Analysis of the desialidation process of the haemagglutinin protein of influenza B virus: the host-dependent desialidation step

The haemagglutinin (HA) and neuraminidase (NA) proteins of influenza virus are transmembrane glycoproteins. The HA protein has many carbohydrate chains and some of these contain sialic acid as the terminal residue. The NA protein removes sialic acid from the HA protein as well as cellular glycolipids or glycoprotein, thereby preventing the aggregation of virus particles and permitting the release of virus from host cell receptors (Palese et al., 1974 ; Shibata et al., 1993 ). The process of desialidation of the HA protein by the NA protein has not yet been made clear. We used haemadsorption activity as a means of studying the desialidation process. Some of the carbohydrate chains on the HA protein must have their sialic acid removed to allow for the haemadsorption of influenza A viruses H7 (Ohuchi et al., 1995 ) and H1 (Tong et al., 1998 ) as well as influenza B viruses (Brassard & Lamb, 1997 ; Luo et al., 1999 ). In our previous report, we showed that desialidation of the HA protein of influenza B virus was necessary for haemadsorption to occur on MDCK cells infected with this virus (Luo et al., 1999 ). An NA inhibitor, Zanamivir, is widely known to be useful for inhibition of replication of influenza A and B viruses by inhibiting the enzymatic activity of the NA protein (Itzstein et al., 1993 ). Morris et al. (1999) revealed that Zanamivir did not penetrate MDCK cells. We showed that Zanamivir inhibited haemadsorption on these cells infected with influenza B virus. Therefore, we concluded that, in MDCK cells, desialidation of the HA protein occurred on the cell surface (Luo et al., 1999 ). However, the reason why desialidation does not occur in infected cells remains unknown. Recently, however, we found that Zanamivir did not inhibit haemadsorption on COS cells infected with influenza B virus. Comparison of the localization of the HA and NA proteins in both cell types and experiments carried out at low temperatures led us to suggest that the accumulation of the HA and NA proteins in the Golgi was a key point as to whether desialidation occurred within the cell or on the cell surface. Cells and viruses.
COS and MDCK cells (1x10⁵ cells per 18 mM coverslip), prepared on coverslips 18 h earlier, were washed with serum-free Eagles minimum essential medium (MEM0). The cells were then inoculated with 50 µl of virus (0·91·2 p.f.u. per cell) and the coverslips were incubated at room temperature for 30 min. After the cells were washed twice with PBS, MEM0 was added and the preparations were incubated at 3437·5 °C, as indicated in the text. Haemadsorption with 0·5% chicken red blood cells (CRBCs) was performed on ice for 15 min. Samples were then washed with cold PBS to remove unadsorbed erythrocytes and fixed with ethanolacetone (1:1). Influenza viruses B/Lee/40, B/Kanagawa/73 and ts-7, which is a temperature-sensitive NA mutant of influenza B/Kanagawa/73 virus (Yamamoto-Goshima et al., 1994 ), were propagated in MDCK cells at 34 °C.

Expression of the HA gene on COS cells.
Transfection of the HA cDNA was carried out with lipofectamine as described previously (Morishita et al., 1996 ).

Antisera and immunostaining.
Anti-HA (rabbit) and anti-NA (mouse) antisera against influenza B/Kanagawa/73 virus were kindly supplied by T. Morishita (Aichi Prefectural Institute of Public Health, Japan). Indirect immunofluorescent staining by rhodamine (for NA) or FITC (for HA) was carried out, as described previously (Nobusawa & Nakajima, 1988 ), and observations were made using confocal fluorescent microscopy (MRC-1000, Bio-Rad).

Treatment of infected cells with Zanamivir.
Zanamivir (4-guanidino-2,3-dehydro-N-acetylneuraminic acid) was kindly supplied by Glaxo Wellcome. Following virus infection, MDCK cells were washed with PBS, after which maintenance medium containing different amounts of Zanamivir was added. The cells were then incubated at 3537 °C for 820 h.

Assay of the NA activity on the infected cells.
Infected cells were harvested together with maintenance medium and NA activities were assayed using fetuin as a substrate in an 18 h incubation at 37 °C, as described previously (Palmer et al., 1975 ).

Two-dimensional (2D) gel electrophoresis.
After infection of MDCK cells (3x10⁵ cells per 3 cm dish) with influenza B virus, infected cells were incubated at 37 °C for 8 h and medium was replaced with methionine-free MEM supplemented with [³⁵S]methionine (100 µCi per dish). After a 15 min labelling period and a 30 min chase period, the HA proteins were immunoprecipitated with anti-HA antibody as follows. Precipitated cells were solubilized with 1 ml lysis buffer [20 mM TrisHCl, pH 7·4, 2 mM EDTA, 2% (w/v) Triton X-100, 0·3 M NaCl and 10 µg/ml each of chymostatin, leupeptin, antipain and pepstatin] by shaking at room temperature for 10 min. The samples were then centrifuged to collect the supernatant. Subsequently, 20 µl of anti-HA monoclonal antibody and protein ASepharose were added to the supernatant and incubated at 4 °C overnight. The antibody complex was precipitated by centrifugation. The pellet was washed three times with 1 ml 20 mM TrisHCl, pH 8·6, and 0·5 M NaCl. Isoelectrofocusing with a Bio-Rad IPG strip and subsequent SDSPAGE were carried out according to the directions of the manufacturer (Bio-Rad).

Host-dependent inhibition of haemadsorption activity by Zanamivir
COS and MDCK cells were inoculated with influenza B/Kanagawa/73 virus and incubated at room temperature for 30 min. After the cells were washed, MEM0 containing different amounts of Zanamivir was added and the cells were incubated at 37 °C. Fig. 1 shows the time-course of haemadsorption and NA activities of MDCK (Fig. 1a) and COS (Fig. 1b) cells after infection with influenza B/Kanagawa/73 virus. Haemadsorption reached maximal levels at around 8 and 11 h post-infection (p.i.) on MDCK and COS cells, respectively. Therefore, the effect of Zanamivir on MDCK cells was measured at 8 h p.i. (Fig. 1c) and on COS cells at 11 h (Fig. 1d). Haemadsorption activity of virus-infected MDCK cells was inhibited in a dose-dependent manner but that of virus-infected COS cells was unaffected. In the presence of Zanamivir, the HA and NA proteins on the cell surface of infected MDCK and COS cells were examined at 8 h (MDCK cells) and 11 h (COS cells) p.i. by immunofluorescent staining after paraformaldehyde fixation. Zanamivir did not affect the migration of either type of protein to the surface of either MDCK or COS cells (data not shown). In order to eliminate the possibility that the binding of CRBCs to Zanamivir-treated infected COS cells was an artefact, CRBCs were treated with Vibrio cholera NA (bNA) (25 mU/ml) for 30 min at 37 °C to remove sialic acid from their surfaces. bNA-treated CRBCs abrogated haemadsorption on the infected COS cells with or without Zanamivir (Fig. 1d), suggesting that haemadsorption of these infected cells in the presence of this compound depended on their binding to the sialic acid components of CRBCs.

(28K): Fig. 1. Time-course of haemadsorption and NA activity after B/Kanagawa/73 virus infection of (a) MDCK and (b) COS cells. The effects of Zanamivir on haemadsorption of (c) MDCK and (d) COS cells are also shown. Haemadsorption and NA activity were measured at 8 h (MDCK cells) and 11 h (COS cells) p.i. Haemadsorption was carried out at 4 °C for 15 min with CRBCs, then haemadsorbed cells were counted under a microscope. NA activity was measured as described in Methods. Haemadsorption activity, NA activity and haemadsorption activity with bNA-treated CRBCs are represented as closed, shaded and open bars, respectively.

The carbohydrate chains of the HA protein may not be the same in COS and MDCK cells. Therefore, it is possible that desialidation of the HA protein synthesized in the COS cells was not necessary for haemadsorption to occur. However, we have already demonstrated that transfection of pME18S-HAcDNA of influenza B/Kanagawa/73 virus to COS cells did not enable these cells to haemadsorb without the co-transfection of pME18S-NAcDNA of influenza B/Lee/40 virus or treatment of the cells with bNA (Luo et al., 1999 ), which indicated that desialidation was required for the HA protein synthesized in COS cells to have haemadsorption activity.

NA activity of MDCK and COS cells was not different
NA activity of infected MDCK and COS cells was inhibited by Zanamivir in a similar manner (Fig. 1c, d). Therefore, the NA protein synthesized in COS cells was not resistant to this compound. As shown in Fig. 1, the NA activity of infected MDCK or COS cells was nearly the same. Therefore, it seems reasonable to exclude the possibility that a difference in the content of the NA protein in these two types of cells may make a difference in haemadsorption activity in the presence of Zanamivir.

Sensitivity to desialidation of the HA protein on COS and MDCK cells measured with bNA was not different
If the HA protein of COS cells was more highly sensitive to desialidation by the NA protein than that of MDCK cells, removal of the sialic acid from the HA protein on COS cells might occur before Zanamivir could act to inhibit NA activity. To test this possibility, the sensitivity of the HA protein to desialidation was measured on COS and MDCK cells (Fig. 2). MDCK cells infected with influenza ts-7 virus, a temperature-sensitive mutant virus containing a mutated NA gene that causes a defective NA protein to be synthesized at 37·5 °C (Yamamoto-Goshima et al., 1994 ), were incubated at 37·5 °C for 8 h. Then, different amounts of bNA were added, incubation was continued for 20 min and haemadsorption activity was measured. COS cells, however, did not display a temperature-sensitive phenotype with this mutant (data not shown). For COS cells, after the transfection of HAcDNA, cells were incubated for 40 h, then different amounts of bNA were added. Incubation continued for 20 min and haemadsorption activity was determined. After fixing the cells with ethanolacetone, the percentages of haemadsorbed cells and cells stained with anti-HA antibody were assessed. Recovery of haemadsorption on both types of cells was dependent on the concentration of bNA (Fig. 2). A total of 40 µl bNA (25 mU/ml) was sufficient for haemadsorption activity on 50% of both COS and MDCK cells. The decline in recovery with dilutions of bNA was also similar in both cell types. Therefore, the sensitivity of the HA protein to desialidation in MDCK and COS cells might not explain the difference in sensitivity observed using Zanamivir to inhibit haemadsorption. Based on these experiments, we suggest that, in MDCK cells, desialidation of the HA protein occurs on the cell surface, whereas in COS cells, desialidation occurs while the HA protein is still within the cells.

(14K): Fig. 2. Sensitivity to desialidation of the HA protein on MDCK and COS cells. Two sets of cells were fixed after adsorption and the HA-expressing cells were detected with an anti-HA mouse antibody and FITC-conjugated anti-mouse IgG antibody. The vertical line indicates the percentage of haemadsorbed cells per HA-expressing cells. •, MDCK cells; , COS cells.

Localization of the HA and NA proteins in infected COS and MDCK cells
Localization of the HA and NA proteins in infected COS and MDCK cells was analysed by confocal microscopy. As shown in Fig. 3, the HA and NA proteins existed at similar locations in both infected cells. However, staining of the HA and NA proteins in the Golgi looked stronger in COS cells than in MDCK cells at each sampling time. These observations led us to suggest that the accumulation of these proteins in the Golgi might be a key point as to whether desialidation occurred in the cell or on the cell surface. Therefore, if the accumulation of the HA and NA proteins in MDCK cells occurs in the Golgi, desialidation of the HA protein might be observed in MDCK cells even in the presence of Zanamivir.

(101K): Fig. 3. Localization of the HA and NA proteins in infected MDCK and COS cells by double staining. Infected cells were stained at 5, 6 and 7 h p.i. (for MDCK cells) and 6, 7 and 9 h p.i. (for COS cells). Staining with FITC and rhodamine and observation with confocal microscopy were described in Methods.

Accumulation of the HA and NA proteins in the trans-Golgi network of MDCK cells allowed for desialidation in the presence of Zanamivir
In order to examine why desialidation of the HA protein by the NA protein did not occur before both proteins had migrated to the MDCK cell surface, we analysed desialidation using low-temperature treatment and a 2D gel electrophoresis method. Fig. 4(a) shows the 2D gel electrophoresis patterns of immunoprecipitated HA protein from infected MDCK cells incubated with or without Zanamivir. Migration of the HA protein of cells incubated without Zanamivir shifted to a basic position. Fig. 4(b) shows a comparison of the isoelectric point (IP) of the HA protein under various conditions. When cells were not exposed to Zanamivir, the HA protein from the NA temperature-sensitive mutant in cells treated using non-permissive temperature conditions migrated to an acidic position (Fig. 4b, lane 4) compared with that of cells infected with the wild-type virus (Fig. 4b, lane 1). In the presence of Zanamivir, the HA protein migrated to an acidic IP (Fig. 4b, lane 2); however, the addition of bacterial NA to medium containing Zanamivir caused the IP of the HA protein to shift to a basic position (Fig. 4b, lane 3). These results revealed that migration patterns and haemadsorption activity on the cell surface were similar. We then analysed desialidation of the HA proteins with low-temperature treatment. After 15 min of labelling with [³⁵S]methionine, infected MDCK cells were incubated at 20 °C for 30 min with MEM. Under these conditions, the HA and NA proteins accumulated in the trans-Golgi network (Fig. 5), as reported by Griffiths et al. (1985) . Cells were further incubated at 37·5 °C for 1 h to allow the migration of the HA protein to the cell surface in the presence of Zanamivir. The HA proteins were then analysed by 2D gel electrophoresis. As shown in Fig. 6, the IP of the HA protein shifted to a basic position (Fig. 6, lane 2). In order to exclude the possibility that the migration pattern with low-temperature treatment was an artefact, we used a temperature-sensitive mutant. MDCK cells were infected with influenza ts-7 virus and incubated at a non-permissive temperature (37·5 °C). Infected cells were then treated under low-temperature conditions, as described above. The migration pattern of the HA protein in ts-7-infected cells is shown in Fig. 6 (lane 3). Without NA activity, the migration pattern did not reveal a shift to a basic position. Therefore, when accumulation of the HA and NA proteins at the trans-Golgi network was allowed, desialidation of the HA protein by the NA protein was accomplished prior to the migration of these proteins to the cell surface. Wandinger-Ness et al. (1990) reported that the NA protein can interact with the HA protein in the Golgi apparatus of MDCK cells. Our results confirmed their findings. However, we have shown in these experiments that, under normal conditions, desialidation of the HA protein occurred after the protein had migrated to the cell surface but, under certain conditions, it could occur before the HA protein had reached the cell membrane. Our results suggested that, under normal conditions, the HA and NA proteins in MDCK cells would be too distant from each other in the trans-Golgi network or sorting vesicles to allow for the interaction to remove sialic acid from the HA protein. This is thought to be more likely than another possibility that some virus protein or cell factor interfered with the interaction between these proteins in the trans-Golgi network. After moving to the plasma membrane, the HA and NA proteins would be near enough for the NA enzymatic site to cleave sialic acid from the HA protein. We observed previously that in reassortant strains of H1N1 influenza virus, the presence of the M gene product correlated to the haemadsorption activity with the NA protein assisting in the reaction (Tong et al., 1998 ). Enami & Enami (1996) showed that the HA and NA proteins stimulate movement of the M1 protein to the plasma membrane. As for influenza B viruses, we do not have any information about a correlation between the M1 protein and haemadsorption. However, some interactions of the HA and NA tails with the M1 protein should be considered after movement to the cell membrane.

(34K): Fig. 4. (a) Analysis of the HA protein on MDCK cells by 2D gel electrophoresis following incubation with (left panel) and without (right panel) Zanamivir. (b) Lanes 1, wild-type HA protein without Zanamivir; 2, wild-type HA protein with Zanamivir; 3, wild-type HA protein with Zanamivir bNA; 4, wild-type HA protein with Zanamivir and low-temperature treatment (see Fig. 5).

(67K): Fig. 5. Immunostaining of the HA and NA proteins after low-temperature treatment. Infected MDCK cells were incubated for 8 h at 37 °C and then moved to a waterbath at 20 °C. After incubation at 20 °C for 30 min, cells were fixed with ethanolacetone and the HA and NA proteins were stained as described previously. (a) HA at 37 °C; (b) NA at 37 °C; (c) HA at 20 °C; (d) NA at 20 °C.

(39K): Fig. 6. NA activity was required for the migration of the HA protein to a basic position after a 20 °C treatment. MDCK cells were infected with either a temperature-sensitive mutant or wild-type virus and incubated at 37·5 °C. Labelling conditions are described in Methods. After labelling, cells were transferred to a 20 °C waterbath. Cells were incubated for 30 min and then moved to a 37·5 °C waterbath. Further incubation was carried out for 30 min. Cells were harvested and 2D gel electrophoresis was carried out as described in Methods. Lanes 1, wild-type HA protein with Zanamivir but no low-temperature treatment; 2, wild-type HA protein with Zanamivir and low-temperature treatment; 3, ts-7 HA protein with low-temperature treatment.

From results of the low-temperature experiments, it was suggested that, in COS cells, the HA and NA proteins might accumulate in the trans-Golgi network or sorting vesicles to a higher degree than in MDCK cells, thus allowing the interaction of these proteins and desialidation of the HA protein before their migration to the cell surface. The difference in the effect of Zanamivir on the haemadsorption activity of COS and MDCK cells may reflect a difference in the migration patterns of glycoproteins from the trans-Golgi network to cell surface.