Lipid Raft Disruption by Cholesterol Depletion Enhances Influenza A Virus Budding from MDCK Cells^▿

Cell lines, viruses, and plaque assay.

MDCK cells were maintained in minimum essential medium (Invitrogen, Rockville, MD), and influenza virus (A/WSN/33, H1N1) stock was prepared in MDCK cells as described previously (4). Assays for PFU were described earlier (4).

[³H]-cholesterol labeling of cells and extraction with cyclodextrins (CDs).

MDCK cells were grown on 24-well plates for 20 h, labeled with 5 μCi/ml [1α,2α-³H]-cholesterol (Sigma, St. Louis, MO) in virus growth medium (VGM; minimum essential medium containing 0.2% bovine serum albumin, 4% basal medium Eagle vitamins, 10 mM HEPES [pH 7.2], 0.155% NaHCO₃, 0.0015% DEAE-dextran, 100-U/ml penicillin G, 100-μg/ml streptomycin) for 18 h, and treated with different concentrations of MβCD or HβCD [(2-hydroxypropyl)-β-CD] or HγCD [(2-hydroxypropyl)-γ-CD] (Sigma) in VGM. Levels of ³H-cholesterol released to the media and ³H-cholesterol remaining in cells (unreleased) were assayed by liquid scintillation counting (8). Percentages of ³H-cholesterol released were calculated by the following equation: 100 × (released fraction)/(released fraction + unreleased fraction).

Release of infectious virus particles from virus-infected cells after MβCD treatment.

MDCK cell monolayers (35 mm dish, 2 × 10⁶ cells) were infected with influenza virus at a multiplicity of infection (MOI) of 3 and treated with 10 mU/ml of bacterial neuraminidase (NA) (Calbiochem, San Diego, CA) in VGM from 10 to 12 h p.i. to remove virus particles attached to the cell surface (4); they were washed five times with PBS⁺ (PBS supplemented with 0·5 mM MgCl₂ and 1 mM CaCl₂) and once with VGM. The cells were then mock treated or treated with different concentrations of MβCD in 1.5 ml of VGM for various lengths of time, and infectious virus particles released into the medium were quantified by PFU assay.

Analysis of ³⁵S-labeled virus particles released from infected cells.

MDCK cells infected with virus at an MOI of 3 were labeled with ³⁵S-protein labeling mix (Perkin-Elmer Life Sciences, Boston, MA) for 8 h (4 to 12 h p.i.) and treated with bacterial NA from 10 to 12 h p.i. At 12 h p.i., the infected cells were washed six times and either mock treated or treated with different concentrations of MβCD, HβCD, or HγCD in 1.5 ml of VGM. Media were harvested at indicated times, cell debris was removed by microcentrifuge (16,000 × g, 10 min at 4°C), and virus particles were concentrated and purified by ultracentrifugation (Beckman SW 50.1Ti rotor, 35,000 rpm, 2.5 h) through a 25% sucrose cushion (4). Viral proteins in the pellet were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and autoradiographed. Radioactive protein bands were detected by storage phosphorimaging with a PhosphorImager (Typhoon 9410) and quantified using ImageQuant software (Amersham Biosciences, Piscataway, NJ). To examine the release of virus particles from MβCD-treated cells but without MβCD in the medium, the infected cells were labeled with ³⁵S-protein labeling mix (4 h p.i. to 11 h 15 min p.i.) and treated with bacterial NA (10 h p.i. to 11 h 15 min p.i.); at 11 h 15 min p.i., they were treated with different concentrations of MβCD along with bacterial NA in 1.5 ml of VGM for 45 min. At 12 h p.i., the cells were washed as before and incubated in 1.5 ml of VGM without MβCD. After 45 min of incubation, cell culture supernatants were harvested; released infectious viruses were quantified by PFU assay; and total particle release was assayed by protein analysis as mentioned above.

MβCD treatment of virus particles.

MDCK cells infected with virus at an MOI of 3 were metabolically labeled with ³⁵S-protein labeling mix for 8 h (4 to 12 h p.i.), and released virions were purified (4) and resuspended in VGM. ³⁵S-labeled virus particles were either mock treated or treated with different concentrations of MβCD or HβCD for various lengths of time at 37°C; aliquots were diluted in ice-cold virus dilution buffer (VDB, phosphate-buffered saline [PBS] supplemented with 0.5 mM MgCl₂, 1.0 mM CaCl₂, 50 μg of DEAE-dextran/ml, and 0.2% bovine serum albumin) for PFU assay. Other aliquots were diluted with ice-cold TNE buffer (10 mM Tris-HCl [pH 7.4], 100 mM NaCl, and 1.0 mM EDTA) and were pelleted again by a second ultracentrifugation (Beckman SW 50.1Ti rotor, 4°C, 40,000 rpm, 2.5 h). The pellets and the supernatants were immunoprecipitated with the respective antibodies, analyzed by SDS-PAGE, and quantified.

Assay for TX-100 insolubility.

At 12 h p.i. ³⁵S-labeled virus-infected MDCK cells were mock treated or treated with different concentration of MβCD or cholesterol-MβCD for 45 min. For some experiments, cells were treated first with 20 mM MβCD at 11 h 15 min p.i. for 45 min and then further mock treated or treated with cholesterol (0.62 mg/ml)-MβCD (10 mM). To evaluate the Triton X-100 (TX-100) insolubility of cell surface proteins, cells were biotinylated and then extracted with 1% TX-100 (Roche, Mannheim, Germany) in TNE buffer on ice for 10 min as described previously (4, 5). Both TX-100-soluble and -insoluble fractions were immunoprecipitated with specific antibodies. Biotinylated cell surface HA and NA were isolated by a second precipitation with streptavidin, analyzed by SDS-PAGE, autoradiographed, and quantified.

To determine the TX-100 insolubility of different proteins in the virus particles, 0.1% TX-100 (instead of 1.0% as used for treating infected cells) was used, since it was shown earlier that the TX-100 resistance of different proteins in virus particles was much lower than that of proteins present in cellular membranes (4, 5, 48). Virus particles released from MβCD-treated cells either with or without MβCD in medium were purified by ultracentrifugation through a sucrose cushion. Purified virions were extracted with 0.1% TX-100 (200 μl final volume in TNE buffer) on ice for 10 min, diluted to 5 ml ice-cold TNE buffer, and ultracentrifuged (Beckman SW 50.1Ti rotor, 4°C, 40,000 rpm, 2.5 h) (4). TX-100-soluble and -insoluble proteins were immunoprecipitated using the respective antibodies, analyzed by SDS-PAGE, and autoradiographed or quantified.

Treatment with exogenous cholesterol.

In these experiments, water-soluble cholesterol (Sigma; 45 mg cholesterol complexed with 955 mg of MβCD [formula weight, 1320]) was used. To prepare different concentrations of cholesterol combined with a fixed concentration (10 mM) of MβCD, a cholesterol solution (0.62 mg/ml) containing 10 mM MβCD in VGM was mixed with various amounts of 10 mM MβCD alone in VGM. At 12 h p.i.; ³⁵S-labeled virus-infected cells were mock treated or treated with various concentrations of cholesterol containing MβCD in 1.5 ml of VGM for 45 min. Released virus particles were quantified by both PFU assay and protein analysis as described above. To replenish cholesterol in cholesterol-depleted cells, ³⁵S-labeled cells were treated (at 11 h 15 min p.i.) with 20 mM MβCD along with bacterial NA in 1.5 ml of VGM for 45 min. At 12 h p.i., the cells were further incubated with (or without) 0.62 mg/ml cholesterol containing 10 mM MβCD in 1.5 ml of VGM for 45 min. The released virus particles were quantified by both PFU assay and protein analysis.

Cellular cholesterol content measurement.

Cellular cholesterol content was measured using an Amplex Red cholesterol assay kit (Invitrogen) according to manufacturer protocols and as described elsewhere (11). Briefly, mock- or MβCD- or cholesterol-MβCD-treated MDCK cells (35 mm dish, 2 × 10⁶ cells) were harvested in 75 μl of PBS, lysed by three cycles of freezing and thawing, and extracted with 400 μl of chloroform-methanol (2:1). After 10 min of microcentrifugation, the bottom (chloroform) layer was collected and 200 μl aliquots were evaporated in a vacuum. The sample was dissolved in 25 μl of 2-propanol containing 10% TX-100, and a 0.8 μl sample was assayed with an Amplex Red cholesterol assay kit using a 96-well plate. All reactions were performed in duplicate.

Negative-stain EM.

Electron microscopy (EM) was performed essentially as described elsewhere (19). Purified virions in PBS were mixed with latex beads (polystyrene; Sigma) that had an average diameter of 1.0 μm and were absorbed onto carbon-coated copper grids for 5 min. Grids were negatively stained with 2% phosphotungstic acid (pH 6.6) and examined with a JEOL JEM-100CX electron microscope (JEOL Ltd., Tokyo, Japan). For quantification, random fields were photographed at a magnification of ×19,000, and a portion of that field was further photographed at a magnification of ×72,000 to study virus morphology.

To examine the effect of MβCD or HβCD treatment on virus morphology, purified virions were treated with different concentrations of drugs and examined (magnification of ×36,000) as described above.

Thin-section electron microscopy.

Cells (grown on polycarbonate filter) infected with virus at an MOI of 3 were treated with bacterial NA (10 to 12 h p.i.) and, at 12 h p.i., were mock treated or treated with 30 mM MβCD for 45 min. These cells were then cross-linked in 2% glutaraldehyde (EM grade) in PBS⁺ and postfixed with 1% osmium tetraoxide in PBS⁺. Filters were dehydrated, cut out from filter units, and embedded in Epon. Ultra-thin (60 nm) sections were stained with uranyl acetate and lead citrate and then examined with a JEOL JEM-100CX electron microscope (4).

Removal of cholesterol from MDCK cells by MβCD treatment.

To determine the effect of MβCD on the removal of cholesterol, ³H-cholesterol-labeled MDCK cells were treated with different concentrations of MβCD for various lengths of time, and the amounts of ³H-cholesterol released in the media and remaining in cells (unreleased) were assayed as described in Materials and Methods. The results (Fig. 1) showed that increasing amounts of cholesterol were released into the medium with increasing concentrations and longer treatment of MβCD. Fifty percent of the cholesterol was released after 45 min of treatment with 30 mM MβCD (Fig. 1). Similar reductions of cellular cholesterol content were observed when total cholesterol was measured using a cholesterol assay kit (data not shown). Cell viability, integrity, and morphology and total cell numbers remained essentially unaffected when MDCK cells were treated with 30 mM MβCD for a maximum duration of 45 min. These results are consistent with previous observations of the effects of MβCD in MDCK cells (8, 9, 12, 25, 34, 38).

Cholesterol depletion in virus-infected MDCK cells by MβCD treatment caused enhancement of virus budding.

To investigate the effect of cholesterol depletion by MβCD treatment on influenza virus budding, virus-infected MDCK cells at the late phase of infection (12 h postinfection [p.i.]) were treated with different concentrations of MβCD for various lengths of time, and amounts of infectious virus released into the medium were quantified by PFU assay. The results (Fig. 2A) show that the PFU titer of the released virus particles varied depending on the concentration and duration of MβCD treatment. Treatment with 10 mM MβCD caused an increase in the yield of infectious virus titer throughout the duration (45 min) of treatment. However, higher concentrations of MβCD caused an initial increase followed by a decrease in PFU titers with longer duration of treatment (Fig. 2A).

Since cholesterol depletion from virions was shown to reduce significantly the infectivity of influenza A viruses (46, 47) as well as of human immunodeficiency virus type 1 (HIV-1) and simian immunodeficiency virus (SIV) (14), it is possible that released virus particles became inactivated because of the continued presence of MβCD in the medium and that PFU titers may not have reflected total virus particle release. We therefore examined the amounts of particle released after MβCD treatment by protein analysis of ³⁵S-labeled released virus particles. The results (Fig. 2B) show that the particle release increased with increasing concentration and length of MβCD treatment as measured by HA and NP band intensity. Quantification of either HA (Fig. 2C) or NP (Fig. 2D) showed that particle release continued to increase with increasing concentrations and durations of MβCD treatment. It should be noted that NA, which migrates as a diffuse band of low intensity, comigrates with NP. It should be further noted that in the absence of exogenous proteases like trypsin, less than 10% of HA in released virus particles become cleaved. Therefore, we routinely quantified only the uncleaved HA (HA0) (this will be referred to as HA throughout the article). However, the amounts of M1 in the released virus particles varied and the ratio of M1/HA decreased in released virus particles with increasing concentrations and durations of MβCD treatment (Fig. 2B). Essentially, results were similar when ratios of M1/NP were determined (data not shown). The reason for the decreased M1/HA (or M1/NP) ratios in released virus particles is discussed later. Overall, the data presented in Fig. 2 show that based on the analysis of HA and NP levels, budding of virus particles increased with increasing concentrations and durations of MβCD treatment, although the titer of infectious viruses decreased with prolonged treatment at higher concentrations of MβCD.

Effect of MβCD on the infectivity of virus particles.

Since infectious virus titers decreased for cells treated with higher concentrations and longer durations of MβCD, we examined the effect of the presence of MβCD on the infectivity of cell-free virus particles. The results (Fig. 3) show that MβCD treatment significantly reduced virus infectivity (more than 2 logs after 45 min treatment with 30 mM MβCD), indicating that the reduction of PFU titers at higher concentrations with longer treatment of MβCD (Fig. 2A) was at least partly because of the loss of infectivity of the released virus particles in the continued presence of MβCD in the medium. Furthermore, it has been reported that cholesterol depletion caused permeabilization of virions and release of viral proteins from HIV-1 and SIV particles (14). Therefore, we examined whether MβCD treatment caused release of viral proteins from influenza virus particles. The results (Table 1) show that unlike mock-treated (0 mM) virus sample results, increasing amounts of HA, NA, NP, and M1 were released from virus particles upon treatment with increasing concentrations of MβCD. M1 was released most (Table 1), which was the likely cause for lower M1/HA (or M1/NP) ratios in virus particles released from MβCD-treated cells (Fig. 2B). Furthermore, it should be noted that the amount of virus particles released after MβCD treatment was probably higher than that observed by viral protein analysis of the released particles (Fig. 2B, C, and D) because even HA and NP were also partially released from virus particles after MβCD treatment (Table 1).

Therefore, we used another independent method, that is, negative-stain EM, to quantify particle release and examine the morphology of released particles. EM micrograph assays of virus particles collected after 10 mM MβCD treatment (45 min) showed an increased release of virus particles (Fig. 4). The EM micrographs also show that the majority of particles released from cells even after treatment with low concentrations (10 mM) of MβCD exhibited partially disrupted morphology (Fig. 4); such disruption was severe at higher concentrations (30 mM) of MβCD (data not shown). Quantification of the virus particles in those micrographs showed (Table 2) that after treatment of virus-infected cells with 10 mM MβCD for 45 min, virus particle release was significantly higher than that estimated by protein analysis (12.6-fold by EM versus 7.5-fold by protein analysis; Table 2 and 3).

Data presented in Fig. 2, 3, and 4 indicate that the lower infectivity of the virus particles released from MβCD-treated cells was at least partly caused by the continued presence of MβCD in the medium during drug treatment. Therefore, we examined the release of virus particles from MβCD-treated cells without MβCD in the medium. The results (Table 3) show that increased particle release from MβCD-treated cells continued even when MβCD was absent in the medium, albeit at a lower rate than that seen during MβCD treatment. The relative infectivity of the released particles, on the other hand, was higher when MβCD was absent in the medium (Table 3), although the relative infectivity of these virus particles still did not reach the same level (1.0) as that seen with the mock-treated cells. These results indicate that particles released from cholesterol-depleted cells also had reduced infectivity.

Cholesterol depletion by MβCD treatment caused disruption of lipid rafts.

Depletion of cholesterol is expected to disrupt lipid raft microdomains, which can be distinguished from intact lipid rafts by a decrease in TX-100 insolubility of raft-associated proteins. Therefore, we examined the effect of cholesterol depletion on TX-100 insolubility of viral proteins in infected cells as well as in released virions. The results (Table 4) show that in cells treated with 10 mM, 20 mM, and 30 mM MβCD for 45 min, the levels of TX-100 insolubility of cell surface HA were reduced to 42%, 33%, and 27%, respectively, compared with the 61% insoluble HA seen in mock-treated cells. Similarly, the TX-100 insolubility of cell surface NA was also reduced (Table 4). These data indicate that cholesterol depletion by MβCD treatment caused disruption of lipid rafts. On the other hand, cholesterol depletion did not significantly affect the TX-100 insolubility of NP and M1 in infected cells. It is known that the TX-100 insolubility of NP is due to its interaction with actin cytoskeletons. Similarly, the TX-100 insolubility of M1 in infected cells is mostly due to its interaction with actin via vRNP (3) and a relatively small fraction of M1 becomes raft associated due to its interaction with HA and NA (1).

To determine whether MβCD treatment of infected cells also affected lipid rafts of released virus particles, the TX-100 insolubility of different viral proteins in virions collected after 45 min of treatment with different concentrations of MβCD was determined as described in Materials and Methods. The results (Table 4) show that HA, NA, and M1 were highly TX-100 soluble in the virions collected from virus-infected MβCD-treated cells. Because NP inside the virus particles remained as a vRNP complex (3), its TX-100 insolubility was not markedly affected by MβCD treatment. On the other hand, our data indicate that the TX-100 insolubility of M1 in virus particles depended on its association with HA and NA. In any case, the TX-100 insolubility of NP and M1 does not have much relevance to the interpretation of data presented here since those are not lipid raft-associated proteins.

Because MβCD was also present in the medium, it was not clear whether the increased TX-100 solubility of HA and NA in virions resulted from cholesterol depletion of the budding cellular membrane or cholesterol depletion of the virions after their release into the medium or both. Therefore, we analyzed the TX-100 insolubility of the proteins in virions that were collected from MβCD-treated cells without MβCD in the medium. The results (Table 4) show that the TX-100 insolubility of HA and NA was significantly higher in particles collected in the absence of MβCD in the medium compared to that of the virus particles collected with MβCD in the medium. The total cholesterol content of the cells as well as the TX-100 insolubility of surface HA and NA did not change significantly (see Table 7) after 45 min incubation of cholesterol-depleted cells in VGM, indicating that any endogenous synthesis of cholesterol within this short time (45 min) was not responsible for the increased TX-100 insolubility of HA and NA in virions released during a drug-free chase of cholesterol-depleted cells (i.e., virions collected without MβCD in the medium). Rather, these data support the idea that the increased TX-100 solubility of proteins in the virions collected with MβCD in the medium (Table 4) was partly caused by the effect of MβCD on released virus particles.

Effect of exogenous cholesterol on particle release and lipid raft integrity.

To determine whether the effects of the drug (MβCD) were due solely to cholesterol depletion, we examined the effect of exogenous cholesterol on virus release during MβCD treatment of infected cells. Accordingly, ³⁵S-labeled virus-infected cells were treated at 12 h p.i. with different concentrations of cholesterol in the presence of 10 mM MβCD and virus release was examined by PFU assay (for infectious particles) as well as by protein analysis of released labeled virus particles. The results show (Table 5) that exogenous cholesterol inhibited the enhanced release of virus particles in an essentially dose-dependent manner and that 0.62 mg/ml of cholesterol containing 10 mM MβCD inhibited virus release by more than 2 logs compared to mock-treated cell results. Furthermore, cholesterol-MβCD caused a significant increase in the relative infectivity of released virus particles compared to that seen when cells were treated with MβCD only, indicating that exogenous cholesterol could partially restore the infectivity of released virus particles. We further investigated whether exogenous cholesterol can inhibit virus release from cholesterol-depleted cells. Accordingly, ³⁵S-labeled virus-infected cells were treated with 20 mM MβCD for 45 min. Cells were then washed and further treated with cholesterol (0.62 mg/ml)-MβCD (10 mM) or mock treated (no cholesterol, no MβCD) for 45 min. Virus release was examined by PFU assay as well as by protein analysis. The results (Table 6) show that the exogenous cholesterol also inhibited the release of virus particles from cholesterol-depleted cells. These results show that the increased release of virus particles after MβCD treatment (Fig. 2) was due to cholesterol depletion.

The above-described results show that lipid raft disruption by MβCD-mediated cholesterol depletion caused an increased release of virus particles. On the other hand, exogenous cholesterol inhibited virus release from cholesterol-depleted cells, which suggests reestablishment of lipid raft integrity. Therefore, we examined the effect of exogenous cholesterol on the TX-100 insolubility of cell surface HA and NA. The results show that after incubation with cholesterol (0.62 mg/ml)-MβCD (10 mM) for 45 min, the TX-100 insolubility of cell surface HA and NA increased and was even higher than that seen with the mock-treated cells (Table 7). Since exogenous cholesterol caused higher TX-100 insolubility of cell surface HA and NA compared to mock-treated cell results (Table 7) and reduced virus budding lower than that of mock-treated cells (Tables 5), we determined the total cholesterol content of the cells after treatment with exogenous cholesterol. The results show that after treatment with 20 mM MβCD for 45 min, the total cholesterol content was reduced to 55.9% relative to mock-treated cell levels, which is similar to the results presented in Fig. 1. After incubation of these cholesterol-depleted cells with cholesterol (0.62 mg/ml)-MβCD (10 mM) for 45 min, the total cholesterol content was increased to 293%. On the other hand, when cells were chased for 45 min without cholesterol-MβCD, the total cholesterol content (57.7%) did not change significantly compared to that of cells without chase (55.9%; Table 7). These results indicate that the higher cholesterol content of the cells after incubation with exogenous cholesterol was responsible for the higher TX-100 insolubility of cell surface HA and NA as well as the reduced release of virus particles.

Effect of HβCD and HγCD on virus budding.

To examine whether the enhanced release of virus particles was mainly caused by cholesterol depletion rather than by other effects caused by the presence of cyclodextrin, we used HβCD, which releases cholesterol from cells but with lower efficiency than MβCD (2, 30), and HγCD, which does not release cholesterol from cells in significant amounts (31). The results (Table 8) show that HβCD treatment increased the release of virus particles, although less efficiently than MβCD treatment. Also, the relative infectivity of virus particles released from HβCD-treated cells was higher that that seen with MβCD-treated cells (Table 8). On the other hand, HγCD treatment did not significantly affect either the PFU titer or virus particle release compared to mock-treated cell results. As expected, HβCD (2, 30) was also less efficient than MβCD in removing cholesterol from cells and HγCD (31) did not cause any significant release of cholesterol from MDCK cells (Table 8). These results support the idea that the increased release of virus particles from MβCD-treated cells and the reduced infectivity of virus particles were caused by cholesterol depletion rather than other effects from CD.

Virus budding from MβCD-treated cells by thin-section electron microscopy.

Since the budding process and release of virus particles involve the fusion of apposing cellular and viral membranes, disruption of lipid raft microdomains may affect virus morphology. Previously it was reported that MβCD treatment does not grossly affect virion morphology (46). However, as mentioned earlier, particles released from cells after exposure to a low concentration (10 mM) of MβCD treatment exhibited partially disrupted morphology (Fig. 4); such disruption was severe at higher concentrations (30 mM) of MβCD (data not shown). This can result either from defective budding or disruption of virus particles after release from the infected cells in the continued presence of MβCD in the medium or both. We therefore investigated virus budding in the presence of MβCD by thin-section EM. A representative EM micrograph (Fig. 5) shows that MβCD treatment did not affect the overall spherical morphology of the cell surface-attached virus particles. Furthermore, it should be noted that fewer virus particles are present at the cell surface of cholesterol-depleted cells (Fig. 5), as expected, because the majority of particles would be released from MβCD-treated cells. These results indicate that disruption of virus particles likely occurred after release of virus particles in the medium containing MβCD.

Therefore, we examined the effect of MβCD or HβCD treatment on cell-free virus particles by negative-stain EM. The results (Fig. 6) show that upon treatment with 10 mM MβCD for 30 min, a significant fraction of particles exhibited a nick or hole on the envelope. These findings were similar to those reported for HIV-1 and SIV (14). With higher concentrations of MβCD, a larger fraction of particles exhibited disruption of viral envelopes (Fig. 6). However, HβCD treatment had less effect on virus particle structures. After 30 mM HβCD treatment only a small fraction of particles showed a possible nick or hole on the envelope (Fig. 6). It should also be noted that treatment with 10 mM, 20 mM, and 30 mM MβCD for 30 min reduced virus infectivity levels to about 7.1%, 1.4%, and 0.5%, respectively (Fig. 3), whereas treatment with 30 mM HβCD reduced virus infectivity to only about 31% (data not shown).