Pseudotypes of Vesicular Stomatitis Virus with CD4
http://www.100md.com
病菌学杂志 2005年第11期
Department of Microbiology and Immunology
Department of Biochemistry, Wake Forest University School of Medicine, Winston-Salem, North Carolina 27157
ABSTRACT
Many plasma membrane components are organized into detergent-resistant membrane microdomains referred to as lipid rafts. However, there is much less information about the organization of membrane components into microdomains outside of lipid rafts. Furthermore, there are few approaches to determine whether different membrane components are colocalized in microdomains as small as lipid rafts. We have previously described a new method of determining the extent of organization of proteins into membrane microdomains by analyzing the distribution of pairwise distances between immunogold particles in immunoelectron micrographs. We used this method to analyze the microdomains involved in the incorporation of the T-cell antigen CD4 into the envelope of vesicular stomatitis virus (VSV). In cells infected with a recombinant virus that expresses CD4 from the viral genome, both CD4 and the VSV envelope glycoprotein (G protein) were found in detergent-soluble (nonraft) membrane fractions. However, analysis of the distribution of CD4 and G protein in plasma membranes by immunoelectron microscopy showed that both were organized into membrane microdomains of similar sizes, approximately 100 to 150 nm. In regions of plasma membrane outside of virus budding sites, CD4 and G protein were present in separate membrane microdomains, as shown by double-label immunoelectron microscopy data. However, virus budding occurred from membrane microdomains that contained both G protein and CD4, and extended to approximately 300 nm, indicating that VSV pseudotype formation with CD4 occurs by clustering of G protein- and CD4-containing microdomains.
INTRODUCTION
For most viruses, envelope assembly involves association of internal virion components with specialized regions of the host plasma membrane containing locally high concentrations of viral transmembrane envelope glycoproteins. These membrane regions then bud from the host membrane and pinch off to form the virus envelope. As a general rule, host proteins are excluded from budding virus envelopes while viral glycoproteins are preferentially included. However, some heterologous membrane proteins are incorporated into virus envelopes at very high levels. This phenomenon is referred to as pseudotype formation or phenotypic mixing (31). There has been intense interest in pseudotype formation because of its importance for understanding the mechanism of virus envelope assembly. In addition, pseudotype formation is widely used in biotechnology applications to alter the host range of viruses and for vaccine development (16, 28).
Pseudotype formation was originally defined by a mixed infection of two enveloped viruses, in which progeny virions contained glycoproteins of both viruses (31). More recently, pseudotype formation has also been observed in recombinant viruses that express foreign glycoproteins directly from the viral genome (12, 24, 26). The exclusion of most host proteins from virus envelopes makes the process of envelope assembly appear to be specific for homologous viral glycoproteins. However, the phenomenon of pseudotype formation suggests that the process of envelope assembly is nonspecific with regard to the incorporation of viral glycoproteins. This apparent contradiction is known as the pseudotype paradox (31). The goal of the experiments presented here was to determine the mechanism behind this apparent paradox for the prototype rhabdovirus, vesicular stomatitis virus (VSV), which has been widely used in studies of viral envelope assembly and pseudotype formation. The VSV core consists of a helical nucleocapsid containing the genome RNA, which is condensed into a tightly coiled, bullet-like shape by the viral matrix (M) protein. The envelope contains primarily a single species of viral transmembrane glycoprotein (G protein) but can readily form pseudotypes with other membrane glycoproteins.
A key insight into the mechanism of virus envelope assembly was provided by the concept of membrane microdomains. A membrane microdomain can be defined as a locally high concentration of a membrane component compared to the average density in the membrane as a whole. By this definition, budding virus envelopes can be viewed as excellent examples of assembly of functional membrane microdomains. The most extensively characterized membrane microdomains are those that are enriched in sphingolipids and cholesterol, referred to as lipid rafts. These membrane microdomains have been characterized largely by their relative resistance to solubilization with Triton X-100 at low temperatures (2) and by copatching with glycolipids such as GM1 (8). The evidence that several viruses bud from lipid rafts that are enriched in their viral glycoproteins has reinforced the idea that virus envelope assembly is in fact a case of assembly of functional membrane microdomains (1, 15, 17, 20, 23, 32).
The concept that viruses can bud from lipid rafts has also led to the idea that pseudotype formation occurs by virus budding from lipid rafts that contain a mixture of glycoproteins (20). This could also account for the exclusion of host proteins that are not present in lipid rafts containing viral proteins. However, it has been difficult to test this hypothesis, since there are few approaches to determine whether different membrane components are colocalized in microdomains as small as lipid rafts, which are estimated to be on the order of 100 nm or less (29). Indeed, even the existence of lipid rafts under conditions other than detergent extraction or copatching has been difficult to detect (9, 10, 25, 29).
Although some viruses bud from lipid raft microdomains, there are likely to be many viruses that bud from nonraft areas of plasma membranes. VSV was the original example of a virus that buds from nonraft areas of membranes (22). Thus, the techniques used to analyze lipid rafts cannot be used to determine whether the VSV G protein is organized into membrane microdomains or whether such microdomains serve as sites of virus assembly. To address this question, we recently developed a new analytical method for immunoelectron microscopy data to determine whether membrane proteins are organized into microdomains and to determine the sizes of these microdomains (3). This method is based on analyzing the distribution of pairwise distances between immunogold particles averaged over a large number of electron micrographs. This method can be applied to any immunolabeled membrane protein regardless of whether or not it is raft associated. Using this approach, we showed that the VSV G protein is primarily organized into membrane microdomains with a diameter of 100 to 150 nm, both when expressed in transfected cells in the absence of other viral components and in the plasma membranes of virus-infected cells (3, 4).
One of the strengths of our new method is that it can also be used to determine colocalization of different membrane components in membrane microdomains in double-labeling experiments. In the experiments reported here, we analyzed the membrane microdomains involved in pseudotype formation. We used a recombinant VSV that expresses the T-cell antigen CD4 from the viral genome (VSV-CD4) (24). Virions produced from VSV-CD4-infected cells incorporate CD4 into their envelopes nearly as efficiently as G protein. The data presented here show that in VSV-CD4-infected cells, G protein- and CD4-containing microdomains have very similar properties. Both were approximately 100 to 150 nm in diameter, and both were primarily detergent soluble. In areas of the plasma membrane outside of virus budding sites, G protein and CD4 were present in separate microdomains, as shown by double-labeling experiments. However, virus budding occurred from membrane microdomains that contained both G protein and CD4, and extended to approximately 300 nm. The result that G protein and CD4 are in separate microdomains outside of budding sites but together within budding sites leads to the conclusion that VSV pseudotype formation with CD4 occurs by clustering of G protein- and CD4-containing microdomains.
MATERIALS AND METHODS
Cells and viruses. The recombinant virus VSV-CD4 was provided by John K. Rose of the Yale University School of Medicine (24). VSV-CD4 was grown at 37°C in BHK cells following two rounds of plaque purification as described previously (14).
Immunogold-labeling electron microscopy. BHK cells were infected with VSV-CD4 at a multiplicity of infection of 20 PFU/cell for 4.5 or 8 h as described elsewhere (3). Cells were fixed with buffered formalin as described elsewhere and labeled with a mouse anti-VSV G protein monoclonal antibody, I1 (13), or an anti-CD4 monoclonal antibody, MHCD0400 (Caltag Laboratories), diluted 1:100 for 1 h. Following three washes with phosphate-buffered saline (PBS) containing 10% bovine serum albumin (BSA), cells were labeled with a secondary goat anti-mouse immunoglobulin G antibody conjugated to 5-nm gold beads (Jackson ImmunoResearch Laboratories) for 1 h. For double labeling, formalin-fixed cells were permeabilized with 0.05% saponin for 10 min, then washed three times with PBS containing 0.1 M glycine, and incubated in PBS containing 0.1 M glycine and 10% BSA at 4°C overnight. Samples were incubated with a rabbit anti-VSV-G Tag antibody (Research Diagnostics Incorporated) and the mouse anti-CD4 monoclonal antibody. Following three washes with PBS containing 10% BSA, cells were labeled with a secondary goat anti-mouse antibody conjugated to 5-nm gold particles to indirectly label CD4 and a secondary goat anti-rabbit antibody conjugated to 15-nm gold particles to indirectly label G protein (Jackson ImmunoResearch Laboratories). Immunolabeled cells were postfixed in 1% osmium tetroxide for 15 min, washed three times in PBS, scraped from the dish, and collected by centrifugation. Cells were embedded in Spurr resin using the standard embedding method. Thin sections (80 nm) of the embedded material were viewed using a Philips EM400 electron microscope operating at 80 keV.
A series of 25 electron micrographs of arbitrarily chosen areas of plasma membrane were collected from two separate experiments (total of 50 micrographs). Each micrograph was obtained from a separate cell. The magnification of micrographs was x43,000. Negatives were digitized with a scanner, and the micrographs were analyzed with Image Pro 3.0 software (Media Cybernetics). A pairwise measurement was made of the distance traced along the plasma membrane between each gold particle and every other gold particle to its right in the image, excluding gold particles within a virus budding site. These measurements generated a data set of distances between all of the gold particles in all 50 images. This data set was plotted as a histogram of the average number of gold particles per 20 nm on the y axis versus the distance of separation on the x axis in 20-nm increments as described previously (3). For double-label experiments, a series of 25 electron micrographs of arbitrarily chosen 15-nm gold particles that were not in budding virus envelopes was collected from two separate experiments (total of 50 micrographs). Each micrograph was obtained from a separate cell. The distance to each 5-nm gold particle within a distance of 1 μm on each side of the reference 15-nm gold particle was measured along the trace of the plasma membrane. The data were plotted as a histogram of the density of anti-CD4 gold particles on the y axis versus the distance from anti-G protein reference particles in 20-nm increments on the x axis. The y values were converted to average densities by dividing the number of gold particles at each distance by the number of 15-nm gold particles that contained data at that distance (3).
Analysis of virus budding sites. A series of 25 electron micrographs of arbitrarily chosen VSV budding sites was collected from each of two separate experiments (total of 50 micrographs), and virus budding sites were analyzed as described previously (3). Since several micrographs contained more than one virus budding site, the actual number of budding sites analyzed in the 50 micrographs was 61 (anti-CD4 at 4.5 h postinfection [hpi]), 74 (anti-CD4 at 8 hpi), 62 (anti-G protein at 4.5 hpi), or 73 (anti-G protein at 8 hpi). Briefly, virus budding sites were analyzed by measuring the distance of each gold particle from the center of the leading edge of the virus budding site along the trace of the plasma membrane, within a distance of 1 μm of the budding site. For budding sites separated by distances <2 μm, the distance between them was divided at the midpoint. Data were presented as a histogram of the average density of gold particles on the y axis versus the distance from the center of the budding site in 20-nm increments on the x axis. The length of the virus envelope was determined by tracing the membrane from the tip of the budding virion to the end of the electron density of the densely stained nucleocapsid-M protein (NCM) complex. The left and right sides of the virus envelope were measured separately.
Western blot analysis of CD4 incorporation into VSV-CD4 virions. BHK cells infected with VSV-CD4 for 4 or 7.5 h were washed three times with PBS and then incubated in medium for 1 h at 37°C. Culture supernatants were collected, centrifuged at 2,000 rpm for 5 min at 4°C to remove cellular debris, and then centrifuged over a 15% sucrose cushion at 50,000 rpm for 40 min in a Beckman TLS-55 rotor. Virus pellets were resuspended in a total volume of 100 μl of radioimmunoprecipitation assay (RIPA) buffer (0.15 M NaCl, 1% deoxycholate, 1% Triton X-100, 0.1% sodium dodecyl sulfate [SDS], 10 mM Tris, pH 7.4). Equal amounts of purified virions were analyzed by Western blotting with anti-CD4 antibody H370 (Santa Cruz Biotechnology, Inc.) and anti-VSV-G Tag antibody (Research Diagnostics Incorporated) as described elsewhere (5). The ratio of CD4 to G protein was determined using Quantity One software (Bio-Rad).
SDS-PAGE analysis of G protein incorporation into VSV-CD4 virions. BHK cells infected with VSV-CD4 for 3.5 or 7 h were incubated in methionine-deficient medium containing 2% fetal bovine serum for 10 min at 37°C, then pulse labeled with [35S]methionine for 10 min, and incubated in nonradiolabeled medium for 1 h at 37°C. Virions released during the chase were harvested as described above and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) on 10% polyacrylamide gels, followed by phosphorescence imaging using Quantity One software.
Analysis of Triton X-100-resistant G protein and CD4 microdomains. BHK cells infected with VSV-CD4 for 4.5 or 8 h were washed with PBS and then extracted with 1% Triton X-100 in 25 mM Tris, 150 mM NaCl, 5 mM EDTA (pH 7.4) for 30 min at 4°C. Cell lysates were sedimented at approximately 100,000 x g in a Beckman Airfuge for 20 min at 4°C to separate soluble and insoluble material. Pellets were resuspended in RIPA buffer to contain the same volume as the supernatant fractions. Supernatant and pellet fractions were analyzed by Western blot analysis with anti-G protein and anti-CD4 antibodies. The volume of supernatant analyzed was one-fourth the volume of the pellet. The relative percentages of G protein and CD4 in the pellet fractions were determined using Quantity One software.
For flotation-based analysis of detergent-resistant G protein and CD4, VSV-CD4-infected cells were washed with cold PBS and resuspended in 1% Triton X-100 for 30 min. Lysates were subjected to 15 strokes of a Dounce homogenizer. Nuclei were removed by centrifugation at 2,000 rpm at 4°C. A 500-μl volume of supernatant was mixed with 1.5 ml of 60% (wt/vol) sucrose containing 1% Triton X-100 and then overlaid with 2 ml of 30% sucrose and 1 ml of 5% sucrose. Gradients were centrifuged in a Beckman SW50.1 rotor at 35,000 rpm at 4°C for 16 h. Fractions (0.6 ml) were collected from the top of the gradient and were analyzed by Western blotting as described above.
RESULTS AND DISCUSSION
G protein- and CD4-containing microdomains in BHK cells are primarily detergent soluble. The observation that VSV efficiently forms pseudotypes with CD4 (24) raises the question of whether G protein-containing and CD4-containing microdomains have similar properties. Little if any VSV G protein is found in detergent-insoluble (lipid-raft) membrane fractions (22, 30, 32). However, the reported extent of CD4 association with detergent-insoluble lipid rafts in T cells ranges from nearly 100% to less than 20%, depending on the source of the T cells (7, 11, 18, 19, 21). We determined the extent of association of G protein and CD4 with detergent-insoluble membrane fractions in BHK cells infected with VSV-CD4 at two different time points, 4.5 hpi, which is relatively early in the period of virus assembly, and 8 hpi, which is a late time when viral proteins have accumulated to near-maximum levels. Cells infected with VSV-CD4 were treated with buffer containing 1% Triton X-100 at 4°C, and cell lysates were sedimented to pellet detergent-resistant material. The resulting supernatant and pellet fractions were analyzed by Western blotting for G protein and CD4 (Fig. 1A), and the percentages of G protein and CD4 in the pellets were quantitated from six independent experiments. At both 4.5 and 8 hpi, there was little if any G protein detected in the pellet (<2%), consistent with earlier data. Likewise, there was only a small amount of CD4 detectable in the pellet fractions (12 to 14%). This result is similar to that obtained for T cells that lack the CD4-associated protein kinase Lck, supporting the idea that in the absence of Lck, only small amounts of CD4 are associated with detergent-resistant membrane fractions (7, 19). These results were confirmed by performing a flotation-based analysis of detergent-resistant G protein and CD4. Detergent-treated lysates were mixed with dense sucrose, overlaid with a discontinuous sucrose gradient, and centrifuged to float the detergent-resistant membranes. Fractions were analyzed by Western blotting (Fig. 1C). Little if any G protein and only small amounts of CD4 were detected in the raft fractions (lanes 1 to 4). These results indicate that G protein and CD4 are similar, though not identical, in their detergent solubility and that neither G protein nor CD4 partitions into detergent-insoluble microdomains to a great extent.
G protein- and CD4-containing microdomains are similar in size. Since most CD4 did not partition into lipid rafts, this raised the question of whether CD4 was organized into membrane microdomains in BHK cells or whether it was distributed randomly in the plasma membrane. To distinguish between these two possibilities, we used our recently described method for the analysis of immunoelectron microscopy data to detect the organization of proteins into membrane microdomains (3). BHK cells were infected with VSV-CD4, and at either 4.5 or 8 hpi, cells were fixed and immunolabeled with an antibody against CD4 and a secondary antibody conjugated to 5-nm gold particles. The amount of gold-conjugated secondary antibody was adjusted to low levels so that no labeling was detectable in the negative controls (3). This also prevents interference between gold particles in heavily labeled areas of samples with positive immunoreactivity. Sets of 25 electron micrographs of arbitrarily chosen areas of plasma membrane were collected in each of two independent experiments (50 micrographs total at each time point), with each image arising from a different cell. A representative image is shown in Fig. 2A. A pairwise measurement was made of the distance traced along the plasma membrane between each gold particle and every other gold particle to its right in the image, excluding gold particles within a virus budding site. These measurements generated a data set of distances between all of the gold particles in all 50 images. This data set was plotted as a histogram of the average number of gold particles per 20 nm on the y axis versus the distance of separation on the x axis in 20-nm increments (3).
Figure 2B shows the analysis of pairwise distance measurements of immunolabeled CD4 in the plasma membranes of VSV-CD4-infected cells at 4.5 and 8 hpi. The average density of CD4 labeling in the plasma membrane at 4.5 hpi, determined from the 50 micrographs (0.11 particle/20 nm), is shown. For distances of separation between 100 nm and 1,000 nm, the histogram shows densities that are close to this average, as expected. However, for distances less than 100 nm, the histogram shows densities that are higher than the average, which indicates that CD4 is organized into microdomains in this size range. If CD4 is organized into circular microdomains with only one size, this analysis yields a straight line with a negative slope that intersects the average density at a distance that is 64% (2/) of the diameter of the microdomain (3). The data obtained at 4.5 hpi are consistent with only a single dramatic change in the slope at approximately 100 nm. This indicates that CD4 is predominantly organized into microdomains with a diameter of 100 to 150 nm.
At 8 hpi, the average density of CD4 (0.21 particle/20 nm) was approximately twofold higher than the average density at 4.5 hpi. If the sizes of all of the CD4-containing microdomains at 8 hpi had remained at 100 to 150 nm, the graph would intersect the average density at the same point as at 4.5 hpi, but the negative slope and the y intercept would be twofold higher (3). For example, the y intercept in Fig. 2B should increase from approximately 0.4 at 4.5 hpi to approximately 0.8 at 8 hpi. However, the densities near the y intercept at 8 hpi increased to only approximately 0.5 particle/20 nm. Furthermore, densities higher than the average extended to at least 400 nm and were not consistent with a single change in slope. These data indicate that much of the CD4 was organized into microdomains in the range of 100 to 150 nm but that some of the CD4 was populating microdomains with a wider variety of sizes. From these data we conclude that CD4 does partition into membrane microdomains that are predominantly 100 to 150 nm but that, when expressed at high levels, CD4 is also distributed among microdomains of larger sizes.
The CD4 microdomains were similar in size to the G protein-containing microdomains we described earlier in wild-type VSV-infected cells (3, 4). To determine if G protein was organized to the same extent in VSV-CD4-infected cells as in wild-type virus-infected cells, the pairwise distance analysis of Fig. 2B was applied to G protein in VSV-CD4-infected cells in Fig. 2C. In contrast to the findings of our earlier study of wild-type VSV (3), the average density of G protein in the plasma membranes of cells infected with VSV-CD4 did not increase between 4.5 and 8 hpi. The average density of G protein at both time points was 0.08 particle/20 nm. We attribute this difference from our results with wild-type VSV to the faster depletion of G protein from the plasma membrane by budding of VSV-CD4 virions compared to wild-type VSV. An earlier electron microscopy study showed that VSV-CD4 virions are longer than wild-type virions due to their extra genetic information (24), and therefore budding of VSV-CD4 would be expected to deplete G protein from the plasma membrane at a higher rate. The important result from Fig. 2C is that the distribution of G protein in the plasma membrane at 4.5 and 8 hpi was characterized by a high density of G protein at short distances of separation, with a single dramatic change in slope at approximately 100 nm. These data indicate that the G protein-containing microdomains are 100 to 150 nm, similar in size to the CD4-containing microdomains. The microdomains containing G protein and CD4 are slightly smaller than those containing the lipid-raft-associated influenza virus hemagglutinin, which were determined to be 200 to 280 nm using the same technique (27). This reinforces the idea that G protein- and CD4-containing microdomains are similar in principle to lipid rafts but differ in detail from lipid rafts in their detergent solubility and overall size.
G protein and CD4 are in separate microdomains outside of budding sites. The similarity in size and detergent solubility of G protein- and CD4-containing microdomains suggested that these proteins might exist as a mixture within a single type of microdomain. Alternatively, they may exist in separate microdomains with similar properties. We distinguished between these two hypotheses by analyzing the distribution of CD4 relative to G protein in VSV-CD4-infected cells by double-label immunogold electron microscopy. If CD4 and G protein formed mixed microdomains, this would be apparent as a higher-than-average density for small distances of separation between anti-G protein gold particles and anti-CD4 gold particles, a result similar to those for CD4 alone and G protein alone (Fig. 2B and C). Alternatively, if G protein and CD4 exist in separate microdomains, the density of CD4 would be similar to the average density at all distances of separation from G protein.
BHK cells were infected with VSV-CD4 for 4.5 or 8 h. Cells were then immunolabeled with antibodies against G protein and CD4. A secondary antibody conjugated to 15-nm gold particles was used to indirectly label G protein, while an antibody conjugated to 5-nm gold particles was used to indirectly label CD4. A representative image depicting the labeling of G protein and CD4 at 4.5 hpi is shown in Fig. 2D. This image contains two 15-nm gold particles that label G protein. Most 5-nm gold particles that label CD4 are clearly separated from the 15-nm gold particles that label G protein, although one 5-nm gold particle is in close proximity to a 15-nm gold particle. To quantitate the density of CD4 in relation to G protein, a series of 25 electron micrographs of arbitrarily chosen 15-nm anti-G protein gold particles (excluding gold particles associated with virus budding sites) was collected at 4.5 and 8 hpi in two separate experiments (50 micrographs total at each time point). Images were analyzed by measuring the distance along the trace of the membrane of each 5-nm gold particle from the 15-nm reference gold particle to a distance of 1 μm on each side.
The results of the quantitation are represented in Fig. 2E as a histogram of the density of CD4 labeling versus the distance from a 15-nm reference gold particle in 20-nm increments for 4.5 and 8 hpi. There was no preferential labeling of CD4 at short distances of separation from anti-G protein gold particles, which would have been expected if CD4 were present in G protein-enriched microdomains. Instead, the density of CD4 labeling was similar to the average density for all distances of separation from anti-G protein gold particles. This is the expected result if CD4 does not partition into G protein-enriched microdomains. Taken together, the results in Fig. 2 indicate that both CD4 and G protein exist in discrete microdomains outside of virus budding sites, as shown in Fig. 2B and C. However, these microdomains are separate and nonoverlapping, as shown in Fig. 2E.
G protein- and CD4-containing microdomains cluster at VSV budding sites. A previous immunoelectron microscopy study showed that nearly all VSV-CD4 virions contain both G protein and CD4, suggesting that both proteins are present in the membrane microdomains that form virus budding sites (24). In Fig. 3, the extent to which CD4 and G protein were concentrated in membrane microdomains at sites of virus budding was determined by comparing the densities of immunogold labeling in budding virus envelopes to the average density in the plasma membrane as whole. Representative images of budding virions labeled with antibody to CD4 are shown in Fig. 3A. The densely stained core of the budding virus consists of the tightly coiled NCM complex. The envelopes of budding virions were labeled with gold particles, while there was little labeling of the surrounding plasma membrane. To quantitate the levels of labeling, infected BHK cells immunolabeled for CD4 or G protein in Fig. 2 were used to generate a separate series of electron micrographs of arbitrarily chosen VSV-CD4 budding sites (50 micrographs total for each sample). The criteria for identifying virus budding sites were (i) electron density created by the NCM complex, (ii) extension of the budding particle into the extracellular space, and (iii) visible attachment to a cell (3). The density of labeling was determined by counting the number of gold particles associated with the virus envelope and dividing by the distance along the trace of the envelope from the base of the densely stained NCM complex on one side to the base on the other. The extent of labeling in each budding virus envelope was expressed as the ratio of the density in the envelope to the average density in the plasma membrane as a whole (determined from the micrographs in Fig. 2). The data are shown as histograms in Fig. 3B (CD4) and C (G protein), in which the x axis is the density of labeling in the budding virus envelope and the y axis is the number of budding sites with each density. Statistical analysis of these data is given in Table 1.
The densities of CD4 labeling in budding virus envelopes ranged from near the density in the membrane as a whole to approximately ninefold higher (Fig. 3B), with an average density that was approximately threefold higher than in the membrane as a whole (Table 1). At 4.5 hpi, 64% of budding virions (39 out of 61) contained CD4 at levels that were at least twofold higher than in the membrane as a whole, and at 8 hpi, 76% of budding virions had twofold or higher levels of CD4 labeling (Table 1). Thus, the majority of virions bud from CD4-enriched areas of the plasma membrane.
The densities of labeling of G protein in budding virions, relative to the membrane as a whole (Fig. 3C), were higher than those for CD4 (notice the difference in the x axis scale between Fig. 3B and 3C). The average density of G protein labeling in budding virus envelopes was approximately ninefold above that in the membrane as a whole at 4.5 hpi and approximately sixfold higher at 8 hpi (Table 1). As expected, >90% of budding virions contained a high level of G protein labeling (Table 1).
These data show that the majority of budding sites were enriched in both G protein and CD4. This result, combined with the finding that G protein and CD4 are in separate microdomains outside of virus budding sites (Fig. 2E), leads to the conclusion that these two types of microdomains must come together to form the virus budding sites. In other words, G protein- and CD4-containing microdomains cluster at VSV-CD4 budding sites.
Sizes of G protein- and CD4-containing microdomains at sites of virus budding. Further evidence for clustering of G protein- and CD4-containing microdomains at sites of virus budding was obtained by analysis of the sizes of the microdomains at virus budding sites. The methods for analysis of microdomains at budding sites have been described previously (3) and are similar to the analysis outside of budding sites, except that the tip of the budding virion is used as a reference point. Briefly, micrographs were analyzed by measuring the distance along the trace of the membrane of each gold particle from the center of the leading edge of the budding virion to a distance of 1 μm on each side. The results of the quantitation for CD4 are represented in Fig. 3D as a histogram of the density, expressed in particles/20 nm, versus the distance from the tip of the virus envelope in 20-nm increments. The vertical lines represent the average length of the virus envelope in all of the budding sites analyzed, as defined by the distance traced from the tip of the budding virion to the end of the densely stained NCM complex. Therefore, data points to the left of the vertical lines arise from gold particles that appear to be associated with the virus envelope. The average density of gold labeling in the plasma membrane as a whole is also shown.
As shown in Fig. 3D, the density of CD4 labeling within the virus envelope was twofold higher at 8 hpi (approximately 0.6 particle/20 nm) than at 4.5 hpi (approximately 0.3 particle/20 nm), and at both time points the labeling density in virus envelopes was higher than the average in the membrane as a whole, as shown in the previous analysis (Fig. 3B; Table 1). However, at both 4.5 and 8 hpi, the CD4-containing membrane microdomains extended beyond the base of the budding virus envelope to a distance of approximately 300 to 400 nm. Thus, the CD4-containing microdomains involved in formation of virus envelopes were considerably larger than the 100- to 150-nm microdomains observed outside of virus budding sites (Fig. 2B). This result indicates that the CD4-containing microdomains at virus budding sites must have been assembled from multiple smaller microdomains. At 8 hpi, the CD4-containing microdomains at virus budding sites would be consistent with the largest microdomains observed outside of virus budding sites (Fig. 2B), leaving open the possibility that viruses bud selectively from these larger microdomains. However, it is more likely that virus budding sites are assembled from multiple smaller microdomains at both 4.5 and 8 hpi.
A similar analysis of G protein labeling in budding virus envelopes is shown in Fig. 3E. As shown previously (3, 4), the highest densities of G protein labeling were observed near the tip of the budding virions (near the y axis), and the densities declined toward the base. At both 4.5 and 8 hpi, densities of G protein labeling higher than the average extended to a distance of approximately 300 nm (Fig. 3E). Thus, the G protein-containing microdomains at virus budding sites extend to approximately 300 nm, a finding similar to the results with CD4 (Fig. 3D). As in the case of CD4, this size is considerably larger than the 100- to 150-nm G protein-containing microdomains observed outside of virus budding sites (Fig. 2C). The larger size of G protein-containing microdomains at virus budding sites (Fig. 3C) than outside of budding sites (Fig. 2C) was observed at both 4.5 and 8 hpi. These data further support the conclusion that the membrane microdomains involved in virus assembly are assembled from multiple smaller membrane microdomains and include both CD4-containing and G protein-containing membrane microdomains.
The density of G protein in the VSV-CD4 envelope decreases slightly, while the density of CD4 increases, at late times postinfection. The data in Fig. 3E and Table 1 indicate that the density of G protein labeling in the VSV-CD4 envelope was slightly lower at 8 hpi than at 4.5 hpi. Despite the broad distribution of densities (Fig. 3C), the difference in the average density at 8 versus 4.5 hpi was statistically significant (Table 1). This result was not expected, since the density of G protein labeling in the membrane as a whole did not decline from 4.5 to 8 hpi. We confirmed the analytical electron microscopy result that the density of G protein in the VSV-CD4 envelope decreased slightly at 8 hpi by SDS-PAGE analysis of G protein incorporation into virions. Cells were infected with VSV-CD4 for either 4 or 7.5 h, at which times the cells were pulse labeled with [35S]methionine for 10 min. Following a 1-h incubation at 37°C to allow for transport to the cell surface and incorporation into virions, virus particles were isolated from the supernatant by centrifugation and were analyzed by SDS-PAGE and phosphorescence imaging (Fig. 4A). Data from three independent experiments were quantitated as the ratio of G protein to M protein (Fig. 4B). The VSV proteins L, G, N/P, and M were heavily labeled, but CD4 was not readily apparent (Fig. 4A), since it is labeled inefficiently with [35S]methionine (24). The amounts of internal viral proteins (N/P and M) released into virions were similar at the two time points, but the amount of G protein incorporated into virions decreased from 4.5 to 8 hpi, from a G/M ratio of 0.63 to 0.42. This decrease in the G/M ratio was consistent with the results of the electron microscopy analysis (Fig. 3).
In contrast to G protein, the amount of CD4 in the VSV-CD4 envelope increased from 4.5 to 8 hpi (Fig. 3 and Table 1). This result was confirmed by Western blot analysis. Cells were infected with VSV-CD4, virions released for a 1-h period surrounding the 4.5- and 8-h time points (4 to 5 hpi and 7.5 to 8.5 hpi, respectively) were isolated, and equal amounts of purified virions were analyzed by Western blotting for CD4 and G protein (Fig. 4C). The ratio of CD4 to G protein in virions was quantitated from three independent experiments and normalized to the data for 8 hpi (Fig. 4D). The amount of CD4 incorporated into virions at 4.5 hpi was 37% of the level incorporated at 8 hpi. These data confirm the data obtained by electron microscopy showing that the amount of CD4 incorporated into virions increases approximately twofold at late times postinfection.
A new model of VSV pseudotype formation. Our basic model to explain the formation of VSV pseudotypes that contain CD4 is that G protein and CD4 partition into distinct microdomains that are similar in size and detergent solubility. This point is supported by the data in Fig. 1 and 2. These separate microdomains then cluster to form the virus envelope, leading to the incorporation of both G protein and CD4. This point is supported by the data in Fig. 3. This model raises the question of what drives the clustering of these microdomains. We propose that the clustering is driven by the formation of the NCM complex. This proposal is illustrated in Fig. 5.
Previous results from our laboratory have suggested that assembly of the NCM complex occurs in at least two steps, which we have called initiation and recruitment (6). The initiation event involves the binding of one or a few molecules of M protein to the nucleocapsid, while in the recruitment phase, the majority of M protein is recruited from both the membrane and cytosol, and condenses the nucleocapsid into a tightly coiled NCM complex. We propose that initiation almost always occurs at a G protein-enriched membrane microdomain (Fig. 5, step 1), which accounts for the higher densities of G protein at the tip of a budding virus envelope relative to the base. We also propose that recruitment involves M protein that is associated with a variety of different microdomains (Fig. 5, step 2), which are all similar in lipid composition to the G protein-containing microdomains. This step of virus assembly would, therefore, account for the clustering of CD4-containing microdomains with G protein-containing microdomains in the virus envelope. It is likely that the efficiency of incorporation of membrane proteins is dependent on the extent to which their microdomains resemble G protein-enriched microdomains and therefore associate with M protein. This would account for the fact that VSV forms pseudotypes with a wide variety of membrane proteins, but the efficiency of their incorporation into the VSV envelope varies over a considerable range (12, 24, 26). For example, VSV forms pseudotypes with the influenza virus hemagglutinin, which is present in lipid rafts, but the efficiency is less than that of CD4 (12).
Our model would also account for the lower level of G protein incorporation and the higher level of CD4 incorporation into the envelopes of VSV-CD4 virions at later times postinfection. At early times postinfection, most of the membrane microdomains that cluster to form virus envelopes would be G protein-containing microdomains, with smaller amounts of CD4-containing microdomains. At later times postinfection, the number of CD4-containing microdomains in the plasma membrane increases due to the higher level of viral gene expression (Fig. 2B). In contrast, the number of G protein-containing microdomains remains approximately the same (Fig. 2C), since depletion of G protein by virus budding balances the increase in viral gene expression. The initiation step is still likely to occur in a G protein-containing microdomain, but the recruitment (clustering) step would be more likely to include CD4-containing microdomains than G protein-containing microdomains, since the G protein microdomains have been depleted faster than the CD4 microdomains, leading to a lower level of G protein in the envelope. This effect could account for the observation that expression of high levels of membrane proteins that form pseudotypes with VSV leads to a reduction in viral infectivity (24), since virions containing lower levels of G protein are likely to be less infectious. A similarly high level of incorporation of alternative membrane microdomains into virus budding sites might occur in cell types that express G protein inefficiently on their cell surfaces.
In summary, our data have shown that G protein and CD4 exist in distinct membrane microdomains outside of virus budding sites but that both are present in budding virus envelopes. These results lead to the conclusion that VSV pseudotype formation with CD4 is due to clustering of membrane microdomains. The ability to come to this conclusion was critically dependent on our new methods of analysis of immunoelectron microscopy data. These results have led to a new model of viral pseudotype formation, which will be further tested in future experiments.
ACKNOWLEDGMENTS
We acknowledge the Microscopy Core Laboratory, especially Ken Grant and Paula Moore, for assistance with the preparation of the immunogold-labeling electron microscopy samples. We also thank Jack Rose for providing the VSV-CD4 and Jay Jerome, Griffith Parks, David Ornelles, and Mark Willingham for critical advice.
This research was supported by Public Health Service grant AI15892 from the National Institute of Allergy and Infectious Diseases. E.L.B. was supported by National Institutes of Health training grant T32-AI07401. The Microscopy Core Laboratory was supported in part by the core grant for the Comprehensive Cancer Center of Wake Forest University (CA12197) from the National Cancer Institute.
REFERENCES
Ali, A., R. T. Avalos, E. Ponimaskin, and D. P. Nayak. 2000. Influenza virus assembly: effect of influenza virus glycoproteins on the membrane association of M1 protein. J. Virol. 74:8709-8719.
Brown, D. A., and J. K. Rose. 1992. Sorting of GPI-anchored proteins to glycolipid-enriched membrane subdomains during transport to the apical cell surface. Cell 68:533-544.
Brown, E. L., and D. S. Lyles. 2003. A novel method for analysis of membrane microdomains: vesicular stomatitis virus glycoprotein microdomains change in size during infection, and those outside of budding sites resemble sites of virus budding. Virology 310:343-358.
Brown, E. L., and D. S. Lyles. 2003. Organization of the vesicular stomatitis virus glycoprotein into membrane microdomains occurs independently of intracellular viral components. J. Virol. 77:3985-3992.
Connor, J. H., and D. S. Lyles. 2002. Vesicular stomatitis virus infection alters the eIF4F translation initiation complex and causes dephosphorylation of the eIF4E binding protein 4E-BP1. J. Virol. 76:10177-10187.
Flood, E. A., and D. S. Lyles. 1999. Assembly of nucleocapsids with cytosolic and membrane-derived matrix proteins of vesicular stomatitis virus. Virology 261:295-308.
Fragoso, R., D. Ren, X. Zhang, M. W. Su, S. J. Burakoff, and Y. J. Jin. 2003. Lipid raft distribution of CD4 depends on its palmitoylation and association with Lck, and evidence for CD4-induced lipid raft aggregation as an additional mechanism to enhance CD3 signaling. J. Immunol. 170:913-921.
Harder, T., P. Scheiffele, P. Verkade, and K. Simons. 1998. Lipid domain structure of the plasma membrane revealed by patching of membrane components. J. Cell Biol. 141:929-942.
Kenworthy, A. K., and M. Edidin. 1998. Distribution of a glycosylphosphatidylinositol-anchored protein at the apical surface of MDCK cells examined at a resolution of <100 ? using imaging fluorescence resonance energy transfer. J. Cell Biol. 142:69-84.
Kenworthy, A. K., N. Petranova, and M. Edidin. 2000. High-resolution FRET microscopy of cholera toxin B-subunit and GPI-anchored proteins in cell plasma membranes. Mol. Biol. Cell 11:1645-1655.
Kozak, S. L., J. M. Heard, and D. Kabat. 2002. Segregation of CD4 and CXCR4 into distinct lipid microdomains in T lymphocytes suggests a mechanism for membrane destabilization by human immunodeficiency virus. J. Virol. 76:1802-1815.
Kretzschmar, E., L. Buonocore, M. J. Schnell, and J. K. Rose. 1997. High-efficiency incorporation of functional influenza virus glycoproteins into recombinant vesicular stomatitis viruses. J. Virol. 71:5982-5989.
Lefrancois, L., and D. S. Lyles. 1982. The interaction of antibody with the major surface glycoprotein of vesicular stomatitis virus. I. Analysis of neutralizing epitopes with monoclonal antibodies. Virology 121:157-167.
Lyles, D. S., M. O. McKenzie, P. E. Kaptur, K. W. Grant, and W. G. Jerome. 1996. Complementation of M gene mutants of vesicular stomatitis virus by plasmid-derived M protein converts spherical extracellular particles into native bullet shapes. Virology 217:76-87.
Manie, S. N., S. Debreyne, S. Vincent, and D. Gerlier. 2000. Measles virus structural components are enriched into lipid raft microdomains: a potential cellular location for virus assembly. J. Virol. 74:305-311.
McKenna, P. M., J. P. McGettigan, R. J. Pomerantz, B. Dietzschold, and M. J. Schnell. 2003. Recombinant rhabdoviruses as potential vaccines for HIV-1 and other diseases. Curr. HIV Res. 1:229-237.
Nguyen, D. H., and J. E. Hildreth. 2000. Evidence for budding of human immunodeficiency virus type 1 selectively from glycolipid-enriched membrane lipid rafts. J. Virol. 74:3264-3272.
Parolini, I., S. Topa, M. Sorice, A. Pace, P. Ceddia, E. Montesoro, A. Pavan, M. P. Lisanti, C. Peschle, and M. Sargiacomo. 1999. Phorbol ester-induced disruption of the CD4-Lck complex occurs within a detergent-resistant microdomain of the plasma membrane. Involvement of the translocation of activated protein kinase C isoforms. J. Biol. Chem. 274:14176-14187.
Percherancier, Y., B. Lagane, T. Planchenault, I. Staropoli, R. Altmeyer, J. L. Virelizier, F. Arenzana-Seisdedos, D. C. Hoessli, and F. Bachelerie. 2003. HIV-1 entry into T-cells is not dependent on CD4 and CCR5 localization to sphingolipid-enriched, detergent-resistant, raft membrane domains. J. Biol. Chem. 278:3153-3161.
Pickl, W. F., F. X. Pimentel-Muinos, and B. Seed. 2001. Lipid rafts and pseudotyping. J. Virol. 75:7175-7183.
Popik, W., T. M. Alce, and W. C. Au. 2002. Human immunodeficiency virus type 1 uses lipid raft-colocalized CD4 and chemokine receptors for productive entry into CD4+ T cells. J. Virol. 76:4709-4722.
Scheiffele, P., A. Rietveld, T. Wilk, and K. Simons. 1999. Influenza viruses select ordered lipid domains during budding from the plasma membrane. J. Biol. Chem. 274:2038-2044.
Scheiffele, P., M. G. Roth, and K. Simons. 1997. Interaction of influenza virus haemagglutinin with sphingolipid-cholesterol membrane domains via its transmembrane domain. EMBO J. 16:5501-5508.
Schnell, M. J., L. Buonocore, E. Kretzschmar, J. E. Johnson, and J. K. Rose. 1996. Foreign glycoproteins expressed from recombinant vesicular stomatitis viruses are incorporated efficiently into virus particles. Proc. Natl. Acad. Sci. USA. 93:11359-11365.
Sharma, P., R. Varma, R. C. Sarasij, Ira, K. Gousset, G. Krishnamoorthy, M. Rao, and S. Mayor. 2004. Nanoscale organization of multiple GPI-anchored proteins in living cell membranes. Cell 116:577-589.
Takada, A., C. S. Robison, H. Goto, A. Sanchez, K. G. Murti, M. A. Whitt, and Y. Kawaoka. 1997. A system for functional analysis of Ebola virus glycoprotein. Proc. Natl. Acad. Sci. USA 94:14762-14769.
Takeda, M., G. P. Leser, C. J. Russell, and R. A. Lamb. 2003. Influenza virus hemagglutinin concentrates in lipid raft microdomains for efficient viral fusion. Proc. Natl. Acad. Sci. USA 100:14610-14617.
VandenDriessche, T., L. Naldini, D. Collen, and M. K. Chuah. 2002. Oncoretroviral and lentiviral vector-mediated gene therapy. Methods Enzymol. 346:573-589.
Varma, R., and S. Mayor. 1998. GPI-anchored proteins are organized in submicron domains at the cell surface. Nature 394:798-801.
Vincent, S., D. Gerlier, and S. N. Manie. 2000. Measles virus assembly within membrane rafts. J. Virol. 74:9911-9915.
Zavada, J. 1982. The pseudotype paradox. J. Gen. Virol. 63:15-24.
Zhang, J., A. Pekosz, and R. A. Lamb. 2000. Influenza virus assembly and lipid raft microdomains: a role for the cytoplasmic tails of the spike glycoproteins. J. Virol. 74:4634-4644.(Erica L. Brown, and Dougl)
Department of Biochemistry, Wake Forest University School of Medicine, Winston-Salem, North Carolina 27157
ABSTRACT
Many plasma membrane components are organized into detergent-resistant membrane microdomains referred to as lipid rafts. However, there is much less information about the organization of membrane components into microdomains outside of lipid rafts. Furthermore, there are few approaches to determine whether different membrane components are colocalized in microdomains as small as lipid rafts. We have previously described a new method of determining the extent of organization of proteins into membrane microdomains by analyzing the distribution of pairwise distances between immunogold particles in immunoelectron micrographs. We used this method to analyze the microdomains involved in the incorporation of the T-cell antigen CD4 into the envelope of vesicular stomatitis virus (VSV). In cells infected with a recombinant virus that expresses CD4 from the viral genome, both CD4 and the VSV envelope glycoprotein (G protein) were found in detergent-soluble (nonraft) membrane fractions. However, analysis of the distribution of CD4 and G protein in plasma membranes by immunoelectron microscopy showed that both were organized into membrane microdomains of similar sizes, approximately 100 to 150 nm. In regions of plasma membrane outside of virus budding sites, CD4 and G protein were present in separate membrane microdomains, as shown by double-label immunoelectron microscopy data. However, virus budding occurred from membrane microdomains that contained both G protein and CD4, and extended to approximately 300 nm, indicating that VSV pseudotype formation with CD4 occurs by clustering of G protein- and CD4-containing microdomains.
INTRODUCTION
For most viruses, envelope assembly involves association of internal virion components with specialized regions of the host plasma membrane containing locally high concentrations of viral transmembrane envelope glycoproteins. These membrane regions then bud from the host membrane and pinch off to form the virus envelope. As a general rule, host proteins are excluded from budding virus envelopes while viral glycoproteins are preferentially included. However, some heterologous membrane proteins are incorporated into virus envelopes at very high levels. This phenomenon is referred to as pseudotype formation or phenotypic mixing (31). There has been intense interest in pseudotype formation because of its importance for understanding the mechanism of virus envelope assembly. In addition, pseudotype formation is widely used in biotechnology applications to alter the host range of viruses and for vaccine development (16, 28).
Pseudotype formation was originally defined by a mixed infection of two enveloped viruses, in which progeny virions contained glycoproteins of both viruses (31). More recently, pseudotype formation has also been observed in recombinant viruses that express foreign glycoproteins directly from the viral genome (12, 24, 26). The exclusion of most host proteins from virus envelopes makes the process of envelope assembly appear to be specific for homologous viral glycoproteins. However, the phenomenon of pseudotype formation suggests that the process of envelope assembly is nonspecific with regard to the incorporation of viral glycoproteins. This apparent contradiction is known as the pseudotype paradox (31). The goal of the experiments presented here was to determine the mechanism behind this apparent paradox for the prototype rhabdovirus, vesicular stomatitis virus (VSV), which has been widely used in studies of viral envelope assembly and pseudotype formation. The VSV core consists of a helical nucleocapsid containing the genome RNA, which is condensed into a tightly coiled, bullet-like shape by the viral matrix (M) protein. The envelope contains primarily a single species of viral transmembrane glycoprotein (G protein) but can readily form pseudotypes with other membrane glycoproteins.
A key insight into the mechanism of virus envelope assembly was provided by the concept of membrane microdomains. A membrane microdomain can be defined as a locally high concentration of a membrane component compared to the average density in the membrane as a whole. By this definition, budding virus envelopes can be viewed as excellent examples of assembly of functional membrane microdomains. The most extensively characterized membrane microdomains are those that are enriched in sphingolipids and cholesterol, referred to as lipid rafts. These membrane microdomains have been characterized largely by their relative resistance to solubilization with Triton X-100 at low temperatures (2) and by copatching with glycolipids such as GM1 (8). The evidence that several viruses bud from lipid rafts that are enriched in their viral glycoproteins has reinforced the idea that virus envelope assembly is in fact a case of assembly of functional membrane microdomains (1, 15, 17, 20, 23, 32).
The concept that viruses can bud from lipid rafts has also led to the idea that pseudotype formation occurs by virus budding from lipid rafts that contain a mixture of glycoproteins (20). This could also account for the exclusion of host proteins that are not present in lipid rafts containing viral proteins. However, it has been difficult to test this hypothesis, since there are few approaches to determine whether different membrane components are colocalized in microdomains as small as lipid rafts, which are estimated to be on the order of 100 nm or less (29). Indeed, even the existence of lipid rafts under conditions other than detergent extraction or copatching has been difficult to detect (9, 10, 25, 29).
Although some viruses bud from lipid raft microdomains, there are likely to be many viruses that bud from nonraft areas of plasma membranes. VSV was the original example of a virus that buds from nonraft areas of membranes (22). Thus, the techniques used to analyze lipid rafts cannot be used to determine whether the VSV G protein is organized into membrane microdomains or whether such microdomains serve as sites of virus assembly. To address this question, we recently developed a new analytical method for immunoelectron microscopy data to determine whether membrane proteins are organized into microdomains and to determine the sizes of these microdomains (3). This method is based on analyzing the distribution of pairwise distances between immunogold particles averaged over a large number of electron micrographs. This method can be applied to any immunolabeled membrane protein regardless of whether or not it is raft associated. Using this approach, we showed that the VSV G protein is primarily organized into membrane microdomains with a diameter of 100 to 150 nm, both when expressed in transfected cells in the absence of other viral components and in the plasma membranes of virus-infected cells (3, 4).
One of the strengths of our new method is that it can also be used to determine colocalization of different membrane components in membrane microdomains in double-labeling experiments. In the experiments reported here, we analyzed the membrane microdomains involved in pseudotype formation. We used a recombinant VSV that expresses the T-cell antigen CD4 from the viral genome (VSV-CD4) (24). Virions produced from VSV-CD4-infected cells incorporate CD4 into their envelopes nearly as efficiently as G protein. The data presented here show that in VSV-CD4-infected cells, G protein- and CD4-containing microdomains have very similar properties. Both were approximately 100 to 150 nm in diameter, and both were primarily detergent soluble. In areas of the plasma membrane outside of virus budding sites, G protein and CD4 were present in separate microdomains, as shown by double-labeling experiments. However, virus budding occurred from membrane microdomains that contained both G protein and CD4, and extended to approximately 300 nm. The result that G protein and CD4 are in separate microdomains outside of budding sites but together within budding sites leads to the conclusion that VSV pseudotype formation with CD4 occurs by clustering of G protein- and CD4-containing microdomains.
MATERIALS AND METHODS
Cells and viruses. The recombinant virus VSV-CD4 was provided by John K. Rose of the Yale University School of Medicine (24). VSV-CD4 was grown at 37°C in BHK cells following two rounds of plaque purification as described previously (14).
Immunogold-labeling electron microscopy. BHK cells were infected with VSV-CD4 at a multiplicity of infection of 20 PFU/cell for 4.5 or 8 h as described elsewhere (3). Cells were fixed with buffered formalin as described elsewhere and labeled with a mouse anti-VSV G protein monoclonal antibody, I1 (13), or an anti-CD4 monoclonal antibody, MHCD0400 (Caltag Laboratories), diluted 1:100 for 1 h. Following three washes with phosphate-buffered saline (PBS) containing 10% bovine serum albumin (BSA), cells were labeled with a secondary goat anti-mouse immunoglobulin G antibody conjugated to 5-nm gold beads (Jackson ImmunoResearch Laboratories) for 1 h. For double labeling, formalin-fixed cells were permeabilized with 0.05% saponin for 10 min, then washed three times with PBS containing 0.1 M glycine, and incubated in PBS containing 0.1 M glycine and 10% BSA at 4°C overnight. Samples were incubated with a rabbit anti-VSV-G Tag antibody (Research Diagnostics Incorporated) and the mouse anti-CD4 monoclonal antibody. Following three washes with PBS containing 10% BSA, cells were labeled with a secondary goat anti-mouse antibody conjugated to 5-nm gold particles to indirectly label CD4 and a secondary goat anti-rabbit antibody conjugated to 15-nm gold particles to indirectly label G protein (Jackson ImmunoResearch Laboratories). Immunolabeled cells were postfixed in 1% osmium tetroxide for 15 min, washed three times in PBS, scraped from the dish, and collected by centrifugation. Cells were embedded in Spurr resin using the standard embedding method. Thin sections (80 nm) of the embedded material were viewed using a Philips EM400 electron microscope operating at 80 keV.
A series of 25 electron micrographs of arbitrarily chosen areas of plasma membrane were collected from two separate experiments (total of 50 micrographs). Each micrograph was obtained from a separate cell. The magnification of micrographs was x43,000. Negatives were digitized with a scanner, and the micrographs were analyzed with Image Pro 3.0 software (Media Cybernetics). A pairwise measurement was made of the distance traced along the plasma membrane between each gold particle and every other gold particle to its right in the image, excluding gold particles within a virus budding site. These measurements generated a data set of distances between all of the gold particles in all 50 images. This data set was plotted as a histogram of the average number of gold particles per 20 nm on the y axis versus the distance of separation on the x axis in 20-nm increments as described previously (3). For double-label experiments, a series of 25 electron micrographs of arbitrarily chosen 15-nm gold particles that were not in budding virus envelopes was collected from two separate experiments (total of 50 micrographs). Each micrograph was obtained from a separate cell. The distance to each 5-nm gold particle within a distance of 1 μm on each side of the reference 15-nm gold particle was measured along the trace of the plasma membrane. The data were plotted as a histogram of the density of anti-CD4 gold particles on the y axis versus the distance from anti-G protein reference particles in 20-nm increments on the x axis. The y values were converted to average densities by dividing the number of gold particles at each distance by the number of 15-nm gold particles that contained data at that distance (3).
Analysis of virus budding sites. A series of 25 electron micrographs of arbitrarily chosen VSV budding sites was collected from each of two separate experiments (total of 50 micrographs), and virus budding sites were analyzed as described previously (3). Since several micrographs contained more than one virus budding site, the actual number of budding sites analyzed in the 50 micrographs was 61 (anti-CD4 at 4.5 h postinfection [hpi]), 74 (anti-CD4 at 8 hpi), 62 (anti-G protein at 4.5 hpi), or 73 (anti-G protein at 8 hpi). Briefly, virus budding sites were analyzed by measuring the distance of each gold particle from the center of the leading edge of the virus budding site along the trace of the plasma membrane, within a distance of 1 μm of the budding site. For budding sites separated by distances <2 μm, the distance between them was divided at the midpoint. Data were presented as a histogram of the average density of gold particles on the y axis versus the distance from the center of the budding site in 20-nm increments on the x axis. The length of the virus envelope was determined by tracing the membrane from the tip of the budding virion to the end of the electron density of the densely stained nucleocapsid-M protein (NCM) complex. The left and right sides of the virus envelope were measured separately.
Western blot analysis of CD4 incorporation into VSV-CD4 virions. BHK cells infected with VSV-CD4 for 4 or 7.5 h were washed three times with PBS and then incubated in medium for 1 h at 37°C. Culture supernatants were collected, centrifuged at 2,000 rpm for 5 min at 4°C to remove cellular debris, and then centrifuged over a 15% sucrose cushion at 50,000 rpm for 40 min in a Beckman TLS-55 rotor. Virus pellets were resuspended in a total volume of 100 μl of radioimmunoprecipitation assay (RIPA) buffer (0.15 M NaCl, 1% deoxycholate, 1% Triton X-100, 0.1% sodium dodecyl sulfate [SDS], 10 mM Tris, pH 7.4). Equal amounts of purified virions were analyzed by Western blotting with anti-CD4 antibody H370 (Santa Cruz Biotechnology, Inc.) and anti-VSV-G Tag antibody (Research Diagnostics Incorporated) as described elsewhere (5). The ratio of CD4 to G protein was determined using Quantity One software (Bio-Rad).
SDS-PAGE analysis of G protein incorporation into VSV-CD4 virions. BHK cells infected with VSV-CD4 for 3.5 or 7 h were incubated in methionine-deficient medium containing 2% fetal bovine serum for 10 min at 37°C, then pulse labeled with [35S]methionine for 10 min, and incubated in nonradiolabeled medium for 1 h at 37°C. Virions released during the chase were harvested as described above and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) on 10% polyacrylamide gels, followed by phosphorescence imaging using Quantity One software.
Analysis of Triton X-100-resistant G protein and CD4 microdomains. BHK cells infected with VSV-CD4 for 4.5 or 8 h were washed with PBS and then extracted with 1% Triton X-100 in 25 mM Tris, 150 mM NaCl, 5 mM EDTA (pH 7.4) for 30 min at 4°C. Cell lysates were sedimented at approximately 100,000 x g in a Beckman Airfuge for 20 min at 4°C to separate soluble and insoluble material. Pellets were resuspended in RIPA buffer to contain the same volume as the supernatant fractions. Supernatant and pellet fractions were analyzed by Western blot analysis with anti-G protein and anti-CD4 antibodies. The volume of supernatant analyzed was one-fourth the volume of the pellet. The relative percentages of G protein and CD4 in the pellet fractions were determined using Quantity One software.
For flotation-based analysis of detergent-resistant G protein and CD4, VSV-CD4-infected cells were washed with cold PBS and resuspended in 1% Triton X-100 for 30 min. Lysates were subjected to 15 strokes of a Dounce homogenizer. Nuclei were removed by centrifugation at 2,000 rpm at 4°C. A 500-μl volume of supernatant was mixed with 1.5 ml of 60% (wt/vol) sucrose containing 1% Triton X-100 and then overlaid with 2 ml of 30% sucrose and 1 ml of 5% sucrose. Gradients were centrifuged in a Beckman SW50.1 rotor at 35,000 rpm at 4°C for 16 h. Fractions (0.6 ml) were collected from the top of the gradient and were analyzed by Western blotting as described above.
RESULTS AND DISCUSSION
G protein- and CD4-containing microdomains in BHK cells are primarily detergent soluble. The observation that VSV efficiently forms pseudotypes with CD4 (24) raises the question of whether G protein-containing and CD4-containing microdomains have similar properties. Little if any VSV G protein is found in detergent-insoluble (lipid-raft) membrane fractions (22, 30, 32). However, the reported extent of CD4 association with detergent-insoluble lipid rafts in T cells ranges from nearly 100% to less than 20%, depending on the source of the T cells (7, 11, 18, 19, 21). We determined the extent of association of G protein and CD4 with detergent-insoluble membrane fractions in BHK cells infected with VSV-CD4 at two different time points, 4.5 hpi, which is relatively early in the period of virus assembly, and 8 hpi, which is a late time when viral proteins have accumulated to near-maximum levels. Cells infected with VSV-CD4 were treated with buffer containing 1% Triton X-100 at 4°C, and cell lysates were sedimented to pellet detergent-resistant material. The resulting supernatant and pellet fractions were analyzed by Western blotting for G protein and CD4 (Fig. 1A), and the percentages of G protein and CD4 in the pellets were quantitated from six independent experiments. At both 4.5 and 8 hpi, there was little if any G protein detected in the pellet (<2%), consistent with earlier data. Likewise, there was only a small amount of CD4 detectable in the pellet fractions (12 to 14%). This result is similar to that obtained for T cells that lack the CD4-associated protein kinase Lck, supporting the idea that in the absence of Lck, only small amounts of CD4 are associated with detergent-resistant membrane fractions (7, 19). These results were confirmed by performing a flotation-based analysis of detergent-resistant G protein and CD4. Detergent-treated lysates were mixed with dense sucrose, overlaid with a discontinuous sucrose gradient, and centrifuged to float the detergent-resistant membranes. Fractions were analyzed by Western blotting (Fig. 1C). Little if any G protein and only small amounts of CD4 were detected in the raft fractions (lanes 1 to 4). These results indicate that G protein and CD4 are similar, though not identical, in their detergent solubility and that neither G protein nor CD4 partitions into detergent-insoluble microdomains to a great extent.
G protein- and CD4-containing microdomains are similar in size. Since most CD4 did not partition into lipid rafts, this raised the question of whether CD4 was organized into membrane microdomains in BHK cells or whether it was distributed randomly in the plasma membrane. To distinguish between these two possibilities, we used our recently described method for the analysis of immunoelectron microscopy data to detect the organization of proteins into membrane microdomains (3). BHK cells were infected with VSV-CD4, and at either 4.5 or 8 hpi, cells were fixed and immunolabeled with an antibody against CD4 and a secondary antibody conjugated to 5-nm gold particles. The amount of gold-conjugated secondary antibody was adjusted to low levels so that no labeling was detectable in the negative controls (3). This also prevents interference between gold particles in heavily labeled areas of samples with positive immunoreactivity. Sets of 25 electron micrographs of arbitrarily chosen areas of plasma membrane were collected in each of two independent experiments (50 micrographs total at each time point), with each image arising from a different cell. A representative image is shown in Fig. 2A. A pairwise measurement was made of the distance traced along the plasma membrane between each gold particle and every other gold particle to its right in the image, excluding gold particles within a virus budding site. These measurements generated a data set of distances between all of the gold particles in all 50 images. This data set was plotted as a histogram of the average number of gold particles per 20 nm on the y axis versus the distance of separation on the x axis in 20-nm increments (3).
Figure 2B shows the analysis of pairwise distance measurements of immunolabeled CD4 in the plasma membranes of VSV-CD4-infected cells at 4.5 and 8 hpi. The average density of CD4 labeling in the plasma membrane at 4.5 hpi, determined from the 50 micrographs (0.11 particle/20 nm), is shown. For distances of separation between 100 nm and 1,000 nm, the histogram shows densities that are close to this average, as expected. However, for distances less than 100 nm, the histogram shows densities that are higher than the average, which indicates that CD4 is organized into microdomains in this size range. If CD4 is organized into circular microdomains with only one size, this analysis yields a straight line with a negative slope that intersects the average density at a distance that is 64% (2/) of the diameter of the microdomain (3). The data obtained at 4.5 hpi are consistent with only a single dramatic change in the slope at approximately 100 nm. This indicates that CD4 is predominantly organized into microdomains with a diameter of 100 to 150 nm.
At 8 hpi, the average density of CD4 (0.21 particle/20 nm) was approximately twofold higher than the average density at 4.5 hpi. If the sizes of all of the CD4-containing microdomains at 8 hpi had remained at 100 to 150 nm, the graph would intersect the average density at the same point as at 4.5 hpi, but the negative slope and the y intercept would be twofold higher (3). For example, the y intercept in Fig. 2B should increase from approximately 0.4 at 4.5 hpi to approximately 0.8 at 8 hpi. However, the densities near the y intercept at 8 hpi increased to only approximately 0.5 particle/20 nm. Furthermore, densities higher than the average extended to at least 400 nm and were not consistent with a single change in slope. These data indicate that much of the CD4 was organized into microdomains in the range of 100 to 150 nm but that some of the CD4 was populating microdomains with a wider variety of sizes. From these data we conclude that CD4 does partition into membrane microdomains that are predominantly 100 to 150 nm but that, when expressed at high levels, CD4 is also distributed among microdomains of larger sizes.
The CD4 microdomains were similar in size to the G protein-containing microdomains we described earlier in wild-type VSV-infected cells (3, 4). To determine if G protein was organized to the same extent in VSV-CD4-infected cells as in wild-type virus-infected cells, the pairwise distance analysis of Fig. 2B was applied to G protein in VSV-CD4-infected cells in Fig. 2C. In contrast to the findings of our earlier study of wild-type VSV (3), the average density of G protein in the plasma membranes of cells infected with VSV-CD4 did not increase between 4.5 and 8 hpi. The average density of G protein at both time points was 0.08 particle/20 nm. We attribute this difference from our results with wild-type VSV to the faster depletion of G protein from the plasma membrane by budding of VSV-CD4 virions compared to wild-type VSV. An earlier electron microscopy study showed that VSV-CD4 virions are longer than wild-type virions due to their extra genetic information (24), and therefore budding of VSV-CD4 would be expected to deplete G protein from the plasma membrane at a higher rate. The important result from Fig. 2C is that the distribution of G protein in the plasma membrane at 4.5 and 8 hpi was characterized by a high density of G protein at short distances of separation, with a single dramatic change in slope at approximately 100 nm. These data indicate that the G protein-containing microdomains are 100 to 150 nm, similar in size to the CD4-containing microdomains. The microdomains containing G protein and CD4 are slightly smaller than those containing the lipid-raft-associated influenza virus hemagglutinin, which were determined to be 200 to 280 nm using the same technique (27). This reinforces the idea that G protein- and CD4-containing microdomains are similar in principle to lipid rafts but differ in detail from lipid rafts in their detergent solubility and overall size.
G protein and CD4 are in separate microdomains outside of budding sites. The similarity in size and detergent solubility of G protein- and CD4-containing microdomains suggested that these proteins might exist as a mixture within a single type of microdomain. Alternatively, they may exist in separate microdomains with similar properties. We distinguished between these two hypotheses by analyzing the distribution of CD4 relative to G protein in VSV-CD4-infected cells by double-label immunogold electron microscopy. If CD4 and G protein formed mixed microdomains, this would be apparent as a higher-than-average density for small distances of separation between anti-G protein gold particles and anti-CD4 gold particles, a result similar to those for CD4 alone and G protein alone (Fig. 2B and C). Alternatively, if G protein and CD4 exist in separate microdomains, the density of CD4 would be similar to the average density at all distances of separation from G protein.
BHK cells were infected with VSV-CD4 for 4.5 or 8 h. Cells were then immunolabeled with antibodies against G protein and CD4. A secondary antibody conjugated to 15-nm gold particles was used to indirectly label G protein, while an antibody conjugated to 5-nm gold particles was used to indirectly label CD4. A representative image depicting the labeling of G protein and CD4 at 4.5 hpi is shown in Fig. 2D. This image contains two 15-nm gold particles that label G protein. Most 5-nm gold particles that label CD4 are clearly separated from the 15-nm gold particles that label G protein, although one 5-nm gold particle is in close proximity to a 15-nm gold particle. To quantitate the density of CD4 in relation to G protein, a series of 25 electron micrographs of arbitrarily chosen 15-nm anti-G protein gold particles (excluding gold particles associated with virus budding sites) was collected at 4.5 and 8 hpi in two separate experiments (50 micrographs total at each time point). Images were analyzed by measuring the distance along the trace of the membrane of each 5-nm gold particle from the 15-nm reference gold particle to a distance of 1 μm on each side.
The results of the quantitation are represented in Fig. 2E as a histogram of the density of CD4 labeling versus the distance from a 15-nm reference gold particle in 20-nm increments for 4.5 and 8 hpi. There was no preferential labeling of CD4 at short distances of separation from anti-G protein gold particles, which would have been expected if CD4 were present in G protein-enriched microdomains. Instead, the density of CD4 labeling was similar to the average density for all distances of separation from anti-G protein gold particles. This is the expected result if CD4 does not partition into G protein-enriched microdomains. Taken together, the results in Fig. 2 indicate that both CD4 and G protein exist in discrete microdomains outside of virus budding sites, as shown in Fig. 2B and C. However, these microdomains are separate and nonoverlapping, as shown in Fig. 2E.
G protein- and CD4-containing microdomains cluster at VSV budding sites. A previous immunoelectron microscopy study showed that nearly all VSV-CD4 virions contain both G protein and CD4, suggesting that both proteins are present in the membrane microdomains that form virus budding sites (24). In Fig. 3, the extent to which CD4 and G protein were concentrated in membrane microdomains at sites of virus budding was determined by comparing the densities of immunogold labeling in budding virus envelopes to the average density in the plasma membrane as whole. Representative images of budding virions labeled with antibody to CD4 are shown in Fig. 3A. The densely stained core of the budding virus consists of the tightly coiled NCM complex. The envelopes of budding virions were labeled with gold particles, while there was little labeling of the surrounding plasma membrane. To quantitate the levels of labeling, infected BHK cells immunolabeled for CD4 or G protein in Fig. 2 were used to generate a separate series of electron micrographs of arbitrarily chosen VSV-CD4 budding sites (50 micrographs total for each sample). The criteria for identifying virus budding sites were (i) electron density created by the NCM complex, (ii) extension of the budding particle into the extracellular space, and (iii) visible attachment to a cell (3). The density of labeling was determined by counting the number of gold particles associated with the virus envelope and dividing by the distance along the trace of the envelope from the base of the densely stained NCM complex on one side to the base on the other. The extent of labeling in each budding virus envelope was expressed as the ratio of the density in the envelope to the average density in the plasma membrane as a whole (determined from the micrographs in Fig. 2). The data are shown as histograms in Fig. 3B (CD4) and C (G protein), in which the x axis is the density of labeling in the budding virus envelope and the y axis is the number of budding sites with each density. Statistical analysis of these data is given in Table 1.
The densities of CD4 labeling in budding virus envelopes ranged from near the density in the membrane as a whole to approximately ninefold higher (Fig. 3B), with an average density that was approximately threefold higher than in the membrane as a whole (Table 1). At 4.5 hpi, 64% of budding virions (39 out of 61) contained CD4 at levels that were at least twofold higher than in the membrane as a whole, and at 8 hpi, 76% of budding virions had twofold or higher levels of CD4 labeling (Table 1). Thus, the majority of virions bud from CD4-enriched areas of the plasma membrane.
The densities of labeling of G protein in budding virions, relative to the membrane as a whole (Fig. 3C), were higher than those for CD4 (notice the difference in the x axis scale between Fig. 3B and 3C). The average density of G protein labeling in budding virus envelopes was approximately ninefold above that in the membrane as a whole at 4.5 hpi and approximately sixfold higher at 8 hpi (Table 1). As expected, >90% of budding virions contained a high level of G protein labeling (Table 1).
These data show that the majority of budding sites were enriched in both G protein and CD4. This result, combined with the finding that G protein and CD4 are in separate microdomains outside of virus budding sites (Fig. 2E), leads to the conclusion that these two types of microdomains must come together to form the virus budding sites. In other words, G protein- and CD4-containing microdomains cluster at VSV-CD4 budding sites.
Sizes of G protein- and CD4-containing microdomains at sites of virus budding. Further evidence for clustering of G protein- and CD4-containing microdomains at sites of virus budding was obtained by analysis of the sizes of the microdomains at virus budding sites. The methods for analysis of microdomains at budding sites have been described previously (3) and are similar to the analysis outside of budding sites, except that the tip of the budding virion is used as a reference point. Briefly, micrographs were analyzed by measuring the distance along the trace of the membrane of each gold particle from the center of the leading edge of the budding virion to a distance of 1 μm on each side. The results of the quantitation for CD4 are represented in Fig. 3D as a histogram of the density, expressed in particles/20 nm, versus the distance from the tip of the virus envelope in 20-nm increments. The vertical lines represent the average length of the virus envelope in all of the budding sites analyzed, as defined by the distance traced from the tip of the budding virion to the end of the densely stained NCM complex. Therefore, data points to the left of the vertical lines arise from gold particles that appear to be associated with the virus envelope. The average density of gold labeling in the plasma membrane as a whole is also shown.
As shown in Fig. 3D, the density of CD4 labeling within the virus envelope was twofold higher at 8 hpi (approximately 0.6 particle/20 nm) than at 4.5 hpi (approximately 0.3 particle/20 nm), and at both time points the labeling density in virus envelopes was higher than the average in the membrane as a whole, as shown in the previous analysis (Fig. 3B; Table 1). However, at both 4.5 and 8 hpi, the CD4-containing membrane microdomains extended beyond the base of the budding virus envelope to a distance of approximately 300 to 400 nm. Thus, the CD4-containing microdomains involved in formation of virus envelopes were considerably larger than the 100- to 150-nm microdomains observed outside of virus budding sites (Fig. 2B). This result indicates that the CD4-containing microdomains at virus budding sites must have been assembled from multiple smaller microdomains. At 8 hpi, the CD4-containing microdomains at virus budding sites would be consistent with the largest microdomains observed outside of virus budding sites (Fig. 2B), leaving open the possibility that viruses bud selectively from these larger microdomains. However, it is more likely that virus budding sites are assembled from multiple smaller microdomains at both 4.5 and 8 hpi.
A similar analysis of G protein labeling in budding virus envelopes is shown in Fig. 3E. As shown previously (3, 4), the highest densities of G protein labeling were observed near the tip of the budding virions (near the y axis), and the densities declined toward the base. At both 4.5 and 8 hpi, densities of G protein labeling higher than the average extended to a distance of approximately 300 nm (Fig. 3E). Thus, the G protein-containing microdomains at virus budding sites extend to approximately 300 nm, a finding similar to the results with CD4 (Fig. 3D). As in the case of CD4, this size is considerably larger than the 100- to 150-nm G protein-containing microdomains observed outside of virus budding sites (Fig. 2C). The larger size of G protein-containing microdomains at virus budding sites (Fig. 3C) than outside of budding sites (Fig. 2C) was observed at both 4.5 and 8 hpi. These data further support the conclusion that the membrane microdomains involved in virus assembly are assembled from multiple smaller membrane microdomains and include both CD4-containing and G protein-containing membrane microdomains.
The density of G protein in the VSV-CD4 envelope decreases slightly, while the density of CD4 increases, at late times postinfection. The data in Fig. 3E and Table 1 indicate that the density of G protein labeling in the VSV-CD4 envelope was slightly lower at 8 hpi than at 4.5 hpi. Despite the broad distribution of densities (Fig. 3C), the difference in the average density at 8 versus 4.5 hpi was statistically significant (Table 1). This result was not expected, since the density of G protein labeling in the membrane as a whole did not decline from 4.5 to 8 hpi. We confirmed the analytical electron microscopy result that the density of G protein in the VSV-CD4 envelope decreased slightly at 8 hpi by SDS-PAGE analysis of G protein incorporation into virions. Cells were infected with VSV-CD4 for either 4 or 7.5 h, at which times the cells were pulse labeled with [35S]methionine for 10 min. Following a 1-h incubation at 37°C to allow for transport to the cell surface and incorporation into virions, virus particles were isolated from the supernatant by centrifugation and were analyzed by SDS-PAGE and phosphorescence imaging (Fig. 4A). Data from three independent experiments were quantitated as the ratio of G protein to M protein (Fig. 4B). The VSV proteins L, G, N/P, and M were heavily labeled, but CD4 was not readily apparent (Fig. 4A), since it is labeled inefficiently with [35S]methionine (24). The amounts of internal viral proteins (N/P and M) released into virions were similar at the two time points, but the amount of G protein incorporated into virions decreased from 4.5 to 8 hpi, from a G/M ratio of 0.63 to 0.42. This decrease in the G/M ratio was consistent with the results of the electron microscopy analysis (Fig. 3).
In contrast to G protein, the amount of CD4 in the VSV-CD4 envelope increased from 4.5 to 8 hpi (Fig. 3 and Table 1). This result was confirmed by Western blot analysis. Cells were infected with VSV-CD4, virions released for a 1-h period surrounding the 4.5- and 8-h time points (4 to 5 hpi and 7.5 to 8.5 hpi, respectively) were isolated, and equal amounts of purified virions were analyzed by Western blotting for CD4 and G protein (Fig. 4C). The ratio of CD4 to G protein in virions was quantitated from three independent experiments and normalized to the data for 8 hpi (Fig. 4D). The amount of CD4 incorporated into virions at 4.5 hpi was 37% of the level incorporated at 8 hpi. These data confirm the data obtained by electron microscopy showing that the amount of CD4 incorporated into virions increases approximately twofold at late times postinfection.
A new model of VSV pseudotype formation. Our basic model to explain the formation of VSV pseudotypes that contain CD4 is that G protein and CD4 partition into distinct microdomains that are similar in size and detergent solubility. This point is supported by the data in Fig. 1 and 2. These separate microdomains then cluster to form the virus envelope, leading to the incorporation of both G protein and CD4. This point is supported by the data in Fig. 3. This model raises the question of what drives the clustering of these microdomains. We propose that the clustering is driven by the formation of the NCM complex. This proposal is illustrated in Fig. 5.
Previous results from our laboratory have suggested that assembly of the NCM complex occurs in at least two steps, which we have called initiation and recruitment (6). The initiation event involves the binding of one or a few molecules of M protein to the nucleocapsid, while in the recruitment phase, the majority of M protein is recruited from both the membrane and cytosol, and condenses the nucleocapsid into a tightly coiled NCM complex. We propose that initiation almost always occurs at a G protein-enriched membrane microdomain (Fig. 5, step 1), which accounts for the higher densities of G protein at the tip of a budding virus envelope relative to the base. We also propose that recruitment involves M protein that is associated with a variety of different microdomains (Fig. 5, step 2), which are all similar in lipid composition to the G protein-containing microdomains. This step of virus assembly would, therefore, account for the clustering of CD4-containing microdomains with G protein-containing microdomains in the virus envelope. It is likely that the efficiency of incorporation of membrane proteins is dependent on the extent to which their microdomains resemble G protein-enriched microdomains and therefore associate with M protein. This would account for the fact that VSV forms pseudotypes with a wide variety of membrane proteins, but the efficiency of their incorporation into the VSV envelope varies over a considerable range (12, 24, 26). For example, VSV forms pseudotypes with the influenza virus hemagglutinin, which is present in lipid rafts, but the efficiency is less than that of CD4 (12).
Our model would also account for the lower level of G protein incorporation and the higher level of CD4 incorporation into the envelopes of VSV-CD4 virions at later times postinfection. At early times postinfection, most of the membrane microdomains that cluster to form virus envelopes would be G protein-containing microdomains, with smaller amounts of CD4-containing microdomains. At later times postinfection, the number of CD4-containing microdomains in the plasma membrane increases due to the higher level of viral gene expression (Fig. 2B). In contrast, the number of G protein-containing microdomains remains approximately the same (Fig. 2C), since depletion of G protein by virus budding balances the increase in viral gene expression. The initiation step is still likely to occur in a G protein-containing microdomain, but the recruitment (clustering) step would be more likely to include CD4-containing microdomains than G protein-containing microdomains, since the G protein microdomains have been depleted faster than the CD4 microdomains, leading to a lower level of G protein in the envelope. This effect could account for the observation that expression of high levels of membrane proteins that form pseudotypes with VSV leads to a reduction in viral infectivity (24), since virions containing lower levels of G protein are likely to be less infectious. A similarly high level of incorporation of alternative membrane microdomains into virus budding sites might occur in cell types that express G protein inefficiently on their cell surfaces.
In summary, our data have shown that G protein and CD4 exist in distinct membrane microdomains outside of virus budding sites but that both are present in budding virus envelopes. These results lead to the conclusion that VSV pseudotype formation with CD4 is due to clustering of membrane microdomains. The ability to come to this conclusion was critically dependent on our new methods of analysis of immunoelectron microscopy data. These results have led to a new model of viral pseudotype formation, which will be further tested in future experiments.
ACKNOWLEDGMENTS
We acknowledge the Microscopy Core Laboratory, especially Ken Grant and Paula Moore, for assistance with the preparation of the immunogold-labeling electron microscopy samples. We also thank Jack Rose for providing the VSV-CD4 and Jay Jerome, Griffith Parks, David Ornelles, and Mark Willingham for critical advice.
This research was supported by Public Health Service grant AI15892 from the National Institute of Allergy and Infectious Diseases. E.L.B. was supported by National Institutes of Health training grant T32-AI07401. The Microscopy Core Laboratory was supported in part by the core grant for the Comprehensive Cancer Center of Wake Forest University (CA12197) from the National Cancer Institute.
REFERENCES
Ali, A., R. T. Avalos, E. Ponimaskin, and D. P. Nayak. 2000. Influenza virus assembly: effect of influenza virus glycoproteins on the membrane association of M1 protein. J. Virol. 74:8709-8719.
Brown, D. A., and J. K. Rose. 1992. Sorting of GPI-anchored proteins to glycolipid-enriched membrane subdomains during transport to the apical cell surface. Cell 68:533-544.
Brown, E. L., and D. S. Lyles. 2003. A novel method for analysis of membrane microdomains: vesicular stomatitis virus glycoprotein microdomains change in size during infection, and those outside of budding sites resemble sites of virus budding. Virology 310:343-358.
Brown, E. L., and D. S. Lyles. 2003. Organization of the vesicular stomatitis virus glycoprotein into membrane microdomains occurs independently of intracellular viral components. J. Virol. 77:3985-3992.
Connor, J. H., and D. S. Lyles. 2002. Vesicular stomatitis virus infection alters the eIF4F translation initiation complex and causes dephosphorylation of the eIF4E binding protein 4E-BP1. J. Virol. 76:10177-10187.
Flood, E. A., and D. S. Lyles. 1999. Assembly of nucleocapsids with cytosolic and membrane-derived matrix proteins of vesicular stomatitis virus. Virology 261:295-308.
Fragoso, R., D. Ren, X. Zhang, M. W. Su, S. J. Burakoff, and Y. J. Jin. 2003. Lipid raft distribution of CD4 depends on its palmitoylation and association with Lck, and evidence for CD4-induced lipid raft aggregation as an additional mechanism to enhance CD3 signaling. J. Immunol. 170:913-921.
Harder, T., P. Scheiffele, P. Verkade, and K. Simons. 1998. Lipid domain structure of the plasma membrane revealed by patching of membrane components. J. Cell Biol. 141:929-942.
Kenworthy, A. K., and M. Edidin. 1998. Distribution of a glycosylphosphatidylinositol-anchored protein at the apical surface of MDCK cells examined at a resolution of <100 ? using imaging fluorescence resonance energy transfer. J. Cell Biol. 142:69-84.
Kenworthy, A. K., N. Petranova, and M. Edidin. 2000. High-resolution FRET microscopy of cholera toxin B-subunit and GPI-anchored proteins in cell plasma membranes. Mol. Biol. Cell 11:1645-1655.
Kozak, S. L., J. M. Heard, and D. Kabat. 2002. Segregation of CD4 and CXCR4 into distinct lipid microdomains in T lymphocytes suggests a mechanism for membrane destabilization by human immunodeficiency virus. J. Virol. 76:1802-1815.
Kretzschmar, E., L. Buonocore, M. J. Schnell, and J. K. Rose. 1997. High-efficiency incorporation of functional influenza virus glycoproteins into recombinant vesicular stomatitis viruses. J. Virol. 71:5982-5989.
Lefrancois, L., and D. S. Lyles. 1982. The interaction of antibody with the major surface glycoprotein of vesicular stomatitis virus. I. Analysis of neutralizing epitopes with monoclonal antibodies. Virology 121:157-167.
Lyles, D. S., M. O. McKenzie, P. E. Kaptur, K. W. Grant, and W. G. Jerome. 1996. Complementation of M gene mutants of vesicular stomatitis virus by plasmid-derived M protein converts spherical extracellular particles into native bullet shapes. Virology 217:76-87.
Manie, S. N., S. Debreyne, S. Vincent, and D. Gerlier. 2000. Measles virus structural components are enriched into lipid raft microdomains: a potential cellular location for virus assembly. J. Virol. 74:305-311.
McKenna, P. M., J. P. McGettigan, R. J. Pomerantz, B. Dietzschold, and M. J. Schnell. 2003. Recombinant rhabdoviruses as potential vaccines for HIV-1 and other diseases. Curr. HIV Res. 1:229-237.
Nguyen, D. H., and J. E. Hildreth. 2000. Evidence for budding of human immunodeficiency virus type 1 selectively from glycolipid-enriched membrane lipid rafts. J. Virol. 74:3264-3272.
Parolini, I., S. Topa, M. Sorice, A. Pace, P. Ceddia, E. Montesoro, A. Pavan, M. P. Lisanti, C. Peschle, and M. Sargiacomo. 1999. Phorbol ester-induced disruption of the CD4-Lck complex occurs within a detergent-resistant microdomain of the plasma membrane. Involvement of the translocation of activated protein kinase C isoforms. J. Biol. Chem. 274:14176-14187.
Percherancier, Y., B. Lagane, T. Planchenault, I. Staropoli, R. Altmeyer, J. L. Virelizier, F. Arenzana-Seisdedos, D. C. Hoessli, and F. Bachelerie. 2003. HIV-1 entry into T-cells is not dependent on CD4 and CCR5 localization to sphingolipid-enriched, detergent-resistant, raft membrane domains. J. Biol. Chem. 278:3153-3161.
Pickl, W. F., F. X. Pimentel-Muinos, and B. Seed. 2001. Lipid rafts and pseudotyping. J. Virol. 75:7175-7183.
Popik, W., T. M. Alce, and W. C. Au. 2002. Human immunodeficiency virus type 1 uses lipid raft-colocalized CD4 and chemokine receptors for productive entry into CD4+ T cells. J. Virol. 76:4709-4722.
Scheiffele, P., A. Rietveld, T. Wilk, and K. Simons. 1999. Influenza viruses select ordered lipid domains during budding from the plasma membrane. J. Biol. Chem. 274:2038-2044.
Scheiffele, P., M. G. Roth, and K. Simons. 1997. Interaction of influenza virus haemagglutinin with sphingolipid-cholesterol membrane domains via its transmembrane domain. EMBO J. 16:5501-5508.
Schnell, M. J., L. Buonocore, E. Kretzschmar, J. E. Johnson, and J. K. Rose. 1996. Foreign glycoproteins expressed from recombinant vesicular stomatitis viruses are incorporated efficiently into virus particles. Proc. Natl. Acad. Sci. USA. 93:11359-11365.
Sharma, P., R. Varma, R. C. Sarasij, Ira, K. Gousset, G. Krishnamoorthy, M. Rao, and S. Mayor. 2004. Nanoscale organization of multiple GPI-anchored proteins in living cell membranes. Cell 116:577-589.
Takada, A., C. S. Robison, H. Goto, A. Sanchez, K. G. Murti, M. A. Whitt, and Y. Kawaoka. 1997. A system for functional analysis of Ebola virus glycoprotein. Proc. Natl. Acad. Sci. USA 94:14762-14769.
Takeda, M., G. P. Leser, C. J. Russell, and R. A. Lamb. 2003. Influenza virus hemagglutinin concentrates in lipid raft microdomains for efficient viral fusion. Proc. Natl. Acad. Sci. USA 100:14610-14617.
VandenDriessche, T., L. Naldini, D. Collen, and M. K. Chuah. 2002. Oncoretroviral and lentiviral vector-mediated gene therapy. Methods Enzymol. 346:573-589.
Varma, R., and S. Mayor. 1998. GPI-anchored proteins are organized in submicron domains at the cell surface. Nature 394:798-801.
Vincent, S., D. Gerlier, and S. N. Manie. 2000. Measles virus assembly within membrane rafts. J. Virol. 74:9911-9915.
Zavada, J. 1982. The pseudotype paradox. J. Gen. Virol. 63:15-24.
Zhang, J., A. Pekosz, and R. A. Lamb. 2000. Influenza virus assembly and lipid raft microdomains: a role for the cytoplasmic tails of the spike glycoproteins. J. Virol. 74:4634-4644.(Erica L. Brown, and Dougl)