Binding of Semliki Forest Virus and Its Spike Glycoproteins to Cells

Transcription

1 Eur. J. Biochem. 97, (1979) Binding of Semliki Forest Virus and Its Spike Glycoproteins to Cells Erik FRIES and Ari HELENIUS European Molecular Biology Laboratory, Heidelberg (Received January 17, 1979) We have studied the binding of the Semliki Forest virus and its isolated spike glycoproteins, in the form of water-soluble octameric complexes, to various cells at 5 "C. The number of viruses bound per cell increased strongly with increasing free concentrations of virus up to about.2 nm. At higher concentrations smaller increases in binding were observed but saturation was not achieved. The number of viruses bound at a given free concentration was widely different for different cells. For some cells the binding of the virus was maximal at ph 6.8 with little decrease at lower ph, for other cells it was maximal around ph 6.. The spike protein complexes were used at 1 times higher molar concentrations than the virus. The binding increased strongly with increasing free concentrations up to about 5 nm and saturation was obtained at higher concentrations. Up to 1.3 x lo6 spike protein complexes could be bound per cell but great variation could be seen between different cell types. For all cells maximal binding was found below ph 6.. Together with earlier observations, our results suggest that the virus can bind to a cell by two different modes. Around neutral ph the virus binds to specific gycoproteins and at low ph unspecifically to the lipids of the plasma membrane. The possible physiological roles of these two types of binding are discussed. Semliki Forest virus (SFV) is a simple membranebounded animal virus belonging to the Toga family. Although its structure and replication are known in great detail [1,2] little is known about the mechanism by which it enters a cell. This lack of knowledge extends to most other animal viruses [3,4]. Studies on several virus/cell systems suggest that the entry of viruses occurs in three steps: (a) binding of the virus to specific cell surface structures (receptors); (b) penetration of the membrane barrier; (c) uncoating of the nucleic acid to yield the replicative form. The initial binding of viruses is mediated by the viral surface proteins. Receptors for a certain virus may be present on a limited number of cell types or they may occur widely, reflecting the host cell specificity of that particular virus. In a few cases, the receptors have been sho& to be glycoproteins [5] but for most viruses the chemical nature of the receptor is not known [6]. Semliki Forest virus infects a broad range of vertebrate and invertebrate cells [7,8]. The viral particle Abbreviations. SFV, Semliki Forest virus; Mes, 2-(N-morpholino)-ethane sulfonic acid; Hepes, 4-(2-hydroxyethyl)-l-piperazineethane sulfonic acid. is spherical with a diameter of 65 nm [9]. On the surface there are about 26 spikes 7-8-nm long [9,1]. Each spike comprises three glycopolypeptides : El (apparent molecular weight 49 x lo3), E2 (52 x lo3) and E3 (1~1~) [11,12]. Both El and E2 have hydrophobic segments which are inserted in the lipid bilayer [13]. Removal of the viral glycoproteins by proteolytic digestion leads to the loss of infectivity [13,14] indicating a role for the viral glycoproteins in cell binding and entry. The spikes can be isolated lipid-free in intact form using mild detergents. Subsequent removal of the detergent leads to the formation of water-soluble octameric complexes [I 51. We have recently shown that the spike protein octamers bind to the major histocompatibility antigens of man (the HLA-A and HLA-B antigens) and of mouse (the H2-K and H2-D antigens) [16]. These antigens are glycoproteins present on the surface of all nucleated cells of these species and homologous proteins are also found on cells of other origin. In this paper we have characterized the binding of the virus and of its isolated glycoproteins in octameric form to several different cells by electron microscopy and by quantitative binding assays.

2 214 Binding of Semliki Forest Virus to Cells MATERIALS AND METHODS Chemicals RPMI 164 medium without NaHC3 was prepared from its ingredients in powder form (Gibco, Scotland) and cleared by filtration. Bovine serum albumin (Armour Pharm. Comp., U.S.A.) and either 2(N-morpholino)-ethane sulfonic acid (Mes, Sigma) or 4-(2-hydroxyethyl)-l -piperazineethanesulfonic acid (Hepes, Sigma) were added at 5 g/l and 2 mm respectively. The desired ph was obtained by titration with NaOH with the medium on ice. Virus A prototype strain of SFV was grown in monolayers of baby hamster kidney cells (BHK-21) and purified as previously described [17]. Virus was quantified with the Lowry method using bovine serum albumin as standard. Preparation of [35S]methioninelabelled virus (specific activity about 1 x lo6 counts min-' pg-') was performed as described earlier [17]. The purity of both virus preparations was better than 95 "/, as judged by polyacrylamide gel electrophoresis. After its preparation the virus was stored in aliquotes at - 8 'C. For binding experiments the virus was thawed and used within 2 h. Longer storage or suspension in phosphate buffer caused aggregation. For calculations we have assumed that the molecular weight of the virus is 64 x lo6 and that 56 % by weight is protein [lo]. When the amount of virus is given in the text this refers to the amount of viral protein. Preparation of' Spike Protein Complexes Unlabelled virus was mixed with 35S-labelled virus to give a specific activity of either ~ 1' or 4-1 x 1' counts min-' pg-'. To these mixtures Triton Rohm & Haas, U.S.A.), buffer and water were added and detergent-free octameric complexes (29-S complexes) were prepared by centrifugation in a sucrose density gradient as described earlier [15]. The fractions containing the complexes were pooled and transferred to a 1-ml Amicon ultrafiltration cell with a PM5 membrane. The original buffer was removed by repeated concentration and dilution with RPMI 164 medium lacking albumin and Mes. The spike proteins were finally collected in ml with a recovery of %. The solution was frozen in liquid N2 as aliquots and stored at - 8 "C. For competition experiments 3H-labelled 29-S complexes were prepared as above with a specific activity of 7 counts min-' pg-'. Cells Baby hamster kidney cells (BHK-21) were grown as monolayers in plastic bottles (Falcon) in Glasgow minimal essential medium supplemented with 17; fetal calf serum and 5 g/l tryptose broth. S3 HeLa cells were grown as suspension cultures in spinner bottles in Eagles' minimal essential medium supplemented with 1% fetal calf serum and were harvested when the density was 2-3 x 1' cells/ml. JY, Daudi (transformed human lymphocytes) and P815 (a murine mastocytoma cell) cells were grown in suspension in a 5% C2 atmosphere using RPMI 164 medium supplemented with 1 fetal calf serum and harvested at a density of 3-7 x lo5 cells/ml. Eb, a transformed murine lymphocyte, was grown in the same conditions but harvested at a density of 1-2 x lo6 cells/ml. Chick erythrocytes were obtained from a 2-month-old Leghorn chicken. Electron Microscopy Cells from a confluent BHK-21 cell culture were released using trypsin and suspended in growth medium. About 5 x lo4 cells in.1 ml were added to glass cover slips (11-mm diameter) and allowed to grow in a C2 incubator for 48 h. The cell layer was rinsed with cold binding medium and virus (.8 pg in 1 p1) in the same medium was then added. After 1 h at C the medium was removed and the cells were washed and fixed for 3 min by the addition of ice-cold 2.5 % glutaraldehyde in.5 M sodium cacodylate ph 7.2, 5 mm KCl and 2.5 mm MgC12. After dehydration in graded ethanol solutions an epon-filled gelatine cup was inverted on the cover slip. After polymerisation of the epon the cover slip was removed by immersion in liquid N2. The block was sectioned into ultrathin slices and double-stained with uranyl acetate in 5% methanol and lead citrate [18,19]. The sections were examined with a Philips 4 electron microscope. Binding Assay fbr BHK-21 Cells Cells in a confluent culture were released using trypsin, suspended in growth medium and 4-6 x lo5 cells were added to plastic petri dishes (35 x 1 mm) (Falcon, U.S.A.). The cells were allowed to grow for about 24 h in a COZ incubator at which time more than 95% of the surface was covered by cells. The dishes were then brought to a cold room (5 "C) where the medium was withdrawn and the cells were washed with three 1.O-ml portions of binding medium. Binding medium (.5 ml) and appropriate volumes with viral particles were added to give a final volume of.6 ml.

3 E. Fries and A. Helenius 21 5 Fig. 1. Electron nzicroscopy ofsernliki Forest virus hound to BHK-21 cells at ph 6.8. (A) Most of the bound virus occurs in patches and these are frequently localized in areas containing microvillii. (B) Viruses are frequently observed to be associated with 'coated pits'. (C) Higher magnifications show no change in the morphology of the virus nor of the plasma membrane at the contact site. The distance between the unit membranes measured from the outer surface was 9 7t 2 nm The dishes were then rocked for 15 min and a sample was taken from the supernatant to determine the level of free viral protein. The remaining medium was removed and saved to determine the ph. The cells were then washed with three.6-ml portions of binding medium and incubated for 1-15 min with.15 ml trypsin/edta solution (Gibco, Scotland) at 37 "C. By vigorous pipetting the cells were then suspended and transferred to a scintillation vial together with.15 ml binding medium that was used to rinse the dish. A duplicate plate was used for cell counting. More than 95 :d of the cells were viable. Binding Assay for Cells in Suspension An appropriate number of cells was collected from the culture medium by centrifugation and washed with three 1.O-ml portions of binding medium without Mes. The cells were then suspended in.5 ml medium and added to a series of 1.5-ml capped polyethylene test tubes (Eppendorf). Finally ml binding medium and appropriate volumes of solutions with viral particles were added to give a final volume of.3 ml. The tubes were then rocked for 15 min and then centrifuged for 4 s at 3 xg. An aliquot was taken from the supernatant for determination of the free concentration of viral protein. In some experiments the supernatant was saved for ph determination. The cells were then washed with three.6-ml portions of the binding medium and suspended in.1 ml of the same buffer. The suspension was transformed to a scintillation vial followed by.1 ml medium that was used to rinse the test tube. A duplicate sample was used to determine the number of cells and their viability as judged by dye exclusion (> 9 %>. RESULTS Electron Microscopy of Cell-Bound Virus The distribution of bound virus on BHK-21 cells was studied after incubating the cells with SFV at - 4 "C and examining thin sections of the cells using electron microscopy. More than 98% of the viruses were present on the external side of the plasma membrane (Fig. 1). The virus occurred in patches distributed over the entire cell. The patches were frequently associated with microvillii (Fig. 1 A and C). Many viruses were bound to regions of the membrane which

4 216 Binding of Semliki Forest Virus to Cells A I Fraction number Fig. 2. Analjsis of the honiogeneity of the virus and the 29-S complc..wes by sucrose gradient velocity centrijugation. (A) 1 pg of [35S]- methionine labelled virus were incubated for 2 h at 5 "C in 3 pl binding medium (see text), ph 6.8. The sample was then layered on to a 5-ml 5-2",!, (w/w) sucrose gradient containing the same medium with a.75-ml 6"/, sucrose cushion in a Beckman SW5 rotor tube. After 25 min of centrifugation at 5xg at 4"C, 2-p1 fractions were collected from below and 24 aliquots were taken for radioactive determinations. Fractions are numbered from the bottom. (B) 3 pg of [35S]methionine-labelled 29-S complexes were incubated for 2 h at 5 "C in 3 p1 medium (see text), ph 6.3, and then layered on to a 12-ml 5-2"/, sucrose gradient with a 1.-ml 6% cushion. After 6 h of centrifugation at 15 xg, 5-p1 fractions were collected from below and 25-p1 aliquots taken for radioactivity determination. The sucrose solutions were prepared in the same buffers as those used for the incubations had a dense underlying structure, so-called 'coated pits' (Fig. 1 B). Occasionally vacuoles containing virus were seen in the cytoplasm indicating that some endocytosis had occurred despite the low temperature. No apparent structural changes in either the virus or the plasma membrane (Fig. 1 C) could be observed upon binding. The distance between the outer surfaces of the unit membrane images of the virus and of the cell was 9. f 2 nm. This is 2 nm more than the length of the viral spike alone. Kinetics and ph Dependence for Binding of Semliki Forest Virus For quantitative binding studies we used [3sS]- methionine-labelled virus and unlabelled carrier virus. Under the experimental conditions used in this study, at least 75 % of the radioactively labelled virus occurred in monomeric form (Fig.2A). A similar result was obtained for unlabelled virus (not shown). The time course of virus binding to monolayers of BHK-21 cells is shown in Fig. 3. After a rapid initial phase the Time (rnin) Fig. 3. Kinetics qf'binding qf Semliki Forest virus and elution at 5 'C. To a BHK-21 culture in a 3.5-cm petri dish 2.2 pg [35S]methioninelabelled virus (spec. act. 2 counts min-' pg-') in 6 p1 binding medium, ph 6.8, were added. The dish was slowly rocked and after certain intervals aliquots of the medium were taken for radioactive determination. After 18 min the cell layer was washed twice with 1.O ml of the incubation medium and 6 p1 new medium were added (arrow). After certain time intervals the medium was replaced and the released radioactivity determined. The dashed line indicates the total amount of virus added 1. c PH Fig.4. ph dependence of binding of Semliki Forest virus to various cells. To 5-3 x lo5 tissue culture cells and 9 x lo6 human red blood cells in media buffered with Mes or Hepes to different ph were added 1-2 pg [35S]methionine labelled virus. After incubation for 2.5 h at 5 'C the amount of bound and free virus was determined. Note that as the number of cells and the amount of virus was different for the different cell types, these data cannot be used to male comparisons between the binding capacity of different cells. The amount of bound virus is given in arbitrary units rate of binding levelled off and after 2 h further binding was negligible. The rate at which the virus was released when the cells were repeatedly washed with fresh medium was very slow (Fig. 3). Addition of unlabelled virus (2 mg/ml) in the medium increased this rate only slightly (data not shown). An increase in ph of the medium from 6.8 to 7.8 was, however, found to release 5 of the virus within 6 min. The ph of the medium had a marked effect on the binding as shown in Fig. 4. For HeLa, Eb (not shown) and BHK-21 cells the binding was optimal around ph 6.8 and there was little decrease in binding at

5 E. Fries and A. Helenius human red blood cells,. - " v 1 2 Free virus concn (nm) P815 Fig. 5. I.sisotlierms.for ihe binding of' Semliki Forest virus to various idls. To 2-3 x lo6 tissue culture cells and 9 x lo6 human red blood cells in medium of ph 6.8 were added increasing amounts of [35S]- methionine-labelled virus. After 2.5-h incubation at 5 "C the number of viruses bound per cell and the free virus concentration were determined as described in the text lower ph values. For the other cells, however, the amount of virus bound at low ph was at least 5 times higher than that bound at neutral ph. Human red blood cells bound no detectable amounts of virus at neutral ph, but at ph 5.8 low binding could be observed. Virus Binding Isotherms Fig. 5 shows data obtained when five different cell types were incubated with increasing amounts of virus at ph 6.8. The extent of binding was determined as a function of the free virus concentration. Except for the human red blood cells binding could be detected to all the cells. It varied between the cells but could not be correlated with cell size. The shapes of the binding isotherms were similar for all cells: the amount of virus bound increased steeply up to nm concentrations of virus ; at higher free virus concentrations the increase was smaller and linear. When the data for BHK-21 cells and Eb cells were represented in Scatchard plots the low binding values could be fitted with straight lines (Fig.6). In a separate experiment with carrier-free (35S)methionine-labelled virus and BHK-21 cells it was found that the linearity extended down to binding values with about 5 viruses bound per cell, which is close to the number of viruses normally used in infection. For BHK-21 cells the apparent binding constant for the high-affinity binding was m 4 Bound virus /cell Fig. 6. Scatchard plot qf'data obtained as described in Fig.S,for BHK and Eb cells 3 x 1" M-' and the apparent number of sites 5 x lo3. For Eb cells the corresponding values were 5 x 1" M-' and 22 x lo3. The data for the HeLa cells were too scattered to allow calculation of these parameters. Binding of 29-S Complexes to Cells We have not been able to release the external polar moiety of the SFV spikes in an active monomeric form as has been done for RNA tumor virus and for influenza virus [2,21]. We have, however, been able to produce water-soluble octameric complexes (293 complexes) in which the spikes are held together by their hydrophobic segments [15]. Fig. 2 B shows that these complexes remain monodisperse under the conditions used in our assays. Becuase of the high adsorption of the 29-S complexes to plastic petri dishes, experiments were done only with cells in suspension. The time course for the binding of the 29-S complexes was similar to that of the virus (not shown). At least 6% of the bound complexes could be released from HeLa cells by repeated washes with the binding medium of ph 6.8. The ph dependence for the binding was similar for all cells with maximal binding around ph 6. or lower. Fig. 7 shows this for three of the cells studied. Binding isotherms for five different cells were obtained at ph 6.3 (Fig.8). With the exception of human red blood cells, all cells bound 293 complexes (see also [16] for data on JY and Daudi cells). At higher free concentrations than those shown in the graph, there was no further increase in binding. When the binding data was represented in Scatchard plots the highest binding values from all cells studied seemed to approach linearity with slopes of the order of 5 x lo7 M-'. The data for Eb and HeLa cells are

6 218 Binding of Semliki Forest Virus to Cells x - Y) x.2 - c O.1 m U P815 A- " %O ph Fig. I ph dependence for the binding of octameric spike protein comp1eue.c (29-S complexes) to various cells. To I- 15 x 1' cells in media of different ph were added 2-5 pg of [3sS]methionmelabelled 29-S complexes. After incubation for 2.5 h at 5'C the amount of bound and free complexes was determined lo*.bound complexes /cell Fig. 9. Scatchard plot ojdata obtained as described in Fig. X,fi,r H@L.a and Eb cells - a, U. ro a, Q E? a r..5 c 3 I% 'f' Concentration of free 29-5 Complexes (nm) Fig. 8. Isotlierms,for the binding oj29-s complexes to various cells. Tissue culture cells (2-3 x lo')), chicken red blood cells (9 x lo6) and human red blood cells (16x lo6) in a medium of ph 6.3, were incubated with increasing amounts of "S-labelled 29-S complexes. After 2.5-h incubation at 5 "C the concentration of free complexes and the number of complexes bound per cell were determined as described in the text - 1. v._ U!: '.5 m *n shown in Fig. 9. The binding constants and the number of binding sites could not be derived with confidence. We can only conclude that the apparent affinity of the 29-S complexes for the cells decreases at higher concentrations and that the apparent binding constants in the concentration region studied was 1-1- fold lower than that found for the whole virus. Competition between Virus and 29-S Complexes Since the binding of virus and 29-S complexes to HeLa cells have different ph dependencies, the extent plexes included of competition for binding was expected to be different at different ph. Fig. 1A shows the effect that preincubation with unlabelled virus had on the subsequent binding of [35S]methionine-labelled 29-S complexes to HeLa cells. The amount of 29-S complexes bound

7 E. Fries and A. Helenius 219 increased steadily with decreasing ph. The amount bound remained, however, at a level of 3-45 "/, of that bound in the absence of added virus. In the reverse experiment (Fig. 1B) the inhibition of binding of [35S]methionine-labelled virus by unlabelled 29-S complexes was studied. Inhibition of virus binding was reproducibly observed at all ph values, but the decrease in virus binding was most clearly seen at ph values below 6.5. We conclude from these results that 29-S complexes and virus share a large proportion of their binding sites on the HeLa cell surface even though the ph for optimal binding is different for the two ligands. DISCUSSION As a starting point for our studies of the mechanism by which the Semliki Forest virus enters cells we have characterized the binding of the virus and of its isolated membrane glycoproteins (the 29-S complexes) to different cells. The experiments were performed at 5 "C, at which temperature the uptake of viral particles into the cell is very slow. Our data therefore represent the initial binding of the virus through its membrane glycoproteins to the cell surface. We have found that the ph dependence for the binding of the virus was different for different cells and that at neutral ph, which is the optimum for viral infection [22], all cells except human red blood cells bound detectable amounts of virus. For some of the cells we determined the number of viruses bound per cell as a function of the free virus concentration. These data showed that the number of viruses bound to different cells at a given free virus concentration varied greatly. With the concentrations of virus used, saturation was not obtained. We have analyzed these data with Scatchard plots although we are aware that they may not be amenable to this type of analysis. The valency of virus binding, the reversibility of the binding, and the accessability of the receptor sites are some of the critical factors of which we do not know enough and whose influence cannot by assessed with confidence. The values obtained for the binding parameters must, therefore, be taken as operational. The apparent bindingconstantswereoftheorder of 1"- 1l1 M-', and the apparent number of binding sites per cell were up to 5x lo4. Similar determinations with whole viruses have not previously been described but the binding constant for isolated spike glycoproteins of murine leukemia virus has recently been reported to be 8 x lo9 M-*, and the number of binding sites 4 x los [23]. At least for one of the cells studied (the BHK-21 cells) linearity in the Scatchard plot extended down to the virus concentrations normally used for infection. This fact makes it likely that our binding isotherms represent binding to the receptors active in infection. We have previously shown that the viral spike proteins bind to human HLA-A and HLA-B and murine H2K and H2D histocompatibility antigens [16]. Whether the virus binds specifically only to these, and to related glycoproteins on cells of other origin, remains to be investigated. The 29-S complexes also bound to all cells at neutral ph, except human red blood cells. However, maximal binding to the different cells occurred at ph values well below neutral. Binding isotherms were therefore determined at ph 6.3. The 29-S complexes were used at about 1 times higher molar concentrations than the virus and saturation could be obtained. To one of the cell types (the HeLa cells) as many as 1.8 x lo6 complexes could be bound per cell. This number is comparable to the number of concanavalin A binding sites found on cells Furthermore, a calculation shows that this number of 29-S complexes probably covers the entire cell surface. It is, therefore, clear that the binding of the spike proteins at low ph cannot be accounted for only by specific glycoprotein receptors. The possible nature of this additional binding is suggested by previous studies on hemagglutination of Toga viruses. The viruses bind to, and agglutinate, avian erythrocytes at ph below 6.2 [25,26]. These viral component responsible for the binding has been identified as the El polypeptide in both SFV and the closely related Sindbis virus [27,28]. Furthermore, it has been shown that phospholipids are effective inhibitors of the hemagglutination reaction [29,3] and Mooney et al. [31] have demonstrated that the Sindbis virus binds to phospholipid-cholesterol liposomes with a ph dependence very similar to that of hemagglutination. From these observations it seems likely that the binding we observe for the spike proteins at low ph represents an interaction between El and the lipid moiety of the cellular plasma membranes. It is suggestive that the curvilinearity of the Scatchard plots at high concentrations of the virus (Fig. 6) is also due to this type of nonspecific binding. Whether those cells to which the virus binds optimally around ph 6. have fewer receptors than those with an optimum around ph 6.8 remains to be determined. It is not clear whether the binding of the viral spike proteins to a cell surface at low ph has any physiological role in the infection. An interesting possible function is suggested by our preliminary electron microscopic studies which show that at 37 "C surfacebound virus is rapidly taken up into the lysosomes. From the data discussed earlier, it is conceivable that the low ph in the lysosomes [32] could induce an interaction between one of the viral polypeptides (El) and the lipids of the lysosomal membrane. This interaction could initiate the penetration of the viral genome into the cytoplasm, perhaps by fusion of the viral and lysosomal membranes. A lysosomal route

8 ~ -~ 22 E. Fries and A. Helenius: Binding of Semliki Forest Virus to Cells of entry has been demonstrated for re virus 133,341. Several other types of animal viruses, for which the infection route is still unknown, are also taken up into the lysosomal vacuoles [3]. We thank K. Simons and J. A. Reynolds for advice, G. Warren for critical reading of the manuscript, H. Virta for growing the virus, A. D Arcy for providing HeLa cells and E. Bolzau for the electron microscopy. E.F. received support from the O.E. and Edla Johansson Scientific Foundation. REFERENCES 1, KBiriBinen. L. & Renkonen,. (1978) in Memhrane Synthesis und Turnover (Poste, C. & Nicholson, D. L., eds) vol. IV, pp , Elsevier/North Holland, Amsterdam. 2. Simons, K., Garoff, H. & Helenius, A. (1978) in Togaviruses (Schlesinger, R. W., ed.) Academic Press, New York. 3. Dales, S. (1973) Bacteriol. Rev. 37, Lohnberg-Holm, K. & Philipson, L. (1974) Monogr. Virol. 9, Hennacke. B. & Boulanger, P. (1974) Biochem. J Gallaher, W. R. & Howe, C. (1976) Immunol. Commun. 5, Davey, M. R., Dennet,. P. & Dalgarno, P. (1973) J. Gen. Virol. 2, Hilfenhaus, J. (1976) J. Gen. Virol. 33, von Bonsdorff, C. H. (1973) Commentat. Bid. Soc. Sci. Fenn. 75, Laine, R., Soderlund, H. & Renkonen,. (1973) Intervirology, I, Garoff, H., Simons, K. & Renkonen,. (1974) Virology, 61, Ziemiecki, A. & Garoff, H. (1978) J. Mol. Bid. 122, Utermann, G. & Simons, K. (1974) J. Mol. Biol. 85, Kennedy, S. 1. T. (1974) J. Gen. Virol. 23, Helenius, A. & von Bonsdorff, C. H. (1976) Biochim. Biophys. Acta, 436, Helenius, A., Morein, B., Fries, E., Simons, K., Robinson, P., Schirrrnacher, V., Terhorst, C. & Strominger, J. L. (1978) Proc. Nail Acad. Sci. U.S.A. 75, Kaariiinen, L., Simons, K. & von Bonsdorff, C. H. (1969) Ann. Med. Exp. Bid. Fenn. 47, Venable, J. H. & Coggeshall, R. (1965) J. Cell Biol. 25, Reynolds, E. S. (1963) J. Cell Bid. 17, Moenning, V., Frank, H., Hunsmann, G., Schneider, I. & Schafer, W. (1974) Virology, 61, Brand, C. M. & Skehel, J. J. (1972) Nut. New Bid. 238, Osterrieth, P. M. (1966) Acta Virol. 1, Bishayee, S., Strand, M. & August, J. T. (1978) Arch. Biochem. Biophys. 189, Hughes, R. C. (1976) Glycoproteins, p. 146, Butterworths, London. 25. Clarke, D. H. & Casals, J. (1958) Am. J. Trop. Med. Hyg. 7, Gorman, 3. (197) J. Gen. Virol. 6, Helenius, A., Fries, E., Garoff, H. & Simons, K. (1976) Biochim. Biophys. Acta, 436, Dalrymple, J. M., Schlesinger, S. & Russell, P. K. (1976) Virology, 29, Porterfield, J. S. & Rowe, C. E. (196) Vidogj~, 11, Salrninen, A. (1962) Virology, 16, Mooney, J. J., Dalryrnple, J. M., Alving, C. R. & Russell, P.K. (1975) J. Virol. 15, Ohkuma, S. &Poole, B. (1978) Proc. NatlAcad. Sci. U.S.A. 75, Silverstein, S. C. & Dales, S. (1968) J. Cell Bid. 36, Silverstein, S. C., Astell, C., Levin, D. H., Schonberg, M. & Acs, G. (1972) Virology, 47, E. Fries, Department of Biochemistry, Stanford University School of Medicine, Stanford, California, U.S.A A. Hehius*, European Molecular Biology Laboratory, Postfach 1229, D-69 Heidelberg, Federal Republic of Germany * To whom correspondence should be addressed.