|Home | About | Journals | Submit | Contact Us | Français|
To explore the association of the Newcastle disease virus (NDV) fusion (F) protein with cholesterol-rich membrane domains, its localization in detergent-resistant membranes (DRMs) in transfected cells was characterized. After solubilization of cells expressing the F protein with 1% Triton X-100 at 4°C, ca. 40% of total, cell-associated F protein fractionated with classical DRMs with densities of 1.07 to l.14 as defined by flotation into sucrose density gradients. Association of the F protein with this cell fraction was unaffected by the cleavage of F0 to F1 and F2 or by coexpression of the NDV attachment protein, the hemagglutinin-neuraminidase protein (HN). Furthermore, elimination by mutation, of potential palmitate addition sites in and near the F-protein transmembrane domain had no effect on F-protein association with DRMs. Rather, specific deletions of the cytoplasmic domain of the F protein eliminated association with classical DRMs. Comparisons of deletions that affected fusion activity of the protein and deletions that affected DRM association suggested that there is no direct link between the cell-cell fusion activity of the F protein and DRM association. Furthermore, depletion of cholesterol from cells expressing F and HN protein, while eliminating DRM association, had no effect on the ability of these cells to fuse with avian red blood cells. These results suggest that specific localization of the F protein in cholesterol-rich membrane domains is not required for cell-to-cell fusion. Paramyxovirus F-protein cytoplasmic domains have been implicated in virus assembly. The results presented here raise the possibility that the cytoplasmic domain is important in virus assembly at least in part because it directs the protein to cholesterol-rich membrane domains.
Plasma membranes contain liquid-ordered lipid microdomains called “lipid rafts.” These small domains are enriched in cholesterol and sphingolipids, as well as in specific transmembrane raft-organizing proteins and proteins containing covalently attached, long-chain acyl groups (reviewed in references 5, 6, and 36). It has also been recently recognized that there are different classes of lipid rafts (11). Protein recruitment to lipid rafts and raft clustering are both regulated during physiologically relevant signaling processes. As a result, lipid rafts have been proposed to function during assembly of biological complexes, such as the immunological synapse (11, 36). Recent studies of lipid rafts in different cell types also have suggested that these domains can be directly connected to underlying cytoskeletal elements (25, 41). Further, localized reorganization of the cortical actin cytoskeleton has been implicated in raft-mediated signaling events during cell stimulation (17, 28, 37, 39).
In unstimulated cells, lipid raft domains are estimated to be ca. 50 nm in diameter by biophysical techniques (27, 38). However, aggregation of raft components with, for example, bivalent antibody, increases their size so that they appear as patches on the plasma membrane that can be visualized by immunofluorescence microscopy (36). After extraction with the nonionic detergent Triton X-100 at 4°C, lipid raft components coalesce into lipid-rich, detergent-resistant membranes (DRMs) (36) that may be enriched in specific proteins. For example, glycosylphosphatidylinositol-anchored proteins, signaling proteins such as Src family kinases and heterotrimeric G proteins (11, 36), as well as cytoskeletal proteins, may be found in this cell fraction (25). Classical DRMs have densities of 1.07 to l.14. In neutrophils, a fraction of DRMs have densities higher than the classical DRMs and have been called heavy DRMs or DRM-H (25). The existence of such DRMs in other cell types has yet to be documented.
Lipid rafts have also been implicated in the assembly of many different enveloped viruses. Ebola virus and Marburg virus glycoproteins (3), as well as the Env proteins of human immunodeficiency virus (HIV) (26, 29) and murine leukemia virus (15), have been recovered in DRMs from infected cells, and the raft-associated lipid GM1 has been found in virions (3). Influenza is reported to bud from lipid rafts (31), and both the influenza virus HA and NA glycoproteins (2, 31) have been reported to be associated with DRMs. Paramyxovirus proteins, including Sendai virus F protein, HN protein, and M protein (1), and measles virus proteins, including the F protein, also have been found in this cell fraction (18, 40). Respiratory syncytial virus proteins are associated with DRMs, and respiratory syncytial virus is proposed to assemble in caveolae, which are cholesterol-rich invaginations of the plasma membrane organized by the raft-associated protein, caveolin (13, 19, 23).
Plasma membranes of paramyxovirus-infected cells modified with viral proteins are not only sites of virus assembly but also sites involved in cell to cell fusion, resulting in syncytium formation characteristic of paramyxoviruses (14). To investigate the potential biological significance of lipid raft localization of paramyxovirus glycoproteins in assembly as well as during syncytia formation, we characterized the DRM association of the Newcastle disease virus (NDV) F protein. NDV, a prototype paramyxovirus, encodes two transmembrane glycoproteins, the fusion protein (F) and hemagglutinin-neuraminidase (HN) protein (14). Although the HN protein is the viral attachment protein binding sialic acid containing receptors, the F protein directs the fusion of the viral and cellular membranes required for viral penetration, as well as cell-cell fusion required for syncytium formation. The F protein, synthesized as a precursor F0, must be cleaved into two disulfide-linked subunits, F1 and F2, to activate fusion activity (14). The fully glycosylated NDV fusion protein consists of an extracellular domain of ca. 470 amino acids, a transmembrane domain (TM domain) located near the carboxyl terminus and a cytoplasmic or intravirion domain (CT domain) of ca. 29 amino acids (9). Like many paramyxovirus fusion proteins, the NDV F protein is palmitoylated, presumably by covalent modification of one or both of two cysteine residues located in the TM domain and at the TM-CT junction (8). Palmitoylation of glycoproteins is often directly linked to lipid raft localization (3, 15, 29).
We report that in cells transfected with the cDNA of the NDV F protein on average 38% of total, steady-state, cell-associated NDV F protein is indeed found in DRMs with densities of 1.07 to 1.14. However, in contrast to previous reports about many other glycoproteins, mutation of cysteine residues that are potential palmitoylation sites did not affect localization in DRMs. Rather, specific deletions within the CT domain of the protein affected classical DRM association. Comparisons of this DRM association and fusion activities of these mutants show little correlation between classical DRM association and cell-cell fusion. Furthermore, cholesterol depletion of cells expressing F protein, as well as HN protein, had no effect on fusion with red blood cells (RBC).
COS-7 cells obtained from the American Type Culture Collection were maintained in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, nonessential amino acids, vitamins, and antibiotics (Invitrogen Corp.). NDV F genes were inserted into pSVL (Pharmacia) as previously described (34). The mutant F-K115Q has a lysine-to-glutamine change to eliminate the furin recognition site and was generated as previously described (16). NDV fusion protein genes with mutations in the CT domain were generated as previously described (35).
To raise anti-HR2 antibody, sequences encoding amino acids 470 to 500 were prepared by PCR with a primer containing a BamHI site and a primer with an EcoRI site, as well as the appropriate F-protein gene sequences. The PCR product was cloned into a BamHI-EcoRI-cut pGex-2T (Pharmacia), and the ligated product transformed into BL21 cells (Stratagene). BL21 cells containing the plasmid were induced with IPTG (isopropyl-β-d-thiogalactopyranoside; 0.1 mM) for 3 h at 37°C. The cells were pelleted and then lysed with BugBuster (Novagen) by protocols recommended by the manufacturer, and the glutathione S-transferase (GST)-F fusion protein was purified by using a GST-Bind Resin (Novagen) and standard protocols. The purified, concentrated fusion protein was used as an antigen to raise polyclonal rabbit antisera (Capralogics, Hardwick, Mass.).
Lipofectamine (Invitrogen Corp.) was used to deliver plasmids into the cells, as recommended by the manufacturer. Briefly, 3 × 105 COS-7 cells were cultured in 35-mm plates and 20 to 24 h later the cells were transfected. For each 35-mm plate, mixes of DNA (0.5 μg) in 0.1 ml of OptiMEM (BRL/Gibco) and 10 μl of transfection reagent in 0.2 ml of OptiMEM were incubated at room temperature for 40 min, diluted with 0.7 ml of OptiMEM, and added to a plate previously washed twice with 2 ml of OptiMEM. Cells were incubated with DNA for 5 h, the transfection mix was replaced with complete medium, and the cells were cultured for another 48 h at 37°C in 5% CO2.
At 48 h after transfection, cells from each 35-mm plate were washed with ice-cold phosphate-buffered saline (PBS) and lysed in 250 μl of TNE buffer (25 mM Tris-HCl [pH 7.4], 150 mM NaCl, 5 mM EDTA) supplemented with 2.5 mg of N-ethylmaleimide (Sigma Corp.)/ml and 1% Triton X-100 on ice for 30 min. The extracts were homogenized and subjected to centrifugation for 15 s at 5,000 rpm. The postnuclear supernatant was kept on ice for an additional 30 min, mixed with ice-cold sucrose to produce a final sucrose density of 1.24 g/ml, and placed on top of 250 μl of sucrose (1.29 g/ml) in a polyallomer SW50.1 tube and over layered with 45, 43, 38, 32, 28, 18, and 5% sucrose dissolved in TNE buffer. Amounts of each solution varied with experiments. The samples were centrifuged in an SW 50.1 rotor at 100,000 × g for 18 h at 4°C. Fractions were collected from the bottom of the gradients (each fraction was 0.25 ml except for the second fraction that had a volume of 1 ml), and proteins present in each fraction were detected by Western analysis. After centrifugation, the densities of all sucrose fractions were measured by using a refractometer and are indicated in each figure.
First, 25 μl of each gradient fraction was blotted onto an Immobilon-P membrane and incubated with a horseradish peroxidase-conjugated cholera toxin B subunit (Calbiochem Corp). The presence of the cholera toxin B subunit was then detected by using enhanced chemiluminescence (Amersham Biosciences).
COS-7 cells transfected for 48 h were washed with serum-free DMEM (Gibco) and incubated with 10 mM methyl-β-cyclodextrin(MβCD; Sigma) in DMEM for 1 h at 37°C. After incubation, cells were washed with ice-cold PBS, and cytoplasmic extracts were prepared as described above.
A total of 30 μl of each density gradient fraction was mixed with 25 μl of gel sample buffer (125 mM Tris-HCl [pH 6.8], 2% sodium dodecyl sulfate, 10% glycerol) containing 1 M β-mercaptoethanol, and the proteins were separated in 8% sodium dodecyl sulfate-polyacrylamide gels. After electrophoresis, the gels were equilibrated in transfer buffer (25 mM Tris-HCl [pH 8.2], 192 mM glycine, 12% methanol) and transferred for 12 to 15 h to Immobilon-P (Millipore) membranes. The membrane was blocked in PBS containing 0.5% Tween 20 and 10% nonfat dried milk overnight at 4°C. Membranes were washed in PBS-Tween 20 and incubated with anti-HR2 antibody diluted in PBS-Tween 20 and 0.5% nonfat milk for 2 h at room temperature. Membranes were washed and then incubated with anti-rabbit immunoglobulin G (IgG) coupled to horseradish peroxidase (1:1,000 dilution in PBS-Tween) (Amersham Biosciences) for 2 h at room temperature. Membranes were washed extensively, and bound antibody was detected by using the enhanced chemiluminescence Western blotting detection reagent system (Amersham Biosciences). Densitometric analysis of the protein bands was performed by using Fluor-S Multi-Imager (Bio-Rad) and MultiAnalyst software (Bio-Rad).
Transfections were done as described above. At 48 h after transfection, cells were washed with PBS, detached from the plates by using trypsin-EDTA (Sigma), stained with primary antibody followed by secondary fluorescence-labeled antibody (Alexa Fluor 488 goat anti-rabbit IgG; Molecular Probes), fixed in 2% paraformaldehyde-PBS, and analyzed by flow cytometry as previously described (20, 22). Five thousand cells were collected by using a FACSCalibur apparatus (BD Bioscience) and then analyzed with CellQuest or FlowJo software.
COS-7 cells were cotransfected with wild-type or mutant F-protein genes (0.75 μg/plate), and the wild-type HN-protein gene (0.75 μg/plate). At 12 h posttransfection, cells were removed from the plate with trypsin (Gibco) and mixed with twice the number of untransfected COS-7 cells and replated at 5 × 105 cells/35-mm plate. The number of nuclei in 40 fusion areas was counted to determine the average size of syncytia at each time point as previously described (34). Values obtained after transfection of the vector alone were subtracted.
Content mixing was measured by using modifications of a protocol previously described (22, 33). Briefly, a plasmid encoding a Tet-responsive transcriptional activator, tTA (Clontech), was transfected (1 μg/35-mm plate) with pSVL-HN (0.75 μg/35-mm plate) and pSVL-F DNAs (0.75 μg/35-mm plate). A separate population of cells was transfected 24 h later with a plasmid encoding the β-galactosidase protein under the control of the Tet-responsive transcriptional activator (Clontech) at 1 μg/35-mm plate. After 20 h, cells transfected with the plasmid encoding the β-galactosidase protein were removed from the plate with trypsin and added on top of the HN- and F-protein-expressing cells. At 47 h posttransfection of the HN- and F-protein-expressing cells, when fusion was evident, the monolayers were lysed (Promega cell lysis buffer), and extracts assayed for β-galactosidase activity. Activity due to background fusion typical of COS-7 cells was measured after transfection of the cells with comparable amounts of vector alone. Values obtained were subtracted from values obtained with cells expressing wild-type or mutant HN and wild-type F proteins.
The protocol used was similar to that previously described (21). Briefly, avian RBC (Crane Laboratories) were washed in PBS and then incubated with 15 μg of R18 (octadecyl rhodamine B chloride; Molecular Probes)/ml for 30 min at room temperature in the dark. Three volumes of complete medium (DMEM with 10% fetal calf serum) were added, and incubation was continued for 30 min. The RBC were then washed four times in ice-cold PBS, resuspended in PBS containing CaCl2 (0.01%), and added to transfected cells, grown on coverslips, that had been washed in PBS with CaCl2. Transfected cells were incubated with labeled RBC for 30 min on ice. Cells were washed with ice-cold PBS containing CaCl2 and then incubated at 37°C. After incubation, cells were washed in cold PBS containing CaCl2 and immediately visualized and photographed by using a Nikon Diaphot 300 fluorescence microscope.
To characterize DRM association of the NDV F protein, the Triton X-100-insoluble membranes from cells transfected with F-protein cDNA and cells transfected with an empty vector were isolated by flotation in a sucrose gradient. Cells were extracted with Triton X-100 at 4°C, and cytoplasmic extracts were placed in the bottom of centrifuge tubes and overlaid with sucrose solutions with decreasing densities. After centrifugation to equilibrium, F-protein distribution across the density gradient was determined by Western analysis. Figure Figure1A1A shows the Western blot, whereas Fig. Fig.1B1B shows quantification of the amount of F protein (both F0 and F1) in each fraction, as well as the densities of each fraction. This figure, as well as other similar experiments, showed that 38% ± 6% of the total cell-associated F protein floated to densities characteristic of classical DRMs, i.e., densities of 1.07 to 1.14 when the DRMs were prepared as described in Materials and Methods. The majority of the protein remained at the bottom of the gradient in soluble fractions, whereas some F protein was found in the gradient at densities of 1.20 to 1.15 g/ml. The material in the 1.15- to 1.20-g/ml density fractions was not characterized further.
A characteristic of lipid rafts is a high concentration of GM1, and this lipid can be identified by its binding of cholera toxin subunit B. With this reagent, GM1 was found preferentially in gradient fractions with densities of 1.07 to 1.14 g/ml (Fig. (Fig.1C).1C). These results suggest that the wild-type fusion protein is recovered with DRMs with densities of 1.07 to 1.14 and are likely associated with lipid rafts. Vesicular stomatitis virus protein G is reported to be a non-raft-associated protein (31), and indeed, we did not find this protein in the classical DRM fraction (not shown).
To support the conclusion that F protein found in fractions with densities of 1.07 to 1.14 was associated with lipid rafts, we determined the localization of the F-protein DRM after treatment of cells with MβCD. This compound depletes cholesterol from plasma membranes and disrupts lipid rafts (30, 36). Figure 2A and B show that wild-type F protein was soluble in Triton X-100-extracted cells when cells were treated with 10 mM MβCD for 1 h prior to Triton X-100 extraction. Thus, cholesterol depletion disrupts association of the F protein with DRMs.
To verify that flotation of the F protein was due to association with DRMs, two conditions known to solubilize DRMs were used. DRMs are solublized in the detergent octyl-β-glucoside (36). F protein present in extracts prepared by using this detergent was found in the soluble fraction of cells (bottom of the gradient) (Fig. 2C and D). DRMs are also solubilized in Triton X-100 at room temperature. The F protein present in extracts prepared by cell lysis with Triton X-100 at room temperature was minimally associated with the classical DRM fraction (Fig. (Fig.2E).2E). These results indicate that conditions known to disrupt classical DRMs inhibited flotation of F protein into the gradient.
Figure Figure11 shows that the majority of the F protein associated with DRMs was the cleaved F protein, whereas minimal F0 was detected in this cell fraction. Cleavage of paramyxovirus F proteins, which occurs in the trans-Golgi membranes (24), results in a conformational change (12) that could influence DRM association. Alternatively, cleavage may occur prior to localization in rafts. To distinguish between these possibilities, the DRM association of an uncleaved F protein was determined. A single point mutation, K115Q, encoding glutamine instead of lysine, inhibits F-protein cleavage and results in an F protein expressed at the cell surface but inactive in cell-cell fusion (16; unpublished results). Figure Figure3A3A shows that the uncleaved F protein was detected in the classical DRM fraction at levels comparable to the cleaved F protein (33% ± 9%). Thus, cleavage per se had little influence upon the association of the F protein with the DRM fraction.
Most paramyxovirus-mediated fusion requires the coexpression of HN protein with the F protein (14). Indeed, many models for paramyxovirus-mediated fusion suggest that HN-protein coexpression alters the conformation of the F protein either before or after attachment of the HN protein to receptors. We therefore sought to determine whether HN-protein coexpression had any influence upon association of the F protein with DRMs. Figure Figure3B3B shows that, just as we observed with F protein expressed alone, 38% ± 3.5% of F protein expressed with HN protein was found in DRMs with a density of 1.07 to 1.14.
To understand the implications of F-protein DRM association, we set out to identify the F-protein sequences required for DRM association. Our approach was to utilize mutants of F protein with a focus on mutations in and around the TM domain. Since DRM association of transmembrane glycoproteins is often correlated with covalent fatty acid modifications on cysteine residues near the TM domain (23) and since we had previously reported that the NDV F protein was modified with palmitate (8), we characterized the DRM association of a mutant altered in the two cysteine residues found within the TM domain and at the TM-CT junction (F-C514S,C523S) (Fig. (Fig.4).4). Figure Figure55 shows that elimination of these potential palmitate addition sites did not affect the association of F protein with DRMs.
We next sought to determine whether mutations in the CT domain (amino acids 523 to 553) of the fully glycosylated F protein have any effect on DRM localization. Mutant proteins used for this analysis are shown in Fig. Fig.4.4. As previously described (35), all mutant proteins except the protein missing the entire CT domain (d523-540)were proteolytically cleaved in the Golgi membranes. All of these mutant proteins were expressed on cell surfaces at wild-type or near-wild-type levels (35).
The DRM association of proteins with overlapping deletions of the CT domain is shown in Fig. Fig.6.6. Interestingly, deletion of all 29 amino acids (d523-553) of the CT domain significantly inhibited DRM association. Furthermore, deletion of the last 13 amino acids of the CT domain (d540-553) also inhibited association with DRMs with densities of l.07 to 1.14. However, deletions of the last seven (d547-553) or four (d550-553) amino acids had minimal effects and fractionated just as the wild-type protein. These results suggest that the CT domain of the F protein is important for classical DRM association.
The contrasting results with F proteins missing the last 13 amino acids and the last 7 amino acids suggested that the sequence between amino acids 540 and 547 may be important for DRM association. Alternatively, the length of the CT domain may be important or some conformational property of the CT domain may be necessary. To explore these alternatives, the DRM associations of F proteins with two internal CT deletions were characterized (Fig. (Fig.7).7). One mutant, F-d540-546, was similar in length to the DRM-associated mutant F-d547-553 but was missing the sequences present in the F-d547-553 mutant protein. This mutant protein was clearly associated with DRMs at near-wild-type levels (Fig. 7A and B). Multiple experiments showed that, on average, 35% ± 7% of the total mutant F protein localized in DRMs. The results obtained with this mutant show that the specific sequence between amino acids 540 and 547 is not critical to classical DRM association. The DRM association of an F protein with another internal deletion, F-d525-531, is shown in Fig. 7C and D. Although this protein retained the length of proteins found in the DRM fraction, as well as in the sequence between amino acids 540 and 553, this mutant protein minimally fractionated with DRMs with densities of 1.05 to 1.14 g/ml. Some mutant protein was found in fractions with densities of 1.15 to 1.20 g/ml. The significance of this material is under investigation.
The different DRM associations of these two mutants cannot be accounted for by differential expression at cell surfaces. Figure Figure88 shows the results of flow cytometry of cells expressing wild-type and F-d525-531 proteins (Fig. (Fig.8A)8A) and wild-type F and F-d540-546 proteins (Fig. (Fig.8B).8B). The numbers of positive cells and the intensity of fluorescence of cells expressing F-d525-531 mutant protein were virtually identical to cells expressing the wild-type protein. The expression level of the F-d540-546 mutant protein was slightly less than that of the wild type.
To explore the relationship between DRM association and cell-cell fusion, two different approaches were taken. First, the ability of cholesterol-depleted cells expressing F protein and HN protein to fuse with avian RBC was compared to that of untreated cells. The membranes of avian RBC were loaded with the fluorescent dye R18. It has been previously shown that these labeled RBC will attach to syncytia expressing the F and HN proteins of NDV at 4°C (Fig. (Fig.9A)9A) (21) and are visualized as individual RBC. Upon incubation at 37°C, the RBC membranes fuse with F- and HN-protein-expressing cells transferring the fluorescent dye to the syncytia (Fig. (Fig.9B)9B) (21). Cells expressing fusion-negative mutants of F protein do not fuse (22). Furthermore, anti-NDV antibody added after RBC binding blocked the transfer of R18 into the syncytia (Fig. (Fig.9E)9E) showing that there was little nonspecific dye transfer to the syncytia. After incubation at 37°C, 78% ± 4% of untreated HN- and F-protein-expressing cells were positive for fusion with the RBC. Figure Figure9C9C shows that MβCD-treated cells bound RBC at levels comparable to those for untreated cells. Incubation of the cells at 37°C resulted in the transfer of R18 from the bound RBC into MβCD-treated cells. A total of 75% ± 8% of HN- and F-protein-expressing cells were positive for fusion with RBC. Thus, cholesterol depletion of glycoprotein-bearing cells significant enough to eliminate classical DRM association of the F protein had no negative effect on the ability of the F protein to direct hemifusion detected in this assay.
In a second approach to exploring the relationship between cell-cell fusion and DRM association, the fusion activities of selected F proteins with mutations in the CT domain were characterized to determine whether there was a correlation between DRM association and fusion activities of these mutants. We have previously reported that the F-d525-531 had significant syncytium-forming activity, whereas F-d540-546 had some activity and F-d540-553 had no activity (35). These results were confirmed (Fig. 10A). In addition, the abilities of F-d525-531 and F-d540-546 mutants to direct pore formation were determined (Fig. 10B) by using a measure of content mixing of fused cells previously reported (20). Both mutants had significant fusion activity in this assay although the activity of the F-d525-531 protein, which was not associated with classical DRMs, was significantly higher than the activity of F-d540-546, which was associated with DRM. These combined results show little correlation between DRM association of the viral F protein and cell-cell fusion activity.
The NDV F protein expressed in the absence of other viral proteins localizes to a DRM fraction of cells, a cell fraction thought to represent lipid raft association in cells. That F protein is associated with DRMs was shown by the flotation of F protein into a sucrose density gradient after cell disruption with Triton X-100 at 4°C into fractions that also contain the ganglioside GM1, a component of lipid rafts. Furthermore, the flotation of F protein into these gradients was inhibited by conditions known to disrupt lipid raft domains and DRMs: cholesterol depletion, solublization of cells with octylglucoside, and solubilization of cells with Triton X-100 at room temperature (36). These results confirm the association of F protein with DRMs with densities of l.07 to 1.14.
Because NDV F protein expressed in the absence of other viral proteins is localized in DRMs, some intrinsic property of the F protein itself is likely to be responsible for this localization. With the goal of determining the functional significance of raft association, we sought to determine what property of the F protein was responsible for DRM localization.
Paramyxovirus F proteins, synthesized as a precursor F0, must be proteolytically cleaved for fusion activity, and this cleavage is reported to result in significant conformational changes in the F protein (12). However, cleavage had no role in DRM association. A point mutation in the F-protein cleavage site that eliminated cleavage of the molecule had little effect on DRM localization of the protein. Furthermore, coexpression of F protein with HN protein, which is also reported to result in conformational differences in the F protein (14, 21), had no effect on F-protein localization in DRMs. Similarly, coexpression of the measles virus attachment protein with measles virus F protein had no effect on the F-protein association with DRMs (18, 40).
Many proteins associated with lipid rafts are modified by covalent addition of the fatty acid palmitate, and the DRM association of many of these proteins is attributed to this modification (23). This saturated fatty acid likely has a preference for the liquid-ordered lipids characteristically found in lipid rafts. Indeed, the DRM association of several viral glycoproteins can be prevented by eliminating this modification. Mutations of palmitate addition sites in the HIV Env protein (29), the murine leukemia virus Env protein (15), influenza virus HA (23), and the Ebola virus glycoprotein(3) all eliminated their DRM associations. The F proteins of many paramyxoviruses, including the F protein of NDV (8), are modified by palmitate. However, mutation of the candidate palmitoylation sites in the F protein had no effect on localization of the fully glycosylated protein in DRMs.
Mutations in the CT domain of the NDV F protein did, however, significantly affect localization with DRMs with densities of l.07 to 1.14 g/ml. Deletion of the entire domain (31 amino acids) (d523-553) or deletion of the most carboxyl-terminal 14 amino acids (d540-553) virtually eliminated classical DRM association, whereas deletion of the last seven amino acids (d547-553) had little effect on DRM association. This result might indicate that amino acids 540 to 547 are crucial. However, deletion of only seven amino acids (d540-546) was not sufficient to disrupt association with classical DRMs. Furthermore, the deletion mutant d525-531, which contains the seven amino acids from 540 to 547, was defective in association with these DRMs. These results argue that neither a specific linear sequence nor a specific length is critical for DRM association. Rather, a conformational determinant of the CT domain may be important for this property of the F protein, a determinant disrupted by the loss of the last 13 amino acids or by the loss of 6 amino acids near the TM domain. Alternatively, mutations in the CT domain of the F protein may have an indirect role in classical DRM localization due to effects on the conformation of other regions in the protein. However, with the exception of d523-553, which is poorly cleaved, no structural defects have been detected in any of the CT mutant proteins. Indeed, all are normally cleaved, and all are precipitated with a conformation-specific monoclonal antibody (35). Thus, these results are most consistent with the notion that the conformation of the CT domain itself is important for classical DRM localization and that mutations that affect this conformation inhibit DRM association. For instance, this domain may interact with host proteins or lipids specific to the intracellular side of lipid rafts.
The percentage of total cell-associated NDV F protein in the DRM fraction of cells prepared as we have described above was quite similar to that reported for the measles virus F protein (18). It is interesting that only a fraction of total measles virus or NDV F proteins are found in DRMs. It is possible that paramyxovirus F proteins are not tightly associated with this cell fraction. Indeed, increased amounts of Sendai F protein were found in classical DRMs after solubilization in lower concentrations of Triton X-100 (1). Alternatively, F protein may be in two different populations or two different domains on cell surfaces. Indeed, cell surface F protein participates in two different pathways, virion assembly and cell-cell fusion (14). We have previously reported that mutations in the CT domain of the NDV F protein can affect cell-cell fusion as measured by syncytium formation (35). To determine the role of lipid raft association in cell-cell fusion, we compared the fusion activities of two of these mutants, d540-546 and d525-531. Fusion activities of the mutant proteins were measured by syncytium formation as well as content mixing. Our results, presented here as well as previously (35), showed no obvious correlation with classical DRM association and fusion activity. F-d525-531 mutant protein is not associated with these DRMs but has approximately half the syncytium-forming activity and nearly 60% the content mixing activity of wild-type protein. In contrast, mutant F-d540-546 is associated with DRMs at nearly wild-type levels but has lower fusion activities than F-d525-531. Thus, the lipid raft association of the F protein does not appear to be directly correlated with cell-cell fusion or syncytium formation. Our finding that cholesterol depletion of cells expressing HN and F proteins fused normally to avian RBC supports this conclusion.
The lipid raft association directed by the F-protein CT domain may, however, be important for virus assembly, as has been suggested in numerous virus systems (reviewed in reference 4). In several systems, it has been reported that mutations of F-protein CT domains affected virus assembly (7, 10, 32, 42). In other studies, F proteins are reported to associate with lipid raft domains (1, 18) as we report here. Our results raise the possibility that the paramyxovirus F-protein CT domain may be involved in virus assembly at least in part because it directs the protein to lipid raft domains, a possibility currently under investigation.
In summary, we have found that the NDV F protein is localized in classical DRMs, i.e., DRMs with a density of 1.07 to 1.14. Furthermore, the CT domain of the F protein is important for association with these DRMs. The association of F protein with heavy-density DRMs, as previously defined in neutrophils (25), is currently under investigation.
This study was supported by grants AI30572 (T.G.M.) and GM33048 (E.J.L.) from the National Institutes of Health.