Infection of Human Tracheal Epithelial Cells by H5 Avian Influenza Virus Is Regulated by the Acid Stability of Hemagglutinin and the pH of Target Cell Endosomes

Despite the possible relationships between tracheal infection and concomitant infection of the terminal part of the lower respiratory tract (bronchioles/alveoli), the behavior of avian influenza viruses (AIVs), such as H5N1, in the conducting airways is unclear. To examine the tropism of AIVs for cells lining the conducting airways of humans, we established human tracheal epithelial cell clones (HTEpC-Ts) and examined their susceptibility to infection by AIVs. The HTEpC-Ts showed differing susceptibility to H5N1 and non-zoonotic AIVs. Viral receptors expressed by HTEpC-Ts bound all viruses; however, the endosomal pH was associated with the overall susceptibility to infection by AIVs. Moreover, H5N1 hemagglutinin broadened viral tropism to include HTEpC-Ts, because it had a higher pH threshold for viral–cell membrane fusion. Thus, H5N1 viruses infect human tracheal epithelial cells as a result of their higher pH threshold for membrane fusion which may be one mechanism underlying H5N1 pathogenesis in human airway epithelia. Efficient replication of H5N1 in the conducting airways of humans may facilitate infection of the lower respiratory tract.

Collectively, these studies suggest that cells lining the human conducting airways (e.g., trachea) are susceptible to infection by H5N1. They also suggest a possible relationship between tracheal

Focus-Forming Assay to Measure Infectious Titers
The MDCK cells, in 96 well plates, were washed three times with PBS (supplemented with calcium/magnesium) (PBS (+)) and then inoculated for 1 h at 37 • C with sample fluid containing virions. After that, the virus inoculum was removed and the cells were washed three times with PBS (+) and overlaid with 1% methylcellulose in minimum essential medium supplemented with 0.2% bovine serum albumin and standard antibiotics (penicillin (100 units/mL), streptomycin (100 µg/mL), and amphotericin B (250 ng/mL)). At 16 h post-infection, the cells were fixed with 4% paraformaldehyde in PBS (−). After washing three times with PBS (−), the cells were stained, as described in Section 2.8, to detect viral antigens. The titers of individual samples (FFU/mL) were determined by counting the number of fluorescent foci in the well under a fluorescence microscope fitted with filters to detect an Alexa Fluor 488-conjugated secondary antibody (see also Section 2.8.).

Establishment of HTEpC-Derived Cell Clones
The HTEpCs were immortalized by transformation with the SV40 large T-antigen gene as described previously [34]. Briefly, the packaging cell line GP2-293 (Takara Bio, Shiga, Japan) was grown in DMEM supplemented with 10% FCS and standard antibiotics. Next, GP2-293 cells in 10 cm dishes were transfected with pVSV-G and pLNCX2 (Takara Bio) using polyethylenimine (Polysciences, Warrington, PA); pLNCX2 contains the gene encoding the SV40 large T-antigen. The medium was replaced at 24 h post-transfection. At 72 h post-transfection, the supernatant containing the retrovirus was collected, passed through a syringe filter (0.45 µm pore size), and purified by ultracentrifugation (112,500× g for 2 h) on PBS (−) containing 20% sucrose (w/v). The resulting pellets were suspended in PBS (−). Next, a monolayer of primary HTEpCs was exposed to medium (AECGM supplemented with polybrene (8 µg/mL)) containing the recombinant retrovirus harboring the SV40 large T-antigen gene. After 24 h, Viruses 2020, 12, 82 4 of 24 the medium was replaced with fresh medium. After another 48 h, G-418 sulfate (100 µg/mL) was added to the medium to isolate the immortalized cells. After 2 weeks of culturing the immortalized cells with G-418 sulfate, single-cell clones were isolated using Scienceware®cloning discs (Merck, Darmstadt, Germany) or isolated by limiting dilution in a 96 well microplate to establish HTEpC-T clones. The HTEpC-T clones were then cultured in DMM prior to their use in experiments.

Viral Infection of Cells
The HTEpC-Ts, SAEC-Ts, and primary HTEpCs were cultured in 96 well plates (2.0 × 10 4 to 3.0 × 10 4 cells/well), washed twice with PBS (+), and infected with viruses at a multiplicity of infection (m.o.i.) of 10, 1, and 0.1. The m.o.i. was calculated on the basis of the cell number and the titer of the viruses used (the virus titer was determined in focus-forming assays on MDCK cells as described in Section 2.4). The cells were then incubated at 37 • C for 1 h with virus suspended in PBS (+). The viral solution was removed, and the cells were washed twice with PBS (+). The HTEpC-Ts and SAEC-Ts were cultured in DMM containing 5% FBS. The HTEpCs were overlaid with 1% methylcellulose in DMM containing 5% FBS. Virus-infected cells were subjected to immunofluorescence analysis to detect viral antigens.

Immunofluorescence Analysis
Cells were cultured in 96 well plates (2.0 × 10 4 to 3.0 × 10 4 cells/well). At 16 h post-infection with the virus, cells were fixed for 30 min at room temperature with PBS (−) containing 4% paraformaldehyde/0.1% Triton X-100, followed by washing three times with PBS (−). Viral antigens Viruses 2020, 12, 82 5 of 24 were detected by staining the cells with a rabbit polyclonal antibody raised against A/duck/Hong Kong/342/78 (H5N2) (1:1000 dilution in PBS (−)/1% bovine serum albumin) which recognizes the NP and M1 proteins of both avian and human influenza strains. Binding of the primary antibody to viral proteins was detected by an Alexa Fluor 488-conjugated secondary antibody (Thermo Fisher Scientific, Waltham, MA, USA) diluted 1:500 in PBS (−). Cell nuclei were counterstained with Hoechst 33342 (Merck). An IN Cell Analyzer 2200 (GE Healthcare, Chicago, IL, USA) apparatus was used to determine the percentage of each HTEpC-T/SAEC-T clone and primary HTEpC infected with the virus (presented as "% infectivity" in the figures). Images of infected cells were acquired by the IN Cell Analyzer 2200, and the number of antigen-positive cells and cell nuclei visible in the same field was calculated (the photographs comprised 16 different visual fields analyzed at the same time in each experiment using individual viral strains) by IN Cell Developer Toolbox software (GE Healthcare). The percentage of cells infected by the virus was calculated by dividing the number of antigen-positive cells by the total number of nuclei in the same field (×100). The number of cells counted per test well was >500 (at an m.o.i. of 10), >1000 (at an m.o.i. of 1), and >1500 (at an m.o.i. of 0.1) (the number of countable cells decreased in line with the strength of the cytopathic effects against infected cells). The results are expressed as the mean ± SD of at least three independent cultured wells.

Western Blot Analysis of Virus Binding to HTEpC-T and SAEC-T Clones
Viruses were allowed to bind to established human cell clones cultured in 12 well plates (2.0 × 10 5 cells/well). Briefly, cells were washed twice with PBS (+), inoculated with viruses at an m.o.i. of 10, and then incubated at 4 • C for 1 h to prevent endocytosis. The cells were then washed five times with ice-cold PBS (+) and lysed in PBS (−) containing 2% SDS. The protein concentration of the resulting cell lysates was measured using the bicinchoninic acid protein quantification kit (Thermo Fisher Scientific). Next, 10 µg of protein per lane was subjected to electrophoresis in a 10% SDS-polyacrylamide gel. The proteins were then transferred to PVDF membranes and blocked overnight at 4 • C in PBS (−) containing 5% non-fat milk and 0.1% Tween 20. The membrane was then exposed for 1 h at room temperature to the aforementioned rabbit polyclonal antivirus antibody (diluted 1:1000 in PBS (−) containing 5% non-fat milk and 0.1% Tween 20), followed by horseradish peroxidase-conjugated donkey anti-rabbit IgG (diluted 1:10,000 in PBS (−) containing 0.1% Tween 20; Jackson ImmunoResearch, West Grove, PA, USA). Between steps, the membranes were washed three times with PBS (−) containing 0.1% Tween 20. Reactive bands were visualized using a chemiluminescence system (Thermo Fisher Scientific) and Fuji XR film.

Evaluation of the Endosomal pH
The pH within the endosomal compartments of the HTEpC-T (LA-9) and SAEC-T (1A5) clones was measured by flow cytometry analysis as described previously [34]. Briefly, HTEpC-T and SAEC-T clones were cultured in 6 cm dishes and then loaded for 10 min at 37 • C with LysoSensor Green DND-189 (1 µm; Thermo Fisher Scientific) prepared in PBS (+). The cells were washed twice with PBS (+) and then cultured in DMM containing 5% FBS for at least 0.5 h to allow the dye to accumulate within acidic vesicles. After staining with LysoSensor Green DND-189, the cells were carefully removed from the culture plates by trypsinization. The fluorescence intensity of 10,000 events was measured using a FACSCalibur flow cytometer (BD Biosciences), and the data were analyzed using CellQuest software (BD Biosciences).

Virus Infection Inhibition Assay
The effect of increasing the endosomal pH on virus infection was examined by incubating HTEpC-T (LA-9) and SAEC-T (1A5) clones for 2 h in DMM containing 5% FBS plus bafilomycinA1 (BafA1) (1.56, 3.125, 6.25, 12.5, or 25 nM; Merck) prior to virus infection (BafA1 increases endosomal pH). After two washes with PBS (+), cells were infected with Cw/Ky (H5N1). The percentage of infected LA-9 and 1A5 cells was determined using an IN Cell Analyzer 2200 (GE Healthcare) after immunostaining at 12 h post-infection as described above.

Production of Progeny Viral Particles in HTEpCs and HTEpC-T Clones
The HTEpCs were infected (in triplicate) with viruses at an m.o.i. of 0.1. The cells were then incubated at 37 • C for 1 h with virus suspended in PBS (+). The viral solution was removed, and the cells were washed twice with PBS (+). The HTEpCs were cultured in DMM containing 0.2% bovine serum albumin and trypsin (1 µg/mL). At the indicated times post-infection, viral RNA (vRNA) titers in the cell culture supernatants of HTEpCs were measured by quantitative real-time RT-PCR as described below. Infectious viral titers in the cell culture supernatant of both HTEpCs and the HTEpC-T clone were also determined in focus-forming assays on MDCK cells as described in Section 2.4.

Quantitative Real-Time RT-PCR
Progeny vRNA was extracted from the culture supernatant of infected cells using a QIAamp Viral RNA Mini Kit (Qiagen, Hilden, Germany) and measured by quantitative real-time RT-PCR (THUNDERBIRD®Probe One-step qRT-PCR Kit, Toyobo, Osaka, Japan) using primers targeting the M gene as described previously [40]. The primer sequences were as follows: 5'-CCMAGGTCGAAACGTAYGTTCTCTCTATC-3' (forward), 5'-TGACAGRATYGGTCTTGTCTTTAGCCAYTCCA-3' (reverse). A TaqMan probe, which is an oligonucleotide with a fluorescent reporter dye attached to the 5' end and a non-fluorescent quencher (NFQ) attached to the 3' end, was used for the assay. The oligonucleotide sequence was FAM-ATYTCGGCTTTGAGGGGGCCTG-MGB-NFQ. The sequences of the primers and TaqMan probes were derived from "The Diagnosis Manual for Influenza Viruses, 3rd Edition" by the National Institute of Infectious Disease in Japan. We conducted one-step real-time RT-PCR using the CFX96 Real-Time PCR System (Bio-Rad, Hercules, CA). The cycling conditions were as follows: a reverse transcription step at 50 • C for 10 min prior to an initial denaturation step at 95 • C for 1 min, followed by amplification for 40 cycles (denaturation at 95 • C for 15 s and annealing/extension at 60 • C for 45 s).

Statistical Analysis
All data are expressed as the mean and standard deviation of at least three determinations per experimental condition. Student's t-tests were used for statistical analysis. One-way analysis of variance (ANOVA) was used, followed by Dunnett's post-hoc test for multiple comparisons. A p-value < 0.05 or < 0.01 was considered significant. Statistical analysis was performed using GraphPad Prism Version 8 software (GraphPad Software Inc.).

Establishment of Human HTEpC-T Clones
To examine the mechanism(s) underlying the susceptibility of cells in the human conducting respiratory tract to AIV infection, we generated cell clones by transforming primary HTEpC with SV40 large T-antigen and then established HTEpC-Ts after cell cloning. We isolated seven HTEpC-T clones, although one of these died out within ten passages. The other six HTEpC-T clones were sub-cultured successfully for ten passages and stably expressed the SV40 large T-antigen (Supplementary Materials Figure S1). The morphology of each cell clone was similar to that of the parent HTEpCs, although the size and shape of the clones (especially LD-2G3) were slightly different from those of the primary cells ( Figure S1).

Susceptibility of HTEpC-Ts to Infection by AIVs
To examine the susceptibility of each HTEpC-T clone to infection by AIVs, we infected them with highly pathogenic H5N1 viruses Cw/Ky (H5N1) and Ck/Eg (H5N1) or with previously circulating (non-zoonotic) AIVs Dk/Hk (H5N3) and Tk/Ont (H5N9). Human influenza viruses (Beijing (H1N1) and Panama (H3N2)) were used as controls. After infection by avian or human viruses, HTEpC-Ts expressed viral antigens ( Figure S2); however, the infectivity (the ratio of antigen-expressing cells to total cells in the same field) was different among the tested viruses ( Figure 1A-C). The HTEpC-Ts showed differing susceptibility to infection by non-zoonotic AIVs and H5N1/human influenza viruses; however, the overall pattern was similar for all clones. All HTEpC-T clones were highly susceptible to infection by H5N1/human influenza viruses (although the degree of susceptibility to H5N1 varied from clone to clone), but they were significantly less susceptible to infection by non-zoonotic AIVs (H5N3 and H5N9) than to infection by Beijing (H1N1) under all conditions (m.o.i. of 0.1, 1, and 10) ( Figure 1D-F). In addition, to examine whether the results in HTEpC-Ts can be replicated in parental HTEpCs, which are primary cultured cells, we infected the latter with human influenza, H5N1, and non-zoonotic AIVs ( Figure 1A-C). Primary HTEpCs were highly susceptible to infection by two H5N1 strains, as well as human influenza viruses, at an m.o.i. of 1 and 10 (there was no statistical significance between Beijing (H1N1) and Cw/Ky (H5N1) or Ck/Eg (H5N1)), despite the statistically lower susceptibility of HTEpCs to infection by these two H5N1 strains relative to infection by Beijing (H1N1) at an m.o.i. of 0.1 ( Figure 1D-F). However, primary HTEpCs were significantly less susceptible to infection by non-zoonotic AIVs (H5N3 and H5N9) than to infection by Beijing (H1N1) as were HTEpC-T clones under all infectious conditions (m.o.i. of 0.1, 1, and 10) ( Figure 1D-F). Notably, H5N1 viruses showed higher infectivity than non-zoonotic AIVs, even in chicken-derived cells (DF-1) known to be broadly susceptible to AIVs [41] (although DF-1 cells were markedly susceptible to infection by non-zoonotic AIVs) ( Figure 1A-F).
In addition, to examine differences within different regions of the human airway, we compared the susceptibility of HTEpC-Ts with that of other clones (i.e., 1A5 and 21E5 cells derived from primary human bronchiolar epithelial cells). Hence, 1A5 and 21E5 cells were also transformed by the SV40 large T-antigen and named SAEC-Ts; the susceptibility of 1A5 and 21E5 cells to non-zoonotic AIVs is "adequate" and "inadequate", respectively [34], although both clones were derived from the same donor. As previously reported, 1A5 cells were highly susceptible to infection by H5N1; the infection was similar to (m.o.i. of 10 and 1) or significantly higher (m.o.i. of 0.1) than that by Beijing (H1N1). The cells were also susceptible to non-zoonotic AIVs at an m.o.i. of 10 and 1, although the susceptibility to non-zoonotic AIVs was not as high as that to Beijing (H1N1) ( Figure 1A-F) [34]. The 21E5 cells were also susceptible to H5N1 viruses; their susceptibility was similar to (m.o.i. of 10) or significantly higher (m.o.i. of 1 and 0.1) than that to Beijing (H1N1), despite the fact that 1A5 and 21E5 cells differ in their susceptibility to infection by non-zoonotic AIVs ( Figure 1A-F). As seen in 1A5 and 21E5 cells, variable susceptibility to non-zoonotic AIVs was observed among the tested cell clones ( Figure 1A-F). Notably, the susceptibility of HTEpC-T clones to infection by non-zoonotic AIV Dk/Hk (H5N3) was similar to that of 21E5 cells ( Figure 1A-F) under all infectious conditions (m.o.i. of 10, 1, and 0.1) but different from that of 1A5 cells. This suggests that the mechanism(s) that determines the susceptibility of HTEpC-Ts to infection by non-zoonotic AIVs is similar to that in 21E5 cells but different from that in 1A5 cells ( Figure 1A-F). , which were derived from a region different from that of the tracheal tract, chicken-derived cells (DF-1), and primary HTEpCs, was compared with that of the HTEpC-T clones. Data are expressed as the mean ± SD of 11 (LD-2G3 and LD-3H4) or 10 (other cell clones) independent results. (D-F) Relative cell susceptibility to infection was normalized using the infectivity of Beijing (H1N1). Asterisks indicate that the infectivity of each virus was significantly different from that of Beijing (H1N1) within the same graph. A p-value < 0.01 (asterisk) was considered significant (one-way ANOVA followed by Dunnett's multiple comparisons post-hoc test).
We focused on the observation that all clones showed uniformly highly susceptibility to infection by H5N1 viruses. The presence of multiple basic amino acids within the HA cleavage site correlates with the virulence of highly pathogenic AIVs in both birds and mammals [42,43]. To examine whether differences in this amino acid motif within the HA cleavage sites of Cw/Ky (H5N1) and Dk/Hk (H5N3) contribute to their differing ability to infect HTEpC-Ts and 21E5 cells, we examined the susceptibility of these cell clones to a recombinant H5N3 virus harboring a sequence of multiple basic amino acids from Cw/Ky (H5N1) (N'-RRKKR-C': corresponding to amino acid position 325-329 in HA, according to H3 numbering [44]) within the HA cleavage site (rDk/Hk-RRKKR-HA). As shown in Figure 1, both rDk/Hk-RRKKR-HA and wild-type Dk/Hk (H5N3) infected HTEpC-Ts and 21E5 cells to a similar extent, suggesting that the multiple basic amino acid sequences within H5N1 HA are not responsible for its ability to infect these cell lines.
To assess the cell characteristics associated with susceptibility to viral infection, we also examined the expression of viral receptors SAα2,3Gal and SAα2,6Gal on HTEpC-Ts by staining cells with MAA-I/II and SNA. Both SAα2,3Gal and SAα2,6Gal were expressed by all HTEpC-T clones as well as SAEC-Ts (1A5 and 21E5 cells) and MDCK cells (virus-susceptible controls) ( Figure 2).

Binding of Viruses to HTEpC-T Clones
Next, we examined the mechanism by which AIVs infect HTEpC-Ts by examining the binding of H5N1, non-zoonotic AIVs, and human influenza viruses (used as a control) to the cells. Representative HTEpC-T (LA-9) and SAEC-T (1A5) clones (1A5 is adequately susceptible to AIVs [34], so it was used as a control) were tested. Briefly, 2.0 × 10 5 cells were inoculated with the virus at an m.o.i. of 10, lysed, and then immunoblotted with an antibody specific to the viral antigen (see Section 2.). We found no marked difference in the number of virions bound to 1A5 or LA-9 cells. Specifically, the number of virions from each strain that bound to each clone was similar, although the number of Dk/Hk (H5N3) virions bound to 1A5 and LA-9 cells was significantly higher than the number of Beijing (H1N1) virions ( Figure 3A,B). Notably, non-zoonotic H5 viruses (Dk/Hk (H5N3), rDk/Hk-RRKKR-HA, and Tk/Ont (H5N9)), which do not usually infect humans, bound to HTEpC-T clone LA-9. Cells were inoculated with viruses for 1 h at 4 • C and harvested immediately. Viral protein was detected with an anti-H5N2 polyclonal antibody. GAPDH was used as a control. The intensity of the bands representing each viral protein and GAPDH was measured using ImageJ. Representative images of the viral protein (M1) and GAPDH are shown. Uncropped images are shown in Figure S3. (B) Summary of the relative intensities of bands corresponding to the viral protein (M1) and GAPDH (shown in (A)). The intensities of the bands representing each viral protein were normalized using those of Beijing (H1N1) viral proteins in 1A5 cells; the intensities of the bands representing GAPDH were normalized using those of 1A5 cells infected with Beijing (H1N1). The intensities of Beijing (H1N1) viral proteins/GAPDH in 1A5 cells were set to 1. Data are expressed as the mean ± SD of three independent results. Asterisks indicate that the relative intensity of the bands representing each viral protein was significantly higher than that representing Beijing (H1N1) in 1A5 cells. A p-value < 0.01 (asterisk) was considered significant (one-way ANOVA followed by Dunnett's multiple comparisons post-hoc test).

Comparison of the Endosomal pH in HTEpC-T and SAEC-T Clones
After cell binding and endocytosis, influenza viruses were trafficked to the late endosomes, in which the acidic milieu induces virus-cell membrane fusion and subsequent release of the viral genome into the cytoplasm [45][46][47]. Because the number of Cw/Ky (H5N1) and Ck/Eg (H5N1) virions bound to the cells was not higher than that of the virions of other viral strains, including non-zoonotic AIVs (Dk/Hk (H5N3) and Tk/Ont (H5N9)), we hypothesized that differences in the pH of late endosomes may determine susceptibility to viral infection. Therefore, we used LysoSensor Green DND-189 (which accumulates in acidic organelles, such as late endosomes and lysosomes, and releases green fluorescence when the pH is low) to examine the endosomal pH of clones LA-9 and 1A5 [48][49][50]. Flow cytometry analysis revealed that the fluorescence intensity in acidic organelles of 1A5 cells was higher than that within those of LA-9 cells ( Figure 4A). Collectively, these data suggest that the pH in the late endosomes of LA-9 cells is higher than that in those of 1A5 cells.
which the acidic milieu induces virus-cell membrane fusion and subsequent release of the viral genome into the cytoplasm [45][46][47]. Because the number of Cw/Ky (H5N1) and Ck/Eg (H5N1) virions bound to the cells was not higher than that of the virions of other viral strains, including non-zoonotic AIVs (Dk/Hk (H5N3) and Tk/Ont (H5N9)), we hypothesized that differences in the pH of late endosomes may determine susceptibility to viral infection. Therefore, we used LysoSensor Green DND-189 (which accumulates in acidic organelles, such as late endosomes and lysosomes, and releases green fluorescence when the pH is low) to examine the endosomal pH of clones LA-9 and 1A5 [48][49][50]. Flow cytometry analysis revealed that the fluorescence intensity in acidic organelles of 1A5 cells was higher than that within those of LA-9 cells ( Figure 4A). Collectively, these data suggest that the pH in the late endosomes of LA-9 cells is higher than that in those of 1A5 cells.

Viral Infectivity Is Regulated by Gradual Changes in Endosomal pH
The different endosomal pH values observed for HTEpC-T clone LA-9 and SAEC-T clone 1A5 prompted us to examine the relationship between endosomal pH values and cell susceptibility to infection by AIVs. To do this, we increased the endosomal pH of both LA-9 and 1A5 cells by treatment with BafA1 which increases the endosomal pH by inhibiting vacuolar-type H+-ATPase activity. Even in 1A5 cells, which are highly susceptible to AIVs [34], BafA1 prevented infection by the H5N1 virus in a dose-dependent manner ( Figure 4B). Much lower concentrations of BafA1 inhibited viral infection of LA-9 cells ( Figure 4B,C), confirming that the endosomal pH of LA-9 cells is higher than that of 1A5 cells.  Figure 1). Representative micrographs of antigen-positive cells treated with BafA1 are shown. Cell nuclei were also counted. Viral infectivity was determined as described in Figure 1. Data are expressed as the mean ± SD of three independent results. Scale bars, 100 µm. (C) Summary of relative infectivity in panel (B). The infectivity in the absence of reagent was set to 1. Asterisks indicate an infectivity ratio that was significantly lower than that for 1A5 cells at the same concentration of reagent. A p-value < 0.01 (asterisk) was considered significant (Student's t-test).

Viral Infectivity Is Regulated by Gradual Changes in Endosomal pH
The different endosomal pH values observed for HTEpC-T clone LA-9 and SAEC-T clone 1A5 prompted us to examine the relationship between endosomal pH values and cell susceptibility to infection by AIVs. To do this, we increased the endosomal pH of both LA-9 and 1A5 cells by treatment with BafA1 which increases the endosomal pH by inhibiting vacuolar-type H+-ATPase activity. Even in 1A5 cells, which are highly susceptible to AIVs [34], BafA1 prevented infection by the H5N1 virus in a dose-dependent manner ( Figure 4B). Much lower concentrations of BafA1 inhibited viral infection of LA-9 cells ( Figure 4B,C), confirming that the endosomal pH of LA-9 cells is higher than that of 1A5 cells.

The Effect of HA Acid Stability on the Susceptibility of HTEpC-Ts to Infection
Next, we examined viral factor(s) that may affect infectivity. Because the infection of cells requires viral-cell membrane fusion, which is induced by a pH-dependent HA conformational change, we investigated whether the acid stability of the HA molecule affects the susceptibility of HTEpC-Ts to viral infection. Previously, we reported that H5N1 viruses undergo HA-dependent viral-cell membrane fusion at higher pH values than non-zoonotic AIVs [34] (the pH threshold for the HA protein to induce membrane fusion is summarized in Supplementary Materials as rDk/Hk-ThailandHA, rDk/Hk-Indonesia, and rDk/Hk-ShanghaiHA, respectively) was able to infect LA-9 cells with a mildly acidic endosomal pH environment as did wild-type Cw/Ky (H5N1) and Ck/Eg (H5N1) ( Figure 5). The infectivity of recombinant Dk/Hk (H5N3) harboring H5N1 HA genes was significantly higher than that of the parent virus (Dk/Hk (H5N3)) in LA-9 cells. By contrast, recombinant Dk/Hk (H5N3) harboring sequences with multiple basic amino acids (N'-RRKKR-C') within the HA cleavage site (rDk/Hk-RRKKR-HA) derived from Cw/Ky (H5N1) infected LA-9 cells poorly as did the parent virus Dk/Hk (H5N3). These viruses were able to infect 1A5 cells, which have an acidic endosomal pH, although the infectivity of recombinant Dk/Hk (H5N3) harboring H5N1 HA genes was significantly higher than that of the parent virus Dk/Hk (H5N3) in 1A5 cells ( Figure 5).  1.2)). Viral infectivity was determined as described in Figure 1 Data are expressed as the mean ± SD of eight independent results. Asterisks indicate that the infectivity of each virus was significantly different from that of Dk/Hk (H5N3) within the same graph. A p-value < 0.01 (asterisk) was considered significant (one-way ANOVA followed by Dunnett's multiple comparisons post-hoc test).

The Effect of HA Acid Stability on the Susceptibility of Primary HTEpCs to Infection
To examine whether the effect of HA acid stability on the susceptibility of HTEpC-Ts can be replicated in HTEpCs, which are primary cultured cells, we infected the latter with recombinant Dk/Hk (H5N3) harboring the H5N1 HA genes described above. The susceptibility of HTEpCs to infection by these AIVs was similar to that of HTEpC-Ts (Figures 5 and 6). The HTEpCs were highly susceptible to infection by H5N1 and recombinant Dk/Hk (H5N3) harboring H5N1 HA genes; however, they were only moderately susceptible to infection by H5N3 and rDk/Hk-RRKKR-HA (as were HTEpC-T clones) ( Figure 6A,B). In HTEpCs, the infectivity of recombinant Dk/Hk (H5N3) harboring H5N1 HA genes was significantly higher than that of the parent virus Dk/Hk (H5N3) (Figure 6A,B). By contrast, the susceptibility of primary HTEpCs and HTEpC-T clones to infection by non-zoonotic AIVs Dk/Hk (H5N3) and rDk/Hk-RRKKR-HA showed some differences; primary HTEpCs were more susceptible to infection by Dk/Hk (H5N3) and rDk/Hk-RRKKR-HA than HTEpC-T clones, especially at an m.o.i. of 10 ( Figures 1, 5 and 6A,B).  Figure 5). Data are expressed as the mean ± SD of four independent results. Asterisks indicate that the infectivity of each virus was significantly different from that of Dk/Hk (H5N3) within the same graph. A p-value < 0.01 (asterisk) was considered significant (one-way ANOVA followed by Dunnett's multiple comparisons post-hoc test).

Viral Growth Kinetics in Primary HTEpCs and Their Established Clones
In the final series of experiments, we examined the possible relationship between susceptibility to infection and viral growth kinetics in both HTEpCs and their established clones. We infected HTEpCs with H5N1, non-zoonotic virus, and its recombinant AIVs at a low m.o.i. (0.1). The replication kinetics of the H5N1 virus and recombinant Dk/Hk (H5N3) harboring H5N1 HA genes were significantly higher than those of Dk/Hk (H5N3), regardless of the H5N1 strain (Figure 7). The kinetics of rDk/Hk-RRKKR-HA were similar to those of the parent virus Dk/Hk (H5N3) (Figure 7A,B). We observed a similar pattern between the copy number and infectious viral titer of virions released from the infected cells, although the copy number was higher than the infectious titer ( Figure S4). This result was reflected by the positive correlation between the copy number and infectious viral titer for each recombinant viral strain ( Figure S4). We also evaluated viral replication kinetics in LA-9 cells and compared the results for primary HTEpCs with those for HTEpC-T. The viral growth curve obtained for LA-9 cells was similar to that for primary HTEpCs, despite the lower infectious titer in LA-9 cells (Figure 7). The replication kinetics of the H5N1 virus in LA-9 cells were significantly higher than those of Dk/Hk (H5N3). The growth kinetics of recombinant Dk/Hk (H5N3) harboring H5N1 HA genes were also significantly higher than those of the parent virus Dk/Hk (H5N3) in LA-9 cells, although statistical significance was rarely observed for some recombinant viral strains harboring H5N1 HA genes ( Figure 7C).  The infectious viral titers in infected cells were measured by infecting HTEpCs with AIVs and recombinant viruses, as described in (A). At the indicated times post-infection, the infectious titer of progeny virions in the culture supernatants was determined in a focus-forming assay. (C) The representative HTEpC-T clone LA-9 was infected with the AIVs described in (A) under the same infectious conditions. At the indicated times post-infection, the infectious titer of progeny virions in the culture supernatants was determined in a focus-forming assay. Data are expressed as the mean ± SD of three independent results. Asterisks indicate virus titers that are significantly higher than that of Dk/Hk (H5N3). A p-value < 0.05 (single asterisk) or < 0.01 (double asterisk) was considered significant (one-way ANOVA followed by Dunnett's multiple comparisons post-hoc test).

Discussion
Here, we report four main findings with respect to the mechanism underlying the susceptibility of human tracheal cells to infection by AIVs: (i) HTEpC-Ts showed differing susceptibility to infection by H5N1 and non-zoonotic AIVs; (ii) viral receptors expressed by HTEpC-Ts were able to bind viral particles at similar levels; (iii) the pH in the endosomes of HTEpC-Ts determined their susceptibility to infection by AIVs; and (iv) the H5N1 HA protein, with a higher pH threshold for HA-mediated membrane fusion, broadened the viruses' tropism for HTEpC-Ts.
Both SAEC-T clone 21E5 and HTEpC-Ts were highly susceptible to infection by human influenza and H5N1 viruses, but they were less susceptible to infection by non-zoonotic AIVs, such as H5N3 and H5N9, regardless of the presence of multiple basic amino acids in the HA cleavage site (Figure 1).
The susceptibility of HTEpC-Ts is in agreement with the results of other studies on receptor binding and/or replication of AIVs [51][52][53][54]. For example, van Riel et al. [53,54] showed that AIVs, including H5 subtypes, bound weakly to sections of the human trachea and bronchus but moderately to bronchioles. By contrast, Matrosovich et al. [51] confirmed that AIV strain A/mallard/Alberta/119/98 (H1N1) infects differentiated nasal and tracheobronchial cells of the human airway. Furthermore, Nicholls et al. [52] used ex vivo tissue culture models to demonstrate the presence of novel viral antigens in the human upper respiratory tract (i.e., nasopharynx, adenoid, and tonsil) with A/Vietnam/3046/04 (H5N1) at 24 h post-infection. Moreover, they showed that the number of progeny virions in both nasopharyngeal biopsy and primary nasopharyngeal epithelial cells infected with the same virus strain increased over time. The results reported herein, together with those in the studies described above, suggest that the human upper respiratory or conducting airways (e.g., the trachea) may be susceptible to AIVs, although the degree of infectivity may vary. By contrast, we found here that individual HTEpC-T clones show slightly different susceptibility to infection, even by the same viral strain (Figure 1). This variable susceptibility among HTEpC-Ts could be due to the different cellular origins (e.g., ciliated epithelial cells, goblet cells, Clara cells, and basal cells) or a different differentiation stages within the same cell. The variability observed in HTEpC-T clones is likely to correlate with the different morphology of HTEpC-Ts (i.e., morphological differences between LD-2G4 and other cell clones) ( Figure S1).
The patterns of viral receptor expression in the human airways could be associated with differences in susceptibility within different regions of the human airway. Some studies have reported that the avian receptor SAα2,3Gal (stained by MAA-II) is expressed mainly in the terminal part of the human LRT [17,55], suggesting that human infection by AIVs could be restricted to bronchioles and alveoli. However, the SAα2,3Gal avian receptor can also be stained by another lectin, MAA-I. In this study, staining HTEpC-Ts with lectins revealed abundant expression of SAα2,3Gal (stained by either MAA-I or MAA-II) and SAα2,6Gal residues (stained by SNA), which are recognized by avian and human influenza viruses, respectively ( Figure 2). However, the pattern of SA expression by HTEpC-Ts does not correspond completely with the pattern in human respiratory tissue (in vivo). Generally, SAα2,6Gal is expressed in the upper to lower airway; by contrast, the analysis of the expression of glycanic structures (i.e., the sub-terminal residues of SAα2-3 oligosaccharides) suggests that expression of SAα2,3Gal differs according to variations in the sub-terminal SAα2-3 oligosaccharide residues [52,[56][57][58][59]. For example, SAα2,3-Galβ(1-4)GlcNAc, which is an N-linked glycan recognized by MAA-I, is likely to be expressed throughout the whole respiratory tract (although expression in the tracheal region has not been well analyzed), whereas that of SAα2,3-Galβ(1-3)GalNAc, which is an O-linked glycan recognized by MAA-II, is detected mainly in the bronchioles and alveoli [17,55,57,59]. As shown by both Matrosovich et al. [51] and Nicholls et al. [52], as well as the data presented herein, one reason that AIVs infect epithelial cells lining the upper and conducting airways is their expression of SAα2,3-Galβ(1-4)GlcNAc which is recognized by MAA-I and expressed broadly throughout the human respiratory tract. The presence of SAα2,3-Galβ(1-4)GlcNAc suggests that AIVs may bind to and/or infect epithelial cells lining the human conducting airways in vivo and in vitro; however, our finding that AIVs infect HTEpC-Ts may also be due to the fact that the HTEpC-T clones express high levels of the avian receptor SAα2,3-Galβ(1-3)GalNAc, which is rarely expressed in the human trachea in vivo [17,55], as well as SAα2,3-Galβ(1-4)GlcNAc. This unexpected pattern of viral receptor expression may be caused by the conditions used to culture HTEpC-Ts. Unfortunately, the characteristics of cells in liquid culture do not always mirror those of their in vivo counterparts, because the differentiation levels are different. Usually, traditional culture systems do not enable complete cell differentiation; indeed, full differentiation requires an air-liquid interface (ALI) culture system which allows for the generation of fully differentiated epithelial cells such as ciliated cells, goblet cells, and Clara (club) cells [60][61][62][63]. Therefore, the unexpected (and sufficient) expression of SAα2,3-Galβ(1-3)GalNAc by the HTEpC-Ts examined herein may be due to the lack of differentiation. Another possibility is individual differences in the expression of SAα2,3-Galβ(1-3)GalNAc by the primary epithelial cells from which the HTEpC-Ts were derived. Indeed, one limitation of this study was that the HTEpC-Ts were derived from only one donor. Further studies should examine HTEpC-Ts derived from multiple donors and use both "traditional" and ALI culture systems.
The HTEpC-Ts maintained in liquid culture expressed high levels of SA (SAα2,3Gal and SAα2,6Gal) residues; this explains the high numbers of viral particles bound to the cells in the virus-binding assay ( Figure 3). The analysis of cell surfaces showed that H5N1, H5N3, and H5N9 viruses bound similarly to HTEpC-T (LA-9) and SAEC-T (1A5) cells which are marginally and adequately susceptible, respectively, to infection by AIVs. These results suggest that an event(s) arising after viral internalization by endocytosis is responsible for differences in the susceptibility of LA-9 and 1A5 cells to infection by AIVs. After internalization via endocytosis, influenza viruses undergo a low-pH-dependent conformational change in the HA molecule, leading to membrane fusion and subsequent infection [45][46][47]. Previously, we showed that H5N1 viruses harboring an HA gene from a different clade have a higher pH threshold (pH 5.625-5.75) for HA-mediated membrane fusion than non-zoonotic avian H5 viruses such as Dk/Hk (H5N3) and Tk/Ont (H5N9) (pH 5.125-5.375) (see Table S1) [34]. The higher pH threshold of H5N1 viruses may account for the results reported herein, i.e., all HTEpC-T and SAEC-T (1A5 and 21E5) clones were highly susceptible to H5N1. By contrast, differences in the susceptibility of LA-9 and 1A5 cells to infection by Dk/Hk (H5N3) and Tk/Ont (H5N9) were associated with cellular endosomal pH values; H5N3 and H5N9 viruses do not undergo viral-cell membrane fusion until the endosomal pH becomes acidic (pH threshold of HA: pH 5.125-5.375) [34]. This prompted us to compare the endosomal pH values in different clones. Flow cytometry analysis using LysoSensor Green DND-189 revealed that the pH in the late endosomes of LA-9 cells is higher than that in those of 1A5 cells ( Figure 4A), which supports our hypothesis that the endosomal pH in individual epithelial cell clones determines their susceptibility to infection by AIVs. The role of different endosomal pH values in 1A5 and LA-9 cells was highlighted by treating cells with BafA1. The infection of LA-9 cells was more sensitive to the alkalinizing effect of BafA1 than that of 1A5 cells ( Figure 4B,C); this finding indicates that the pH environment in the endosomes differs considerably among these cells, and it also confirms the data from the flow cytometry experiments using a pH-dependent indicator which showed that the endosomal pH in LA-9 cells was higher than in 1A5 cells. From a physiological point of view, epithelial cells (such as LA-9 cells) with a mildly acidic endosomal pH environment may help to defend tissues from some pathogens including viruses and toxins that can be activated by an acidic environment. Thus, human tracheal epithelia, including cells with a mildly acidic endosomal pH, might protect the conducting airways from invasion, although the ratio of cells with an acidic/mildly acidic endosomal pH in tracheal epithelia is unknown at this moment. Thus, we found that the less acidic endosomal environment within HTEpC-Ts enables them to resist infection by H5N3 and H5N9 viruses. By contrast, HTEpC-Ts may succumb to infection by H5N1 viruses because the higher pH threshold of this virus allows it to overcome the mildly acidic endosomal pH.
This marked ability of H5N1 to infect HTEpC-Ts prompted us to undertake functional analyses of the H5N1 HA protein which may be associated with broader tropism. Recombinant H5N3 viruses possessing the HA protein from H5N1 were able to infect HTEpC-T clone LA-9 as were wild-type H5N1 viruses ( Figure 5). These results led us to surmise that the H5N1 HA protein plays an important role in broadening cell tropism to include HTEpCs as was described previously for human epithelial cells lining the bronchioles [34]. Beare et al. [64] reported that non-zoonotic AIVs (not the H5 subtype) caused mild or no clinical symptoms in experimentally infected human volunteers, and they detected a 4-fold or greater increase in HI titer in only a few volunteers. This resistance to AIV infection could be ascribed to their possession of a stable HA protein because non-zoonotic AIVs have a lower pH threshold for HA-mediated membrane fusion [34]. It may be that a different outcome would be observed if AIVs of other subtypes, such as H5N1 (which has a higher pH threshold), were used in such an experiment. However, conducting an experiment in which human volunteers are experimentally infected with harmful viruses, including H5N1, is unlikely because of ethical concerns. It would be interesting to examine possible relationships between the viral infection of HTEpC-T clones and the pH threshold for HA-mediated membrane fusion in other AIV strains (H5N6, H6N1, H7N2, H7N3, H7N4, H7N7, H7N9, H9N2, H10N7, and H10N8) that can be transmitted directly from birds to humans [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16].
Recent reports, including our own, have shown that the acid stability of the H5N1 HA protein is more vulnerable to pH (i.e., the pH threshold of H5N1 viruses is higher) than that of non-zoonotic AIVs [34,[65][66][67][68][69][70][71]. After viral internalization via endocytosis, this biochemical property (i.e., acid stability) of H5N1 may support infectivity as shown by the results in human tracheal (present results) and bronchiolar (previous results) epithelial cells [34]. An acid-destabilized HA protein of H5N1 could increase the infection of target cells once the virus enters the target cell; by contrast, acid destabilization is likely to disturb their persistence in host environments. Usually, AIVs replicate in the intestinal tract of waterfowl [72][73][74][75]. The contents of the avian intestine are relatively acidic, although there are slight differences among species [76][77][78][79]. In this case, AIVs possessing acid-stable HA molecules (e.g., the H5N3 and H5N9 strains used in this study) have an advantage in that they can persist in the intestinal tract of waterfowl, although the degree to which viruses infect intestinal cells is another aspect to consider. Further studies are required to reveal the endosomal environment of avian intestinal cells and the relationship between their endosomal pH values and susceptibility to infection by AIVs. Should AIVs find their way into the human airway after close contact with birds, then the virus may attach to the surface of airway epithelial cells. Indeed, the human nasal mucosa is relatively acidic [80][81][82]. Such an acidic host environment (e.g., the avian intestinal tract and the human upper airway) may prevent H5N1 from persisting on the surface of host cells, although the virus can replicate efficiently once it has entered the target cell. This might be one reason that cases of human infection by H5N1 have been sporadic to date, although human infection by this virus is more common in some areas including Egypt.
We also evaluated the susceptibility of primary HTEpCs to infection by AIVs. The susceptibility of primary HTEpCs to infection by AIVs broadly reflected that of HTEpC-Ts (Figures 1, 5 and 6), suggesting that HTEpC-Ts can be used as a cell model for the detailed analysis of AIV infection of human conducting airways. Regarding the susceptibility of primary HTEpCs to infection by AIVs, the antigen-positive/negative ratio of AIVs in HTEpCs was similar to that in HTEpC-T clones, although this ratio for non-zoonotic AIVs in primary HTEpCs (especially in cells infected at an m.o.i. of 10) was different from that observed in HTEpC-T clones (i.e., primary HTEpCs were more susceptible than HTEpC-T clones to infection by non-zoonotic AIVs) ( Figure 1A-C, Figure 5, and Figure 6A,B). These phenomena suggest two possibilities with respect to the endosomal pH in primary HTEpCs. First, we failed to generate HTEpC-T clones with an acidic endosomal pH, although primary HTEpCs may also include cells with an acidic endosomal pH. Second, the pH in primary HTEpCs is often mildly acidic; however, endogenous proteases may support viral infection of HTEpCs because tracheal proteases, such as transmembrane protease serine 2 (TMPRSS2) [83] and human airway trypsin-like protease (HAT) [84], which are required to process released virions, allow for multiple cycles of viral infection [85]. Further studies should generate more than six HTEpC-T clones and analyze both the endosomal pH and susceptibility to infection by AIVs.
Even if the difference in infectivity between H5N1 and non-zoonotic AIVs in primary HTEpCs is not as marked as that in HTEpC-Ts, the difference seems to reflect the amount of virions released from infected HTEpCs upon long-term cultivation at a low m.o.i; this is because influenza viruses exhibit exponential growth as a result of multiple cycles of viral infection during long-term cultivation. During the long-term cultivation of virus within primary HTEpCs, the amount of H5N1 virions released at each time point was significantly higher than the amount of non-zoonotic AIV virions (Figure 7). In addition, the replication kinetics of recombinant Dk/Hk (H5N3) harboring H5N1 HA genes were similar to those of the H5N1 virus (Figure 7). These results suggest that the H5N1 virus can replicate continuously, because it has a higher pH threshold for membrane fusion; it can then spread to the entire tracheal region within the respiratory tract. This could be one mechanism underlying H5N1 pathogenesis in human airway epithelia.

Conclusions
Because viral replication in the tracheal region of the respiratory tract may result in infection of the terminal part of the LRT (i.e., bronchioles/alveoli) which can result in ARDS, it is important to know the degree to which H5N1 infects tracheal epithelial cells. The results presented herein suggest that the susceptibility of tracheal epithelial cells to infection by H5N1 is determined by the balance between the acid stability of the HA molecule (i.e., pH threshold of HA-mediated membrane fusion) and the pH of the endosomal compartment. Further research into the HA-governed mechanisms underlying membrane fusion will help us to develop effective strategies for the surveillance and control of AIVs.
Supplementary Materials: The following are available online at http://www.mdpi.com/1999-4915/12/1/82/s1. Figure S1. Establishment of human tracheal epithelial cell clones. Figure S2. Viral antigen expression in human tracheal epithelial cell clones. Figure S3. Uncropped images of the Western blots in Figure 3. Figure S4. Graphs showing the correlation between copy number and infectious titer of virus particles released from infected HTEpCs. Table S1. Summary of the pH thresholds for viral-cell membrane fusion for all AIV strains used in this study.