated mutations in the receptor binding domain (RBD) have increased the binding affinity of HA(QH) to the α2–6 linked SA receptors on the human cells. Structural and binding studies of a closely related HA (A/Vietnam/1203/2004, or Viet04 HA) in a recent report are somewhat consistent with this notion. It was found that Viet04 HA is more related to the 1918 and other human HAs than to a 1997 duck H5 HA (DK97 HA) and there are only two noticeable substitutions in the conserved residues of RBDs, E216 and P221 in DK97 HA
vs R216 and S221 in Viet04 HA. Viet04 HA was shown to bind α2–6 glycans with some affinity, while DK97 HA did not display any binding [
49]. Since the corresponding residues in HA(QH) are K216 and S221 at these positions, it can be assumed that HA(QH), just like Viet04 HA, can bind the α2–6 SA receptors on the human cells to initiate viral entry. However, Viet04 HA was shown to bind to the α2–3 glycans with higher affinity than the α2–6 glycans [
49], which is most likely true for HA(QH). At the same time, HA(QH) is able to mediate more efficient transduction in human cells than avian cells. Further, there is no obvious correlation between the HA-mediated transduction efficiency and the surface levels of 2,3 or 2,6 SAs displayed by several cell lines, and HA substitutions of the SA-binding residues do not greatly impact the HA-mediated transduction or target cell binding. It is possible that an alternative linkage of sialic acid, such as 2,8 or 2,9, may play a role in mediating viral attachment and entry. Here we propose that SA is necessary, but not sufficient to act as the cellular receptor and another surface molecule (or molecules) in addition to SA is also required to mediate efficient H5N1 entry. Indeed it was reported that SA specificity of avian influenza viruses may not restrict initial avian-to-human transmission [
50]. Although no such molecule(s) has been identified up to date, a recent report indeed suggests that a host N-linked glycoprotein is required for human influenza virus entry [
27]. Therefore, a similar (or different) protein on the human cells such as A549 can act as the co-factor in mediating efficient H5N1 entry, while the same protein on the avian cells is not as efficient as the co-factor either due to a low level of surface expression or low binding affinity to HA. In contrast, the resistant cells such as L2 (rat lung), Lec 1 (Chinese hamster ovary), or Jurkat (human T lymphocyte) may not express a functional homolog on the cell surface. This conclusion is in agreement with a recent report that human tissues lacking the SA receptor can be infected with H5N1 viruses [
51].
Finally, it is important to emphasize that the HIV-based pseudotyping system, together with the numerous susceptible and resistant cell lines and a functional recombinant HA protein reported in this study, will greatly facilitate identification of the co-factor(s) for H5N1 and likely other influenza viruses by genetic and biochemical techniques. Identification and characterization of such host co-factor(s) will have a huge impact on our understanding of the host restriction of H5N1 and on development of potential therapeutics against H5N1 pathogenicity.
Methods
Plasmids and cell lines
HA (QH) gene is from a highly pathogenic H5N1 influenza virus in goose (Goose/Qinghai/59/05)[
16]. HA (QH) and HA (codon optimized) were cloned into pcDNA3 and sequences were confirmed. T4-pMV7 and pc-CCR5 were from the National Institute of Health AIDS Research and Reference Reagent Program. NA (A/PR/8/34) influenza virus (H1N1) in the vector pEF6/V5-His-TOPO was kindly provided by John C Olsen (University of North Carolina, Chapel Hill, USA).
The following cell lines were either purchased from ATCC or provided by other investigators: Vero E6, MDBK, NCI-H661, L2, and Lec1 (ATCC); SAOS-2 and RAW264.7(Bin He, University of Illinois at Chicago), A549 (James Cook, University of Illinois at Chicago), HPAEC (Chinnaswamy Tiruppathi, University of Illinois at Chicago, USA), and DF-1 (Douglas Foster, University of Minnesota, USA). The rest of the cell lines used in the study are maintained in the laboratory. All cell lines were maintained in medium according to the protocols supplemented with 10% FBS and penicillin/streptomycin (50 units/ml).
Antibodies
The rabbit polyclonal antibody ab21297, which recognizes 14 amino acids within residues 100–150 of avian influenza A (H5N1) hemagglutinin protein, was purchased from Abcam Inc. The mouse anti-HIV p24 monoclonal antibody was obtained from the National Institute of Health AIDS Reseach and Reference Reagent Program, which was used to estimate the levels of HIV particle production as measured by p24 levels using Western blotting.
Production of HIV pseudovirions
Human embryonic kidney 293T cells were transiently co-transected with 8 μg hemagglutinin envelope expression plasmid with or without 0.5 μg NA and 10 μg Env-deficient HIV vector (pNL4.3.Luc-R-E- or pNL4-3-GFP-R-E-) in 100-mm plates by a standard Ca3(PO4)2 protocol. Sixteen hours post-transfection, cells were washed by phosphate-buffered saline (PBS) without Ca2+, Mg2+, 10 ml fresh medium was added into each plate. Forty-eight hours post-transfection, the supernatants were collected and filtered through a 0.45-μm-pore size filter (Nalgene) and the pseudovirions were directly used for infection.
Infection assay of HIV pseudovirions
The HIV pseudovirions (0.5 ml/well) prepared above were incubated with various cell types, seeded at 5 × 104/well in 24-well plates. For HIV-Luc virions, the targeted cells were lysed in 50 μl of cell culture lysis reagent (Promega) 48 h post-infection. The luciferase activity was measured with a luciferase assay kit (Promega) and an FB12 luminometer (Berthold detection system) according to the supplier's protocols. The experiments were repeated three times. For HIV-GFP virions infection, the targeted cells were observed by fluorescence microscopy.
Neuraminidase treatment
The 293T producer cells were treated with a commercial neuraminidase (New England Biolabs) 26 h and 46 h post-transfection at concentrations of 0, 5, 10, 20, 50, 100, 200 units/ml to optimize the yield of HIV pseudovirions, and later on at 100 units/ml for generation of pseudovirions. Pseudovirions were harvested 48 h post-transfection.
Detection of HA incorporation in pseudotyped viruses
To examine incorporation of HA(QH) to HIV particles, 4 ml of the collected supernatant was layered onto a 1.5-ml cushion of 20% sucrose (wt/vol) in PBS and centrifuged at 55,000 rpm for 45 min in a Beckman SW55 rotor. The pelleted HIV virions were lysed in 40 μl of Triton X-100 lysis buffer(50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, and a protease inhibitors cocktail consisting of 10 μg of leupeptin per ml, 5 μg of aprotinin per ml, and 2 mM phenylmethylsufonyl fluoride), and a 35 μl sample was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene difluoride membrane. The membrane was first incubated with anti-H5N1 HA1 polyclonal antibody ab21297(1:500) for 1 h and then probed with peroxidase-conjugated goat anti-rabbit antiserum (Pierce) for 40 min. The bands were visualized by the chemiluminescence method according to the protocol of the supplier (Pierce). In these experiments, HIV p24 level was determined by Western blotting as an estimate for the relative amounts of the pseudovirions.
NH4Cl and Bafilomycin A1 treatment
Cells were treated with 3.125, 6.25, 12.5, 25, and 50 nM Bifilomycin A1 (BAF) (Sigma) or 3.125, 6.25, 12.5, 25, and 50 mM ammonium chloride (NH4Cl) 30 min prior to infection. For the infection of the HIV envelope (gp120/gp41) pseudotyped HIV virions, 293T cells were first co-transfected with T4-pMV7 and pc-CCR5 36 h prior to infection.
Lectin Binding Assay
5 × 105 cells (293T, A549, NCI-H661, HeLa, L2, CHO, DF1, and QT6) were washed with PBS, pelleted, and resuspended in phosphate-buffered saline. Each cell type was incubated independently with 100 μg/mL of fluorochrome-conjugated lectins, Maackia amurensis agglutinin (MAA)-TRITC and Sambucus nigra agglutinin (SNA)-FITC (EY Laboratories) for 15 minutes in the dark. Cells were then washed 3 times with PBS and analyzed by flow cytometry.
Receptor-binding Domain Mutagenesis
Mutations were generated in the codon optimized HA backbone using a Site-Directed Mutagenesis Kit (Stratagene) and custom primers.
Fusion Protein Production and Purification
To study HA binding to the target cells, plasmids encoding HA fusion protein (different fragment of HA fused to Fc of human IgG) were constructed and the fusion proteins were purified. NheI-BamHI fragment of S1.hIgG was replaced with the PCR product of different fragments of the codon optimized HA gene. Plasmids of HA.hIgG were tranfected into 293T cells using the calcium phosphate method. After overnight, cells were re-fed with protein-free VP-SFM media supplemented with 4 mM L-glutamine (Gibco). Supernatants were collected twice at 48 and 72 hours post-tranfection and filtered through 0.45 μM membrane (Millipore). The supernatants were applied to a column of protein A beads (Santa Cruz Biotechnology) followed by three washing with 10 ml PBS. The proteins were eluted three times with 1 ml 0.1 M glycine (pH 2.8) and immediately neutralized with 60 μl Tris-HCl (pH8.0) each time. Then the proteins were dialyzed in 3.5K Slide-A-Lyzer Dialysis Cassettes (Pierce) and concentrated by centrifuging in YM-10 Centricon (Millipore). The protein concentrations were measured using the BCA Protein Assay Kit (Pierce).
Cells (5 × 105 293T or A549) were blocked in 500 μl volume of PBS/1%BSA on ice for 0.5 hour. Then different amounts of the purified proteins were incubated with cells in 500 μl volume of PBS/1%BSA on ice for 1.5 hour. Cells were incubated with no purified proteins as a negative control. Next the cells were washed twice with PBS/1%BSA and incubated with anti-human antibody conjugated with FITC (1:100 dilution, Sigma) in 500 μl volume of PBS/1%BSA on ice for 40 minutes. Then cells were washed three times with PBS/1%BSA followed by one washing with PBS and subjected to flow cytometry. Relative mean fluorescent intensity (MFI) was calculated by subtracting MFI of the negative control from MFI of cells incubated with different purified proteins.
Acknowledgements
We thank John Olson, Douglas Foster, Bin He, James Cook, Lili Wang, Chinnaswamy Tiruppathi, and Douglas Foster for cell lines or reagents. The laboratory research was supported by National Institutes of Health grants AI 059570 and CA 092459 (L.R.). Jizhen Wang was supported by a University of Illinois at Chicago Fellowship. H.X. and G.F.G. were partly supported by grants from Ministry of Science & Technology, China (2004BA519A29, 2005CB523001) and National Science Foundation of China (NSFC 30599434, 30525010). J. Y. and Y.C were partly supported by the National Natural Science Foundation of China (30270308, 30599432) and the Ministry of Science and Technology of China (2006CB910103, 2007CB914800).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
YG, ERB, JW, MC, and LR participated in the design of the study, YG, ERB, JW, HX, JY, LG performed the experiments, JW, GFG, YC provided key reagents, and all authors participated in drafting the manuscript.