Budding of Rous sarcoma virus and vesicular stomatitis virus from localized lipid regions in the plasma membrane of chicken embryo fibroblasts.

The origin of the envelope lipids acquired by Rous sarcoma virus (RSV) and vesicular stomatitis virus (VSV) during budding from the plasma membrane of chicken embryo fibroblasts was examined. Several differences were observed between the lipid composition of RSV and the plasma membrane. When the phospholipid composition of the cells was modified by growing them in the presence of the choline analogues, N,N-dimethylethanolamine or l-2-amino-1-butanol, the phospholipid composition of the virus was subsequently altered but in a very different manner than the plasma membrane. In the plasma membrane, the increase in the analogue-containing phospholipid was at the expense of phosphatidylcholine and phosphatidylethanolamine while the amount of sphingomyelin remained constant. In RSV, however, there was a decrease in sphingomyelin and phosphatidylethanolamine while there was only a small change in the amount of phosphatidylcholine. Phospholipid polar head group modification did not significantly alter the fatty acid composition or the cholesterol content. Membranes of phagosomes isolated after the cells had ingested latex beads had essentially the same phospholipid composition as the plasma membrane. The phospholipid composition of VSV was different from RSV, but it also did not reflect the composition of the plasma membrane. The composition of the plasma membrane was intermediate between the viruses and the endoplasmic reticulum, but contamination of the plasma membrane fraction with the endoplasmic reticulum could not account for the observed differences. These results show that the viruses bud from localized lipid regions that do not reflect the average properties of the plasma membrane.

The origin of the envelope lipids acquired by Rous sarcoma virus (RSV) and vesicular stomatitis virus (VSV) during budding from the plasma membrane of chicken embryo fibroblasts was examined. Several differences were observed between the lipid composition of RSV and the plasma membrane. When the phospholipid composition of the cells was modified by growing them in the presence of the choline analogues, N,i% dimethylethanolamine or I-2-amino-1-butanol, the phospholipid composition of the virus was subsequently altered but in a very different manner than the plasma membrane. In the plasma membrane, the increase in the analogue-containing phospholipid was at the expense of phosphatidylcholine and phosphatidylethanolamine while the amount of sphingomyelin remained constant. In RSV, however, there was a decrease in sphingomyelin and phosphatidylethanolamine while there was only a small change in the amount of phosphatidylcholine. Phospholipid polar head group modification did not significantly alter the fatty acid composition or the cholesterol content. Membranes of phagosomes isolated after the cells had ingested latex beads had essentially the same phospholipid composition as the plasma membrane. The phospholipid composition of VSV was different from RSV, but it also did not reflect the composition of the plasma membrane. The composition of the plasma membrane was intermediate between the viruses and the endoplasmic reticulum, but contamination of the plasma membrane fraction with the endoplasmic reticulum could not account for the observed differences. These results show that the viruses bud from localized lipid regions that do not reflect the average properties of the plasma membrane.
Enveloped viruses have a relatively simple structure and provide a good means for studying the assembly of a biological membrane. Many enveloped viruses, including Rous sarcoma virus and vesicular stomatitis virus, obtain their membrane envelopes from preformed lipids of the host cell plasma mem-brane during the budding process (see Refs. 1 and 2 for recent reviews). The limited number of proteins found in the envelope are virally coded and the host proteins are largely excluded from the virion. Many studies have compared the lipid composition of enveloped viruses with the plasma membrane isolated from their host cells. The general conclusions from these studies are that the viruses acquire their lipids randomly from the plasma membrane and the viral lipid compositions are similar to the plasma membrane composition of their host cells. Several relatively small differences have been noted but their interpretation is clouded by the difficulty in isolating and assessing the purity of the plasma membrane. That is, the differences may be due to contamination of the plasma membrane fraction by other cellular membranes (1).
Chicken embryo fibroblasts offer a good cell system for studying the budding process as well as for studying the consequences of viral infection. One hundred per cent of the cells can be transformed by RSV' and the cells continually produce viral particles without cell death or dramatic cytopathic effects (3). Electron microscopic studies of the budding process show evaginations of the cellular membrane forming a spherical coat surrounding the viral core followed by release of the intact virus into the extracellular space (4). The processing of viral envelope proteins and inclusion into the mature particle has been well investigated recently (5-8). Chicken embryo fibroblasts can be infected by a variety of other enveloped viruses, including VSV. A large amount of information has been acquired about VSV protein synthesis, processing, and assembly (9-19). In order to investigate the role of lipids in virus infection and transformation-associated membrane changes in chicken embryo fibroblasts, methods were developed to manipulate the lipid composition of these cells during growth under defined conditions (20). RSV-infected cells with altered lipid compositions still produced comparable amounts of infectious virus. In this study, the rates of incorporation of phospholipid polar head group analogues into the endoplasmic reticulum, plasma membrane, RSV, and VSV were determined. The results indicate that the viruses bud from localized lipid regions, either pre-existing or virally induced, that do not represent the average composition of the plasma membrane. The existence of localized regions in the membrane whose size is on the order of a virus particle (approximately 0.1 pm) or larger has important implications for understanding me?brane structure and function.

RESULTS
The characterization of isolated plasma membrane and endoplasmic reticulum fractions from RSV-transformed chicken embryo fibroblasts is shown in Table I. Plasma membrane was prepared by two different methods, the standard preparation and phagosomes, as described under "Materials and Methods." Phagosomes were prepared from cells that had ingested latex beads. The membrane surrounding the beads was compared with the standard preparation in order to determine if the phagosomal membrane was representative of the plasma membrane. Similar results were found for the two preparations from cells grown in the presence of choline, N,Ndimethylethanolamine, or I-2-amino-1-butanol. Although substantial analogue incorporation into phospholipids occurred (Table 11), no significant effect on enzymatic activities or purification of the membrane markers was observed. In the standard preparation, an approximately 8-fold purification (over the particulate lysate) of the (Na',K')-ATPase, a plasma membrane marker, was found. In the case of phagosomes, a 5-fold purification was obtained, with a 5% yield as compared to a 15% yield for the standard preparation. This is indicative of the small amount of phagocytosis which occurs in chicken embryo fibroblasts compared with, for example, LM cells .3 In the standard preparation and the phagosome preparation, the endoplasmic reticulum marker, NADPH-dependent cytochrome c reductase, was substantially reduced in the plasma membrane fraction. In the endoplasmic reticulum fraction from the standard preparation, a decrease in (Na', K')-ATPase was found and approximately 3.5-fold purification of NADPH-dependent cytochrome c reductase was observed. Similar results were found for uninfected cells and cells infected by VSV for 12 h before harvest (data not shown). After 12 h of infection with VSV, no cytopathic effects caused by the virus were observed.
The phospholipid composition of cell membranes and RSV grown in the presence of choline or choline analogues for 48 h is shown in Table 11. For choline-grown cells, qualitatively similar differences were found between the plasma membrane and RSV as reported by Quigley et al. (21,22). Primarily, the virus contained less phosphatidylcholine and more sphingomyelin than the plasma membrane. The plasma membrane was intermediate in the content of all the phospholipids between the endoplasmic reticulum and RSV. For example, the endoplasmic reticulum had 7.9% sphingomyelin, the plasma membrane had 15.5%, and the virus had 25.3%. If one were to use only these data, the argument could be made that the virus really has the same phospholipid composition as the plasma membrane and the observed differences are due to contamination of the plasma membrane primarily with endoplasmic reticulum which is the major membrane fraction of the cell.
After 48 h in the presence of N,N-dimethylethanolamine, 56.2% of the phospholipids were phosphatidyldimethylethanolamine in the endoplasmic reticulum and 42.2% in the plasma membrane. The increase in phosphatidyldimethyl-

TABLE I1 Phospholipid composition of endoplasmic reticulum (ER), plasma membrane (PM), and Rous sarcoma virus (RSV) isolated from cells grown for 48 h in the presence of choline or choline analogues
Phospholipid composition ( 15.5% decrease in the amount of phosphatidyldimethylethanolamine found in the plasma membrane. In the virus, the analogue replaced sphingomyelin and phosphatidylethanolamine, with only a small change in phosphatidylcholine.
In the case of 1-2-amino-I-butanol supplementation, the plasma membrane and RSV had almost identical amounts of phosphatidylbutanolamine. This analogue, as in the case of N,N-dimethylethanolamine, replaced phosphatidylcholine and phosphatidylethanolamine in the plasma membrane and in the endoplasmic reticulum. In the virus, a drop in the amount of sphingomyelin and phosphatidylethanolamine was also found with no change in phosphatidylcholine.
A more detailed analysis of the incorporation of choline and its analogues into endoplasmic reticulum, plasma membrane, and RSV is shown in Fig. 1. The amount of phosphatidylcholine in choline-supplemented cells for the membrane fractions and virus was constant with time of growth ( Fig. lA), as expected. When choline was replaced by N,N-dimethylethanolamine or 1-2-amino-1-butanol, there was an increase in phosphatidyldimethylethanolamine ( Fig. 1B) or phosphatidylbutanolamine ( Fig. IC), respectively, in the endoplasmic reticulum and in the plasma membrane. The rate and extent of incorporation were greater for the endoplasmic reticulum than the plasma membrane. The rate of incorporation of 1-2amino-1-butanol into RSV was similar to the rate of incorporation into the plasma membrane. In the case of N,Ndimethylethanolamine, however, the rate of incorporation was markedly different for the virus and plasma membrane.
The rate of change in phosphatidylcholine during the polar head group modifications for endoplasmic reticulum, plasma membrane, and RSV is shown in Fig. 2. For  tions was quite dramatic (Fig. 2, B and C ) . However, only a small decrease in the amount of phosphatidylcholine was found for RSV during the first 30 h of N,N-dimethylethanolamine supplementation, and then it remained constant ( Fig. 2B). No significant change at all was observed for the amount of phosphatidylcholine in RSV isolated from 1-2amino-1-butanol-supplemented cells. The pattern for the changes in sphingomyelin during polar head group modifkations for virus and membranes was opposite to that found for phosphatidylcholine (Fig. 3). Sphingomyelin remained constant in the plasma membrane and endoplasmic reticulum, while sphingomyelin first remained constant in the virus for 30 h and then decreased during supplementation with both choline analogues (Fig. 3, B and C).
Phagosomes were prepared from cells grown for 24 and 48 h in the presence of choline and choline analogues. The "plasma membrane" of the phagosomes had a similar lipid composition to that of the standard plasma membrane preparation. This was true for both the amount of the two analogues incorporated and the changes found in phosphatidylcholine and sphingomyelin (Figs. 1, 2, and 3).
Supplementation had no significant effect on the fatty acid or cholesterol composition, although dramatic differences between plasma membranes and RSV were apparent (see miniprint). The possibility of changes in the lipid composition of RSV after release from the cell was also examined (see miniprint). No alteration in the phospholipid composition was observed.
In order to determine if other enveloped viruses had similar patterns of analogue incorporation, a comparison of RSV with VSV and their respective plasma membranes was made (Table  VI). The phospholipid composition for VSV grown on cells in choline containing medium was basically similar to that found by McSharry and Wagner (25). The differences in the plasma membrane phospholipid compositions between RSV-and VSV-infected chicken embryo fibroblasts in choline-containing medium was similar to that found between RSV-transformed and normal uninfected ceUs (26): That is, VSV infection of normal cells for 12 h did not significantly alter the lipid composition of the host cell plasma membrane. In addition, the rate of analogue incorporation was similar for uninfected and VSV-infected cells for 12 h. When cells grown on media supplemented with choline or choline analogues were transformed with RSV, there was a small decrease in the amount of phosphatidylcholine and an increase in the amount of phosphatidylethanolamine found in the plasma membrane. There was also a small decrease in the amount of phosphatidylinositol in the plasma membrane isolated from cells grown on choline-supplemented medium.
The small differences between the plasma membranes isolated from RSV-and VSV-infected cells cannot account for the large differences found between RSV and VSV. When the viruses were grown on cells in choline-containing medium, for example, RSV had 10.6% less phosphatidylcholine than VSV. There were also changes in the amount of phosphatidylethanolamine, phosphatidylinositol, phosphatidylserine, and sphingomyelin. Similar changes were found between RSV and VSV grown on cells in media supplemented with choline analogues.

The rates of incorporation of N,N-dimethylethanolamine
and I-2-amino-1-butanol into both viruses were qualitatively similar (Table V and

Localized Lipid
Regions in the Plasma Membrane tion was slightly faster for N,N-dimethylethanolamine into VSV, however, while the rate of incorporation of I-f-amino-lbutanol was slightly faster into RSV. It is highly significant that the changes in phosphatidylcholine and sphingomyelin were analogous to those shown for RSV (Figs. 2 and 3). That is, phosphatidylcholine decreased marginally while sphingomyelin decreased dramatically in the viruses during supplementation. Thus, both RSV and VSV did not reflect the phospholipid composition of their respective host cell plasma membranes.

DISCUSSION
The ability to modify the phospholipid composition of cellular membranes and subsequently the composition of enveloped viruses, permits a more careful examination of the role of the plasma membrane in the budding process. This study was designed to determine whether the viruses obtain their lipids randomly from the plasma membrane or whether they bud from localized lipid regions.
There were several large differences in the phospholipid composition between RSV and its host cell plasma membrane. For example, the rate of incorporation of N,N-dimethylethanolamine was markedly different for RSV as compared with the plasma membrane. This is in contrast to the rate of incorporation of I-2-amino-1-butanol which was roughly similar for the virus and the plasma membrane. The endoplasmic reticulum had the highest rate of incorporation for both of the analogues which is consistent with the endoplasmic reticulum being the major site of phospholipid biosynthesis.
When the composition of the virus is compared to the composition of the plasma membrane only at a single time point or when the lipid composition was not changing (i.e. cells grown in normal medium containing choline), any differences are difficult to interpret because of the uncertainty in assessing the purity of the plasma membrane preparation. The differences in phospholipid composition between RSV, plasma membrane, and endoplasmic reticulum are such that the plasma membrane might be identical to RSV and the observed differences due to contamination of the plasma membrane fraction with endoplasmic reticulum. The contamination would have to amount to approximately 55 to 60% in order to account for the results given in Table 11, for example. This interpretation is not consistent with the purity of the plasma membrane as judged by enzyme markers. There was a substantial purification of the (Na',K')-ATPase and a reduction in the NADPH-dependent cytochrome c reductase. If it is assumed that the specific activity of the NADPH-dependent cytochrome c reductase in the endoplasmic reticulum fraction represented the maximum specific activity possible, then it can be calculated that the plasma membrane fraction contained approximately 15% endoplasmic reticulum in order to account for the lower specific activity. A similar calculation for the succinate-dependent cytochrome c reductase indicates that the plasma membrane fraction had less than 1% mitochondrial contamination.
Similar calculations to determine whether contamination of the plasma membrane fraction with endoplasmic reticulum can account for the differences between the plasma membrane and RSV can be carried out when the cells were supplemented with choline analogues. The plasma membrane fraction obtained at 48 h from cells supplemented with N,N-dimethylethanolamine would have had to contain 53% endoplasmic reticulum to account for the difference in phosphatidyldimethylethanolamine. In the case of I-2-amino-1-butanol supplementation, the plasma membrane would have had to contain essentially no endoplasmic reticulum to account for the similar amounts of phosphatidylbutanolamine. Furthermore, during the incorporation of both choline analogues, the amount of sphingomyelin changes dramatically in the virus, but it did not change in the plasma membrane or endoplasmic reticulum. In this case, the plasma membrane fraction would have to be practically 100% endoplasmic reticulum to account for the results. The purification of the plasma membrane fraction did not change with analogue incorporation as judged by enzyme markers. Clearly, the contamination of the plasma membrane fraction cannot account for the differences observed between RSV and its host cell plasma membrane.
In contrast to these differences between RSV and the plasma membrane, the rate of incorporation of choline analogues and the phospholipid composition of phagosomal membranes were similar to the plasma membrane. This supports the conclusion that the plasma membrane is randomly interiorized to give the phagosomal membrane as shown for the lipids of Acanthanoeba castellanii (27) and LM cells (28) and the proteins of L cells (29).
The observations made with RSV also applied to another enveloped virus, VSV. This is a cytotoxic virus and, consequently, experimental conditions were chosen so that there was no detectable effect on phospholipid biosynthesis. When infection was carried out for longer times, greater than 95% of the cells showed cytopathic effects and eventually lysed. The rate of incorporation of the choline analogues into VSV was less than that found for the plasma membrane and the changes in the other phospholipids were similar to that found for RSV. There were, however, substantial quantitative differences in the phospholipid composition between VSV and RSV. Thus, the two viruses were different from each other and both acquired their lipid from localized regions in the plasma membrane.
Further investigation is necessary to determine if the localized lipid regions were induced by the viral proteins or if they pre-existed in the membrane and were selected by the virus for budding. If extensive localized lipid regions exist in cellular membranes, regardless of how they are formed, it would have numerous implications for membrane structure and function. For example, in a previous study using a fluorescent probe, it was shown that the plasma membrane from RSV-transformed cells was slightly less fluid than the plasma membrane from normal cells (26). In this study, and in most studies of this type, only the average properties of the membrane were measured. There could be very different localized lipid regions between normal and transformed cells and they would not have been detected. It is interesting that some of the effects on transformation-associated changes in RSV-transformed chicken embryo fibroblasts (20) and on adenylate cyclase activities in LM cells (30) caused by in vivo lipid modification occurred more slowly than the changes in the lipid composition. One explanation for these observations is that the components exist in localized lipid regions that do not change as rapidly as the general membrane. Thus, the existence of localized regions would not be unique to viral budding and would greatly alter the interpretation of how various membrane components depend on lipids for activity.   Phorphollp7d C a n p r i t m n (I)