Borna Disease Virus Requires Cholesterol in both Cellular Membrane and Viral Envelope for Efficient Cell Entry

Abstract

Borna disease virus (BDV), the prototypic member of the family Bornaviridae within the order Mononegavirales, provides an important model for the investigation of viral persistence within the central nervous system (CNS) and of associated brain disorders. BDV is highly neurotropic and enters its target cell via receptor-mediated endocytosis, a process mediated by the virus surface glycoprotein (G), but the cellular factors and pathways determining BDV cell tropism within the CNS remain mostly unknown. Cholesterol has been shown to influence viral infections via its effects on different viral processes, including replication, budding, and cell entry. In this work, we show that cell entry, but not replication and gene expression, of BDV was drastically inhibited by depletion of cellular cholesterol levels. BDV G-mediated attachment to BDV-susceptible cells was cholesterol independent, but G localized to lipid rafts (LR) at the plasma membrane. LR structure and function critically depend on cholesterol, and hence, compromised structural integrity and function of LR caused by cholesterol depletion likely inhibited the initial stages of BDV cell internalization. Furthermore, we also show that viral-envelope cholesterol is required for BDV infectivity.

Borna disease virus (BDV) is an enveloped virus with a nonsegmented negative-strand RNA genome whose organization (3′-N-p10/P-M-G-L-5′) is characteristic of mononegaviruses (6, 28, 46, 48). However, based on its unique genetics and biological features, BDV is considered to be the prototypic member of a new virus family, Bornaviridae, within the order Mononegavirales (8, 28, 46, 49).

BDV can infect a variety of cell types in cell culture but in vivo exhibits exquisite neurotropism and causes central nervous system (CNS) disease in different vertebrate species, which is frequently manifested in behavioral abnormalities (19, 33, 44, 53). Both host and viral factors contribute to a variable period of incubation and heterogeneity in the symptoms and pathology associated with BDV infection (14, 16, 29, 42, 44). BDV provides an important model for the investigation of both immune-mediated pathological events associated with virus-induced neurological disease and mechanisms whereby noncytolytic viruses induce neurodevelopmental and behavioral disturbances in the absence of inflammation (15, 18, 41). Moreover, serological data and molecular epidemiological studies suggest that BDV, or a BDV-like virus, can infect humans and that it might be associated with certain neuropsychiatric disorders (17, 24), which further underscores the interest in understanding the mechanisms underlying BDV persistence in the CNS and its effect on brain cell functions. The achievement of these goals will require the elucidation of the determinants of BDV cell tropism within the CNS.

BDV enters its target cell via receptor-mediated endocytosis, a process in which the BDV G protein plays a central role (1, 5, 13, 14, 39). Cleavage of BDV G by the cellular protease furin generates two functional subunits: GP1 (GP_N), involved in virus interaction with a yet-unidentified cell surface receptor (1, 39), and GP2 (GP_C), which mediates a pH-dependent fusion event between viral and cellular membranes (13). However, a detailed characterization of cellular factors and pathways involved in BDV cell entry remains to be done.

Besides cell surface molecules that serve as viral receptors, many other cell factors, including nonproteinaceous molecules, can influence cell entry by virus (52). In this regard, cholesterol, which plays a critical role in cellular homeostasis (55), has also been identified as a key factor required for productive infection by different viruses. Accordingly, cholesterol participates in a variety of processes in virus-infected cells, including fusion events between viral and cellular membranes (3), viral replication (23), and budding (35, 37), as well as maintenance of lipid rafts (LR) (12) as scaffold structures where the viral receptor and coreceptor associate (11, 26, 32, 36). LR are specialized microdomains within cellular membranes constituted principally of proteins, sphingolipids, and cholesterol. LR facilitate the close proximity and interaction of specific sets of proteins and contribute to different processes associated with virus multiplication (38). Cholesterol can also influence virus infection by contributing to the maintenance of the properties of the viral envelope required for virus particle infectivity (21, 54). Here, we show for the first time that cholesterol plays a critical role in BDV infection. Depletion of cellular cholesterol prior to, but not after, BDV cell entry prevented productive BDV infection, likely due to disruption of plasma membrane LR that appear to be the cell entry point for BDV. In addition, we document that cholesterol also plays an essential role in the properties of the BDV envelope required for virus particle infectivity.

MATERIALS AND METHODS

Cells and viruses.

Vero E6 cells (ATCC CRL-1586) were maintained in Dulbecco's modified Eagle's medium supplemented with 2 mM glutamine, 10 mM HEPES, and 10% heat-inactivated fetal bovine serum. The recombinant vesicular stomatitis viruses (rVSV) rVSVΔG*/BDVG and rVSVΔG*/VSVG have been described previously (39). The BDV and VSV strains utilized were He80 (47) and Indiana (22), respectively. CHO cells expressing Fc-GP1 were maintained in G-Dulbecco's modified Eagle's medium (10) supplemented with 25 μM methionine sulfoximine.

Quantification of cells infected with rVSV.

rVSV-infected cells were quantified based on green fluorescent protein (GFP) expression by rVSV. For this purpose, infected cells were trypsinized and fixed with 4% paraformaldehyde (PFA) prior to fluorescence-activated cell sorter (FACS) analysis.

Pharmacological treatments of cells.

Cells were treated with methyl-β-cyclodextrin (MβCD) (Sigma) at the indicated concentrations for 1 h, followed by three washes with medium. Cholesterol replenishment experiments were done by incubating cells or virus with MβCD (2.5 mM) loaded with soluble cholesterol (0.5 mM).

Treatment of CF-BDV infectious particles with MβCD.

Cell-free BDV (CF-BDV) was prepared as described previously (5, 14). CF-BDV was treated with MβCD diluted in OptiMem at the indicated concentrations for 1 h at 37°C. After drug treatment, CF-BDV was recovered by ultracentrifugation through a sucrose cushion (20%). The CF-BDV-containing pellet was resuspended in caveolin 1 (CAV-1) and incubated in a second treatment in the presence of cholesterol-loaded MβCD for 1 h at 37°C, and then the medium containing cholesterol-loaded MβCD was removed by ultracentrifugation through a sucrose cushion (20%). The final CF-BDV pellet was resuspended in OptiMem and subjected to titration.

Antibodies.

The rabbit serum to BDV GP1 was described previously (5). To detect the nucleocapsid protein of VSV, we used a murine monoclonal antibody (antibody to VSV N). The antiserum to mouse CAV-1 (BD Biosciences); AffiniPure mouse anti-human immunoglobulin G (IgG), Fcγ fragment specific (Jackson ImmunoResearch); and anti-transferrin receptor (Zymed) were commercially obtained.

Immunoadhesin generation.

Fc-GP1 was constructed by fusing an IgG1 Fc domain N terminus to BDV GP1 lacking the predicted signal peptide. For this, BDV GP1 from the Gln residue at position 21 to the Leu residue at position 244 was cloned in frame with Fc (2). Fc-GP1 was produced in CHO cells using the glutamate synthetase amplification system (10). Fc-GP1 was batch purified from culture supernatants by affinity chromatography over protein A.

Quantification of immunoadhesin-cell surface interaction.

Fc-GP1 immunoadhesin (250 ng) in phosphate-buffered saline (PBS) was incubated with 10⁵ cells in suspension for 1 h at 4°C. After three washes with PBS, the cells were incubated first with an antibody, anti-human Fc, at 4°C for 1 h, followed by three washes with PBS prior to incubation with the secondary anti-mouse (IgG) phycoerythrin-conjugated antibody (Jackson Immunoresearch). The cells were fixed with 4% PFA and subjected to FACS analysis using a FACSCalibur flow cytometer (Becton Dickinson) and analyzed using FlowJo (Treestar) software.

Blockade of BDV G-mediated cell entry by Fc-GP1.

Vero cells were preincubated with increasing concentrations of Fc-GP1 for 45 min at 4°C and subsequently infected (multiplicity of infection [MOI], 0.1) with either rVSVΔG*/VSVG or rVSVΔG*/BDVG in the presence of Fc-GP1. After adsorption for 45 min at 4°C, the cells were washed three times with medium containing 2% fetal bovine serum, and fresh medium was added to the cell cultures. At 7 h postinfection (p.i.) (rVSV), the cells were fixed and examined for expression of viral antigen by immunofluorescence assay (IFA).

Flotation assays.

The conditions used for flotation assays were essentially as described previously (32). Briefly, Vero cells (10⁷) were incubated for 1 h with Fc-GP1 (10 μg) in ice-cold PBS, followed by lysis with 1% Triton X-100 in TNE buffer (25 mM Tris [pH 7.5], 150 mM NaCl, 5 mM EDTA) containing a protease inhibitor cocktail (Roche) for 30 min at 4°C in a final volume of 1 ml. After clarification by centrifugation at low speed, the supernatant was mixed with 1 ml 80% sucrose in TNE buffer to a final sucrose concentration of 40%, placed at the bottom of the tube, and overlaid with 6 ml 30% and 3 ml 5% sucrose in TNE buffer. Samples were centrifuged for 18 h at 180,000 × g using an SW-41 rotor (Beckman) at 4°C. Fractions (1 ml) were collected and subjected to trichloroacetic acid precipitation. The concentrated samples were analyzed by Western blotting using antibodies to BDV-GP1, CAV-1, and transferrin receptor. Autoradiogram films were scanned with a GS-800 densitometer (Bio-Rad) and analyzed with Image Quant image analysis software (Molecular Dynamics, Palo Alto, CA).

RESULTS

Effect of MβCD on BDV infection.

To assess whether successful infection by BDV was sensitive to cellular cholesterol levels, we used MβCD, a drug widely used to sequester cellular cholesterol, mainly from plasma membrane, in a fast and effective manner (20). Moreover, upon MβCD treatment, membrane cholesterol levels remain low for 9 to 12 h (25, 43). Cells were treated with MβCD (10 mM) for 1 h either 1 h prior to or 1 h after infection with BDV (MOI, 0.1), and at 48 h p.i., we examined the number of BDV-infected cells based on expression of BDV N as determined by IFA (Fig. 1). MβCD treatment prior to, but not after, BDV infection caused a strong reduction in the number of BDV-infected cells (Fig. 1C).

FIG. 1.

Sequestering of cellular cholesterol inhibits BDV infection. (A) Schema of the experimental design. Cells were treated first with MβCD (10 mM) 1 h prior to, or 1 h after, infection with BDV (MOI = 0.1 focus-forming unit/cell). (B) At 48 h p.i., cells were fixed and subjected to IFA to identify the cells expressing BDV N. (C) BDV N-positive cells were counted and normalized over the number of BDV-infected cells, with BDV-infected cells without treatment as a control. The data are from three independent experiments. **, P ≤ 0.01. The error bars indicate standard deviations.

BDV G is the sole viral protein strictly required for receptor recognition and cell entry, and we have documented that an rVSV expressing BDVG instead of VSVG (rVSVΔG*/BDVG) mimics the cell tropism and entry pathway of bona fide BDV (39). VSV infection is known to be resistant to MβCD treatment (45, 50). We therefore predicted that if depletion of cellular cholesterol specifically inhibited BDV G-mediated cell entry, then MβCD treatment should also inhibit infection with rVSVΔG*/BDVG. To test this, we treated cells with increasing concentrations of MβCD and subsequently infected them (MOI, 0.5) with rVSVΔG*/BDVG or with rVSVΔG*/VSVG as a control, and at 7 h p.i., we determined the numbers of infected cells (Fig. 2). These two rVSV also expressed GFP (39), which allowed us to use FACS to monitor infection. At the lowest (2 mM) concentration of MβCD used, the infectivities of both rVSVΔG*/VSVG and rVSVΔG*/BDVG were similarly reduced compared to untreated and infected cells. This result likely reflected an initial generalized physiological response of the cell to MβCD exposure that translated into a nonspecific inhibitory effect on virus infection. Importantly, within the 2 to 6 mM range, MβCD exhibited a clear dose-dependent inhibitory effect on rVSVΔG*/BDVG, whereas rVSVΔG*/VSVG infectivity remained unaffected, and we did not observe differences in cell survival within this range of MβCD treatment. At the highest (8 mM) concentration of MβCD used, we observed some degree of cell toxicity, which could explain the observed effect on rVSVΔG*/VSVG.

FIG. 2.

Depletion of cellular cholesterol inhibits BDV G-mediated cell entry. Vero cells were treated for 1 h with increasing concentrations of MβCD, followed by infection with either rVSVΔG*/BDVG or rVSVΔG*/VSVG at an MOI of 0.5 PFU/cell. At 7 h p.i., cells were collected, fixed, and subjected to FACS analysis to detect GFP-positive cells. *, P ≤ 0.05; **, P ≤ 0.01. The error bars indicate standard deviations.

To confirm that the inhibitory effect of MβCD on rVSVΔG*/BDVG infection was due specifically to cholesterol depletion, we conducted a reconstitution experiment (Fig. 3). Cells were treated first with MβCD (5 mM) for 1 h, followed by the removal of MβCD and incubation for 1 h in serum-free medium or cholesterol-loaded MβCD to restore normal levels of cellular cholesterol. Then, the cells were infected with rVSVΔG*/BDVG, and 14 h p.i., a time point at which the effect of subsequent rounds of infection was still not significant (39), GFP-expressing cells were detected by FACS analysis. Cholesterol replenishment restored normal levels of cell susceptibility to rVSVΔG*/BDVG infection.

FIG. 3.

Replenishment of normal cellular cholesterol levels restores BDV G-mediated entry. Cells were treated for 1 h with MβCD (5 mM), followed by cholesterol replenishment by incubation with exogenous cholesterol complexed with MβCD (+Ch) for 1 h (see Materials and Methods for details), and infected with rVSVΔG*/BDVG. At 14 h p.i., the cells were fixed and GFP expression was determined by FACS analysis. The data obtained from three independent experiments are shown. *, P ≤ 0.05. The error bars indicate standard deviations.

Effect of depletion of cellular cholesterol on BDV attachment to its target cells.

Depletion of cholesterol by MβCD treatment has been shown to affect the expression levels of a variety of cell surface proteins (20). Consequently, MβCD-mediated inhibition of BDV infection could reflect reduced levels of cell surface proteins involved in BDV-cell attachment. To examine this, we generated stably transfected CHO cells (CHO/Fc-GP1) expressing an immunoadhesin molecule in which the human Fc domain of IgG was fused to the N terminus of BDV GP1 (Fc-GP1) (Fig. 4A). The Fc domain conferred a tag on the complex to ease purification. CHO/Fc-GP1 cells produced levels of secreted Fc-GP1 that were readily detectable in the cell culture supernatant. Biochemical analysis showed that under denaturing reducing conditions, Fc-GP1 exhibited an electrophoretic mobility corresponding to that of a protein with a mass (circa 90 kDa) predicted for a monomeric Fc-GP1 (Fig. 4B). In contrast, under nonreducing conditions, we also observed species with an electrophoretic migration corresponding to that of predicted dimers generated via Fc-Fc interaction (not shown).

FIG. 4.

Purification and functional characterization of immunoadhesin Fc-GP1. (A) Schema of BDV Fc-GP1 immunoadhesin. The immunoadhesin Fc-GP1 was generated by fusion of the Fc domain of human IgG to the N terminus of a BDV GP1 construct lacking the signal peptide (SP). (B) Expression of Fc-GP1 by CHO cells stably transfected with Fc-GP1. Under denaturing conditions, Fc-GP1 exhibited a mobility of ∼90 kDa, based on both Coomassie staining and Western blotting using a serum against GP1, a migration consistent with a monomeric Fc-GP1. (C) Vero cells were precoated with the indicated concentrations of Fc-GP1 and then infected with rVSVΔG*/BDVG or rVSVΔG*/VSVG (MOI, 0.1). Cells infected with rVSV were detected by IFA using anti-VSV N serum. The error bars indicate standard deviations. (D) Equal amounts of purified Fc-GP1 and Fc control were incubated with cells for 1 h, and immunoadhesin binding was quantified using an anti-Fc serum. PE, phycoerythrin.

To assess whether Fc-GP1 was biologically active, we examined its ability to inhibit BDV G-mediated cell entry. For this, we used increasing amounts of Fc-GP1 to attempt to block the interaction between the BDV-G-bearing rVSV and the cell surface receptor involved in BDV entry. Fc-GP1 exhibited a dose-dependent inhibitory effect on rVSVΔG*/BDVG, but not rVSVΔG*/VSVG, cell entry (Fig. 4C). In addition, we also did binding assays showing that Fc-GP1 specifically bound, via its GP1 domain, to the surfaces of Vero cells, which are known to be highly susceptible to BDV (Fig. 4D). Depletion of cellular cholesterol by treatment with MβCD (5 mM) caused only a very modest (10%) reduction in Fc-GP1 binding to Vero cells compared to untreated control Vero cells (Fig. 5), which did not account for the large (≥80%) reduction in rVSVΔG*/BDVG infectivity observed in MβCD-treated Vero cells (Fig. 2). These findings indicated that initial cell attachment of BDV via interaction of GP1 with a yet-unidentified virus receptor is not significantly affected by levels of plasma membrane cholesterol.

FIG. 5.

Fc-GP1 binding to cholesterol-depleted cells. Vero cells were incubated with Fc-GP1 (250 ng) after cholesterol depletion by MβCD (5 mM). (A) Binding of Fc-GP1 to cells was determined by FACS analysis and quantified. The data obtained are shown in split charts. (B) The results from panel A (gray shading, mock; dashed line, Fc-GP1; solid line, MβCD plus Fc-GP1) were combined in a single graphic to facilitate the assessment of the effect of cellular-cholesterol depletion on Fc-GP1 binding to the plasma membranes of live cells. PE, phycoerythrin.

Role of LR in BDV cell entry.

We next explored whether LR, which have been shown to play an important role in the entry of a variety of both enveloped and nonenveloped viruses (4), were involved in BDV entry. To address this question, we examined whether Fc-GP1 accumulated within LR at the cell surface. For this, we incubated cells with Fc-GP1 for 1 h at 4°C to prevent virus-cell interaction and then prepared cell lysates in the presence of Triton X-100 (1%) at 4°C, conditions known to preserve LR structures unaltered (35, 56). We subjected cell lysates to a flotation assay and analyzed the collected fractions by Western blotting using antibodies against GP1 and CAV-1 (Fig. 6A). Most Fc-GP1 accumulated within the fraction corresponding to LR (Fig. 6B), defined by the presence of CAV-1. However, we also detected low levels of Fc-GP1 in other fractions.

FIG. 6.

Flotation assay of lysates prepared from cells preincubated with Fc-GP1. (A) Vero cells (10⁷) were incubated with Fc-GP1 (10 μg) at 4°C to avoid receptor-mediated internalization of Fc-GP1. Cell lysates were prepared in the presence of Triton X-100 (1%) at 4°C to preserve detergent-resistant domains (DRM). After centrifugation at low speed, supernatant containing the membrane fractions was subjected to flotation centrifugation, and the fractions were collected and subjected to Western blotting using antisera to GP1, as well as to CAV-1 and transferrin receptor (TrfRc), as markers for LR and the soluble fraction, respectively. The schematic illustrates the relative positions of different fractions in the centrifuge tube containing the indicated sucrose concentrations. (B) Specific signals corresponding to Fc-GP1, CAV-1, and TrfRc were quantified by densitometry.

Role of cholesterol in the infectivity of cell-free BDV.

To assess whether cholesterol influenced the properties of the BDV envelope required for optimal virus infectivity, we examined the effect of MβCD treatment on the infectivity of CF-BDV. For this, we treated CF-BDV (10⁶ focus-forming units) with 0, 5, or 10 mM MβCD for 1 h at 37°C, and after drug removal by ultracentrifugation, virus infectivity was determined using an immunofocus assay. MβCD treatment caused a marked dose-dependent reduction in BDV infectivity (Fig. 7A). The infectivity of MβCD-treated virus particles was recovered upon cholesterol replenishment (Fig. 7B), indicating that the effect of MβCD on CF-BDV infectivity was not due to virus particle disruption or other nonspecific effects associated with drug treatment, but rather, specifically to reduced cholesterol levels in the virus envelope.

FIG. 7.

Requirement for cholesterol in the virus envelope for efficient BDV infection. (A) CF-BDV was purified by centrifugation through a sucrose cushion (20%). Then, CF-BDV was resuspended and incubated in OptiMem containing different concentrations of MβCD for 1 h at 37°C. The CF-BDV preparation was subsequently purified by centrifugation through a sucrose cushion to eliminate MβCD, and CF-BDV-containing pellets were resuspended in OptiMem and subjected to BDV titration. (B) CF-BDV virions were purified and treated with MβCD as described for panel A. Aliquots were subsequently left untreated or treated with cholesterol complexed with MβCD (+Ch). CF-BDV preparations were purified through a sucrose cushion (20%) to eliminate MβCD, and the CF-BDV-containing pellets were resuspended in OptiMem and subjected to BDV titration. The error bars indicate standard deviations.

DISCUSSION

The role of cholesterol as an important modulator of viral infection has been well documented (4, 38). Accordingly, altered cellular cholesterol levels have been shown to specifically influence the courses of infection of a variety of viruses. Here, we have demonstrated that depletion of cellular cholesterol had a strong inhibitory effect on BDV infection. As with hepatitis C virus (23), cholesterol depletion could affect BDV replication or transcription, or both. However, only cholesterol depletion prior to, but not after, infection had an effect on BDV infection (Fig. 1), suggesting that cellular cholesterol is required for an early step of BDV cell entry. Because these first steps of BDV infection are driven by BDV G, we investigated the role of cellular cholesterol in BDV G-mediated cell entry by using rVSVΔG*/BDVG, which allowed us to examine the role of BDV G independently of other BDV gene products. VSV cell entry and multiplication are not affected by cholesterol depletion (45, 50). Therefore, our finding that infection with rVSVΔG*/BDVG was inhibited upon depletion of cellular cholesterol by treatment with MβCD indicated that the inhibitory effect of cholesterol on BDV infection is at the cell entry level, a process mediated by the virus surface glycoprotein G. We also documented the rescue of cell susceptibility to BDV infection upon restoration of normal cellular cholesterol levels, confirming a specific role of cellular cholesterol in BDV infection.

The first step of the BDV entry process involved the interaction between a yet-unidentified cell surface receptor and the virus GP1 (14, 40). We observed that binding of BDV GP1, the virus ligand that interacts with the cell surface virus receptor, to BDV-susceptible cells was minimally affected by cholesterol depletion (Fig. 5), but following its initial attachment to the cell surface, BDV GP1 was found to accumulate in LR (Fig. 6). The accumulation of Fc-GP1 within LR (Fig. 6) strongly suggests that recruitment of cell-attached BDV to plasma membrane LR is required prior to cell internalization of virus taking place, a finding similar to that described for other viruses (4). The roles of LR in viral infection have been comprehensively discussed (34, 38). As with other viruses, BDV may use LR structures as a platform to interact with additional host cell factors required for efficient BDV entry, ranging from a variety of coreceptors as described for other viruses (31) to functional proteins, such as tyrosine kinases (51), that can trigger the activation of different intracellular pathways required for effective virus entry. Consequently, inhibition of BDV infection in MβCD-treated cells is likely related to LR disruption. The assembly of LR depends strictly on cholesterol (34). However, we cannot rule out the possibility that cholesterol may also have a role in later steps of the BDV cell entry process, including fusion between viral and cellular membranes in the late endosome, as occurs with poliovirus (7, 30). Nevertheless, our data showed that cholesterol depletion at 1 h p.i. or later did not affect BDV infection, and we have obtained evidence that pH-dependent fusion between BDV and cell membranes occurs within the first 45 min of BDV infection (data not shown). These findings strongly suggest that cholesterol plays its most critical role during early stages of BDV infection, prior to the pH-dependent fusion event that releases the virus ribonucleoprotein core that, upon translocation into the cell nucleus, directs RNA replication and transcription of the BDV genome. It should be noted that cholesterol dependence of endocytosis per se is an operational definition, and the specific pathways affected may vary between cell types.

It has been shown that several viruses depend on cholesterol levels in the viral envelope for efficient cell entry (21, 54), and in some cases, as with human immunodeficiency virus type 1 (9, 26, 27), both the cellular and virus envelope cholesterol levels influence cell entry by virus. We observed that cholesterol also played a critical role in the properties of the envelope of BDV required for efficient cell entry (Fig. 7). It is quite plausible that viral envelope cholesterol may play a role during the fusion between virus and cell membranes in the late endosome, but further analysis is required to elucidate this issue in the case of BDV.

In summary, our results have established that BDV infection depends on the presence of cholesterol in both cellular and viral envelopes. In addition, we have provided evidence suggesting that upon its cell attachment, BDV translocates to LR to initiate its cell internalization process by endocytosis.