Role for Influenza Virus Envelope Cholesterol in Virus Entry and Infection

Enveloped viruses are highly dependent on their lipid envelopesfor entry into and infection of host cells. Here, we have examinedthe role of cholesterol in the virus envelope, using methyl-ß-cyclodextrindepletion. Pretreatment of virions with methyl-ß-cyclodextrinefficiently depleted envelope cholesterol from influenza virusand significantly reduced virus infectivity in a dose-dependentmanner. A nonenveloped virus, simian virus 40, was not affectedby methyl-ß-cyclodextrin treatment. In the case ofinfluenza virus, infectivity could be partially rescued by theaddition of exogenous cholesterol. Influenza virus morphology,binding, and internalization were not affected by methyl-ß-cyclodextrindepletion, whereas envelope cholesterol depletion markedly affectedinfluenza virus fusion, as measured by a specific reductionin the infectivity of viruses induced to fuse at the cell surfaceand by fluorescence-dequenching assays. These data suggest thatenvelope cholesterol is a critical factor in the fusion processof influenza virus.

INTRODUCTION

Top

Introduction

Infection of host cells by enveloped viruses relies on the fusionof the viral envelope with either the endosomal or plasma membraneof the cell (12). As such, the protein and lipid compositionsof both the viral envelope and the host cell membrane play crucialroles in virus infection. For orthomyxoviruses, such as influenzavirus type A, it is well established that the virus is internalizedin the cell prior to membrane fusion (27). Fusion occurs fromwithin endosomes and depends on low pH. Exposure of the viralhemagglutinin (HA) protein to a pH of 5.0 causes extensive conformationalchanges, exposing the fusion peptide, which is responsible forinitiating bilayer fusion (19, 43, 44).

For all enveloped viruses, the envelope is derived from thehost cell during the process of virus budding. In the case ofinfluenza virus, budding takes place at the apical plasma membraneand is heavily dependent on the presence of lipid microdomains,or "rafts," which are enriched in cholesterol and sphingomyelin(29, 39, 42, 51). Both of the major viral glycoproteins, HAand neuraminidase, associate with lipid microdomains (29). Purifiedinfluenza virus particles contain high levels of cholesterol:44% of the total virus lipid is cholesterol, which represents $~$ 11 to 12% of the total mass of the virion (10, 22, 29). In additionto influenza virus, many other viruses are known to utilizelipid microdomains during budding. Human immunodeficiency virustype 1 (HIV-1) is perhaps the best studied (30, 32), but otherviruses that bud out of defined membrane microdomains includeother retroviruses, as well as rhabdoviruses, e.g., vesicularstomatitis virus, and alphaviruses, e.g., Semliki Forest virus(5, 23, 24, 34, 40).

In addition to its involvement in virus budding, cholesterolhas been shown to be important during virus entry. Semliki Forestvirus fusion is absolutely dependent on the presence of cholesterolin the target (endosomal) membrane (1, 35, 48). HIV-1 and herpessimplex virus entry has also been shown to require cholesterol,both in the target (plasma) membrane and in the donor membrane(the viral envelope) (2, 6, 7, 16, 18, 47). Other viruses aredependent only on the cholesterol-containing lipid microdomainsof the cell, e.g., the nonenveloped polyomavirus simian virus40 (SV40), which enters cells via caveolae (8, 33).

It has recently been shown that depletion of cholesterol fromcellular membranes has no effect on influenza virus entry (41).Here, we have examined the role of cholesterol in the virusenvelope, using methyl-ß-cyclodextrin depletion. Methyl-ß-cyclodextrinefficiently depleted envelope cholesterol and significantlyreduced influenza virus infectivity in a dose-dependent manner.Virus morphology, binding, and internalization were not affected,whereas envelope cholesterol depletion markedly affected virusfusion.

MATERIALS AND METHODS

Top

Materials and Methods

Cells, viruses, and infections. MDBK cells (American Type Culture Collection) were maintainedin Dulbecco's modified minimal essential medium (Cellgro) containing10% calf serum, 100 U of penicillin/ml, and 10 µg of streptomycin/mland passaged twice weekly. Cv-1 cells (American Type CultureCollection) were maintained in the alpha modification of minimalessential medium (Cellgro) containing 10% fetal calf serum,100 U of penicillin/ml, and 10 µg of streptomycin/ml andpassaged twice weekly.

Influenza virus A/WSN/33 (H1N1) was grown in 10-day-old embryonatedeggs, (25). SV40 strain 377 was kindly provided by James Pipas,University of Pittsburg, and was grown in Cv-1 cells. The infectionswere performed essentially as described previously (38). Briefly,virus stocks were diluted in RPMI 1640 medium containing 0.2%bovine serum albumin (BSA) and buffered to pH 6.8 with HEPES.Unless otherwise described, the virus was adsorbed for 45 minat 37°C. The cells were then maintained in growth mediumcontaining 2% serum at 37°C before analysis.

Methyl-ß-cyclodextrin treatment of virions. Methyl-ß-cyclodextrin was obtained from Sigma. Typically,50 µl of influenza virus (1 mg/ml, or 0.36 x 10⁹ hemagglutinatingunits/ml) was mixed with phosphate-buffered saline (PBS) ordifferent concentrations of methyl-ß-cyclodextrinand incubated at 37°C for 30 min. For binding and internalizationassays, viruses with a multiplicity of infection (MOI) of $~$ 25focus-forming units/cell were used, and for infection studies, $~$ 1 focus-forming unit/cell was used. Virus was treated with cyclodextrinand ultracentrifuged at 200,000 x g for 30 min to remove thecyclodextrin. Virus was resuspended in 200 µl of RPMI1640 containing 0.2% BSA before infection of cells.

For competition experiments, 2 µl of exogenous cholesterol(Sigma) at a concentration of either 2 mg/ml or 200, 20, or2 µg/ml was added to the virus-cyclodextrin mixture toobtain a final volume of 10 µl. The mixture of exogenouscholesterol, virus, and cyclodextrin was incubated at 37°Cfor 30 min before being used for infection.

Hemagglutination assay. Freshly isolated chicken blood was washed three times with PBS,and the red blood cells were resuspended in PBS at a concentrationof 1%. Virus dilutions (50 µl) were made in PBS, and 50µl of the red blood cell suspension was added to individualwells in a 96-well plate and allowed to settle for 45 min. Thewells were judged to be HA negative (no agglutination of redcells) if a dot or pellet of red cells was present, and theywere judged to be positive (red cells agglutinated by HA) ifa smooth suspension of red cells was present.

Negative-stain electron microscopy. Samples were prepared for negative-stain electron microscopyby fixation with 2% glutaraldehyde-PBS. The samples were thenplaced onto a Formvar-coated hexagonal (300-hex) Cu grid, allowedto stand for 30 s, and immediately stained for 10 s with 2%aqueous uranyl acetate. After a brief drying period, the gridwas viewed on a Philips Tecnai 12 electron microscope at 100kV. Images were acquired with a digital Gatan Multiscan Camera(model 791) before transfer into Adobe Photoshop (version 7.0).

Binding assay. To monitor binding, methyl-ß-cyclodextrin-treatedor untreated influenza virions were bound to cells on ice for90 min. The cells were washed, fixed, and processed as describedbelow for immunofluorescence microscopy or flow cytometry.

Internalization assay. As a measure of virus internalization, cells were incubatedwith methyl-ß-cyclodextrin-treated or untreated influenzavirions on ice for 90 min and either fixed immediately or shiftedto 37°C for 20 min. The cells were then washed with 0.1M glycine-0.1 M NaCl, pH 3.0, buffer for 2 min, fixed, and permeabilized.Virion localization was assayed by immunofluorescence microscopyusing anti-nucleoprotein (NP) antibodies.

Fusion assays. As a measure of virus fusion, we assayed the ability of virusfused at the cell surface to cause infection of cells (26).Briefly, influenza virus was bound to the surfaces of BHK cellson ice for 90 min, and fusion was triggered by lowering thepH of the medium to 5.0 for 5 min. Monensin (10 µM; Sigma)was added to prevent any entry of virus through endosomes, andthe cells were analyzed for infection by immunofluorescencemicroscopy at 6 h postinfection.

For fluorescence-dequenching assays, we used the procedure ofStegmann et al. (45) as modified by Gilbert and coworkers (15).To label influenza virus, 10 µl of 1.7 mM octadecyl rhodamineB chloride (R18; Molecular Probes) was injected slowly into1 ml of influenza virus (1 mg/ml), and the mixture was incubatedin the dark on a rotary shaker at room temperature. For R18fusion assays, 7 µg of labeled virus was bound to 2 x10⁶ MDBK cells at 4°C for 1 h in 600 µl of cell suspensionin the presence of 10 µM monensin to block endocytosis.Unbound virus was removed by washing the cells with bindingbuffer (RPMI 1640 medium containing 0.2% BSA and 10 µMmonensin, pH 6.8). The cells were resuspended in 5 mM HEPES-5mM morpholineethanesulfonic acid-5 mM succinate buffer, pH 7.0,at 37°C. Fusion of virus on the cell membrane was triggeredby adding a predetermined amount of 150 mM HCl to obtain a finalpH of 4.9. Samples were analyzed on a Molecular Devices SpectramaxGemini XS fluorimeter. Fluorescence dequenching was measuredduring a 5-min time course at 3-s intervals with the fluorimeterset to excitation and emission wavelengths of 560 and 590 nm,respectively, with a 579-nm cutoff filter. Fusion efficiencywas calculated following treatment with 1% Triton X-100 to obtain100% dequenching.

Indirect immunofluorescence microscopy. Preparation of cells for indirect immunofluorescence microscopywas performed as described previously (50). Influenza virusNP was detected using the monoclonal antibody H16-L10-4R5 (AmericanType Culture Collection). SV40 was detected using an anti-T-antigenmonoclonal antibody (clone 416; Research Diagnostics Inc.).The secondary antibodies used were Alexa 488-labeled (green)or Alexa 568-labeled (red) goat anti-mouse immunoglobulin G(Molecular Probes). The cells were mounted in MOWIOL 4-88 (Calbiochem)and viewed on a Nikon Eclipse E600 fluorescence microscope,using a 40x Plan Apo objective (NA 0.95) or a 20x Plan Apo objective(NA 0.75), and images were captured with a SPOT RT camera andSPOT software (version 3.5.4) before transfer into Adobe Photoshop(version 7.0). For quantitation, >100 cells were scored foreach sample.

Flow cytometry. For flow cytometry preparation, cells were detached with 0.2%trypsin-EDTA (Cellgro) or cell dissociation buffer (Sigma),washed in PBS, fixed in 3% paraformaldehyde-PBS, and permeabilizedin 0.075% saponin in 10% goat serum-PBS. The cells were incubatedwith the monoclonal antibody to influenza NP for 30 min, followedby incubation with Alexa 488-labeled goat anti-mouse immunoglobulinG for 30 min. The cells were analyzed on a FACSCalibur cytometerusing CellQuest version 3.1f software (Becton Dickinson ImmunocytometrySystems). At least 10,000 cells were analyzed for each sample.

Cholesterol determination. The cholesterol content was determined using an Amplex Red Cholesterolassay kit (Molecular Probes) according to the manufacturer'sinstructions. Typically, 50 µl of virus (1 mg/ml) waspelleted and resuspended in Amplex Red reaction buffer and analyzedon a Molecular Devices Spectramax Gemini XS fluorimeter, usingan excitation wavelength of 544 nm and an emission wavelengthof 590 nm.

RESULTS

Top

Results

Treatment of influenza virions with methyl-ß-cyclodextrin to deplete cholesterol significantly reduces virus infectivity. It was recently shown that depletion of cholesterol from cellshas no effect on influenza virus entry (41). However, the previousexperiments were designed to target the cellular membranes andnot the viral envelope. As a first measure of the role of envelopecholesterol in influenza virus infection, we treated purifiedvirions with various concentrations of methyl-ß-cyclodextrinfor 30 min to deplete the cholesterol. Exposure to methyl-ß-cyclodextrinresulted in a dose-dependent inhibition of influenza virus infectivity,as measured by single-hit infection assays (Fig. 1). By immunofluorescencemicroscopy, $~$ 70% of the cells were infected with influenza virusin the absence of methyl-ß-cyclodextrin, as shownby the fluorescence signal of influenza NP in the nuclei ofinfected cells (Fig. 1A). Upon addition of 5 mM methyl-ß-cyclodextrin,the level of infection dropped to $~$ 55%, and with 10 mM methyl-ß-cyclodextrin,it dropped to $~$ 40%. Treatment of influenza virions with 20 mMmethyl-ß-cyclodextrin reduced infectivity to <20%,and with 50 mM methyl-ß-cyclodextrin, infectivitywas detectable in only $~$ 5% of the cells. Figure 1B shows a graphicalrepresentation of these data, normalized to 100% for the untreatedsample.

View larger version (39K):
[in this window]
[in a new window]
FIG. 1. Treatment of influenza virions with methyl-ß-cyclodextrin to deplete cholesterol significantly reduces virus infectivity. (A) Influenza virus was pretreated with various concentrations of methyl-ß-cyclodextrin, and MDBK cells were infected at an MOI of $~$ infectious unit per cell. The cells were analyzed by indirect immunofluorescence microscopy at 5 h postinfection using anti-NP antibodies (b, d, f, h, and j), or nuclei were stained with Hoechst 33258 (a, c, e, g, and i). (B) Quantitation of data in panel A. At least 100 cells were counted for each sample to determine the percentages of infectivity. The error bars represent the standard errors of the mean. (C) Influenza virus was pretreated with various concentrations of methyl-ß-cyclodextrin, and MDBK cells were infected at an MOI of $~$ 1 infectious unit per cell. The cells were analyzed at 5 h postinfection by FACS analysis using anti-NP antibodies. Negative control, uninfected cells.

We next confirmed our immunofluorescence microscopy data byfluorescence-activated cell sorting (FACS) analysis to obtainmore quantitative results (Fig. 1C and D). By flow cytometry,untreated virus showed an 87% infection level. Treatment with5 mM methyl-ß-cyclodextrin reduced the level of infectionto 62%, 10 mM methyl-ß-cyclodextrin gave a 45% infectionlevel, and 20 mM cyclodextrin resulted in a 17% infection level,while 50 mM methyl-ß-cyclodextrin allowed only 5%of the cells to be infected. In general, the flow cytometrydata closely paralleled the results from immunofluorescence,with a dose-dependent effect of methyl-ß-cyclodextrinthat produced a roughly linear decrease in infectivity withincreasing methyl-ß-cyclodextrin concentrations.

It is important to note that our experimental protocol was designedto minimize the effects of methyl-ß-cyclodextrin oncellular membranes. To control for nonspecific effects on thecell, we treated virions with methyl-ß-cyclodextrinimmediately before adding them to cells. Under these conditions,we retained essentially 100% infectivity, as measured by immunofluorescencemicroscopy (Fig. 2).

View larger version (52K):
[in this window]
[in a new window]
FIG. 2. Methyl-ß-cyclodextrin has no effect on virus infectivity without pretreatment. Influenza virus was exposed to various concentrations of methyl-ß-cyclodextrin, and MDBK cells were infected immediately at an MOI of $~$ 1 infectious unit per cell. The cells were analyzed by indirect immunofluorescence microscopy using anti-NP antibodies, and the percentages of infectivity at 5 h postinfection were determined by scoring at least 100 cells for infection for each sample. The error bars represent the standard errors of the mean.

Methyl-ß-cyclodextrin does not affect influenza virus structure. In order to examine the effects of methyl-ß-cyclodextrintreatment of influenza virus particles, we first examined treatedand untreated virions by negative-stain electron microscopy.As shown previously by other investigators (39), there wereno obvious differences between influenza virus particles aftermethyl-ß-cyclodextrin treatment (Fig. 3) and thosethat were untreated (not shown), confirming that the drug doesnot grossly affect virion morphology. As a more functional measureof virus structure, we performed hemagglutination assays oftreated and untreated virions. The HA titer of the virus (1:1,024)did not change upon the addition of methyl-ß-cyclodextrin(5, 10, 20, and 50 mM) to deplete envelope cholesterol, whereasthe addition of 0.1% Triton X-100 reduced HA activity to backgroundlevels (1:10).

View larger version (100K):
[in this window]
[in a new window]
FIG. 3. Methyl-ß-cyclodextrin does not affect influenza virus structure. Influenza virus was pretreated with 20 mM methyl-ß-cyclodextrin, and virions were analyzed by negative-stain electron microscopy.

Treatment with methyl-ß-cyclodextrin has no effect on SV40 infectivity. As a control virus for the effects of cholesterol depletion,we employed SV40. SV40 is a nonenveloped virus known to entercells via lipid microdomains and to be very sensitive to cholesteroldepletion of host cell membranes by methyl-ß-cyclodextrin(37). SV40 was treated with various doses of methyl-ß-cyclodextrin,and infection was monitored in a single-hit infection assay(Fig. 4). SV40 showed no significant drop in infectivity (Fig.4), confirming that methyl-ß-cyclodextrin treatmentacted specifically on virus envelope cholesterol.

View larger version (48K):
[in this window]
[in a new window]
FIG. 4. Treatment with methyl-ß-cyclodextrin does not reduce SV40 infectivity. SV40 was pretreated with various concentrations of methyl-ß-cyclodextrin, and Cv-1 cells were infected at an MOI of $~$ 1 infectious unit per cell. The cells were analyzed by indirect immunofluorescence microscopy with anti-T-antigen antibodies, and the percentages of infectivity at 24 h postinfection were determined by scoring at least 100 cells for infection for each sample. The error bars represent the standard errors of the mean.

Measurement of cholesterol content of methyl-ß-cyclodextrin-treated influenza viruses. We next assayed the relative depletion of envelope cholesterolmediated by methyl-ß-cyclodextrin treatment. Purifiedinfluenza virus was treated with various concentrations of methyl-ß-cyclodextrinfor 30 min, and the cholesterol content was determined usinga cholesterol assay kit (Fig. 5). Influenza virus showed a dose-dependentdrop in the level of cholesterol, with $~$ 90% depletion obtainedwith 20 mM methyl-ß-cyclodextrin. These data confirmthat methyl-ß-cyclodextrin treatment produces specificand efficient depletion of virion envelope cholesterol.

View larger version (44K):
[in this window]
[in a new window]
FIG. 5. Measurement of cholesterol content of methyl-ß-cyclodextrin-treated viruses. Purified influenza virus was pretreated with various concentrations of methyl-ß-cyclodextrin or was left untreated. The virions were repurified and analyzed using a cholesterol assay kit. The cholesterol contents were plotted based on fluorescence units.

Replenishment of envelope cholesterol can restore influenza virus infectivity. To assess whether the effect of methyl-ß-cyclodextrinwas permanent or reversible, and to show that the effects ofthe drug were due solely to cholesterol depletion, exogenouscholesterol was used to replenish the envelopes of methyl-ß-cyclodextrin-treatedinfluenza virions. We incubated the virions with various amountsof free cholesterol during the methyl-ß-cyclodextrintreatment. This trans-supplementation approach partially restoredinfluenza virus infectivity in a dose-dependent manner (Fig.6). Addition of 2 mg of cholesterol/ml to virions treated with20 mM methyl-ß-cyclodextrin restored infectivity toa level of 42% in our single-hit infection assay. Due to technicaldifficulties, we were unable to use levels of cholesterol above2 mg/ml, which might have restored infection more completely.

View larger version (41K):
[in this window]
[in a new window]
FIG. 6. Replenishment of envelope cholesterol can restore influenza virus infectivity. Influenza virus was pretreated with 20 mM methyl-ß-cyclodextrin in the presence of various concentrations of exogenous cholesterol. MDBK cells were infected at an MOI of $~$ 1 infectious unit per cell. The cells were analyzed by indirect immunofluorescence microscopy at 5 h postinfection using anti-NP antibodies, and the percentages of infectivity were determined by scoring at least 100 cells for infection for each sample. The error bars represent the standard errors of the mean.

Methyl-ß-cyclodextrin does not affect influenza virus binding or internalization. We wished to address the point in the replication cycle wheremethyl-ß-cyclodextrin-treated virions might be defective.Our first assay was to examine influenza virus binding. Cellswere incubated on ice with a high MOI of virus that had beentreated with 20 mM methyl-ß-cyclodextrin or that wasuntreated. Virus binding was assayed by flow cytometry (Fig.7A). We saw no significant differences between the abilitiesof methyl-ß-cyclodextrin-treated and untreated virusesto bind to cells. In both cases, the levels of binding werevirtually indistinguishable.

View larger version (21K):
[in this window]
[in a new window]
FIG. 7. Methyl-ß-cyclodextrin does not affect influenza virus binding or internalization. (A) Influenza virus was treated with 20 mM methyl-ß-cyclodextrin or was left untreated and was bound to the surfaces of MDBK cells on ice at a high MOI. The cells were analyzed by FACS analysis using anti-NP antibodies. Negative control, uninfected cells. (B) Influenza virus was treated with 20 mM methyl-ß-cyclodextrin (c and d) or was left untreated (a and b) and was bound to the surfaces of MDBK cells on ice (binding) at a high MOI, followed by a shift to 37°C for 20 min (internalization). The cells were analyzed by indirect immunofluorescence microscopy with anti-NP antibodies, and the nuclei were stained with Hoechst 33258. Merged images of virus localization (red) and nuclei (blue) are shown.

We next addressed whether methyl-ß-cyclodextrin treatmentaffected virus internalization. Cells were incubated with ahigh MOI of methyl-ß-cyclodextrin-treated or untreatedinfluenza virions on ice and were shifted to 37°C for 20min. Virion localization was then assayed by immunofluorescencemicroscopy (Fig. 7B). As with flow cytometry assays, microscopicanalysis of cells at 4°C showed no significant differencesin virus binding. With both methyl-ß-cyclodextrin-treatedand untreated virions, a 20-min temperature shift to 37°Cresulted in the localization of virions to perinuclear vesicles.These data are consistent with equivalent internalizations ofmethyl-ß-cyclodextrin treated and untreated virusesinto endosomes.

Envelope cholesterol depletion inhibits influenza virus fusion. As an initial measure of influenza virus fusion, we assayedthe ability of virus fused at the cell surface to cause infectionof cells (26). Virions that were methyl-ß-cyclodextrintreated or untreated were bound to cells on ice, and then fusionwas induced with pH 5.0 buffer. Influenza virus infection wassubsequently assayed by single-hit infection assays, with theinclusion of monensin to prevent virus entry through endosomes(Fig. 8). We saw a significant difference in the ability ofmethyl-ß-cyclodextrin-treated viruses to fuse withcell membranes. We first carried out experiments by immunofluorescencemicroscopy. In the absence of methyl-ß-cyclodextrin,we observed efficient infection of cells ( $~$ 54%), albeit at alower level than when the virus entered cells via endocytosis(compare Fig. 8A, a and d, with Fig. 1). In the presence of20 mM methyl-ß-cyclodextrin, virus infection was virtuallyundetectable when the viruses were fused at the cell surfacewith pH 5.0 buffer (Fig. 8A, b and e). Control experiments inwhich viruses were not fused at the cell surface produced almostno detectable infection in the presence of monensin to blockendocytosis (Fig. 8A, c and f), confirming that the infectiondetected with pH 5.0 treatment was due solely to virions fusedat the cell surface.

View larger version (41K):
[in this window]
[in a new window]
FIG. 8. Envelope cholesterol depletion inhibits influenza virus fusion. (A) Influenza virus was pretreated with 20 mM methyl-ß-cyclodextrin or was left untreated and was bound to the surfaces of MDBK cells on ice. Fusion at the cell surface was induced by exposure to pH 5.0 buffer (a to d), or control samples were treated with pH 7.3 buffer (e and f). All samples were treated with monensin to block endocytosis. The cells were analyzed by indirect immunofluorescence microscopy at 5 h postinfection using anti-NP antibodies (b, d, and f), or the nuclei were stained with Hoechst 33258 (a, c, and e). (B) Quantitation of data in panel A with various concentrations of methyl-ß-cyclodextrin. At least 100 cells were counted for each sample to determine the percentages of infectivity. The error bars represent the standard errors of the mean. (C) Influenza virus was pretreated with 20 mM methyl-ß-cyclodextrin or was left untreated and was bound to the surfaces of MDBK cells on ice. Fusion at the cell surface was induced by exposure to pH 5.0 buffer. All samples were treated with monensin to block endocytosis. The cells were analyzed by FACS analysis using anti-NP antibodies. Negative control, uninfected cells. (D) Bar chart of FACS analysis as described for panel C with various concentrations of methyl-ß-cyclodextrin.

As with entry through endocytosis, low-pH-induced influenzavirus fusion was sensitive to methyl-ß-cyclodextrintreatment in a dose-dependent manner (Fig. 8B). Treatment with5 mM methyl-ß-cyclodextrin resulted in $~$ 65% of cellsbeing infected, relative to untreated virions. Methyl-ß-cyclodextrin(10 mM) allowed an infection rate of only 25%, and 20 mM methyl-ß-cyclodextrinallowed an infection rate of <10%. After treatment of virionswith 50 mM methyl-ß-cyclodextrin, virus infectivitywas almost undetectable. These data show that influenza virusfusion is sensitive to cholesterol depletion with methyl-ß-cyclodextrin.

To confirm these results, and to obtain more quantitative data,we carried out FACS analysis of our fusion assay. Under ourflow cytometry conditions, untreated virus gave a 55% infectionlevel (Fig. 8C and D). Treatment with 5 mM methyl-ß-cyclodextrinreduced the level of infection to 20%, 10 mM methyl-ß-cyclodextringave a 6% infection level, and 20 mM cyclodextrin resulted ina 3% infection level, with 50 mM methyl-ß-cyclodextrinallowing only 1% of the cells to be infected. In general, theflow cytometry data closely paralleled results from immunofluorescencemicroscopy, with a dose-dependent effect of methyl-ß-cyclodextrinthat produced a roughly sigmoidal decrease in infectivity withincreasing methyl-ß-cyclodextrin concentrations.

To confirm that the reduced viral infection observed was duesolely to inhibition of fusion, we employed a fluorescence-dequenchingassay (20, 45). Influenza virus (WSN) was treated with methyl-ß-cyclodextrinor remained untreated and was then labeled with R18. The viruswas bound to the cell surface, and fusion was induced with low-pHbuffer (Fig. 9). With untreated virions, we observed rapid dequenchingof the probe as fusion occurred at low pH (Fig. 9). In contrast,at neutral pH, little or no fusion was observed in this assay(Fig. 9). When virions were treated with 20 mM methyl-ß-cyclodextrin,we observed a significant drop in both the extent and initialrate of fusion (Fig. 9). This effect was dose dependent, astreatment with 50 mM methyl-ß-cyclodextrin showedfurther reductions in both the fusion rate and extent (Fig.9). We estimate that a 45% reduction of fusion occurred with20 mM methyl-ß-cyclodextrin and a drop of $~$ 60% occurredwith 50 mM methyl-ß-cyclodextrin. These data confirmthat depletion of envelope cholesterol is important for influenzavirus fusion with host cells.

View larger version (19K):
[in this window]
[in a new window]
FIG. 9. Fluorescence-dequenching assays of virus fusion. Influenza virus was pretreated with methyl-ß-cyclodextrin (MßCD) or left untreated, labeled with R18, and bound to the surfaces of MDBK cells on ice. The cells were warmed to 37°C, and fusion at the cell surface was induced by dropping the pH of the buffer to 4.9 (solid symbols), or control samples were maintained in pH 7.0 buffer (open squares).

DISCUSSION

Top

Discussion

In this paper, we examined the role of envelope cholesterolon influenza virus infection, with a focus on the events occurringduring virus entry, namely, binding, internalization, and fusion.We showed a marked effect of influenza virus envelope cholesteroldepletion on entry and infection, most likely due to an inhibitionof virus fusion.

Several groups have studied the role of cholesterol in influenzaHA-mediated fusion. For fusion between influenza virus and liposomes,the inclusion of cholesterol in the target membrane has beenshown to have little effect on the extent or rate of fusion(46, 49). However, subsequent studies did show a strong effectof target membrane cholesterol on the growth of fusion poreswhen cells expressing HA were fused to planar phospholipid bilayers(36). In addition, cyclodextrin-mediated depletion of HA-expressingcells did not affect fusion to red blood cell ghosts (28). Whileallowing extensive manipulation of the experimental system,the use of HA-expressing cells, red blood cells, or planar bilayersfor fusion studies suffers from the disadvantage that many biologicalparameters are not present, e.g., membrane curvature, densityof HA, and presence of other proteins.

Our present data clearly show an effect of envelope cholesterolon influenza virus fusion with cell membranes; however, ourdata have all been obtained with the WSN (H1N1) strain of influenzavirus. It remains possible that differences may occur with othervirus strains, and it is important to note that the majorityof studies of influenza virus fusion have been performed withH3 serotypes (e.g., X:31). Our fluorescence-dequenching assaysshow values for the initial rate and extent similar to thoseobserved for the PR/8/34 strain of the virus (31), and theyare clearly pH dependent; however, some differences from fusionof the X:31 strain of the virus may be apparent.

What, then, is the role of cholesterol in the virion envelope?Although very limited numbers of HA molecules seem to be requiredto actually trigger the fusion event (3, 17), it is clear thatthere must be cooperativity among HA trimers in order for fusionto occur (3, 4, 11, 13, 14). Based on the known role of cholesterol-containinglipid microdomains in the formation of influenza virions (29,39, 51), we propose that virion cholesterol is important forthe organization of influenza virus HA trimers into fusion-competentdomains and that depletion of cholesterol inhibits virus infectivitydue to inefficient fusion.

Our data show that pH-mediated fusion of virions with the plasmamembrane is sensitive to envelope cholesterol. In our infectivityassays, we found that virus fusion (Fig. 8) is consistentlymore sensitive than overall infection to methyl-ß-cyclodextrindepletion (Fig. 1). We observed a sudden decrease in infectivitybetween viruses treated with 5 mM methyl-ß-cyclodextrinand those treated with 10 mM methyl-ß-cyclodextrin(Fig. 8)—a decrease to the same degree was not observedfollowing entry through endosomes (Fig. 1). It is possible thatthis represents differences in the local environment betweenfusion from within late endosomes and low-pH-induced fusionat the plasma membrane. We have yet to address the effect ofenvelope cholesterol on fusion within the endosome. Also notableis the fact that measurement of virus fusion by fluorescence-dequenchingassays indicated that the virus was somewhat less sensitiveto cholesterol depletion than measurement by infectivity assays.It is possible that the different MOIs employed in the differentassays contribute to differences in the apparent sensitivityof methyl-ß-cyclodextrin (45). Cholesterol depletionof HIV-1 envelopes has been shown to affect virus infectivity,based on inhibition of virus internalization (7, 18). HIV-1fuses with the plasma membrane, so cholesterol was suggestedto play a role in the conformational changes accompanying gp41-mediatedfusion (18, 24). In general terms, HIV-1 entry appears to bemore sensitive to cholesterol depletion than influenza virusentry. It is possible that the complex receptor-primed conformationalchanges accompanying HIV-1 fusion are more dependent on thelipid architecture of the viral envelope than those occurringduring the conceptually simpler pH-triggered fusion of influenzavirus.

Influenza virus is known to selectively incorporate both proteinsand lipids into its envelope (21, 22, 29), and influenza virusHA association with lipid microdomains, or rafts, is well established(39). It remains to be seen whether the effect of methyl-ß-cyclodextrinshown here is specific to the presence of lipid microdomainsin the virus envelope or whether the cholesterol itself is thekey component.

Previous studies have shown that HA-mediated hemifusion is dependenton the lipid composition of the contacting membrane monolayers(9), although cholesterol itself was not analyzed. In this paper,we show a specific dependence on cholesterol in the viral envelope(the donor membrane for fusion); however, the identificationof the precise stage in influenza virus envelope fusion forwhich cholesterol is required awaits further investigation.

ACKNOWLEDGMENTS

We thank Ruth Collins for helpful discussions throughout thecourse of this work, Chris Heger and Kate Beach for help withfluorimetry assays, and Beverley Bauman and Damon Ferguson forexcellent technical assistance. We are indebted to Judy Whitefor advice on fluorescence-dequenching assays. We also thankJim Pipas for his kind contribution of reagents.

This work was supported by a Scientist Development Grant fromthe American Heart Association and by grant R01AI48678 fromthe National Institutes of Health (to G.R.W).

FOOTNOTES

* Corresponding author. Mailing address: C4 127 Veterinary Medical Center, Department of Microbiology and Immunology, Cornell University, Ithaca, NY 14853. Phone: (607) 253-4019. Fax: (607) 253-3384. E-mail: grw7{at}cornell.edu.

REFERENCES

Top