Nonpolarized distribution of glycosylphosphatidylinositols in the plasma membrane of polarized Madin-Darby canine kidney cells.

Glycosylphosphatidylinositols (GPIs) are ubiquitous in eukaryotes and serve to anchor a variety of proteins to the exoplasmic leaflet of cellular membranes. GPIs are synthesized in the endoplasmic reticulum (ER), in excess of the amount needed for protein modification. The fate of the excess GPIs is unknown, but they may be retained in the ER, transported to other membranes, and/or metabolized. In relation to this problem, we were interested in determining whether GPIs were transported to the plasma membrane and whether, like GPI-anchored proteins, their presence was confined to the apical plasma membrane domain in polarized epithelial cells. Polarized Madin-Darby canine kidney epithelial cell monolayers were incubated with [H]mannose or [H]ethanolamine to label GPIs and then infected with enveloped viruses. We used influenza virus (flu) and vesicular stomatitis virus (VSV) for these experiments as these viruses are assembled at the cell surface and acquire their envelope lipids from the plasma membrane. Furthermore, flu and VSV bud specifically from the apical and basolateral plasma membrane domains, respectively. Flu and VSV were isolated from the apical and basolateral media, respectively, and subjected to lipid analysis. Radiolabeled GPIs were found in both viruses. Moreover, the membrane concentration of GPIs (i.e. GPI radioactivity normalized to membrane mass) in the two viruses was essentially the same. These observations suggest that (i) non-protein-linked GPIs are located at the plasma membrane; (ii) since GPIs are synthesized in the ER, they must be transported from the ER to the plasma membrane; and (iii) transport of non-protein-linked GPIs is not influenced by the sorting processes that target GPI-anchored proteins exclusively to the apical plasma membrane.

GPIs are made in excess of the amount needed for protein modification, and eukaryotic cells contain significant pools of non-protein-linked GPIs (10 5 -10 7 molecules/cell depending on GPI structure and cell type) in addition to GPI structures (anchors) covalently linked to protein. Evidence from different experimental systems suggests that a fraction of the nonprotein-linked GPIs may exit the ER and relocate to other organelles, including the plasma membrane (PM) (7)(8)(9). Indeed, some members of a family of GPI-related glycolipids (glycoinositol phospholipids) in Leishmania parasites are found at the cell surface, where they are accessible to antibodies and where they can be modified by exogenously added glycosyltransferases (1,10).
In renal and intestinal epithelial cell lines such as Madin-Darby canine kidney (MDCK) and Caco-2, GPI-anchored proteins are expressed exclusively at the apical PM domain (11)(12)(13)(14). It has been proposed that GPI anchors are dominant apical sorting signals for GPI-anchored proteins, acting through biophysical interactions with glycosphingolipid (GSL) clusters in the trans-Golgi network of epithelial cells (11,15,16). In contrast, very little is known about the subcellular distribution and transport of non-protein-linked GPIs. We have studied the composition and distribution of non-protein-linked GPIs in MDCK cells to investigate if the GPI structure per se contains intracellular sorting information, reflected in a polarized distribution of non-protein-linked GPIs on the epithelial cell surface. By analysis of enveloped RNA viruses that bud from the two cell-surface domains of MDCK cells, we show that GPIs are indeed present in the PM, but that they do not display a polarized distribution between the apical and basolateral domains. In the framework of models for epithelial cell polarity (15)(16)(17)(18), intracellular lipid transport (17,19), and GPI biosynthesis (3)(4)(5), we suggest that non-protein-linked GPIs may be located in the cytoplasmic leaflet of the PM lipid bilayer.
Cell Culture and Glycolipid Labeling-MDCK II cells were grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin in an atmosphere of 5% CO 2 at 37°C. For polarity experiments, 5 ϫ 10 7 MDCK cells were seeded on 75-mm Transwell filters (0.4-or 3.0-m pore size; Costar Corp., Cambridge, MA), and monolayers of cells were used for experiments 3 days after plating.
Virus Infection and Purification-Influenza virus strain WSN (flu) and vesicular stomatitis virus (VSV; wild-type) were used for infections. Virus stocks were prepared as described (20). Inoculation of radiolabeled MDCK cells on 75-mm Transwell filters (0.4-m pore size) with flu was performed for 1 h at 37°C with a multiplicity of infection of 10 -20 plaque-forming units/cell in 3.5 ml of Dulbecco's modified Eagle's medium containing 0.2% bovine serum albumin and 100 g/ml DEAEdextran (infection medium), added to the apical surface. 6.5 ml of infection medium without virus was added to the basolateral side. After removal of the inoculation medium, 6 ml of infection medium was added to the apical surface (supplemented with 2.5 g/ml L-1-tosylamido-2phenylethyl chloromethyl ketone-treated trypsin) and 7 ml to the basolateral side of the monolayer, and infection was continued for 8 h at 37°C. MDCK cells on 75-mm Transwell filters (3.0-m pore size) were inoculated for 1 h at 37°C with 10 -20 plaque-forming units of VSV/cell in infection medium (3.5 ml apical and 6.5 ml basolateral). The inoculation medium was replaced with infection medium (6 ml apical and 7 ml basolateral). To minimize cytopathic effects of VSV infection that result in the loss of polarity, the 8-h incubation was conducted at 32°C (21). Under these conditions, the transepithelial resistance of mockinfected cells remained constant at 110 ohms⅐cm 2 . After 4 and 8 h, the transepithelial resistance of flu-infected cells was 110 and 100 ohms⅐cm 2 , respectively, and for VSV-infected cells, the values were 90 and 80 ohms⅐cm 2 , respectively. After 10 h of infection with VSV, the transepithelial resistance dropped to 65 ohms⅐cm 2 , and leakage of apical medium to the basolateral side was observed. After infection for 8 h, apical and basolateral media were collected, and the monolayers were rinsed for 5-10 min at room temperature with 4 ml of infection medium apically and 3 ml basolaterally by gentle shaking on a rocker to optimize release of remaining viruses sticking to cells or Transwell filters.
Cells on the Transwell filters were scraped in infection medium, pelleted, and taken for glycolipid extraction. Viruses were isolated from the combined apical or the combined basolateral media. First, detached cells and cellular debris were removed from the media by respective centrifugation for 5 min at 1000 rpm and for 25 min at 5000 rpm at 4°C in a Sorvall HS-4 rotor. Viruses in the remaining supernatant were pelleted through a 0.25 M sucrose cushion containing 1 mM Tris-HCl (pH 7.4) and 1 mM EDTA (TE buffer) by centrifugation at 4°C for 3 h at 40,000 rpm in a Beckman SW 41 rotor. The pellet was resuspended in 250 l of TE buffer, layered on top of a 7-52% linear sucrose gradient, and centrifuged at 4°C for 150 min at 30,000 rpm in a Beckman SW 41 rotor. Fractions of 0.5 ml were harvested from the top of the gradient using an Auto Densi-Flow IIc pump (Buchler, Lenexa, KA), and radioactivity was measured in 25-l samples of each fraction by liquid scintillation counting. The sucrose density in each fraction was determined by measuring the refractive index. Fractions containing radiolabeled virus were collected, diluted three to four times with TE buffer, and centrifuged at 4°C for 2.5 h at 23,000 rpm in a Beckman SW 41 rotor. The resulting virus pellets were resuspended in 0.25 ml of TE buffer or directly taken for lipid extraction.
Lipid Extraction, Analysis, and Quantitation-[ 3 H]Mannose-labeled GPIs in cells and viruses were directly extracted with chloroform/ methanol/water (10:10:3, by volume), whereas [ 3 H]ethanolaminelabeled samples were first extracted two to four times with chloroform/ methanol (2:1, v/v) to remove [ 3 H]PE prior to extracting GPIs with chloroform/methanol/water (10:10:3, by volume). Extraction of GPIs from virus pellets in Ultra-Clear TM ultracentifuge tubes (Beckman Instruments), which are not resistant against chloroform, was performed by resuspending the pellets in 300 l of distilled water and transferring the suspension to a glass tube. The centrifuge tube was briefly rinsed with 1 ml of methanol, which was subsequently added to the virus suspension. Following this, 1 ml of chloroform was added while gently vortexing to produce a one-phase mixture. Routinely, 1-3 ml of additional chloroform/methanol/water (10:10:3, by volume) was added to optimize the extraction, which was continued for 1 h at room temperature under repeated gentle mixing. Debris was spun down in an Eppendorf microcentrifuge, lipid extracts were dried using a SpeedVac evaporator (Savant Instruments, Inc., Farmingdale, NY), and extracted products were partitioned between 1-butanol and water. Samples containing 2000 -5000 cpm were dried in a SpeedVac evaporator, dissolved in 15 l of water-saturated 1-butanol, and applied to Silica Gel 60 TLC plates for separation in chloroform/methanol/water (10:10:3, by volume). TLC plates were scanned for radioactivity with a Berthold LB-2842 automatic TLC linear analyzer, and the radioactivity incorporated into individual lipids was determined using the integration software provided with the analyzer in conjunction with liquid scintillation counting.
[ 3 H]Ethanolamine-or [ 3 H]mannose-labeled lipid products were identified by cochromatography (on thin-layer plates) with previously characterized mammalian GPIs (22,23) and by susceptibility to treatment with GPI-specific phospholipase D and nitrous acid (24). Phospholipids on silica TLC plates were visualized by exposure to iodine vapor, identified by cochromatography with standards, and quantified by phosphorus analysis of the appropriate region of silica (25).
GPI-anchored Proteins in Viruses-Experiments to detect GPI-anchored proteins in flu and VSV were performed with MDCK cell clones expressing the chimeric GPI-anchored protein gD1-DAF, consisting of the extracellular domain of the herpes simplex virus glycoprotein D1 and the carboxyl-terminal domain of decay-accelerating factor (DAF) (14). MDCK(gD1-DAF) cells were grown on 100-mm plastic dishes, incubated with 10 mM sodium butyrate for 8 -10 h to induce expression of gD1-DAF, and then inoculated with flu or VSV as described above. Infection was continued for 24 h, and viral and cellular samples were obtained as described above. Samples were dissolved in 500 l of SDS-PAGE sample buffer, and 50-or 100-l aliquots were used for gel electrophoretic analysis. Following transfer to polyvinylidene difluoride membranes, gD1-DAF was probed by Western blotting and chemiluminescence detection (ECL Western blotting kit, Amersham Corp.). Western blotting was performed using either a mouse monoclonal antibody against glycoprotein D1 (Advanced Biotechnologies Inc., Columbia, MD) or rabbit polyclonal antibodies (Dako Corp., Carpinteria, CA). These primary antibodies were used at a 1:2000 dilution in Tris-buffered saline containing 1% non-fat milk. Chemiluminescence detection was performed with the ECL system according to the instructions of the manufacturer (Amersham Corp.).

GPIs in MDCK Cells-MDCK cells were metabolically labeled with [2-3 H]mannose in the presence of tunicamycin.
When glycolipids were extracted from the labeled cells and analyzed by TLC, 8 -11 labeled products were resolved (Fig.  1A). The four most hydrophilic products were identified as GPIs by their sensitivity to GPI-specific phospholipase D (Fig.  1B) and nitrous acid (data not shown) (24). For the purposes of this paper, the structures will be designated GPI-1, GPI-2 (a mixture of two poorly resolved GPIs), and GPI-3, in order of increasing hydrophobicity (Fig. 1A). Cochromatography with mammalian GPI standards indicated that the more hydrophilic component of the GPI-2 doublet was a GPI structure containing 3 phosphoethanolamine residues and that GPI-3 contains 2 phosphoethanolamine residues (22,23). This assignment was supported by the observation that both GPI-2 and GPI-3 could be metabolically radiolabeled via [ 3 H]ethanolamine (data not shown). GPI-1 may be a lyso form of one of the components of GPI-2 or may contain an additional mannose residue; yields of radiolabeled GPI-1 varied between experiments, and in some cases, GPI-1 was not detected. Also in some experiments, an additional GPI structure migrating between GPI-2 and GPI-3 was identified (see Fig. 3, A and B); this may be the lipid identified previously as H7Ј (22). The remaining [ 3 H]mannose-labeled products (running in the upper half of the chromatogram in Fig. 1A) were not sensitive to various tunicamycin pretreatments and were GPIspecific phospholipase D-, nitrous acid-, and mild acid-resistant, but mild base-sensitive. Their identity is currently unknown, but based on their response to the various treatments, it is clear that they are not dolichol-P-monosaccharides, dolichol-PP-oligosaccharides, or GPIs. Work is in progress to identify the nature of these lipids. Following published work on domain-specific biotinylation of cell-surface proteins in MDCK and other polarized cells (11), we investigated whether ethanolamine-containing GPIs could be derivatized at the cell surface by membrane-impermeant amine-reactive probes. Consistent with previous observations on GPI modification under approximately physiological conditions (3,4), these attempts were unsuccessful (data not shown). We therefore decided to use enveloped RNA viruses to investigate the presence of GPIs in the plasma membrane. These viruses obtain their envelope lipids by budding from select membranes in their host cells; flu and VSV bud from the PM, and in polarized MDCK cells, the two viruses bud specifically from the apical (flu) and basolateral (VSV) PM domains (20). Thus, analyses of flu and VSV budded from [ 3 H]mannose-or [ 3 H]ethanolamine-labeled MDCK cells provide an opportunity to investigate whether non-protein-linked GPI structures are present at the PM and whether they display a polarized surface distribution.
Polarized Budding of Flu and VSV in MDCK Cells-Polarized budding of flu and VSV from MDCK cells was confirmed using MDCK monolayers that had been grown for 3 days on large 75-mm Transwell filters. The cells were first incubated for 12 h at 37°C with [ 3 H]galactose to label bulk glycosphingolipids and then infected with flu or VSV. After 8 h, viruses were purified from the apical and basolateral media. In the flu-infected cell samples, 3 H label was quantitatively recovered from the apical medium ( Fig. 2A); no radioactivity was found in the basolateral medium. The majority (65-80%) of 3 H label after VSV infection was found in the basolateral medium (Fig.  2B). For both flu-and VSV-infected cells, the radioactivity recovered in the medium banded at an isopycnic density of 1.18 g/cm 3 (Fig. 2, A-C), identical to that of the [ 35 S]methioninelabeled marker proteins HA2 and VSV-G (Fig. 2C) and intact viruses (21). In mock-infected cells, no release of 3 H label at these densities was observed (Fig. 2C). Therefore, 3 H label recovered in the media after virus infections was specifically incorporated into intact virus particles, representing highly purified plasma membrane samples (21). Analysis of [ 3 H]galactose-labeled GSLs in the two viruses provided another test of the polarity of the MDCK cell monolayers used for our experiments. Viruses budded from [ 3 H]galactose-labeled MDCK cell monolayers were collected from sucrose gradients as in Fig. 2 (A and B) (only basolaterally budded VSV was used for analysis), and GSLs were extracted and analyzed by TLC. As shown in Fig. 3A, there are clear differences in the radiolabeled GSL profiles of flu and VSV. For example, the structures denoted B1, B2, and B3 (comigrating (with 2 phosphoethanolamine residues; the middle mannose is not substituted). The designations GPI-1, GPI-2 (H8 ϩ another poorly resolved GPI), and GPI-3 (H7) are used throughout this paper. EtN, ethanolamine.

FIG. 2. Polarized budding of flu and VSV from the apical and basolateral surface domains of MDCK cells. MDCK cells were seeded on 75-mm Transwell filters, incubated with [ 3 H]galactose (to label GSLs), and infected with flu (A), VSV (B), or no virus (C). Released viruses in the apical (q) or basolateral (E) media were purified and analyzed by centrifugation in 7-52% linear sucrose gradients (A-C).
Gradient fractions (0.5 ml) were collected from the top of the gradients and taken for liquid scintillation counting (A-C, radioactivity). The sucrose density in each fraction was determined by measuring the refractive index (C, sucrose density). As a control, MDCK cells were infected with flu or VSV for 4 h as described under "Experimental Procedures," followed by metabolic labeling for 4 h at 37°C with 150 Ci/ml [ 35 S]methionine. Released viruses were analyzed as described above, and 1.0-ml fractions were monitored for the marker proteins influenza HA2 and VSV-G by immunoprecipitation, followed by SDS-PAGE and densitometry (C, relative intensity percent).
on TLC with G D3 , lactosylceramide, and glucosylceramide, respectively) are found solely or predominantly in VSV, while those marked A1 and A2 (comigrating with G M3 and globosides, respectively) are enriched in flu. Although we did not characterize individual GSL species in any detail, the characteristic differences in GSL expression at the apical (flu) and basolateral (VSV) PM domains are consistent with previous reports (26,27) and confirm that the MDCK cell monolayers exhibit glycolipid polarity under our experimental conditions.

GPIs in Influenza and Vesicular Stomatitis Virus-[ 3 H]
Mannose-labeled MDCK cells were infected with flu or VSV, and viruses were isolated from the apical or basolateral medium as described above. Lipid extraction and TLC analysis showed that both viruses contained radiolabeled GPIs (GPI-1, -2, and -3 plus an additional GPI migrating between GPI-2 and GPI-3 as discussed above) (Fig. 3, B and C) as well as the faster moving unknown lipids identified in Fig. 1A. Similar results were obtained on analysis of viruses derived from infected [ 3 H]ethanolamine-labeled cells (data not shown). The data indicate that free GPIs are present in the PM of MDCK cells. The [ 3 H]mannose-derived radioactivity recovered in lipid extracts of the viruses was 3-4% of that recovered from the infected cell sample, suggesting that at least this fraction of cellular GPIs is located at the PM. For a more accurate calculation of the distribution of GPIs between the PM versus intracellular membranes and between the apical versus basolateral surface domain, it was necessary to obtain estimates of GPI concentration in the viral membranes. This was accomplished by normalizing GPI radioactivity to the amount of PE in each virus sample. PE is an appropriate marker for membrane bulk as (i) it is equally distributed across the membranes of the various compartments in the exocytic and endocytic pathways (17), and (ii) it is uniformly distributed between the apical and basolateral surface domains of polarized MDCK cells. This second point derives from the observation that PE is enriched in the nonpolarized cytoplasmic leaflet of the PM lipid bilayer (17), which in polarized epithelial cells is continuous between the apical and basolateral domains (17,18,21). Therefore, normalization of GPI radioactivity to PE also serves to correct for the differences in budding efficiency of flu and VSV in individual experiments. PE was measured by phosphorus analysis; typically Յ5% of total cellular PE was recovered in the purified viruses (see legend to Table I).
The [ 3 H]mannose-labeled GPI/PE ratio (cpm/nmol) in purified viruses, representing the plasma membrane, was 0 -30% lower than in cells (Table I, "Enrichment"), indicating that the free GPIs are probably not preferentially enriched in the plasma membrane. Moreover, comparison of the GPI/PE ratio in flu versus VSV showed that the GPIs were similarly concentrated in both viruses (polarity values of ϳ1) and hence did not display a polarized distribution in the MDCK cell plasma membrane (Table I, "Polarity").
We also investigated whether GPI-anchored proteins could be recovered in enveloped viruses in a manner reflecting their polarized cell-surface distribution. Although cellular glycoproteins are generally not assembled into enveloped viruses (28), GPI-anchored proteins may be incorporated via the same mechanisms used to recruit plasma membrane lipids into viral envelopes. Indeed, Calafat et al. (28) reported the presence of a GPI-anchored protein, Thy-1 (29), in VSV and murine leukemia virus virions recovered from infected thymoma cells. Since Thy-1 is not found in epithelial cells (30) and as we wished to maximize our chances of detecting GPI-anchored proteins in viruses, we chose to analyze a stably transfected MDCK cell clone expressing high levels of a chimeric GPI-anchored protein, gD1-DAF (14). This cell line has been previously characterized and is known to express gD1-DAF exclusively at the apical cell-surface domain (14). SDS-PAGE and Western blotting of solubilized MDCK(gD1-DAF) cells clearly showed a band corresponding to the 40-kDa gD1-DAF precursor and a smear (50 -55 kDa) corresponding to mature forms of the chimera (data not shown). These bands were not detected in wild-type cells. The gD1-DAF-expressing MDCK cells were then infected with virus (flu or VSV) under conditions chosen to maximize virus yield. Western blotting analysis of purified virus samples recovered from the infected cultures detected none of the gD1-DAF forms, indicating that gD1-DAF is not incorporated into the viral envelopes. Thus, the viral assay is not useful in probing the plasma membrane distribution of GPI-anchored proteins in general. These results contrast with those of Calafat et al. (28) described above. One explanation for these apparently conflicting observations is that relatively small GPI-anchored proteins such as Thy-1 (ϳ24 kDa) may behave essentially as large glycolipids and escape the potential steric constraints or ectodomain interactions that result in the elimination of gD1-DAF (ϳ50 -55 kDa) from virus budding sites (28). DISCUSSION We used enveloped RNA viruses to obtain purified samples of the compositionally distinct PM domains of polarized MDCK cell monolayers in order to assay for the presence of nonprotein-linked GPIs at the cell surface. The observation that particular classes of enveloped viruses bud from specific cellular membranes has been exploited to obtain very pure membrane samples of subcellular compartments (21,31,32). In this way, flu and VSV, which bud from the apical and basolateral PM domains, respectively, have been used to study the phospholipid and GSL composition of the PM domains in [ 32 P]phosphate-labeled (21)

or [ 3 H]galactose-labeled (26) MDCK cells.
Elegant control experiments with non-epithelial cells or MDCK cells under nonpolarized conditions showed that these viruses do not select for the lipids in their envelope, but that their lipid content reflects the composition of the membrane from which they bud (21). The viral assay offers unique advantages for lipid analyses: pure membrane preparations may be obtained without contamination from intracellular membranes, and unlike the amine modification protocols that we attempted without success in preliminary work, information is obtained on lipid constituents present in both leaflets of the parent membrane. The results presented here are based on this assay and show clearly that non-protein-linked GPIs may be found in the PM of polarized MDCK cells and that the GPIs are similarly concentrated in the apical and basolateral PM domains.
The presence of GPIs in flu and VSV and, by implication, in the PM indicates that GPIs are not retained at their site of synthesis in the ER. Consistent with this observation, previous work has shown that GPIs may be found in lipid extracts of rat liver PM preparations (9) and that members of a family of GPI-related glycolipids (glycoinositol phospholipids) are expressed on the surface of the protozoan parasite Leishmania major (10). Since GPIs are synthesized in the cytoplasmic leaflet of the ER (3)(4)(5), they may be transported to the PM via monomeric diffusion or protein-mediated exchange through the cytosol or via a vesicular transfer mechanism. All three transport mechanisms would result in expression of GPIs at the cytoplasmic face of the PM (see below), unless transbilayer movement (flip-flop) occurs. Since differences in lipid composition between the apical and basolateral PM domains are confined to the exoplasmic leaflet of the membrane bilayer (17,18), the nonpolarized distribution of GPIs at the PM (Table I) suggests that GPIs may indeed be located in the cytoplasmic leaflet of the PM. We were unable to confirm this directly by examining the orientation of the various GPI structures in the MDCK cell PM or in purified viruses. The probes (phosphatidylinositol-specific phospholipase C and concanavalin A) that have been used successfully in the analysis of transmembrane orientation of protozoan and mammalian GPIs lacking inositol acyl and side chain phosphoethanolamine groups (3, 4) do not hydrolyze or recognize the mammalian GPIs that we describe. Therefore, it cannot be ruled out that some or all of the nonprotein-linked GPIs in the PM are expressed at the cell exterior, albeit without polarity. This would involve flipping of GPIs into the exoplasmic membrane leaflet of an intracellular organelle such as the ER, followed by vesicular transport without sorting at the trans-Golgi network. The results and hypotheses presented in this paper demonstrate the contrast between the distribution and transport of free GPIs and GPI-anchored proteins. The latter are confined to the exoplasmic leaflet of cellular membranes and are presumably sorted in the trans-Golgi network for selective delivery to the apical PM domain via a vesicular mechanism (11,16). Although we were unable to use our viral assay to confirm the cell-surface polarity of GPIanchored proteins, the exclusive localization of these proteins to the apical plasma membrane domain in MDCK cells has been extensively documented using immunofluorescence (14), domain-selective biotinylation (12,14), domain-selective radioiodination (30), and cell-surface immunoprecipitation (13). in the presence of 50 g/ml tunicamycin and infected with virus as described under "Experimental Procedures". Radiolabeled GPIs in flu (apical) and VSV (basolateral) were extracted, analyzed, and quantified using a TLC linear analyzer. PE was quantified by phosphorus analysis. The amounts of GPI in cells and purified viruses were normalized to the amount of PE and are presented as the ratio GPI/PE (cpm/nmol). The apical to basolateral polarity is calculated by taking the ratio GPI/PE in flu over the ratio GPI/PE in VSV: PE in flu-infected cells, 175 Ϯ 36 nmol, and in flu, 5.8 Ϯ 0.7 nmol (3.5 Ϯ 1.0% of total PE); total PE in VSV-infected cells, 155 Ϯ 11 nmol, and in VSV, 5.1 Ϯ 1.9 nmol (3.8 Ϯ 0.5% of total PE). Data are presented as the mean of two measurements.