|Home | About | Journals | Submit | Contact Us | Français|
Influenza A viruses of the H9N2 subtype are endemic in poultry in many Eurasian countries and have occasionally caused clinical respiratory diseases in humans. While some avian H9N2 viruses have glutamine (Q) at amino acid position 226 of the hemagglutinin (HA) receptor-binding site, an increasing number of isolates have leucine (L) at this position, which has been associated with the establishment of stable lineages of the H2 and H3 subtypes of viruses in humans. Little is known about the importance of this molecular trait in the infection of H9N2 viruses in humans. We show here that during the course of a single cycle of infection in human airway epithelial (HAE) cells cultured in vitro, the L-226-containing H9N2 viruses displayed human virus-like cell tropisms (preferentially infecting nonciliated cells) different from the tropisms showed by Q-226-containing H9N2 isolates (which infect both ciliated and nonciliated cells at ratios of 1:1 to 3:2) or other waterfowl viruses (which preferentially infect ciliated cells). During multiple cycles of replication in HAE cultures, L-226-containing H9N2 isolates grew consistently more efficiently and reached approximately 100-fold-higher peak titers than those containing Q-226, although peak titers were significantly lower than those induced by human H3N2 viruses. Our results suggest that the variation in residue 226 in the HA affects both cell tropism and replication of H9N2 viruses in HAE cells and may have implications for the abilities of these viruses to infect humans.
Avian influenza A viruses of the H9N2 subtype were first detected in the United States in 1966 (14). In North America, there are no reports of H9N2 virus-associated disease in chickens to date, although these viruses can be found in wild ducks and have caused a number of outbreaks in turkeys (12, 16, 36, 37). In Asia, H9N2 viruses were detected only in apparently healthy ducks in limited surveillance studies of live-poultry markets and farms in Hong Kong from 1975 to 1985. However, in the early 1990s, H9N2 viruses became more prevalent in domestic poultry (38). Reports from Korea, South Asia, Middle Eastern countries, South Africa, and Europe since the late 1990s indicate widespread distribution of H9N2 viruses (1, 3, 18, 24, 26, 31). During surveillance of live-poultry markets in Hong Kong from 2001 to 2003, H9N2 was the most prevalent subtype among the isolated influenza viruses (5). At least two lineages of viruses have become established in domestic poultry in Asia, represented by A/Duck/Hong Kong/Y280/97 (Dk/HK/Y280) and A/Quail/Hong Kong/G1/97 (Qa/HK/G1) (10, 20, 28). Viruses of the Dk/HK/Y280 lineage were also isolated from pigs, although their prevalence appeared to be low (28). Furthermore, human cases of H9N2 virus infection have been reported in Hong Kong and other locations of south China in the late 1990s and early 2000s (2, 11, 21, 29). Follow-up serological surveillance suggests that the incidence of human infections with H9N2 viruses might be higher than previously anticipated (4, 11, 29). The prevalence of H9N2 viruses throughout the world, along with their abilities to infect mammals and humans, increases concern about their pandemic potential.
Available data on antigenic and phylogenetic analysis suggest that the H9N2 viruses circulating in poultry continue to evolve. Continued reassortment with other strains has increased the genotypic diversity of these viruses (20). Recent studies also suggest increased abilities of H9N2 viruses to replicate in mice and pigs (5, 19). Numerous recent H9N2 isolates contain L-226 in their hemagglutinins (HAs) (5, 19, 20, 28) and show preferential binding to analogs of receptors with sialic acid (SA) linked to galactose by α2,6 linkage (SAα2,6Gal) (22, 35); both traits are typical of human H3 viruses. Thus, these viruses might possess one of the key elements for infection and sustained transmission in humans. However, there is little evidence regarding the possible enabling role of HA L-226 in human infections by H9N2 viruses. In the present study, we investigated the effect of this amino acid signature on cell tropism and growth of H9N2 viruses in human airway epithelial (HAE) cultures, a well-studied human respiratory tract epithelium in vitro model. HAE cells cultured on semipermeable membrane supports at the air-liquid interface have been used in various studies, from drug discovery to pathogenesis investigation, of many respiratory viruses. These cultures are polarized and pseudostratified, resembling the human respiratory tract epithelium in both morphology and cell type distribution. Our results suggest that the presence of L-226 in the HA allows H9N2 viruses to preferentially infect nonciliated cells and grow more efficiently in HAE cultures, highlighting the possible implications of this molecular trait for infection of humans.
The influenza viruses used in this study, including 10 viruses of avian origin and 2 of human origin (Table (Table1),1), were obtained from the influenza repository at St. Jude Children's Research Hospital, Memphis, TN. The avian influenza viruses were propagated in 10-day-old embryonated, specific-pathogen-free (SPF) chicken eggs, and the human influenza viruses were propagated in Madin-Darby canine kidney (MDCK) cells. The virus-containing allantoic fluids or cell culture supernatants were aliquoted and stored at −80°C. The median tissue culture infectious dose (TCID50) of each virus was determined using primary chicken embryo kidney (CEK) cells or MDCK cells. CEK cells have previously been used in influenza studies and shown to be more optimal for growth of avian viruses (17, 39, 40). We chose CEK cells in this study in order to increase the sensitivities of titrations of avian viruses, especially the viral samples for growth curves (see growth curve data in Results and Fig. 7A and B). To prepare CEK cells, the kidneys of 16- to 18-day-old SPF chicken embryos were trypsinized and grown in M199 medium (Invitrogen Corp., Grand Island, NY) containing 5% fetal bovine serum and 2% chicken serum. MDCK cells were grown in Dulbecco's modified Eagle's medium (Sigma, St. Louis, MO) containing 5% fetal bovine serum. For titration, avian viruses were 10-fold serially diluted with M199 containing 1 μg/ml trypsin and 0.15% bovine serum albumin (BSA; Sigma, St. Louis, MO), and human influenza viruses were 10-fold serially diluted with OPTI-MEM I medium (Invitrogen Corp., Grand Island, NY) containing 1 μg/ml trypsin. Confluent monolayers of CEK or MDCK cells growing in 96-well microplates were inoculated with the serial dilutions, with each dilution being added to four wells (200 μl per well). At 3 days postinoculation (p.i.), 50 μl of supernatant from each well was transferred out and mixed with 0.5% chicken red blood cells (CRBCs) for HA assays. The infective titer of each virus was calculated using the Reed and Muench method (32).
The recovery of influenza viruses was performed using reverse genetics as previously described (13, 25, 30). Briefly, eight plasmids (1 μg each) containing each of the influenza virus genes were used to transfect the 1:1 mixture of 293T human embryonic kidney cells and MDCK cells. The transfection mixture was replaced with OPTI-MEM I after 6 h of incubation at 37°C. Twenty-four hours later, OPTI-MEM I containing 1 μg/ml trypsin was added. At 48 to 72 h posttransfection, the culture supernatant was collected and propagated in 10-day-old embryonated SPF chicken eggs. The recovered viruses were stored and titrated as described above.
The desired mutations were introduced into the HA genes using a QuikChange II site-directed mutagenesis kit (Stratagene, Inc., La Jolla, CA) according to the manufacturer's protocols (primer sequences will be provided upon request). Two H9N2 virus backbones, A/Guinea fowl/Hong Kong/WF10/99 and A/Quail/Hong Kong/A28945/88, referred to herein as RGWF10 and RGQa88, respectively, were used to generate the desired mutant viruses. The mutant viruses were recovered as described above, and the HA genes were sequenced to verify the presence of the introduced mutations and the absence of additional, unwanted mutations. The receptor specificities of wild-type and mutant viruses were determined by using SAα2,3Gal- or SAα2,6Gal-resialylated CRBCs as described previously (8, 27). Briefly, 100 μl of 10% CRBCs was incubated with 50 mU Vibrio cholerae neuraminidase (Sigma, St. Louis, MO) at 37°C for 1 h. After being washed with phosphate-buffered saline (PBS), the desialylated CRBCs were resuspended in 200 μl PBS containing 1% BSA and 1.5 mM CMP-SA (Sigma, St. Louis, MO) and then incubated with 2 mU α2,3-sialyltransferase (EMD Biosciences, Inc., San Diego, CA) or 1.5 mU α2,6-sialyltransferase (EMD Biosciences, Inc., San Diego, CA) at 37°C for 1.5 h. The resialylated CRBCs were washed with PBS, resuspended, and used in HA assays.
Passage 1 HAE cells were purchased from Cell Applications, Inc. (San Diego, CA). The cells were expanded in T-75 flasks with growth medium from Cell Applications, Inc., and stored in liquid nitrogen. Passage 2 cells were expanded similarly. Upon reaching 60% to 80% confluence, the cells were trypsinized and seeded into Transwell-Clear inserts with semipermeable membranes (diameter, 12 mm or 6.5 mm; pore size, 0.4 μm; Corning, Inc., Corning, NY) that were coated with rat tail collagen type I (Upstate, NY) and maintained in Gray's medium (9, 34). At full confluence (usually 2 days after seeding), the apical medium was removed and the cells were maintained at the air-liquid interface to allow the differentiation of the epithelial subtypes. The medium was replaced daily only from the basal compartment. Fully differentiated cultures (4 to 6 weeks old, with a transepithelial electrical resistance of ≥400 Ω/cm2) were used for all experiments.
HAE cultures were thoroughly washed with growth medium, fixed with 4% paraformaldehyde, and then permeabilized with 0.2% Triton X-100. Potential endogenous peroxidase activity was eliminated with 1% H2O2-methanol. After being blocked with 1% BSA-PBS, the cells were incubated with specific monoclonal antibodies against β-tubulin (Sigma, St. Louis, MO), the cellular marker of ciliated cells, followed by incubation with peroxidase-conjugated goat anti-mouse immunoglobulin G (IgG; Sigma, St. Louis, MO). The cilia were visualized by developing the cells in Vector SG substrate (Vector Laboratories, Inc., Burlingame, CA). To examine the presence of goblet cells and Clara cells, parallel HAE cultures were immunostained for the presence of cilia and further incubated with rabbit polyclonal antibodies against mucin 5AC (Muc5AC) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), the cellular marker of goblet cells, or with rabbit polyclonal antibodies against Clara cell secretory protein (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), the cellular marker of Clara cells. The cultures were washed and incubated with peroxidase-labeled goat anti-rabbit IgG (Kirkegaard & Perry Laboratories, Gaithersburg, MD). The goblet cells and Clara cells were visualized by incubating the cultures in a solution of aminoethylcarbozole (AEC) (Sigma, St. Louis, MO). The membranes bearing the cells were cut off and mounted onto slides by using aqueous mounting medium (Sigma, St. Louis, MO) and were en face photographed by using SPOT ADVANCED software (version 4.0.8; Diagnostic Instruments, Inc., Sterling Heights, MI).
To monitor the expression of SAα2,3Gal or SAα2,6Gal receptors on HAE cells, the cilia of HAE cells were stained as described above. The cells were further blocked with 1% BSA in Tris-buffered saline (TBS), followed by incubation with either 5 μg/ml digoxigenin-labeled Maackia amurensis agglutinin (MAA; specific for SAα2,3Gal; Boehringer Mannheim Biochemicals, Germany) or with 1 μg/ml digoxigenin-labeled Sambucus nigra agglutinin (SNA; specific for SAα2,6Gal; Boehringer Mannheim Biochemicals, Germany) diluted with 1% BSA-TBS containing Ca2+, Mg2+, and Mn2+ at room temperature. The cells were washed with TBS and incubated with peroxidase-conjugated anti-digoxigenin Fab fragments (Boehringer Mannheim Biochemicals, Germany) in 1% BSA-TBS. Cells were washed and developed in AEC solution. The cultures were mounted and en face photographed. To count the cells, the stained cultures were observed at ×400 magnification. The cells positively stained by lectins were classified as ciliated, nonciliated, or undefined based on the presence or absence of β-tubulin staining. For each culture, no fewer than 5,000 positive cells in different, nonoverlapping fields were counted. Proportions of ciliated and nonciliated cells with respect to the total number of cells positively stained by lectins were calculated.
HAE cultures were washed with growth medium and inoculated, through the apical side, with virus at a multiplicity of infection (MOI) of 1.0 or 0.02. After 1 h of incubation at 35°C, the inoculum was removed, and the cells were washed with growth medium and incubated further at 37°C in an atmosphere of 5% CO2. At 6 h (for an MOI of 1.0) or 23 h (for an MOI of 0.02) thereafter, the cells were fixed with 4% paraformaldehyde and processed for double immunostaining.
The infected cultures were fixed and stained for the presence of cilia. After being washed with PBS and blocked with 10% normal mouse serum or 1% BSA in PBS, the human influenza virus-infected cells were incubated with peroxidase-conjugated monoclonal antibodies against human influenza virus nucleoprotein (kindly provided by Chinta M. Lamichhane, Synbiotics, Corp., College Park, MD), while the avian influenza virus-infected cells were incubated with chicken antisera against avian influenza viruses prepared in our laboratory (with HA inhibition titers of >320), followed by incubation with peroxidase-conjugated goat anti-chicken IgG (Kirkegaard & Perry Laboratories, Gaithersburg, MD). The viral antigen was visualized by incubating the cells in AEC solution. The cultures were mounted and en face photographed for representative images. To determine the number of infected cells and the tropisms of the viruses, the stained cultures were observed at ×400 magnification. The infected cells were classified as ciliated or nonciliated based on the presence or absence of β-tubulin staining. For each culture, 60 to 80 nonoverlapping fields were counted and the results were averaged. To evaluate the spread of the Q-226- or L-226-containing viruses, the stained cultures were observed at ×400 magnification. The infected cells in 40 to 50 fields were counted, and the results were averaged. Each inoculation and staining was repeated at least two times with different lots of HAE cultures, and the results were further confirmed with ready-to-use HAE cultures purchased from MatTek Corp. (Ashland, MA).
For growth curves, duplicate HAE cultures growing in the 12-mm-diameter inserts were inoculated via the apical side with each virus at an MOI of 0.2. After incubation at 35°C for 1 h, the inoculum was removed. The cells were washed five times with 200 μl of growth medium (the medium from the last wash was collected for titration, as described below) and incubated further at 37°C in 5% CO2. At different time points p.i., 200 μl of growth medium was added to each culture to harvest the progeny viruses. After 10 min of incubation at 37°C, the medium was collected, aliquoted, and stored at −80°C. At each time point, 100 μl of basal medium was also sampled and stored. For the repeat experiment, HAE cultures purchased from MatTek Corp. (growing in 8-mm-diameter inserts) were inoculated at the same MOI (0.2). The released progeny viruses were harvested similarly, with the volumes of the medium for sampling being adjusted accordingly. The viruses in both the apical and the basal medium samples were titrated in CEK cells or MDCK cells by performing both TCID50 assays as described above and immune plaque assays. The results from the two experiments were averaged, and the viral titers were expressed as log10 numbers of TCID50/ml. At the end of the sampling, the infected HAE cultures were washed and fixed. Immunostaining for cilia and viral antigen was performed to show the comparative cytopathic effects induced by infection with different influenza viruses.
To validate the results from the TCID50 assays, one set of samples from the duplicates in each experiment was titrated by immune plaque assays. Briefly, the samples were 10-fold serially diluted with M199 (avian virus samples) or PBS (human virus samples). Confluent monolayers of CEK cells or MDCK cells in 12-well plates were infected with the serial dilutions for 1 h at 37°C (avian virus samples) or for 15 min at 4°C, followed by 45 min at 37°C (human virus samples), with occasional shaking. The inoculum was removed, and the cells were washed with PBS and then overlaid with 0.9% agar solution in M199 or Dulbecco's modified Eagle's medium and supplemented with 1 μg/ml trypsin. The infected cells were incubated at 37°C in 5% CO2. At 3 days p.i., the cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, treated with 1% H2O2-methanol, and blocked with 10% normal mouse serum or 1% BSA in PBS. Then, the human influenza virus-infected cells were incubated with peroxidase-conjugated monoclonal antibodies against human influenza virus nucleoprotein, whereas the avian influenza virus-infected cells were incubated with chicken antisera against avian influenza viruses, followed by incubation with peroxidase-conjugated goat anti-chicken IgG. The viral antigen was visualized by incubating the cells in AEC solution. The positively stained plaques were counted, and the viral titers were expressed as log10 numbers of PFU/ml.
The apical surface of a fully differentiated HAE culture contains both ciliated cells and mucus-secreting, nonciliated cells. Available data show that the nonciliated cells express mainly SAα2,6Gal receptors, while the ciliated cells are rich in SAα2,3Gal receptors (15, 23, 41, 44). To test whether the HAE cells in our hands have similar characteristics, we first identified the specific cell types present at the apical surfaces of our HAE cultures. Four- to 6-week-old HAE cultures were immunostained with different cellular markers specific for ciliated cells and nonciliated, mucin-producing cells. Results showed unambiguously the presence of ciliated cells (Fig. 1A and B), goblet cells (Fig. (Fig.1A),1A), and Clara cells (Fig. (Fig.1B)1B) in our HAE cultures. Next, we examined the expression of SAα2,3Gal and SAα2,6Gal receptors on our HAE cells. Cells were stained for the ciliated-cell marker β-tubulin, followed by staining with SAα2,3Gal-specific lectin (MAA) or SAα2,6Gal-specific lectin (SNA). Results showed that SAα2,3Gal receptors are present predominantly on ciliated cells (Fig. (Fig.2A),2A), while SAα2,6Gal receptors are expressed mainly on nonciliated cells (Fig. (Fig.2B).2B). Similar results were observed in different lots of HAE cultures prepared in our laboratory and the ready-to-use HAE cultures purchased from MatTek Corp. The distribution of the receptors was further quantified by counting the cells that were positively stained by the lectins. As shown in Fig. Fig.2C,2C, among the cells positive for MAA staining (SAα2,3Gal receptor), approximately 91% are ciliated cells, while fewer than 7% are nonciliated cells. Among the cells positive for SNA staining (SAα2,6Gal receptor), approximately 77% are nonciliated cells, and 21% are ciliated cells. Despite the small proportion of cells (1% to 2%) that were difficult to define due to the extensive staining of the cultures by lectins (especially SNA), our data agree with the general pattern of the distribution of SA receptors on HAE cells cultured in vitro: SAα2,3Gal receptors are expressed predominantly on ciliated cells, while SAα2,6Gal receptors are expressed mainly on nonciliated cells and to a lesser extent on ciliated cells.
Matrosovich and colleagues reported that human influenza viruses selectively target nonciliated cells, and avian influenza viruses mainly target ciliated cells during the first round of replication in HAE cultures (23). In addition, two groups recently reported that while avian influenza viruses mainly infect ciliated cells, human viruses can infect both nonciliated and ciliated cells (15, 41).
H9N2 viruses are of interest as their importance in influenza pandemic preparedness has risen since the late 1990s. Although many H9N2 isolates of avian and swine origins have been identified, no studies were conducted to elucidate their abilities to replicate in the human respiratory tract or to determine the specific cell types that they target. To investigate the cell tropisms of avian H9N2 influenza viruses in human respiratory epithelium, eight prototypic H9N2 strains, including three early isolates (isolated before 1990) and five recent ones (isolated since 1993) of different avian species origins (duck, quail, chicken, and guinea fowl) (Table (Table1),1), were tested in HAE cultures. We infected HAE cultures with these H9N2 viruses from the apical side. After a single cycle of virus replication at 7 h p.i., we performed double immunostaining specific for the ciliated-cell markers and viral antigens to show the specific cell types targeted by these viruses, as previously reported (23). All of the tested H9N2 isolates replicated in HAE cells except for the early isolate Dk/HK/448/78. Of the viruses that could infect HAE cells, four isolates (RGWF10, Ck/HK/G9/97, Dk/HK/Y280/97, and Ck/HK/SF3/99) preferentially targeted nonciliated cells (typically more than 80% of the infected cells) (Fig. 3A, B, and G), an outcome resembling that for infection with human A/Bayern/7/95-like H1N1 and recent H3N2 viruses (23, 41) and those in the present study with human H3N2 isolates (Fig. 3E and G). The remaining isolates (RGQa88, Dk/HK/702/79, and Qa/AR/29209-1/93) infected both ciliated and nonciliated cells at ratios of approximately 1:1, 3:2, and 1:1, respectively (Fig. 3C, D, and G). The two mallard isolates, Mal/Alberta/119/98 (H1N1) and Mal/Alberta/24/01 (H7N3), representing typical duck viruses that did not circulate in terrestrial poultry, targeted preferentially ciliated cells (approximately 80% and 90% of the infected cells, respectively), as represented by the image of Mal/Alberta/24/01-infected cultures (Fig. (Fig.3F)3F) and the proportion of ciliated and nonciliated cells infected by these two viruses (Fig. (Fig.3G).3G). These results indicate that the avian H9N2 viruses analyzed in this study have different cell tropisms in HAE cells, in contrast to other waterfowl viruses represented in the present study and those reported previously (23, 41).
The differential cell tropisms of H1N1 influenza viruses in HAE cells have been reported to correlate with their receptor specificities, e.g., the viruses with SAα2,6Gal preferences mainly target nonciliated cells, whereas the viruses with SAα2,3Gal preferences target ciliated cells more efficiently than nonciliated cells (23). To verify whether this is true for H9N2 viruses, we examined the receptor specificities of the tested H9N2 viruses by performing an HA assay with SAα2,3Gal- or SAα2,6Gal-resialylated CRBCs. As shown in Fig. Fig.4,4, H9N2 viruses with nonciliated-HAE-cell preferences (represented by RGWF10 and Ck/HK/G9/97) bound exclusively to SAα2,6Gal-resialylated CRBCs, as did a prototypic human H3N2 virus (A/Panama/2007/99), indicating that they have human virus-like receptor specificities. The isolates that showed dual tropisms for HAE cells (represented by RGQa88 and Dk/HK/702/79) agglutinated both SAα2,3Gal- and SAα2,6Gal-resialylated CRBCs, suggesting that they have dual receptor specificities. Unlike the mallard H7N3 isolate, none of the tested H9N2 viruses showed exclusive preferences for SAα2,3Gal-resialylated CRBCs. Thus, the tropisms of H9N2 viruses for HAE cells correlate well with their receptor specificities.
As the cell tropism of an H9N2 virus depends on its receptor specificity, we compared the amino acids within the receptor-binding pockets of HAs that are directly involved in binding SA receptors. It is noteworthy that the four isolates displaying nonciliated-cell preferences invariably contain L-226 in their HAs, regardless of the changes at positions 183 and 190 (Table (Table2),2), whereas the three isolates that showed dual cell tropisms have Q-226. The residues at position 226 of the HA have been reported to be crucial in determining the receptor specificities of H1N1, H3N2, and H9N2 viruses (6, 22, 23, 33, 42). To investigate whether the differences in the cell tropisms of H9N2 viruses are also determined by the amino acid at position 226, we mutated L-226 in the RGWF10 HA to Q and recovered the virus using reverse genetics. The mutant virus (mWF10, Q-226) showed a dual tropism for HAE cells (infecting both ciliated and nonciliated cells) in the initial round of replication (approximately 60% ciliated and 40% nonciliated cells, respectively) (Fig. 5A and E), similar to those showed by the viruses possessing Q-226 (Dk/HK/702/79, RGQa88, and Qa/AR/29209-1/93). Conversely, when we mutated Q-226 in the RGQa88 HA to L and tested the mutant virus (mQa88, L-226), we found that mQa88 displayed a preference for nonciliated HAE cells (92% of the infected cells) (Fig. 5B and E), similar to what was observed with RGWF10 and other wild-type H9N2 viruses that have L-226 in their HAs (Fig. (Fig.3G).3G). These results indicate that the amino acid at position 226 of HA is an important determinant for the cell tropism of an H9N2 virus in HAE cultures.
The amino acids at positions 183 and 190 in the HAs of H9N2 viruses, which are also part of the receptor-binding pocket and interact with the cell surface receptors, are usually found in two combinations. Some isolates (including RGWF10 and RGQa88, used in the present study) have the histidine/glutamic acid (H/E) pair, while many others have asparagine/alanine (N/A) (Table (Table2).2). We then assessed the contributions of the amino acids at these two positions for the cell tropisms of H9N2 viruses in HAE cells. We mutated the amino acid residues at positions 183 and 190 of RGWF10 HA from H/E to N/A. These mutations did not alter the cell tropism, as the double-mutant virus (dmWF10, N-183, A-190) still targeted a high proportion of nonciliated cells (Fig. 5C and E). Similarly, when the H/E-to-N/A mutations were introduced into mQa88 HA that already contained the Q-to-L mutation at amino acid position 226, the triple-mutant virus (tmQa88, N-183, A-190, L-226) still displayed a preference for nonciliated cells (Fig. 5D and E). The receptor specificities of these mutant viruses (Table (Table2),2), measured by HA assays with SAα2,3Gal- or SAα2,6Gal-resialylated CRBCs, provide further evidence that the cell tropism of an H9N2 virus is mediated by its receptor specificity. Furthermore, mutant viruses carrying the H/A pair showed no change in cell tropism (not shown), and we were unable to produce a viable mutant virus with the N/E pair at positions 183 and 190, which may be related to molecular constraints that prevent appropriate binding to cell receptors. These results, together with the observation that the three field isolates with N/A combinations in their HAs (Ck/HK/G9/97, Ck/HK/SF3/99, and Dk/HK/Y280/97) showed preferences for nonciliated cells, confirm that the cell tropism of an H9N2 virus for HAE cells is determined by the amino acid at position 226 of HA and is not affected by the presence of H or N at position 183 or by E or A at position 190.
To examine the effect of the amino acid at position 226 of HA on the growth of an H9N2 influenza virus in HAE cells, we first investigated the spread of RGWF10, RGQa88, and the mutant viruses mWF10 and mQa88 in HAE cultures infected at a low MOI of 0.02, as previously described for other influenza viruses (23). The infected cultures were stained at 24 h p.i., allowing multiple cycles of replication. RGWF10 showed evident spread among HAE cells, as indicated by some foci consisting of continuous cells in the infected cultures (Fig. (Fig.6A).6A). In contrast, mWF10 showed less efficient spread, the viral antigen was limited mainly in single cells (Fig. (Fig.6B),6B), and the majority of the infected cells at this stage were ciliated cells. Likewise, while RGQa88 did not spread well (Fig. (Fig.6C),6C), mQa88 could spread among nonciliated cells to an extent similar to that for RGWF10 (Fig. (Fig.6D).6D). In both cases, there were more infected cells in HAE cultures inoculated with RGWF10 and mQa88 (both contain L-226; 35.0 ± 19.0 and 44.0 ± 8.0 infected cells per field, respectively) than those inoculated with mWF10 and RGQa88 (both contain Q-226; 9.0 ± 2.0 and 8.0 ± 2.0 infected cells per field, respectively). Notably, the spread of RGWF10 and mQa88 among nonciliated cells was more evident than that among ciliated cells (as shown by the representative images in Fig. 6E and F), although a considerable number of ciliated cells were also infected in the inoculated cultures at this stage. These results suggest that the variation of amino acid at position 226 of HA affects the spread of H9N2 virus among HAE cells.
To better define the effect of amino acid 226 on the growth of H9N2 virus in HAE cells, we quantified the number of progeny viruses produced following inoculation. HAE cultures were inoculated with RGWF10, RGQa88, and the mutant viruses mWF10 and mQa88 through the apical side at a low MOI of 0.2. The progeny viruses released into the apical side were harvested with growth medium at the indicated time points and titrated by performing TCID50 assays (Fig. (Fig.7A,7A, upper panel) and immune plaque assays (Fig. (Fig.7A,7A, lower panel) with primary CEK cells. CEK cells were chosen for the titrations of avian virus samples because both RGQa88 and mQa88 grew in CEK cells significantly better than in MDCK cells (>2-log10-higher titers), although RGWF10 and mWF10 grew to similar titers in these two cell types (Fig. (Fig.7B).7B). In addition, the titrations of avian and human influenza viruses were performed using CEK cells and MDCK cells, respectively, to increase the sensitivities of detection of the viruses produced in HAE cells. In interpreting the data presented in Fig. Fig.7A,7A, we assume that the viral titers are a direct indication of the virus' ability to grow in HAE cells regardless of the cell types used for titrations (regarding the titrations of human virus samples from HAE cells; see data described below). As shown in Fig. Fig.7A,7A, the wild-type and mutant H9N2 viruses grew to similar titers (lower than 4 log10 TCID50/ml or 4 log10 PFU/ml) in infected HAE cells at 12 h p.i. However, from 24 h p.i. onwards, the L-226-containing viruses (RGWF10 and mQa88) grew consistently more efficiently, and reached approximately 100-fold-higher peak titers, than the Q-226-containing viruses (mWF10 and RGQa88). The titration results provide more convincing evidence that the amino acid at position 226 in HA is important for multiple cycles of growth of H9N2 virus in HAE cells. No infectious virus was detected in the basal medium throughout the analysis period (not shown), suggesting that the progeny viruses were shed exclusively through the apical side.
Next, we investigated whether the presence of L-226 in the HA enables H9N2 viruses to grow as efficiently as human viruses in HAE cells by comparing the viral titers induced by L-226-containing H9N2 viruses (RGWF10 and mQa88) to those induced by human H3N2 viruses. HAE cultures were inoculated with human H3N2 isolate Memphis/14/98 or Panama/2007/99 at the same MOI as that for H9N2 viruses (0.2). The released progeny viruses were harvested and titrated using MDCK cells. As shown in Fig. Fig.7A,7A, the L-226-containing H9N2 viruses did not grow as efficiently as human H3N2 viruses, with peak titers between 10- and 100-fold lower than those observed for Memphis/14/98 and Panama/2007/99. The kinetics of virus production was also different, with H9N2 viruses taking more time to reach peak titers than the human viruses. Notably, when observed under the microscope, the vast majority of the superficial layer of cells infected with the human H3N2 viruses detached by 36 to 48 h p.i., leaving only the basal layer of cells. However, the cultures infected with H9N2 viruses showed comparatively less cytopathic change even at 60 h p.i., as cilia beating to various extents could still be seen. This was also reflected by the immunostaining of the H9N2 virus-infected HAE cultures at the end of the sampling, which showed the presence of a significant number of ciliated cells, particularly those infected with the Q-226-containing viruses (mWF10 and RGQa88) (Fig. (Fig.7C).7C). It is unknown whether these observations are indicative of the mild effects of H9N2 virus infection in humans. However, these findings suggest that molecular features other than amino acid 226 in the HA can also dictate the abilities of influenza viruses to replicate in HAE cells.
H9N2 influenza viruses are endemic in poultry, particularly in Asia. Some of these viruses have transmitted to humans (2, 11, 21, 29). The seroprevalence of H9N2 viral antibodies in apparently healthy individuals was reported to be around 2% (29) or higher (4, 11). Similar to human H9N2 isolates, an increasing number of avian and swine isolates are shown to have L at amino acid position 226 of their HAs (5, 19, 20, 22, 28). The acquisition of L-226 in H9N2 viruses was thought to have occurred in the avian hosts and preceded their introduction into mammals (28). The biological importance of this molecular change, however, is poorly understood, and the features of the receptor-binding site that contribute to human infection have yet to be determined. Our studies showed that the presence of L-226 in the HA enables H9N2 viruses to infect nonciliated epithelial cells during the first round of infection of an HAE in vitro model in a manner similar to that for human H1N1 and H3N2 virus infections. This human virus-like cell tropism is likely to allow the viruses to increase their abilities to cross to humans. More importantly, the Q-to-L substitution at amino acid position 226 in the HA allows H9N2 viruses to replicate more efficiently (with 100-fold-higher peak titers) in HAE cultures, indicating a possible increase of infection severity in humans. The mechanism for this increased growth ability remains unknown, although our group and others (23) have noticed that the nonciliated HAE cells seem to be more optimal for the replication and subsequent spread of influenza viruses with preferences for SAα2,6Gal receptors. Taken together, our findings provide insightful clues for explaining the presence of the L-226 signature in all but one of the H9N2 isolates from the documented clinical cases of H9N2 infection in humans (the exceptional isolate, A/Guangzhou/339/00, has an M residue at position 226).
Avian influenza viruses of some subtypes (H1N1, H3N8, H5N1, H5N3, and H7N1) have been reported to target no fewer than 80% of ciliated HAE cells (23, 41). However, the avian H9N2 isolates tested in our study did not show such tight, restricted tropism for ciliated cells. In our studies, we observed that H9N2 viruses showed only modest differences in the proportions of ciliated versus nonciliated cells infected. This is not likely due to the difference between our HAE cultures and those used by other groups, as similar results were repeated with the commercially ready-to-use HAE cultures. In addition, the mallard H1N1 and H7N3 isolates reproducibly targeted predominantly ciliated cells when tested with HAE cultures produced in our laboratory. Consistent with the cell tropisms in HAE cells, the tested H9N2 viruses displayed either exclusive SAα2,6Gal or dual receptor specificities (bound both SAα2,3Gal and SAα2,6Gal), and none of them showed exclusive SAα2,3Gal receptor specificities, in HA assays with resialylated CRBCs. The nonciliated-HAE-cell preferences and SAα2,6Gal receptor specificities of the four recent H9N2 isolates tested were obviously determined by L-226 in their HAs, as the mutation of the residue at HA position 226 resulted in the shift of the cell tropism and receptor specificity. However, it is noteworthy to mention that the three Q-226-containing isolates used in our studies have dual HAE cell tropisms and dual receptor specificities rather than the typical avian virus phenotype observed with other avian influenza viruses (ciliated-HAE-cell preference and exclusive SAα2,3Gal receptor specificity). The residue glycine (G) at amino acid position 225 of HA has been shown to be responsible for the dual receptor specificities of H1N1 viruses (8). We noticed that most avian H9N2 isolates have G and fewer have alanine (A) or aspartic acid (D) at amino acid position 225 of their HAs (22). It remains to be determined whether the dual cell tropisms and dual receptor preferences of some of the tested H9N2 viruses in the present study are due to the existence of G-225 in their HAs.
All but one H9N2 virus tested in our studies replicated to some extent in HAE cells. The exception, the early duck isolate Dk/HK/448/78, did not show any replication in HAE cells even at 24 h p.i. (data not shown). The failure of the virus to replicate in HAE cells was obviously due to the restriction in its surface proteins, as the reassortant carrying RGWF10 surface protein genes and Dk/HK/448/78 internal component genes replicated readily in HAE cells (data not shown). Considering this result and the observation that two additional early isolates (Dk/HK/702/79 and RGQa88) have dual cell tropisms, whereas the more recent isolates (Ck/HK/G9/97, Ck/HK/SF3/99, RGWF10, and Dk/HK/Y280/97) showed cell tropisms similar to those of human viruses, it seems reasonable to speculate that domestic poultry have provided an ideal environment for H9N2 viruses to evolve more proneness to infect humans. This concept is consistent with previous observations that some poultry species, chickens and quail among them, possess also SA receptors for human influenza viruses besides the typical avian virus-like receptors (7, 43).
Since the late 1990s, the H9N2 viruses have been occasionally isolated from patients displaying flu-like illness (2, 11, 29). Fortunately, there is no evidence, so far, of human-to-human transmission, and the human isolates of H9N2 virus were of purely avian origin. Our results for growth kinetics indicated that avian H9N2 viruses, even those with L-226 in their HAs, grew to peak titers significantly lower (10- to 100-fold) than two prototype human H3N2 viruses. This is in agreement with the recent findings indicating that avian influenza viruses of the H5N3 subtype grew less efficiently than human H3N2 isolates in HAE cells (41). Our observations may help explain, albeit in part, the lack of transmission of H9N2 viruses among humans. However, considering their dual- or human virus-like tropisms, H9N2 viruses are more likely to be transmitted to intermediate hosts (like pigs) that possess SA receptors for both avian and human influenza viruses and thus increase the chance of reassortment with viruses of other subtypes. Nowadays, the opportunity for such reassortment continues to exist, as Peiris and colleagues have reported that H9N2 viruses are cocirculating with contemporary human H3N2 influenza viruses in pigs in Asia (28). Furthermore, such reassortment can also occur in humans in close contact with poultry species, in which H9N2 viruses are endemic. Our results are in favor of continuous surveillance and better characterization studies of H9N2 viruses. The use of HAE cells on a routine basis could certainly be helpful in discriminating H9N2 influenza virus isolates that are more prone to infect humans.
This work was supported by grants from NIH-NIAID (RO1-AI052155) and USDA (CSRESS NO 2005-35605-15388).
We are indebted to Ruben Donis at the Centers for Disease Control and Prevention, Atlanta, GA, for critically reading and editing the manuscript. We thank Laurel Glaser from the Mount Sinai School of Medicine and Liqun Zhang from the University of North Carolina for helpful discussion in performing the studies. We express our gratitude to Robert G. Webster and Scott Krauss for providing the viruses used in the study. We especially thank Haichen Song and Erin H. Graf for help in the studies and in preparing the manuscript. We also thank Chinta M. Lamichhane for providing the monoclonal antibody used in the immunostaining.
Published ahead of print on 7 March 2007.