Liposome-mediated transfer of integral membrane glycoproteins into the plasma membrane of cultured cells

E.rperimental

LIPOSOME-MEDIATED GLYCOPROTEINS

Cell Research

129 (1980) 393-408

TRANSFER

OF INTEGRAL

INTO

THE

PLASMA

OF CULTURED G. POSTE,’

‘Departments Memorid

MEMBRANE

CELLS

N. C. LYON,’ P. MACANDER.’ C. W. PORTER: P. REEVE’ and H. BACHMEYER’

oJ’E.~perimental Instittrte,

MEMBRANE

Pathology Btr~fulo,

NY Vienna.

and 14263, A-1235

~Experimentd USA.

und

Therapeutics, 9ando;

Kos~ldl

Park

ForsclllcrlRsinstitrrt.

Atrstritr

SUMMARY Cultured mouse 3T3 cells treated with phosphatidylserine or phosphatidylserinelphosphatidylcholine (3 : 7 mole ratio) liposomes containing ortho- and paramyxovirus envelope glycoproteins become susceptible to killing by virus-specific cytotoxic T lymphocytes indicating that the liposome-derived glycoproteins have been inserted into the cellular plasma membrane. Cells incubated with liposomes of similar lipid composition containing viral antigens plus a dinitrophenylated lipid hapten were killed by both virus- and hapten-specific T lymphocytes indicating that both protein and lipid components are inserted into the plasma membrane. We consider that assimilation df liposome-derived antigens into the plasma membrane results from fusion of liposomes with the plasma membrane. Cells incubated with phosphatidylcholine liposomes containing lipid haptens and viral glycoproteins were not killed by cytotoxic lymphocytes indicating that liposomes of this composition do not fuse with the plasma membrane. Liposome-derived paramyxovirus glycoproteins inserted into the plasma membrane retain their functional activity as shown by their ability to induce cell fusion. These experiments demonstrate the feasibility of using liposomes as carriers for introducing integral membrane (glyco)proteins into the plasma membrane of cultured cells and establish a new approach for studying the role of individual (glyco)proteins in the expression of specific cell surface properties.

Studies in many laboratories have shown that liposomes can be used as carriers to introduce a variety of biologically active materials into cells in vitro [l-2] and in vivo [3-4]. Despite rapid growth in the use of liposomes as carriers, information on the mechanisms by which these structures interact with cells is still incomplete. Several mechanisms have been implicated in liposome-cell interactions in vitro, including: fusion of liposomes with the cellular plasma membrane [5-lo]; endocytosis of intact liposomes [.5-8, 11, 121; selective and reciprocal exchange diffusion of components between liposomes and cells [13-161; and adsorption of liposomes to the cell surface 26-801806

without transfer of liposomal components to cells [17, 181. None of these mechanisms are mutually exclusive and some, or all, may be operating simultaneously. If, as proposed [5-11. 16, 17, 193, liposomes with appropriate membrane properties can fuse with the plasma membrane this could be used experimentally to introduce various types of membrane components into the plasma membrane of specific cell types. However, the evidence for fusion of liposomes with the plasma membrane published to date is largely circumstantial and definitive evidence for insertion of liposomal membrane components into the plasma membrane by any process other

394

Postc

et

trl.

than phospholipid exchange is still lacking. Experimentally, the difficulty in establishing whether liposomes can fuse \vith the plasma membrane lies in the problem ofdefining whether liposomal membrane components recovered in association with the plasma membrane have been inserted into the lipid bilayer of the plasma membrane as a result of liposome-plasma membrane fusion or if they are merely adsorbed to the outer surface of the membrane. Recent evidence indicates that adsorbed liposomes may account for as much as 50-90% of the cell associated liposome population [ 17. 181. Any assay used to detect fusion of liposomes with the plasma membrane must thus be capable of identifying a relatively small number of fusion events against this large background of adsorbed liposomes. In this paper we have studied the feasibility of using liposomes as carriers to introduce integral membrane glycoproteins into the plasma membrane of cultured cells. These experiments show that liposomes can successfully transfer integral membrane glycoproteins into the cellular plasma membrane. lntegral membrane glycoproteins extracted from the envelope of orthomyxoviruses and pat-amyxoviruses were selected for insertion into cells because they offer two useful features for evaluating the feasibility of using liposomes to alter plasma membrane protein composition. First. these viruses grow to high titers in eggs and cell cultures. enabling envelope glycoproteins to be obtained in substantial quantities. Also, the presence of only two major envelope glycoproteins in these viruses eliminates many of the separation and purification steps needed to obtain similar amounts of integral membrane (glyco)proteins from mammalian cells. Second. viral glycoproteins can be reliably distinguished from cellular components and, in the case

of paramyvoviruses. their ahilit\ to induce cell fusion ~~secl to stud!’ u,hrthrr lipa~omederived glycoproteins remain functional after assimilation into the plasma membt-ane.

MATERIALS

AND

METHODS

Animiils Adult male BALB/c (H-Z,“‘,. CS7BL/6 I H-2”“) and CBA (H-?) mice were obtained from the West Seneca Animal Breeding Facilitv of Roswell Park Memorial Institute and the Jackson Laboratories. Bar Harbor. Maine.

Cells Chick embryo fibroblasts. BALBlc mouse 3T3 cells and Madin-Darby bovine kidney (MDBK) cells were grown at 37°C in Dulbecco’s modified Eagle medium (DMEM) plus 10% fetal calf serum (FCS) as described previously [?O, 211. All culture media and sera were purchased from Gibco. Grand Island. N.Y.

Virus

groh+tll

and pw~ication

Sendai virus, Newcastle disease virus (strain Herts) and influenza virus (A/Scotland/74: MRC-II recombinant) were grown in the allantoic cavity of IOday-old embryonated eggs or MDBK cells infected harvested and purified by sucrose with IO’ EID,,,. gradient centrifugation as described previously [20. 221. Virus hemagglutination [?O]. neuraminidase [23]. hemolytic [24]. cell fusion [25] and hemadsorption [25] activities were determined as described elsewhere. Labeling of virus lipids with IzP was done as described previously [22]. In certain experiments. the virus envelope glycoproteins of purified virions were labeled with ly51 using lactoperoxidase-catalysed iodination [26] or with ‘H using the galactose oxidase-tritiated potassium borohydride method [27]. Protein concentrations were determined by the method of Lowry et al. [28] using bovine serum albumin (BSA) as a standard. In certain experiments, allantoic fluids from paramyxovirus-infected eggs were treated with IOq ethanol for I h at room temperature to eliminate viral infectivity. cell fusion and hemolytic activities [29]. Virus preparations were also inactivated by ultraviolet t UV) irradiation or P-propiolactone as described previously [22].

Extraction, prrr$cntion and sepurntion of L>irvs en\*elope glycoproteins Glycoproteins were extracted from the envelope of sucrose gradient-purified virions using cetyl trimethyl ammonium bromide as described in our previous work [22. 301. Unlike the solubilization of virus envelope glycoproteins by detergents such as Triton

X-100 [31]. sodium deoxycholate [32]. Tween 20 [33], NPJO [34]. octylglucoside [35] or Empigen [36] which completely disrupt virions. cetyl trimethyl ammonium bromide extraction maintains the integrity of the lipid bilayer of the virus envelope. Extracted glycoproteins are thus not contaminated by viral nucleoprotein and matrix proteins. Aliquols (50 ml) of gradient-purified virions (0.4-0.6 mg protein/ml in phosphate-buffered saline) were incubated with cetyl trimethyl ammonium bromide (0.03 % influenza virus; 0.05 % Sendai virus and NDV) for 30 min at room temperature followed by centrifugation in a 5-50s (wtlwt) linear sucrose gradient at 4°C for 24 h at 30000 rpm in a Beckman Ti I5 rotor to separate extracted envelope glycoproteins from ‘spikeless’ subviral cores as described previously [22; 301. Our previous work has shown that >95% of the viral lipid remains associated with subviral cores and electronmicroscopy confirmed that the lipid bilayer of the envelope remains intact [22]. SDSPAGE characterization of the extracted glycoproteins and detergent (25 Triton X-lOO)-disrupted subviral cores by iodium dodecyl sulphate (SdS)-polyacrylamide gel electrophoresis as described previously [22] revealed that the extracted glycoproteins were not contaminated by either nucleoprotein or matrix protein which were associated exclusively with the subviral cores.

Lipids Chromatographically pure egg phosphatidylcholine. beef brain phosphatidylserine and [‘H]dipalmitoylphosphatidylcholine (spec. act. 10 Cilmmole) were purchased from Aoolied Science Laboratories, State College. Pa.. [l’Cjphosphatidylcholine (spec. act. I.5 Cilmmole) from New England Nuclear. Boston, Mass., and N-dinitrophenylaminocaproylphosphatidylethanolamine (DNP-CapPE) from Avanti Biochemicals. Birmingham, Ala. All lipids were shown to be chromatographically pure by thin-layer chromatography and were stored under nitrogen in sealed ampoules at a concentration of IO-20 pmoles phosphate in chloroform until use.

Incorporation glycoproteins

of’ purified \*inrs ifrto liposomes

Purified. detergent-extracted orthoand paramyxovirus envelope glycoproteins were incorporated into liposomes at a protein : lipid ratio of I : 5 by a detergent-dialysis procedure using either Nonidet P-40, deoxychblate ‘or octyl-/3+g~copyranoside [30]. As described in detail elsewhere [30], reconstituted liposomes prepared by this method are closed. predominantly unilamellar structures (ia 300-l 200 A) whose surface is studded with ‘spikes’ which represent the viral glycoproteins. Isopyknic centrifugation has established that the viral glycoproteins are associated with a single class of liposomes and protease-digestion and detergent-lysis experiments have demonstrated that >80% of the incorporated proteins are exposed on the outer surface of liposomes [30].

Liposome-cell

interactions

Liposomes were incubated with confluent monolayers of different target cells (see Results) in 60 mm plastic Petri dishes (Falcon Plastics, Oxnard. Calif.) in low Ca’+-DMEM (Gibco) without serum or antibiotics for up to 2 h at 37°C as described previously [21] at a liposbme concentration of I imole/phospholipid (I 160 pg viral protein) per IO cells. The number of liposomes associated with cells after various times was measured using liposomes containing radiolabeled phospholipids ([‘H]dipalmitoylphosphatidylcholine; [“Clphosphatidylcholine) or Ip51- or “H-labeled viral proteins.

Production of lirrrs- and hapten-specific cptotoxic eifector cells Virus-specific cytotoxic effector cells were prepared using a modification of the method of Schrader & Edelman [37]. Mice were immunized against Sendai virus by intraperitoneal (ip) inoculation of UV-irradiated virus (500 hemagglutinating units). The animals were sacrificed 2 wee& later. a-suspension of spleen cells prepared and aliquots of 5 x IO’ spleen cells incubated in 20 ml RPM1 1640 medium containing 10% heatinactivated fetal calf serum (FCS) and 5x IO-’ M 2mercaptoethanol (Sigma. St Louis) in Falcon 3013 plastic tissue culture flasks together with 3~ IO syngeneic stimulator cells. The latter were prepared from spleens of normal syngeneic mice and were incubated in RPM1 1640 medium containing 2.5 pg/ml Sepharose-concanavalin A (Pharmacia. Piscataway. N.J. J for 48 h at 37°C followed by irradiation (2000 R) and treatment with 200 hemagglutinating units of virus. Cytotoxic effector cell populations were harvested 7 days later and used in cytotoxicity assays as described below. Effector cells with hapten-specific cytotoxicity for dinitrophenylated residues were produced bv the method of’ Shearer [38]. BALB/c m&se spleen-cells (2X IO ) were incubated with I5 mM dinitrobenzene sulfonic acid (DNBS) (Eastman Kodak, Rochester, NY) in Hanks’ balanced saline solution. pH 7.6. for 30 min at 37°C after which they were washed three times in serum-free RPM1 1640 and aliquots of 5~ IO” cells used to stimulate Ix 10: syngeneic responder spleen cells in 4 ml RPM1 1640 containing 10% heat-inactivated FCS. 5 x IO-: M 2-mercaptoethanol (Sigma, St Louis, MO) and 40 &ml eentamvcin (Scherina Corp.. Kenilworth. N.J:).-St&lator >nd responder cells were incubated together for 5 days at 37°C after which cells were harvested and used-in cytotoxicity assays.

C.vtotoxicity

assays

“ICr-labeled target cells (IX lO’/well) were incubated with various numbers of cytotoxic effector cells in Linbro microplates with round-bottom wells for 6 h at 37°C. Centrifugation of plates released the radioactivity into the supematant from damaged target cells. The spontaneous release of “‘Cr from all targets was

396

POSIL

1’I

lil.

Lipohome

None (glycoproteins alone)

Sample” Liposome-associated proteins Protein-free liposomes+ isolated glycoproteins

compoGtion’

PS

PS/P(:

17.9lll.3?,”

19.21

PC‘

glycoI .2X (0.83

?i 1

I .56 ( I .O I

7

1

I .44

t12.-lri, (0.93

I7.9htll.6~i) c; J

I ..I.:

10.86Oj

” Confluent monolayer 3T3 cell cultures were incubated with: ( I) liposomes of the indicated lipid composition containing 1Z51-labeled Sendai virus envelope glycoproteins (I .33 pmoles phospholipid/lO: cells and approx. 150 pg viral protein/pmole phospholipid): (2) isolated ‘““I-labeled Sendai virus glycoproteins (approx. 200 @g/lOi cells) containing the same radioactivity as the liposome-associated glycoprotein preparation: or (3) a mixture of protein-free liposomes (I .33 pmoles phospholipid/ IO’ cells) and isolated ““I-labeled virus glycoproteins (200 @g/IO7 cells). The total amount of radioactivity added to cell cultures in each sample was 1.54~ IO5 cpm. All samples were incubated with cells for 2 h at 37°C in serum-free low Ca’+-DMEM after which cultures were washed three times in prewarmed PBS and the cell-associated radioactivity measured. b Mean values derived from three separate experiments. ’ PS. phosphatidylserine: PS/PC. phosphatidylserine-phosphatidylcholine 13 : 7 mole ratio): and PC. phosphatidylcholine. rl The figures in parentheses are cell-associated radioactivity as a 5 of the radioactivity in the total inoculum.

I I-27% (see was calculated

T-S

Results). as follows:

The

percentage

cytotoxicity

Xl00

Max-S where: T is radioactivity released from virusor liposome-treated target cells exposed to effector cells: 5. radioactivity released spontaneously from untreated cells exposed to effector cells: and Max, total radioactivity released from target cells disrupted by detergent. The results listed in the text are mean values for Y cytotoxicity derived from the sum of means from radioactivity measurements on triplicate wells per experiment.

Antiservr Hyperimmune SeI2 to sucrose-gradient purified Sendai virus were prepared in rabbits as described previously [39]. Hyperimmune anti-DNP serum was prepared in rabbits by four intramuscular injections at 2-week intervals of I mg dinitrophenylated keyhole limpet hemocyanin in complete Freund’s adjuvant (Difco Laboratories. Detroit, Mich.). All antisera were heatinactivated at 56°C for 40 min before use.

Antibody-mediated

cytolysis

“lCr-labeled target cells treated with virus particles or liposomes were washed three times with ice-cold phosphate-buffered saline and resuspended in serum-free culture medium. Aliquots of cells (2X IO”) in 100~1 of a I : 4 dilution of rabbit antibodies to the appropriate virus or DNP and 100 ~1 of a I : 9 dilution of guinea pig

complement t Litton Bionetics. Kensington, Md) in Linbro microwell plates and the mixture incubated for I h at 37°C. The amount of radioactivity released into the supernatant was measured and the percentage of specific cell lysis determined by subtraction of the amount of cell-associated radioactivity released spontaneously by target cells. Spontaneous release was 12-235.

Purified IgG fractions from rabbits immunized against purified Sendai virus and unimmunized control animals were conjugated to ferritin as described elsewhere [40]. Untreated control cells and cells 12x 10:) treated with liposomes containing viral glycoproteins for I h at 3PC in serum-free DMEM were incubated with the ferritin-IgG preparations (300 PC(p) for 30 min at 4°C and then washed twice with serum-free RPM1 1630 and suspended in 3% glutaraldehyde in 0. I M phosphate buffer (pH 7.4) for l-2 h at 4°C. A pellet

I. TEMs of the periphery of a iT3 cell incubated for I h at 37°C with unilamellar phosphatidylserinel phosphatidylcholine (3 : 7 mole ratio) liposomes containing Sendai virus (egg-grown) envelope glycoproteins and stained with ferritin-conjugated anti-Sendai virus antibodies showing binding of ferritin to the plasma membrane and possible internalization of ferritin-labeled antibodies in coated pits (urro\~‘.s ). Ferritin molecules were distributed evenly around the entire periphery of liposome-treated ceils and labeling was particularly prominent over microvilli. Control 3T3 cells incubated with ferritin-conjugated antibodies alone did not show cell surface staining (not shown). X88000.

Fig.

Liposot7le-medicttrtl

rratuf‘er oj’integral

t~letnbmne glycoproteins

397

Liposome”

Antibody”

Complemerit”

‘i Spec. “Cr-relrabc (mean 1 S.E.1

RESULTS

,Antivirus” Anti-DNP*’ PC

Antivirus Anti-DNP

PS/PC-Virus GP-DNP Cap PE

PC-Virus GP-DNP Cap PE

+

Antivirus Antivirus Anti-DNP .4nti-DNP

Antivirus Anti-DNP

” S’Cr-labeled 3T3 cells were incubated for 2 h at 3--C with the following lipoaomes before expo\urr to antibody and/or complement: ( I I phosphatid! I\rrinc phosphatidylcholine (PS/PC: 3 : 7. moie mtior lipL)somes (I.33 pmoles phospholipid/lO; cell>,: 121 phosphatidylcholine liposomes (1.33 Fmoles phosphalipiJ IO’ cells) or (3) PS/PC or PC liposomes containing viral glycoproteins (virus-GP) and ,l’-dinitrophen) c aminocaproylphosphatidylethanolamine IDNP Cap PE; 10%. (wt/wtl of total phospholipid,. ” Liposome-treated cells (2~ IO’) incubated irith the indicated antibodies (7 &ml) and/or complement for I h at 37°C as described in the Methods. ’ Determined from release of “‘Cr from lipobomctreated target cells as described in the Methods. The results represent mean values + S.E. from triplicdtr measurements. The 5? spontaneous Wr release from control cells incubated in medium without liposomeb treated with the various antisera and/or complement was 12-23q. No significant increase in spontaneous release was detected when cells were incubated with normal rabbit IgG (25 pglml). ” Anti-Sendai virus (egg-grown) and anti-dinitrophenylated keyhole limpet hemocyanin (anti-DNPI antibodies produced in rabbits as described in Method>. ’ Statistically significant (p
was formed by centrifugation and fixed for an addItional 7 h. The fixed cell pellets were washed o\,emight in phosphate-buffer, post-fixed with I G phobphatebuffered osmium tetroxide, dehydrated in a graded alcohol series, and embedded in Epon-araldite. Semi-

Incorporation of detergent-extracted Sendai virus envelope glycoproteins into liposomes significantly increases their ability to associate with monolayer cell cultures (table 1). Liposomes prepared from phosphatidylserine. phosphatidylcholine or a mixture of phosphatidylserine and phosphatidylcholine (3: 7 mole ratio) were equally efficient in enhancing the association of viral glycoproteins with cells (table 1). Treatment of cells with a mixture of protein-free liposomes and free glycoproteins did not enhance the association of glycoproteins with cells (table I ). This indicates that the increased association seen N ith glycoproteins incorporated into liposomes is not caused by liposome-mediated changes in cell surface properties that facilitate non-specific glycoprotein binding. Experiments using liposomes containing tuo membrane markers ([3H]labeled viral [“Cjphosphatidylglycoproteins and choline) revealed that the ratio of these components in cell-associated liposomes was similar to that in liposomes incubated for 2 h at 37°C without cells (results not shown). This indicates that the cell-associated liposome population is composed of intact liposomes. Liposome-derived virus glycoproteinb can be detected on the surface of treated cells by immunoelectronmicroscopy (fig. I). Although the electronmicrographs in fig. I

Liposome-t?teLliateLl

Table 3. Antibody-

atzd cell-mediated

trcttwfi~r oj’itltegral

tt1etnbtutle glyvproteitn

lysis ctf 3T3 cells treated c/r Spec.

-‘Cr

release

\r.itA Srndai (mean

Virus”

Antibody+ complement’

Cytotoxic lymphocytes”

No virus Sendai (egg)’ Sendai (MDBK)’ Sendai (MDBK-trypsinja UV Sendai (egg? BPL-Sendai (egg)’ Ethanol-Sendai (egg~j

0.113 0.485 0.468 0.492 0.49-l 0.458 0.136

0 48.62 23.42 50.6+ 44.3+ N.D.“’ 18.42

0 69.2+5.9’ 5.4k2.1 61.727.3’ 75.lk7.8’ 68.326.3’ 4.-t+ I .-I

5.3k

\rirus parricies

+ S.E.)

Cell-associated neuraminidase activity (Aas?’

9.3” 5.F I I .? 8.1”

399

T

” 51Cr-labeled monolayer cultures of 3T3 cells incubated for I h at 37°C with the indicated virus preparations (4000 HAU/107 cells=200 pg viral protein= 1.2~ IO” virus particles) before exposure to antibodies or cytotoxic effector cells. ’ Activity per 4x loh cells measured after incubation with 200 HAU of virus for I h at 37°C. I’ Determined as described in table 2’ using rabbit antibodies to egg-grown Sendai virus. Spontaneous “‘Cr represent mean values from two separate experiments. release from cells was 13-22 %. The results ” 3 specific “‘Cr release from virus-treated target cells induced by syngeneic cytotoxic T lymphocytes generated in vitro from BALB/c spleen cells primed in vivo by egg-grown Sendai virus and restimulated in vitro with concanavalin A-treated y-irradiated lymphoblasts exposed to Sendai virus (see Methods). An effector to target cell ratio of 40: I was used in all experiments except for cells incubated with ethanol-treated virions where a ratio of 60 : I was used. Spontaneous 5LCr-release from target cells was I Z-2 I ?. The results represent mean values from three experiments. ” Egg-grown virus. ’ Virus grown in MDBK cells. IJ Virus grown in MDBK cells and incubated with 3x crystalline trypsin (20 pg/ml) for I5 min at 37°C before addition to cells. ’ Virus irradiated with ultraviolet light for 20 min (IWO ergs/cm%ec). ’ Virus treated with 0.2% (v/v) P-propiolactone in isotonic saline for 2 h at 37°C. ’ Virus treated with 10% ethanol for I h at ZYC. ’ Statistically significant (piO.01). ’ Statistically significant (p
cannot show definitively that liposomeplasma membrane fusion has taken place, the distribution of the ferritin molecules suggests that this process has occurred. All of the ferritin molecules shown in fig. 1 are distributed evenly on the outer face of the plasma membrane. If these molecules were associated with liposomes adsorbed to the cell surface then they should also be found at distances up to 800 A (the diameter of the larger liposomes from the outer face of the plasma membrane). Cells treated with liposomes containing Sendai virus envelope glycoproteins and dinitrophenylated-lipid haptens can be lysed by antivirus- and antihapten-antibodies and complement (table 2). However,

this does not establish that liposomederived antigen(s) have been introduced into the plasma membrane. Several studies have shown that complement-dependent antibody-mediated cytolysis can occur when the target antigen is merely adsorbed to the cell surface and not incorporated into the plasma membrane [4144] (also see below). The question of whether liposomederived virus antigens have been inserted into the plasma membrane or are merely adsorbed to it has been resolved by comparing events in liposome-treated cells with those in cells exposed to intact virus particles where fusion of virus envelope with the plasma membrane is known to occur

400

Postc

i’t

lil.

(for reviews see refs [4.5, 461). One immediate consequence of fusion of Sendai virus particles with the plasma membrane is that cells become susceptible to killing by virus-immune cytotoxic T lymphocytes which recognize virus envelope glycoproteins inserted into the plasma membrane. Cytotoxic lymphocytes will only kill cells in which viral antigens have been incorporated into the plasma membrane and will not kill cells under conditions where virus particles are merely adsorbed to the cell surface [29. 50, 511. This is illustrated by the data in table 3. Cells incubated at 37°C with virions that fuse with the plasma membrane (i.e. infectious virions containing the F glycoprotein; see refs [45-49]) are killed by cytotoxic T lymphocytes (table 3). Intracellular virus replication is not needed for the expression of T cell cytotoxicity and cells incubated with virions treated with UV irradiation or /3-propiolactone to eliminate their infectivity are also killed by T cells (table 3). Such particles do. however. fuse with the plasma membrane [45]. This indicates that the lymphocytes are responding to antigens derived from the virus envelope and not to new antigens synthesized as a result of intracellular virus replication. In contrast, under conditions where fusion of virus particles with the plasma membrane does not occur. e.g. treatment of cells with non-infectious virions grown in MDBK cells which contain the F,, glycoprotein. cytotoxic T lymphocytes fail to kill (table 3) even though equal amounts of virus are bound to cells as shown by measurements of cell-associated neuraminidase activity (table 3). Additional evidence that virus envelope glycoproteins must reside within the plasma membrane to provoke killing of target cells by cytotoxic T cells is provided by experiments on the reactivity of cytotoxic lymphocytes to cells incubated with

ethanol-treated virions. Ethanol destroys the viral infectivity and fusion acti\,ities a\sociated with the F envelope glycoprotein but leaves the hemagglutination (HA) and neuraminidase (NJ activities mediated b), the HN glycoprotein unaffected [29]. The functional HN glycoprotein in these \irions enables them to bind to cells. but the absence of a functional F protein dictates that they cannot fuse with the plasma membrane and insert envelope glycoproteins into the plasma membrane. Cells incubated with ethanol-treated virions are completely resistant to killing by cytotoxic T lymphocytes (table 3) despite normal \‘irus binding (table 3). Although resistant to killing by T lymphocytes, cells incubated with noninfectious (F,,) virions grown in MDBK cells or ethanol-treated virions can be killed by antivirus antibodies plus complement (table 3). These data reinforce the reservations expressed above that complement-dependent antibody-mediated cytolysis is not suitable for detecting insertion of exogenous antigens into the plasma membrane. since cytolysis can occur when the target antigen is merely bound to the cell sul-face. The data in table 3 demonstrate that lymphocyte-mediated cytolysis provides a sensitive assay for detecting virus envelope glycoproteins residing within the lipid bilayer of the plasma membrane. Experiments were therefore done to see if this assay could be used to detect liposomemediated insertion of virus glycoproteins into the plasma membrane of cultured cells. As shown in table 4, incubation of 3T3 cells with phosphatidylserine. phosphatidylserine/phosphatidylcholine or phosphatidylethanolamine/phosphatidylcholine liposomes containing glycoproteins from egg-grown Sendai virus renders them susceptible to killing by Sendai-immune T lymphocytes. This indicates that virus glyco-

Liposome-mediated

transfer

oJ’integra1 membrane

glycoproteins

401

Table 4. T lgmphocyte-mediated lysis of 3T3 cells treated with liposomes of differing lipid composition containing Sendai virus and influenza virus envelope glycoproteins andlor N-dinitrophenglaminocapro~lphosplzatid~lethanolamine (DNP Cap PE) Liposome” PS/PC-Sendai GP (egg)-DNP

Cap PE

PS/PC-Sendai GP (MDBK)-DNP

Cap PE

PS-Sendai GP (egg) - DNP Cap PE PE/PC-Sendai GP (egg)-DNP

Cap PE

PC-Sendai GP (egg) - DNP Cap PE PC-Sendai GP (MDBK) PS/PC-Influenza GP PC-Influenza GP

Priming/stimulating antigen for effector cellsb

% Spec. 51Cr release’ (mean k S.E.)

Sendai virus DNP Sendai virus DNP Sendai virus DNP Sendai virus DNP Sendai virus DNP Sendai virus Influenza virus Influenza virus

62.5k6.2” 41.7k4.3’ 49.2k4.7’ 35.3k3.5’ 65.8k4.9d 38.7k2.7’ 57.3k3.9” 44.4f3.6’ 5.3 I .3 3.9t0.9 4.6+ I .2 34.8k3.0’ 7.1k2.0

” “‘Cr-labeled 3T3 cells were incubated for 2 h at 37°C with liposomes of the indicated phospholipid composition (1.33 pmoles phospholipid/lO’ cells) containing glycoproteins (virus GP) extracted from egg-grown Sendai virus (F 2nd HN glycoproteins) or Sendai virus grown in MDBK cells (F, and HN glycoproteins) or influenza virus plus N-dinitrophenylaminocaproylphosphatidylethanolamine (DNP-Cap PE; 10% (wt/wt) of total phospholipid) and then incubated with the indicated cytotoxic effector cells. The identity of the glycoproteins extracted from virus particles and incorporated into liposomes was confirmed by SDS-PAGE as described in the Methods. PS/PC, phosphatidylserine/phosphatidylcholine (3 : 7 mole ratio); PC, phosphatidylcholine; PS. phosphatidylserine; and PE/PC, phosphatidylethanolaminelphosphatidylcholine (3 : 7 mole ratio). ’ Virus-specific effector cells generated as described in table 3,” and DNP-specific effector cells as described in the Methods. ” Y? specific SLCr release from liposome-treated cells incubated with the indicated effector cells measured as described in table 3” except that effector to target cell ratios of 60 : I and 100 : I were used with virus-specific and hapten-specific effector cells, respectively. 51Cr leakage induced by treating cells with protein-free liposomes, isolated Sendai or influenza virus glycoproteins or a mixture of protein-free liposomes plus isolated virus glycoproteins never exceeded 12 %. ’ Statistically significant (p
proteins have been inserted into the cellular plasma membrane. In contrast, virus immune lymphocytes did not kill untreated control cells or cells exposed to isolated virus glycoproteins, liposomes without proteins or a mixture of liposomes and isolated glycoproteins (see footnotes to table 4). The extent of the T lymphocyte-mediated destruction of liposome-treated target cells in table 4 is similar to that seen in cells exposed to preparations of intact virus particles containing a similar concentration of virus envelope glycoproteins (table 3). This suggests that liposomes are equally as

efficient as virus particles in introducing viral antigens into the cellular plasma membrane. T lymphocyte-mediated killing of cells containing liposome-derived viral glycoproteins parallels events in natural virus infection [29, 52, 531 in that killing is limited to lymphocytes which are syngeneic to the target cell and allogeneic virus-immune T cells are not cytotoxic (table 5). These data demonstrate that killing of liposome-treated cells is not due to non-specific alterations in cell surface properties produced by liposome treatment. Exp

Cell

Res

129 (1980~

5 Spec. “Cr release tmean i S.E.) induced by effector cell% from target cells treated with

Effector

cell”

Unstimulated B.4LB/c (H-P I Sendai immune, BALB/c (H-7”“) Sendai immune. CL7BL/6 (H-P) Sendai immune. CB.4 (H-2““) Influenza immune. BALB/c (H-2”“) NDV immune. BALB/c (H-2”“)

Intact

virus”

II 69.3f5.8” 5 .__‘+ I.3 3.x+ I .o 0 2.7kO.X

Liposome-associated virus glycoproteins 0 5X.6+4.8” 2.320.6 7.0+05 I .7+0.7 ?.-ttl.l

’ Effector cell populations generated in vitro from cells primed in viva and restimulated in vitro against the indicated viruses as described in table 3”. h 5’Cr-labeled BALB/c 3T3 cells incubated with egg-grown Sendai virus t-1000 H.L\U/IO’cells) for 2 h at 37°C before incubation with effector cells. 5 specific 5’Cr release was determined as described in table 3”. The results are mean values derived from triplicate cultures. Spontaneous 5’Cr release was 12-21 ‘Y. (I “‘Cr-labeled 3T3 cells incubated with phosphatidylserine/phosphatidylcholine (3 : 7 mole ratio) liposomes ( I .33 @moles phospholipid/lO’ cells) containing virus envelope glycoproteins extracted from egg-grown Sendai virus (160 c(g viral proteinlpmole phospholipid) for 2 h at 37°C before incubation with effector cells at an effector to target cell ratio of 60 : I. The results are mean values derived from measurements on triplicate cultures. Spontaneous Yr release was 17-16 v. ” Statistically significant tp
Liposome-mediated insertion of viral glycoproteins into the cellular plasma membrane is also accompanied by insertion of liposomal lipids. This was demonstrated by showing that cells incubated with phosphatidylserine, phosphatidylserine/phosphatidylcholine or phosphatidylethanolamine/phosphatidylcholine liposomes containing viral antigens and a dinitrophenylated lipid hapten were killed by both virusspecific and hapten-specific cytotoxic effector cells (table 4). The envelope of Sendai virus contains two glycoproteins. HN glycoprotein (hemagglutinin-neuraminidase) is present in virus grown in eggs or mammalian cells. The other protein differs depending on the source of the virus. Egg-grown virus contains the F (fusion) glycoprotein but virus grown in MDBK bovine cells contains a higher molecular weight precursor of F, termed F,, [45-49]. The F glycoprotein of Sendai is believed to be responsible for in-

ducing fusion between virus envelope with the cellular plasma membrane to initiate the infective process [46-49]. This therefore raises the question of whether hposomes containing this molecule might display a greater facility to fuse with the plasma membrane than liposomes of similar lipid composition carrying other membrane glycoproteins. Two pieces of evidence argue against this possibility. First. cells treated with phosphatidylcholine liposomes containing egg-grown Sendai virus glycoproteins (F and HN) are not killed by cytotoxic T cells (table 4). This indicates that the presence of the F glycoprotein does not automatically elicit fusion of liposomes with the plasma membrane. Second. phosphaticontaining dylserine/phosphatidylcholine influenza virus envelope glycoproteins or Sendai virus F,, and HN glycoproteins extracted from virus grown in MDBK cells are able to insert these molecules successfully into the cellular plasma membrane (table 4).

Liposotne-mediated

trcrmfer

Table 6. Fusion

of cells by egg-groitw drti \,irus atId liposomes cotltaitlitlg \,elope glycoproteins frotn egg-grown dcii virus “r Cell

Treatment” Untreated UV Sendai PS/PC-Sendai PC-Sendai

3T3 cells virus

(egg) virus GP (egg)” virus GP (egg)”

3.3 62.5” 43.2” I.9

cells

SettetzSen-

qf integral

membrme

glycoproteirzs

unable to insert these glycoproteins plasma membrane (table 4).

403

into the

Fttnctiotml clcti\*ity qf‘liposotne-deri,,ed glycoproteins inserted into the plastnn tnetnbrme

fusionh Chick embryo cells 7.8 68.5” 47.4” 2.8

” Monolaver cultures of 3T3 cells and chick embryo fibroblasts were incubated for ? hat 37°C in serum-free low Ca’+-DMEM with UV-irradiated egg-grown Sendai virus (4000 hemagglutinating unit$G07 cells) or with liposomes of the indicated lipid composition (I .33 pmoles phospholipid/lOr cells) containing envelope glvcoproteins from the same virus (160 wglrrmole phospholipid) after which the cells were washed three times with prewarmed PBS. fresh DMEM plus IO% calf serum added and the extent of cell fusion measured after incubation for a further 12 h at 37°C. PS/ PC, phosphatidylserinelphosphatidylcholine (3:l mole ratio): PC, phosphatidylcholine. ” Determined from counts on fixed monolayers of virusor liposome-treated cells stained with MayGrunwald-Giemsa. The results represent mean values from two separate experiments. Five replicate cultures were examined in each experiment and a minimum of 500 cells counted from each culture. The % cell fusion in control cell populations incubated with isolated virus glycoproteins and/or protein free liposomes was as follows: PYPC lioosomes. 9.4 (3T3). 10.4 (CEFI: . PC liposomes. 2.7 (3T3), 2.5 (CEF); and Sendai virus glycoproteins, 3.5 (3T3). 3.1 (CEF). ” Extracted from egg-grown virus and characterized by SDS-PAGE. d Statistically significant (p
Since intact virus particles containing these glycoproteins do not fuse with cells [4549] these results indicate that liposomemediated insertion of proteins into the plasma membrane is determined primarily by the lipid composition of the liposome rather than the viral glycoproteins. Further support for this conclusion is provided by data showing that phosphatidylcholine liposomes containing influenza virus glycoproteins or F,, and HN glycoproteins extracted from Sendai virus grown in MDBK cells are

The ability of Sendai virus to induce cell-tocell fusion (review refs [45, 461) offers a convenient assay for studying whether virus envelope glycoproteins inserted into the plasma membrane by liposome carriers retain their function. As shown in table 6. cells incubated with phosphatidylserinel phosphatidylcholine liposomes containing glycoproteins from egg-grown virus undergo extensive fusion to form polykaryocytes (fig. 2) which are indistinguishable from those induced by intact virus particles. In contrast, phosphatidylcholine liposomes containing the same glycoproteins did not induce cell fusion (table 6). Since liposomes of this composition do not insert viral glycoproteins into the plasma membrane (table 4) this indicates that adsorption of virus glycoproteins to the cell surface is not sufficient to elicit cell fusion. DISCUSSION The present experiments have shown that mammalian cells incubated in vitro with liposomes of appropriate phospholipid composition containing virus envelope glycoproteins become susceptible to killing by virus-specific cytotoxic T lymphocytes. Unlike antibody-mediated cytolysis which can occur when the target antigen(s) is bound to the cell surface and not situated within the plasma membrane [41. 42. 441, lymphocyte-mediated cytolysis requires that the target antigen reside within the plasma membrane [29, 50, 511. The present results therefore demonstrate that liposomes of this particular composition can E.rp Cell

Rcs /29 119801

Liposome-mediated

ttwzsfer

successfully insert virus antigens into the plasma membrane of uninfected cells. Lymphocyte-mediated destruction of cells treated with liposomes containing virus envelope glycoproteins is highly specific and occurs only when the lymphocytes are syngeneic with the target cell. This indicates that killing of liposome-treated cells does not result from non-specific alterations in cell surface properties but involves specific recognition of liposome-derived antigens and appropriate cellular histocompatibility determinants situated within the plasma membrane. Additional data presented here showing that cells treated with phosphatidylserine or phosphatidylserinelphosphatidylcholine liposomes containing a lipid hapten as well as virus glycoproteins are killed by virus- and hapten-specific T lymphocytes further indicates that both the protein and lipid components of the liposomal membrane are inserted into the cellular plasma membrane. We conclude that the entire liposome membrane is probably assimilated into the cellular plasma membrane via fusion between the two structures. Liposome-mediated insertion of virus glycoproteins and lipid haptens into the cellular plasma membrane depends on the phospholipid composition of the liposome. Successful insertion was only obtained when liposomes were prepared from phosphatidylserine alone or a mixture of phosphatidylcholine and either phosphatidylserine or phosphatidylethanolamine. In contrast, liposomes prepared from phosFig. 2. Fusion of mouse 3T3 cells by liposomes containing Sendai virus envelooe elvcotxoteins. (a) Monolayer rulture of untreated’co%rol’3T3 cells; ib) 3T3 cell monolayers I8 h after incubation for Z h at 37°C with unilamellar phosphatidylserine/phosphatidylcholine (3 : 7 mole ratio) liposomes (I .33 wmoles phospholipid/lO’ cells) containing Sendai virus (egg-grown) envelope glycoproteins (I50 pg proteinlpmole phospholipid) showing a large polykaryocyte containing multiple nuclei. Stain: May-Griinwald-Giemsa. X520.

oj-integral

membrane

40.5

glycoproteins

phatidylcholine containing similar amounts of viral glycoproteins and lipid hapten failed to fuse with the plasma membrane. This indicates that the presence of the F glycoproteins is not sufficient to automatically elicit membrane fusion These results also argue against current dogma that the F glycoprotein is directly I-esponsible for inducing membrane fusion [46-49]. Our results support the recently expressed alternative view that the F glycoprotein does not induce fusion directly but is instead responsible for enabling virus particles to bind to cell surface glycolipids to establish a suitably close apposition between the virus envelope and the plasma membrane so that fusion between these structures can take place via lipid-lipid interactions [54, 551. Further evidence that fusion of liposomes with the plasma membrane is determined by the properties of the lipids of the liposome membrane rather than the virus F glycoprotein is provided by the present results showing that phosphatidylserine or phosphatidylserine/phosphatidylcholine liposomes containing the F, Sendai virus glycoprotein or influenza virus glycoproteins were able to fuse with the plasma membrane. This contrasts with the behavior of intact Sendai and influenza virus particles containing these glycoproteins which fail to fuse with the plasma membrane [45]. This difference in the fusion capacity of liposomes and virus particles containing the same glycoproteins could be due to several factors: (1) relative enrichment of the liposome membrane in phospholipids such as phosphatidylserine and phosphatidylethanolamine which have been shown to promote fusion in model membrane systems [55]; (2) differences in the topography and/or mobility of glycoproteins in liposomes and virus particles imposed by the presence of the M Exp Cell

Res

129 ,198O)

protein and nucleocapsid in virus particles: or (3) differences in the collisional efficiency of liposomes and virus particles with cells to achieve the close contact inter-action with the plasma membrane needed for fusion to take place. Irrespective of which mechanism(s) is involved. the data presented here indicate that the phospholipid composition of the membrane in which the viral glycoproteins are embedded is crucial in determining whether fusion can take place. The present results on the ability of hapten-specific cytotoxic T lymphocytes to kill cells treated with liposomes containing dinitrophenylated phosphatidylethanolamine differ from the results of Ozato & Henney [58] who found that liposomes containing this hapten rendered cells susceptible to killing by antibodies plus complement but not by lymphocytes. We consider that this difference results from the differences in the lipid composition of the liposomes used in the two studies. Ozato & Henney used phosphatidylcholine liposomes, whereas in our experiments. lymphocyte cytotoxicity was detected using liposomes prepared from phosphatidylserine or a mixture of phosphatidylserine and phosphatidylcholine. However. when phosphatidylcholine liposomes were used we obtained the same results as Ozato & Henney. As discussed below, there is now considerable evidence that phosphatidylcholine liposomes are highly resistant to fusion. We therefore consider that the inability of cytotoxic T cells to kill cells treated with antigen-containing phosphatidylcholine liposomes reported here. and by Ozato & Henney. simply reflects the failure of liposomes of this composition to fuse with the plasma membrane and insert antigens into the lipid bilayer of the plasma membrane in the way needed for antigen

recognition and expression 01’ ~‘1totoictt!, by T lymphocytes. Although cells treated with antigen-containing phosphatidylcholine liposomes were resistant to killing by lymphocytes. our results, and those of Ozato R: Henney [5X]. indicate that they can be killed by antibodies plus complement. This finding is consistent with previous observations showing that antibody-mediated cell damage can occur under conditions where the target antigen is adsorbed to the plasma membrane rather than assimilated into the membrane [4 I-M]. These data show that. contrary to recent claims in the literature [%I. antibodymediated cytolysis does not provide a suitably sensitive assay for detecting liposome-plasma membrane fusion. The present results showing that phosphatidylcholine liposomes have little or no ability to fuse with the plasma membrane is consistent with other evidence showing that liposomes of this composition are unable to fuse with either natural [I. 7. 5. IO. 571 or model membranes [5X--62]. This large body of negative evidence argues against the proposals of Pagan0 et al. [ 16. 171and Weinstein et al. [ 191 who claim that phosphatidylcholine liposomes are incorporated into cells by fusion with the plasma membrane. These investigators show:ed that cells treated with phosphatidylcholine liposomes containing 6-carboxyfluorescein develop diffuse cytoplasmic fluorescence and they interpreted this as indicating that liposomes had fused with the plasma membrane releasing fluorescent dye into the cytoplasm. However, recent work by Szoka et al. [I81 has shown that cytoplasmic uptake of liposome-encapsulated 6-carboxyfluorescein is not caused by liposome-plasma membrane fusion but results from leakage of dye from liposomes followed by uptake into the cell as a result of liposome-induced alterations

Liposonze-mediated

tr-crmfkr

in plasma membrane permeability. There is thus presently no convincing published evidence to suggest that phosphatidylcholine liposomes can fuse. In addition to providing direct evidence that liposomes of appropriate lipid composition can fuse with the cellular plasma membrane, the present experiments are the first to demonstrate the feasibility of using liposomes as carriers to introduce exogenous integral membrane proteins into the plasma membrane of cultured cells. Integral membrane glycoproteins introduced into the plasma membrane of acceptor cells by this method retain their functional properties and immunologic specificity. We have found (unpublished observations) that they can be detected in the plasma membrane of acceptor cells for up to 24 h after their initial insertion. The mechanism by which liposome-donated glycoproteins and lipid haptens are eventually removed from the plasma membrane is not known and further studies on this question are in progress. The use of liposomes to introduce specific phospholipids and cholesterol into the plasma membrane of cultured cells is already well established [ 13-151. The demonstration that liposomes can also be used as carriers to introduce defined proteins into the cellular plasma membrane thus adds a further dimension to the versatility of liposomes as an experimental tool for introducing biologically-active materials into cultured cells and offers a new approach for defining the role of specific integral membrane (glyco)proteins in the expression of particular cell surface functions.

This work was supported in part by Grants CA 18260 (G. P.) and CA 13038 (C. W. P.) from the USPHS and NIH. Core Support Grant CA 17609 to Roswell Park Memorial Institute, Cancer Cell Center.

qf integtxl

men1btwtIe

glycoproteitw

407

REFERENCES I. Poste, G, Papahadjopoulos. D & Vail, W J. Methods in cell biology (ed D M Prescott) vol. 14. pp. 33-7 I. Academic Press, New York (1976). 7. Pbste, G. Liposomes in biology and medicine (ed G Gregoriadis & A C Allison) .pp._ 301-351. Wiley. New York ( 1980). 3. Tyrrell. D A. Heath, T D. Colley. C M & Ryman, B E, Biochim biophys acta 456 (1976) 759. 4. Kimelberg. H K & Mayhew. E G. CRC crit rev toxicol I I (1978) 25. 5 Poste, G & Papahadjopoulos. D. Proc natl acad sci US 79 (1976) 1603. 6. Martin. F J & MacDonald. R C, J cell biol 70 (1976) 515. 7. Batzri. S & Korn. E. J cell bio166 (1975) 621. 8. Wilson. T, Papahadjopoulos. D & Taber, R. Proc natl acad sci US 74 (1977) 3471. 9. Orci. L & Perrelet. A, Membrane fusion (ed G Poste & G L Nicolson). Cell surface reviews, vol. 5, pp. 629-656. North-Holland. New York (1978). IO. Standahl, 0 & Tagesson. C, Exp cell res I08 (1977) 167. II. Jansons, V K. Weis, P. Chen. T-H & Redwood, W R. Cancer res 38 ( 1978) 53 I. I?. Weissmann. G, Korchak. H. Finkelstein. M. Smolen. J & Hoffstein, S. Ann NY acad sci 308 (1978) 235. 13. Cooper, R A, Leslie, M H. Fischkoff, S. Shinitzky. M & Shattil. S J, Biochemistry I7 (1978) 327. 14. Insel, P A. Nirenberg. P. Turnbull. J & Shattil. S .I. Biochemistry I7 (1978) 5269. IS. Claret. M. Garay. R & Girand. F: J physiol 274 (1978) 247. 16. Pagano. R E & Takeichi, M. J cell biol 67 (1977) 49. 17. Huang. L, Ozato, K & Pagano. R E. Membrane biochem I (1978) I. IS. Szoka. F C Jr. Jacobson, K & Papahadiopoulos. D, Biochim biophys acta 551 (1979) 295. - . 19. Weinstein, J N. Yoshikami. S. Henkart. P. Blumenthal. R & Hagins. W A, Science I95 (1977) 489. 20. Poste. G, Waterson. A P. Terry G. Alexander. D J & Reeve, P. J gen virol I6 (197:) 95. 21. Poste, G. Kirsh. R. Fogler. W & Fidler, I J, Cancer res 39 ( 1979) 88 I. 32. Bachmeyer, H. Intervirology 5 (1975) 260. 23. Aminoff. D. Biochem j 81 (1961) 384. 14. Shimizu. Y K, J virol 9 (1972) 841. 25. Poste. G, Methods in cell biology (ed D M Prescott) vol. 7. pp. 211249. Academic Press. New York (1973). 26. Poste. G. Reeve, P & Bachmeyer. H, Negative strand viruses and the host cell (ed B W J Mahy & R D Barry) pp. 815-823. Academic Press. London (1978). 27. Critchley. D R. Cell 3 (1974) 121. 28. Lowry, 0 H, Rosebrough. N J. Farr. A L & Randall. R J, J biol them 193 (1951) 265. 29. Sugamura, K. Shimizu. K & Bach, F H. J exp med 149 (1978) 276. 30. Poste, G. Macander. P. Lyon. N C. Bachmeyer. H & Reeve, P, J virol. In press.

31. 32. 33. 34. 3s. 36. 37. 38. 39. 40. 41. 42. 43. 44.

45.

46.

47.

Scheid. A & Choppin. P W. Virology 61 (19711 125. Helenius. A. Fries. E. Garoff. H & Simonh. K. Biochim biophys acta 436 ( 1976) 3 19. Orvell. C & Norrby. E. J immunol I l9( 19771 1882. Shimizu. K. Shimizu. Y K. Kohama. T & Ishida. H, Virology 62 (1973) 90. Helenius. A. Fries. E & Kartenback. J. J cell biol 7s (1977) 866. Dyke. S F. Williams. W & Seto. J T, J med virol 2 11978) 143. Schrader. J W & Edelman. G M. J exp med I45 (1977) 523. Shearer, G M. Eurj immunol 4 (1974) 527. Bachmeyer. H. Liehl. E & Schmidt. G. Postgrad med j 52 ( 1976) 29. Nicolson. G L & Poste. G. J supramol struct 8 (1978) 235. Forman. J & Finkelstein. R A. J immunol I IX (1977) 1655. Palmer. J C. Lewandowski. L J & Waters. D. Nature 269 ( 1977) 595. Roitt. I. Essential immunology. 3rd edn. p. 166. Blackwell, Oxford (1977). Cochrane. C G & Dixon. F J, Textbook of immunopathology (ed P A Miescher & H J MiillerEberhard) 2nd edn. vol. I. pp. 137-156. Grune & Stratton. New York (1976). Poste, G & Pasternak. C A. Membrane fusion. In Cell surface reviews (ed G Poste & G L Nicolson) vol. 5, pp. 305367. North-Holland. New York (1978). Hosaka. Y & Shimizu. K. Virus infection and the cell surface. In Cell surface reviews (ed G Poste & G L Nicolson) VOI. 2. pp. 129-155. North-Holland. Amsterdam (1977). Homma. M & Ouchi. M. J virol II (1973) 1457.

E.rp Cell

Rrs

129 i/980)

4X. 49. SO. KoszinoLvshi. L’. Grthtng. hl J B I\.Itcrl’ielJ. \l. Nature 267 ( 19771 IhO. 51. Gething. M J. tioszino\bchi. C! S: II atst-field. M. Nature 273 (1978) 689. 32. Blanden. R V. Pang. T E 6t Dunlop. hl B C. Viruy infection and the cell ,u&ce. In Cell surface reviews (ed G Poste & G L Nicolson) vol. 2. pp. 239-290. North-Holland. Amsterdam t 19771. 53. Zinkernagel, R M. Curr topics misrobiol immunol 82 (1978) 113. 54. Poste. G & Waterson. .A P. Negative \trand viruse5 (ed B W J Mahy & R D garry) vol. ?. pp. 905-922. Academic Press. London (19751. 55. Papahadjopoulos. D. Poste. G & Vail, W .I. Methods in membrane biology (cd E Kern) t 19791. In presr. 56. Nakamura. K Y, Homma. M. J gen Lirol 25 (1971) 117. 57. Papahadiopoulos, D. Vail. W J. Pangborn. W .A & Pokte. G-. Biochim biophys acta 448 ;1976) 265. 58. Ozato. K & Hennev. C S. J immunol I!1 ( 19781

2405. 59. 60. 61. 62.

Haywood. A M. J mol biol X7 t 1975) 625. Papahadjopoulos. D. Poste. G. Schaeffer. Vail. W J. Biochim biophvs acta 352 (1971) Kantor. H L & Presieg&d. J. Biochemistry (1975) 1790. Ldwaczek. R. Kainosho. M, Girardet. Chan, S I. Nature 256 (197.5) 584.

Received Revised Accepted

December 27. version received .April 18. 1980

lY79 i\pril

IS. 19X0

B E & IO. I4 J-L

&