Lipid Rafts and Pseudotyping

Abstract

Specific interactions between envelope and core proteins govern the membrane assembly of most enveloped viruses. Despite this, mixed infections lead to pseudotyping, the association of the viral cores of one virus with the envelopes of another. How does this occur? We show here that the detergent-insoluble lipid rafts of the plasma membrane function as a natural meeting point for the transmembrane and core components of a phylogenetically diverse collection of enveloped viruses. As a result, viral particles preferentially incorporate both the envelope components of other viruses as well as the extra- and intracellular constituents of host cell lipid rafts, including gangliosides, glycosyl phosphatidylinositol-anchored surface proteins, and intracellular signal transduction molecules. Pharmacological disruption of lipid rafts interferes with virus production.

The core-forming proteins of many small viruses are created from a polyprotein precursor, whereas the envelope is usually a separate translation product that must interact with the core during assembly. Although it might appear genetically efficient to let one protein incorporate both envelope and core functionality, among enveloped viruses this organization has not, to our knowledge, been described. Instead, envelope proteins meet the corresponding core-forming proteins on the inner side of the virus bud. To date, specific interactions of envelope with core have been shown for alpha-, hepadna-, and lentiviruses (8, 14, 15, 36, 45, 46, 50, 74, 76). Cytoplasmic elements of neuraminidase mediate assembly interactions in influenza virus (16, 44), and vesicular stomatitis virus (VSV) has distributed information in the G protein (VSV-G) cytoplasmic domain that is required for efficient budding (66).

The site of virus assembly depends on the virus type and appears to be governed by the sorting patterns of the respective envelope proteins (52). It has long been known, for example, that in polarized epithelial cells, influenza virus hemagglutinin (HA) and neuraminidase traffic to the apical surface, whereas VSV-G traffics to the basolateral surface (61, 73). Although in many families of enveloped viruses assembly occurs at the plasma membrane (72), some viruses bud from intracellular membranes such as the Golgi cisternae (bunyaviruses) or the inner nuclear membrane (herpesviruses); these are subsequently released after deenvelopment and reenvelopment at the plasma membrane or at post-Golgi vesicles (22).

The notion that viral assembly is a consequence of specific interactions between envelope and core proteins at distinct cellular locations makes the well-documented phenomenon of pseudotyping hard to explain (83). Pseudotyping is the result of a mixed infection of a host cell with two different enveloped viruses which results in the production of virus progeny in which the genome of one virus is packaged within both its own envelope protein and the envelope protein of the second virus (12, 28, 30, 82). The phenomenon has been observed not only between RNA viruses (12, 82), which tend to be similar in size and internal organization, but also between DNA and RNA viruses, which are structurally highly dissimilar (1, 30). Since its first description about 30 years ago (12, 28, 30, 82), pseudotyping has proven a very useful tool for extending the host range of viral vectors (43, 51, 69) and/or improving the physical characteristics of viral particles (9). The molecular mechanism for pseudotyping is, however, poorly understood.

A long-standing observation related to pseudotyping has been that molecules from the host cell can be selectively incorporated into viral particles (4, 10, 34, 35, 41, 63, 71, 81, 84). The list of such molecules is extensive, but an important finding has been that many of the molecules that have been found to be incorporated into one virus have also been found in others. Glycosyl phosphatidylinositol (GPI)-anchored proteins are recurrently represented in the list (10, 41, 63), and many such proteins have been found to localize to plasma membrane lipid microdomains (7, 11). For example the raft-localized (58) CD4 molecule (67, 68, 71) and the associated human immunodeficiency virus (HIV) coreceptor CXCR4 (68, 71) can be incorporated into viral envelopes at a level high enough to allow targeting of HIV-infected cells.

Lipid microdomains comprise both caveolae and related structures (19, 47), variously called glycosphingolipid-enriched microdomains, detergent-insoluble glycosphingolipid-enriched domains (DIGs), or rafts (2, 7, 27, 33, 53, 70). They are resistant to nonionic detergents at low temperatures and can be physically separated from the bulk of soluble membranes by isopycnic sucrose gradient centrifugation (7, 13). In this report, we explore the potential relationship between pseudotyping and lipid rafts by testing the partitioning of Env proteins of different virus families into detergent-resistant membrane fractions. We report that gangliosides, GPI-anchored surface proteins, and intracellular signal transducing molecules are targeted to viral particles. Dispersion of lipid rafts leads to inhibition of virus production.

MATERIALS AND METHODS

Cell lines and reagents.

293T cells, a laboratory isolate expressing simian virus 40 large T antigen, the Epstein-Barr virus (EBV) producer cell line B95-8 (American Type Culture Collection [ATCC]), and the human T-lymphotropic virus type 1 (HTLV-1)-releasing HuT 102 cell line (ATCC) were maintained in Iscove modified Dulbecco medium (Sigma Chemical) plus 10% fetal calf serum (HyClone). Moloney murine leukemia virus (MoMLV) gag-pol-env, gag-env, and viral env cDNAs were expressed using the mammalian expression vector pEAK12 (Edge Biosystems, Gaithersburg, Md.). Retroviral constructs were created in the pMFG (57) derivative pMMP412 (a generous gift from Richard Mulligan). Human placental alkaline phosphatase lacking the C-terminal GPI anchoring sequence was amplified from a placenta cDNA library (B. Seed, unpublished data) and inserted into the expression vector pEAK12 to generate plasmid pEAK12.SEAP. The viral envelope and the CD4, CD14, CD59, and CD99 cDNAs were PCR amplified without their leader sequences and inserted downstream of a CD5 leader HA tag cassette into pEAK12. The following DNA sequences served as a source for PCRs and constructions: syngp160mn for gp160 (24), pΨ⁻ for MoMLV Env, pMD.G (51) for VSV-G (kindly provided by Richard Mulligan), and pDP122 for influenza virus HA (ATCC). Monoclonal antibodies (MAbs) and antisera were as follows: anti-HA (clone 12CA5; Boehringer Mannheim), anti-CD4 (rabbit polyclonal antibody; NIH AIDS Reference Resource and Reagents Program, Rockville, Md.), anti-CD14 (sheep polyclonal antibody AB383; R&D Systems), anti-CD59 (clones MEM43 and MEM43/5; a generous gift from Vaclav Horejsi, Prague, Czech Republic), anti-CD71 (anti-transferrin receptor [TfR] MAb M-A712; Pharmingen, Palo Alto, Calif.), CD99 (clone 3B2/TA8; kindly provided by Otto Majdic, Institute of Immunology, Vienna, Austria), anti-HTLV-1 core matrix protein (MA) (clone P2H9-E9-B7; Chemicon, Temecula, Calif.), MLV sera 74S-454, 78S-282, 80S-013, and 81S-262 (all goat polyclonal antibodies; NCI Biological Carcinogenesis Branch Repository), MLV p30^Gag capsid protein (CA) MAb (rat clone R187; ATCC), EBV envelope gp350/250 MAb (2L 10; Advanced Biotechnologies Inc., Columbia, Md.), c-Src, Grb-2, Vav, protein kinase C-α and FAP-1 rabbit or goat antisera (Santa Cruz Biotechnology), mouse pan-Ras MAb (Ab-3; Oncogen Research Products), mouse Shc MAb (Transduction Laboratories), and mouse phospholipase C-γ MAb (Biodesign). Biotinylated cholera toxin fragment (BCTx) was purchased from Sigma, and horseradish peroxidase (HRP)-conjugated secondary antisera were purchased from DAKO Corp. Influenza virus virus (H1N1) strain A/Taiwan/1/86 and parainfluenza virus strain Sendai (produced in allantoic fluid) were obtained from Advanced Immunochemical Inc., Long Beach, Calif. EBV was produced from B95-8 cells (ATCC), and herpes simplex virus type 1 (HSV-1) strain hRR3 was produced from Vero cells (kindly provided by Deb Schuback, Molecular Neurogenetics, Massachusetts General Hospital, Boston, Mass.). MoMLV was produced as described below.

Cell lysis and fractionation.

DIG fractions were prepared essentially as described elsewhere (77). Briefly, 293T cells were transfected with the indicated mammalian expression plasmids and retroviral vectors (54); 48 h later, cells were lysed in a 1% Triton X-100 containing buffer. Subsequently, lysates were Dounce homogenized, cleared from particulate material by centrifugation at 4°C and 100 × g for 10 min, mixed with an equal volume of 80% (wt/vol in MES-buffered saline) sucrose and placed in an SW55 centrifuge tube (Beckman). Samples were then overlaid with 2 ml of 30% sucrose followed by 1 ml of 5% sucrose and spun at 200,000 × g and 4°C for 16 to 18 h. The gradient was collected in 0.5-ml steps from the top to the bottom. Plasma membranes were prepared as described previously (77). Protein concentrations were determined by the detergent-insensitive DC protein assay (Bio-Rad, Hercules, Calif.) according to the manufacturer's recommendations. Aliquots of 20 μl of individual fractions were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on 4 to 20% gradient gels. Proteins were transferred to polyvinylidene difluoride (PVDF) membranes (Immobilon-P; Millipore, Bedford, Mass.) and subjected to Western blotting using the indicated reagents.

Virus production.

Viral supernatants were harvested from transiently transfected 293T cells (54) 2 days after transfection. Cellular debris was removed by centrifugation at 100 × g for 10 min, followed by filtration through 0.45-μm-pore-size syringe filters (Gelman Sciences). Virus was concentrated by centrifugation through a 20% (wt/vol in phosphate-buffered saline [PBS]) sucrose cushion at 4°C (141,000 × g for 1 h) and resuspended in a small volume of PBS. For some experiments CD4-pseudotyped viral particles were enriched by immunoprecipitation with microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany).

Inhibition of budding.

One day after transfection with pEAK12.MLV, pMMP412.GFP, and pEAK12.SEAP, 293T cells were harvested and replated into 24-well plates (150 × 10³ cells/well) in medium alone or medium containing β-methyl cyclodextrin (β-mCD; stock solution, 50 mM in H₂O; Sigma) or nystatin (stock solution, 3.75 mg ml⁻¹ in methanol; Sigma). Some experiments were performed with cells which had been preincubated with lovastatin (Merck) (prepared as a 4-mg ml⁻¹ solution in H₂O and activated as described previously [18]) for 12 h, and then incubated with lovastatin plus β-mCD for 12 h. Control cells were incubated with the solvents alone at the corresponding dilutions. Reverse transcriptase (RT) activity in the supernatants was determined 12 h after addition of drugs as described by Goff et al. (21). Addition of drugs to mock-treated virus supernatants at the highest concentration of each drug directly before the RT assay did not alter RT activities. After supernatants had been harvested for determination of RT activity, cells were recultured in fresh Iscove modified Dulbecco medium medium for 12 h, and cell viability was determined by propidium iodide exclusion and analyzed by flow cytometry on a Epics XL-MLC flow cytometer (Coulter). Secretory alkaline phosphatase (SEAP) activity in the medium was determined in a colorimetric microplate assay with p-nitrophenyl phosphate as the substrate.

The influence of cholesterol depletion on the raft-dependent and -independent interactions of autologous and heterologous envelopes with MoMLV core was determined as follows. 293T cells were transfected with pMMP412.GFP, pEAK12.SEAP, and pEAK12.MLV(gag-pol-env) or with pEAK12.MLV(gag-pol) and pEAK12.VSV-G and treated with β-mCD and/or lovastatin. Supernatants were harvested, diluted 1:20 with fresh medium containing Polybrene at a final concentration of 8 μg/ml, and added to NIH 3T3 cells (ATCC) at 3 × 10⁴ cells/well in 24-well tissue culture plates. Cells were transduced by centrifugation at 670 × g for 1 h in a Sorvall RT7 centrifuge. Medium was replaced after 12 h, and the cells green fluorescent protein (GFP) fluorescence was analyzed 3 days later by flow cytometry.

Confocal microscopy.

293T cells were seeded at approximately 20% confluence onto tissue culture-treated coverslips (Fisher) and transfected the next day with 1 μg of a mammalian expression vector (pEAK12) containing the indicated HA-tagged versions of envelope proteins or membrane markers. After 36 h, cells were washed twice in cold PBS containing 1% borine serum albumin and then stained for 45 min at 4°C with 2 μg of either anti-HA MAb 12CA5 (Boehringer Mannheim), anti-TfR MAb M-A712 (CD71; PharMingen), or EBV gp350/250 MAb 2L10 (Advanced Biotechnologies) per ml as indicated and simultaneously with 4 μg of biotinylated CTx (Sigma) per ml. After two washes in PBS plus 1% bovine serum albumin, cells were incubated at 4°C in the presence of Oregon Green-conjugated anti-mouse immunoglobulin G antibody (2 μg ml⁻¹) per ml and streptavidin coupled to the Texas red fluorochrome (Molecular Probes) (2 μg ml⁻¹) for 45 min. Patching of the stained membrane components was carried out at 37°C for 30 min as described elsewhere (25). Subsequently, cells were fixed in 4% paraformaldehyde for 5 min and then incubated in methanol at −20°C for 5 min (25). Preparations were mounted in a mixture of Mowiol 4-88 (Calbiochem) and glycerol and analyzed using the 488- and 568-nm bands of argon and argon-krypton lasers, respectively, of a Leica (Heidelberg, Germany) DMRE confocal microscope.

RESULTS

Viral Env proteins of various virus families are targeted to detergent-insoluble lipid rafts in producer cells.

To analyze whether lipid rafts of the cell membrane play a critical role in the assembly and budding of enveloped viruses, we first tested Env proteins of different virus families for their partitioning into detergent-insoluble lipid rafts. Figure 1 shows that the virus surface receptors of ecotropic MoMLV, HIV (49), VSV, and influenza virus (64) appear in the detergent-resistant fractions of cell lysates. The reasons for the varying efficiencies of incorporation into the detergent-resistant sucrose gradient fraction are not fully understood; they may include protein-specific factors that influence the ability of the envelope to assort with lipid during the extended fractionation period (27) or may reflect the influence of unterminated metabolic function, such as depalmitoylation of acylated proteins following cell lysis. The raft lipid constituents themselves may influence the durability of association; for example, Harder et al. (25) have reported VSV-G colocalization with raft-like basolateral structures in copatching experiments in intact cells but not in biochemical experiments. In addition, the ratio of the fraction of envelope in the plasma membrane pool compared to intracellular pools may vary from envelope to envelope. To show the distribution of Env proteins in all of the relevant cellular compartments, we analyzed whole postnuclear supernatants of Triton X-100-lysed cells and loaded equal volumes of each gradient step in lanes 1 to 9 in Fig. 1A to D. Because detergent-resistant membranes represent only a small fraction of the whole plasma membrane, the relative concentration of Env proteins within the detergent-insoluble fractions (i.e., fractions 2 and 3) is significantly higher than the relative concentration of Env proteins within the soluble membrane pool (fractions 5 to 9).

FIG. 1.

Envelope proteins of different virus families are targeted to DIGs and become incorporated into MoMLV particles. 293T cells expressing HA-tagged forms of Env molecules were lysed in buffer containing 1% Triton X-100, and fractionated by flotation on a 5 to 40% sucrose gradient; equal amounts of each fraction were blotted. The bulk of detergent-insoluble material is concentrated in fractions 2 and 3, whereas soluble material ends up in fractions 5 to 9. Membranes were probed with antibody to the HA tag; arrows indicate the positions of mature Env of MoMuLV (A), HIV (B), VSV (C), and influenza virus (flu; D). Precursors and multimers of Env proteins migrate with reduced mobility. (E) Incorporation of recombinant envelope proteins from different virus families in MoMLV particles. 293T cells were transfected with expression plasmids encoding MoMLV gag-pol and HA-tagged env cDNAs and pMMP412.GFP (retroviral vector). Viral particles were passed through 0.45-μm-pore-size filters and concentrated by centrifugation through a 20% sucrose cushion. Viral particles were blotted and probed with antibody to the HA tag. Mock viral particles were prepared from cells expressing an untagged form of CD4 instead of env. Sizes are indicated in kilodaltons.

Subsequent analyses of viral supernatants revealed that all viral surface receptors tested were able to efficiently pseudotype MoMLV particles (Fig. 1E). Surprisingly, also gp160 was incorporated into MoMLV at detectable levels, although it has been reported that the full-length intracytoplasmic tail of the HIV transmembrane glycoprotein is disadvantageous for functional pseudotyping of MoMLV particles (39, 69).

Viral Env proteins colocalize with the lipid raft marker ganglioside G_M1 in vivo.

To provide another measure of raft affiliation not dependent on persistent association through an extended fractionation period, as well as to establish the in vivo distribution of lipid rafts and envelope proteins at the cell surface, we performed colocalization experiments in intact cells by double-immunofluorescence techniques and confocal microscopy. Patching (25) of viral Env proteins and the lipid raft marker ganglioside G_M1, which was detected with CTx, demonstrated the colocalization of G_M1 with viral Env proteins of MoMLV, HIV, VSV, and EBV, as well as with GPI-anchored molecules (7) (Fig. 2A). In contrast, membrane proteins not associated with lipid rafts, such as CD99 or TfR (CD71), did not coassociate with G_M1 patches (25) (Fig. 2A). The extent of colocalization of VSV-G with CTx is somewhat more complete in our experiments than the colocalization of PLAP with VSV-G reported by Harder et al. (25). In part this difference might be attributable to the different behaviors of gangliosides and GPI-anchored molecules. In other work, we have shown that exogenous gangliosides cause the internalization of some raft constituents in T cells, notably CD4, while having less or no effect on GPI-anchored molecules (77).

FIG. 2.

(A) Envelope proteins of RNA and DNA viruses colocalize with G_M1 in vivo. Shown is confocal microscopy of membrane proteins and gangliosides. The upper panels show the binding of MAbs to envelopes or surface receptors in green, the middle panels show binding of cholera toxin to G_M1 in red, and the lower panels show the overlay. 293T cells grown on coverslips were transfected with expression plasmids encoding HA-tagged viral envelopes or cell surface proteins. The marmoset cell line 95B-8 was used to study the EBV envelope protein gp350/250. Cells were stained with HA-specific MAb and biotinylated CTx. Endogenous TfR was stained with MAb M-A712; the EBV envelope protein was stained with MAb 2L10. The subcellular localization of the bound primary reagents was visualized with Oregon Green-labeled anti-mouse immunoglobulin G and Texas red-conjugated streptavidin on a Leica confocal microsope (see Materials and Methods). (B) RNA and DNA viruses incorporate G_M1 into their envelopes. Viral preparations separated by SDS-PAGE (4 to 20%, acrylamide) and then transfered onto PVDF membranes were probed with biotinylated CTx and visualized by incubation with streptavidin coupled to HRP. The position of the dye front (df) is indicated

Viral membranes contain the lipid raft marker ganglioside G_M1.

Consistent with these results, purified retroviruses, parainfluenza virus, influenza virus, EBV, and HSV were all found to contain G_M1 (Fig. 2B), indicating that lipid rafts not only are the sites of concentration of viral constituents but also contribute to the composition of the viral membrane.

GPI-anchored and transmembrane lipid raft-resident cell surface proteins are targeted to viral particles.

Do other raft components appear in viral particles? Lipid raft proteins such as the GPI-anchored molecules CD59 (Fig. 3A), CD55 and CD14 (data not shown), CD4 (Fig. 3C), and gangliosides (Fig. 3D) are targeted to MoMLV particles (Fig. 3G, left), whereas other host cell surface molecules that are excluded from lipid rafts, such as CD99 (MIC2/E2) (Fig. 3B), are excluded from MoMLV particles (Fig. 3G, left). Exclusion appears not to be related to abundance or viral perturbation of expression, since CD59 and CD99 are approximately equally prevalent in the producer cell line (Fig. 3E, top), and their expression does not change upon transfection with retroviral expression constructs (Fig. 3E, bottom). Supernatants derived from control-transfected cells did not yield significant amounts of the copackaged molecules, excluding the possibility that constitutively produced microvesicles (20) might have contributed to the particle-associated proteins detected here (Fig. 3G, lanes labeled −).

FIG. 3.

Specific incorporation of lipid raft-resident molecules into MoMLV particles. Shown is localization of CD59, (A), CD99 (B), CD4 (C), and G_M1 (D). Detergent-resistant membranes were prepared by sucrose gradient centrifugation and probed with anti-CD59, -CD99, and -CD4 antibodies or with biotinylated CTx. Arrows indicate the positions of specific bands. Migration patterns of prestained molecular mass standards are indicated in kilodaltons. (E) Relative expression of CD59 and CD99 on producer cells was determined by flow cytomety before (top) and 48 h after (bottom) transfection of producer cells with retroviral vectors. Overlay histograms represent cells incubated with a nonbinding control MAb (thin line), CD59 MAb MEM43 (thick line), and CD99 mAb 3B2/TA8 (dotted line). (F) Gag proteins of MoMLV and HTLV-1 localize to lipid rafts. Detergent-resistant membranes were prepared by sucrose gradient centrifugation from cells transfected with pEAK12.MLV and the retroviral vector pMMP412.GFP (top and middle) or from HuT 102 cells which are releasing HTLV-1 (bottom). The MoMLV Gag proteins were visualized by incubation of membranes with the Gag-specific MAb R187 (top) or serum 74S-454 (middle). Positions of Pr65^Gag, CA (p30^Gag) and MA (p15^Gag) are indicated. HTLV-1 Gag proteins were detected with the MA-specific MAb P2H9-E9-B7. The positions of Pr55^Gag and MA (p19^Gag) are indicated. (G) Viral particles bear lipid raft-resident host cell molecules. Virus supernatant was collected from 293T cells transfected with pEAK12.MLV (gag-pol-env), pMMP412.GFP (retroviral vector), and pEAK12.CD4 (lanes labeled +) or pEAK12.CD105 (lanes labeled −). Viral particles were purified as described (in Materials and Methods). Products were resolved by SDS-PAGE (4 to 20% polyacrylamide) and transferred onto PVDF membranes; individual strips were probed with MAbs, antisera, or toxin fragments, the binding of which was visualized by HRP-labeled secondary reagents. Positions of prestained molecular mass standards are indicated in kilodaltons.

The viral core proteins of MoMLV and HTLV-1 are concentrated in the detergent-insoluble lipid raft fraction of producer cells.

The gag-encoded core protein MA of B, C, and D type retroviruses is myristoylated (26), and virus particle formation and budding are inhibited by mutations that compromise the N-terminal myristoylation sequences of MoMLV (56) and HIV (23). We found that the viral core protein precursor Pr65^Gag of MoMLV as well as its proteolytic fragments, MA (p15^Gag) and CA (p30^Gag), are targeted to the detergent-insoluble fraction of packaging cells (Fig. 3F). Because only Pr65^Gag and p15^Gag bear myristic acid, the major core protein p30^Gag detected in these fractions is likely stably associated with p15^Gag or Pr65. The core protein precursor Pr55^Gag of a second C-type retrovirus, HTLV-1, is also strongly associated with the detergent-insoluble fraction of HuT 102 producer cells (Fig. 3F). Furthermore, also in HTLV-1-infected cells and in analogy to MoMLV, the mature MA (p19^Gag) remains associated to the lipid raft fraction of producer cells, although less stably than the precursor protein.

Intracellular signaling molecules c-Src, Ras, and Grb-2 are incorporated into virus particles.

One interpretation of these findings is that intracellular molecules that associate with membrane lipid rafts (55, 77, 78) could also be incorporated in virions. Consistent with this, we found that several signal transduction molecules that are known to be associated with lipid rafts could also be detected in purified virions (Fig. 3G, right), whereas other proteins that appear in bulk membrane fractions (Table 1) are excluded (Fig. 3G, right). Interestingly, the adapter protein Grb-2, which associates only with activated signaling centers in lipid rafts, could also be detected in virions, suggesting that the process of virus assembly may induce or mimic receptor activation.

TABLE 1.

Intracellular signaling molecules targeted to MoMLV^a

Molecule	Fraction
	Virus	DIG	Soluble
c-Src	+	++	−
Ras	+	++	+
Grb-2	+	−	++
Shc	−	+/−	++
Protein kinase C-α	−	+/−	++
Phospholipase C-γ	−	−	+
Vav	−	−	+/−
FAP-1	−	−	+

^aViral particles and fractions of producer cells were generated as described in Materials and Methods. Preparations were resolved by SDS-PAGE, transferred to PVDF membranes, and probed with MAbs or antisera recognizing the indicated molecules. DIG and soluble material correspond to fractions 2 and 9, respectively. −, no detectable signal; +/−, weak positive signal; +, positive signal; ++, strong positive signal. 

Dispersion of lipid rafts in producer cells inhibits virus particle formation.

The preceding establishes that lipid rafts are associated with viral assembly or reenvelopment. To explore whether they are functionally important for virus formation, we turned to agents that compromise lipid raft integrity. Nystatin, a polyene antifungal, has been shown to disperse raft contents in the plasma membrane (62). Lipid rafts can also be disrupted by agents that extract cholesterol from the plasma membrane, such as β-mCD (32), and this action is potentiated when de novo synthesis of cholesterol is blocked by the hydroxymethylglutarats coenzyme A reductase inhibitor lovastatin (31). Under conditions in which the producer cells showed no gross changes in either cell viability (data not shown) or de novo secretion of SEAP, cholesterol dispersion by nystatin or depletion by β-mCD led to a clear reduction in supernatant RT activity, a measure of viral titer (Fig. 4A and B). RT activity of existing viral particles was unaffected by coincubation with these agents (data not shown). Inhibition of endogenous cholesterol synthesis by lovastatin further diminished virus particle formation (Fig. 4C) without altering the rate of viral protein synthesis in the packaging cell line (Fig. 4D), but with a modest reduction in SEAP secretion.

FIG. 4.

Inhibition of viral particle formation by nystatin, β-mCD, and lovastatin. At 24 h after transfection with pEAK12.MLV (gag-pol-env), pMMP412.GFP (retroviral GFP expression construct), and pEAK12.SEAP (SEAP), 293T cells were trypsinized and replated into 24-well plates. Cells in panel C were preincubated with lovastatin (4 μg ml⁻¹) for 12 h. Subsequently, individual wells were incubated with the indicated concentrations of nystatin (A), β-mCD (B), or lovastatin (4 μg ml⁻¹) plus β-mCD (C). Supernatant was accumulated for the next 12 h, and the RT activity and SEAP activity released were determined by standard methods (see Materials and Methods). RT activity levels and SEAP secretion are expressed as percentages of values for the solvent-treated controls. (D) Expression levels of retroviral p30^Gag in producer cells of the lovastatin-plus-β-mCD-treated samples shown in panel C.

Heterologous Env proteins are more sensitive to cholesterol depletion than their autologous relatives.

To clarify whether autologous Env proteins, which may have raft-independent mechanisms for interaction with their cognate core proteins, are more stably associated with the autologous viral core than heterologous Env proteins, we produced two types of MoMLV core particles: one expressing MoMLV Env, and the other expressing VSV-G.

Figure 5 shows that the transducing titers of MoMLV particles which express the heterologous Env protein VSV-G are more severely affected by cholesterol depletion than MoMLV particles expressing the ecotropic Env protein. For MoMLV Env-expressing particles, the drop in transducing titers is comparable to the degree of inhibition of RT activity secreted into the supernatant (Fig. 4) and can thus be explained as a general reduction of the number of viral particles. The transducing titers of VSV-G-pseudotyped MoMLV particles were more sensitive to β-mCD treatment and were diminished to nearly zero when producer cells were pretreated with lovastatin.

FIG. 5.

β-mCD and/or lovastatin treatment depletes heterologous Env proteins more efficiently than autologous Env proteins. At 24 h after transfection with pMMP412.GFP (retroviral construct), pEAK12.SEAP (SEAP), and pEAK12.MLV (gag-pol-env) or with pEAK12.VSV-G plus pEAK12.MLV (gag-pol), 293T cells were trypsinized and replated into 24-well plates. A fraction of the cells was preincubated with lovastatin (4 μg ml⁻¹) for 12 h. Subsequently, individual wells were incubated for 12 h with the indicated concentrations of β-mCD (open symbols) or lovastatin (4 μg ml⁻¹) plus β-mCD (closed symbols). Supernatants were harvested, diluted 1:20, and used to transduce NIH 3T3 fibroblasts, which were analyzed for GFP fluorescence after 3 days of culture. The graph shows the mean fluorescence intensity of NIH 3T3 cells transduced with VSV-G-pseudotyped MoMLV particles (circles) or with MoMLV Env-expressing MoMLV particles (squares).

DISCUSSION

The above data support the view that enveloped viruses exploit the natural sequestration of membrane molecules that lipid rafts provide. In the course of viral assembly, other molecules that localize to lipid rafts may be copackaged as adventitious contaminants. Pseudotyping, an important biological principle with a long and distinguished history, appears to be a consequence of this common mechanism for enveloped virus formation. We have presented evidence that members of the C-type retrovirus, lentivirus, paramyxovirus, orthomyxovirus, herpesvirus, and gammaherpesvirus families assemble or reassemble their envelope in the lipid rafts of the plasma membrane. To these findings can be added the recent reports that HIV (51), fowl plague virus (64) (an orthomyxovirus), and measles virus (40) (a paramyxovirus) interact with lipid rafts during budding from the plasma membrane. Utilization of rafts as a structural organizing principle accounts for the otherwise bizarre finding that herpesviruses can pseudotype VSV (29, 30) and vice versa (1). Although it is unclear what features confer a selective advantage on viruses that undergo pseudotyping, it is tempting to speculate that pseudotyping promotes viral spread and may facilitate the adaptation of viruses to new hosts.

The mechanisms whereby viral core and envelope proteins arrive at rafts are likely heterogeneous, but a common theme is lipid acylation. Env molecules of many viruses have been shown to become posttranslationally esterified on intracellular membrane-proximal cysteine(s) with fatty acids (48, 59, 65, 79, 80), and core proteins are typically myristoylated (26). Partitioning of molecules with covalently linked saturated acyl chains into liquid-ordered phase domains has been suggested to be an important mechanism for the targeting of proteins to lipid rafts (42, 85). Although VSV-G is palmitoylated, it has often been assumed that the basolateral targeting of this virus is inconsistent with raft localization (6). However, an important recent discovery has been that of Harder et al. (25), who showed that VSV-G colocalizes with raft-like basolateral structures. VSV-G protein in the presence of VSV membrane protein enhances the formation of sphingomyelin- and cholesterol-rich lipid domains in unilamellar vesicles (37, 38), and VSV itself is enriched in sphingomyelin and cholesterol (75).

How do pseudotypes form between viruses that bud from opposite surfaces of polarized epithelium? The answer appears to lie in two observations: first, that membrane lipid rafts can be found both basolaterally and apically (3, 5, 25); and second, that the appearance of pseudotypes is closely related to loss of polarity (12, 60). Early in the mixed infection of a polarized culture, coinfecting viruses with opposite segregation emerge separately; but following the onset of cytopathic effects, pseudotypes rapidly appear (12, 60). The simplest explanation for this behavior is that pseudotypes are dependent on the functional equivalence of basolateral and apical lipid rafts and that the sorting pathway that leads to polarized viral release cannot distinguish between these lipid microdomains—if they do remain different—once polarity is lost.

The results presented here demonstrate that protein and lipid constituents of lipid rafts are incorporated into virions in the course of viral assembly. The latter may have important consequences for viral physiology. For example, the lipid composition of lipid rafts might permit a smaller radius of membrane curvature, facilitating budding, or might confer stability on the particles they comprise. Whether there are physiological consequences of the presence of both extracellular and intracellular proteins of lipid rafts in viral particles is not known, but it seems plausible that elements of the host response to viral pathogens may be based on the association of enveloped viruses with these domains.

The findings here also have practical implications, providing a general framework for the predictable manipulation of the components of a broad variety of enveloped viruses and accounting for the unusual effectiveness of CD4 as a retargeting molecule in strategies aimed at delivery of viral therapeutics to HIV-infected cells (17, 81). In addition, the identification of lipid raft components by viral colocalization might become a useful experimental tool for identifying signal transduction intermediates.