J. Cell Sci. 28, 167-177 (i977) 167Printed in Great Britain © Company of Biologists Limited 1977

A FLUORESCENCE ENHANCEMENT ASSAY

OF CELL FUSION

PAUL MALCOLM KELLER,* STANLEY PERSON ANDWALLACE SNIPES

Laboratory of Biophysics, Department of Biochemistry and Biophysics, Tlie PennsylvaniaState University, University Park, PA 16802, U.S.A.

SUMMARY

Two probes were synthesized which consist of fluorescent molecules conjugated to saturatedhydrocarbon chains, 18 carbons long, to ensure their localization into cellular membranes.There is an overlap between the emission spectrum of one probe (donor) and the absorptionspectrum of the other probe (acceptor). By the use of appropriate wavelengths it is possible tospecifically excite the donor probe and record the fluorescence of the acceptor probe. Two cellpopulations, each labelled with one of the probes, were infected with a virus that causes cellfusion, mixed in equal proportions, and the fluorescence of the acceptor probe measured asa function of time after infection. An increase in fluorescence was observed beginning at thetime of onset of cell fusion indicating a mixing of the fluorescent membrane molecules. Aninvestigation of the distance dependence indicated that the increase in fluorescence was mainlydue to resonance energy transfer and not to photon emission and reabsorption. Resonanceenergy transfer requires that the 2 probes be close together and that there be an overlap of theemission spectrum of the donor probe and the absorption spectrum of the acceptor probe. Thepossible application of this assay to other types of membrane fusion is noted.

INTRODUCTION

Membrane fusion is a general biological process which is probably involved in suchphenomena as endo- and exocytosis, bone and muscle cell development, fertilizationand the release of neural transmitters. In addition to these examples of membranefusion, virus-induced cell fusion is caused by a wide variety of enveloped animal viruses.We are studying cell fusion using mutants of Herpes Simplex Virus type 1 (HSV-I)that cause cells to fuse in an otherwise normal infection. These mutants are calledsyn mutants and were isolated as mutants which produce large plaques containingcells with many nuclei (syncytia) (Person et al. 1976). Although we have found thatthe wild type virus induces a small amount of cell fusion, the resulting plaque does notcontain syncytia and is mainly characterized by the presence of individual roundedcells.

Current models of biological membranes envision the membrane as a 2-dimensionalfluid of proteins and lipids (Singer & Nicolson, 1972). It was shown by Frye & Edidin(1970) that cell surface proteins intermix in the plasma membranes of neighbouringcells within an hour after cell fusion has occurred. In addition, it has been shown thatsmall lipid molecules have diffusion coefficients of the order of io~8 cm2/s in synthetic

• Present address: Department of Physiology, Rutgers University Medical School,Piscataway, NJ. 18854, U.S.A.

168 P. M. Keller, S. Person and W. Snipes

and natural membranes (Hubbell & McConnell, 1969; Kornberg & McConnell, 1971;Devaux & McConnell, 1972). Therefore, a suitable assay for cell fusion would be onebased on the intermixing of probe molecules from neighbouring cell membranesfollowing fusion.

We have developed such an assay. It employs a donor fluorescent probe incorporatedinto the membranes of one population of cells and an acceptor fluorescent probe in-corporated into the membranes of another population. The labelled cells are infected,harvested, mixed and seeded into Petri dishes. The donor probe is excited and thefluorescence of the acceptor probe is recorded. Following fusion there is an increasein acceptor probe fluorescence due to resonance energy transfer from the donor probe.The theory for resonance energy transfer has been developed by Forster (1948) andhas been documented by the use of compounds with donor and acceptor groupsseparated by known distances (Latt, Cheung & Blout, 1965; Stryer & Haugland, 1967;Gabor, 1968; Haugland, Yguerabide & Stryer, 1969; Becker, Oliver & Berlin, 1975).Two conditions are required for resonance energy transfer to occur: an overlap be-tween the emission spectrum of one probe and the absorption spectrum of the secondprobe, and a closeness of the 2 fluorescent molecules. The first requirement may besatisfied by the use of fluorescein and rhodamine as the donor and acceptor fluorescentmolecules, and the second condition may be satisfied as a result of the cell fusionprocess. Although we have used the assay to measure cell fusion, it should be adaptableto measure the distribution of molecules on a cell surface.

EXPERIMENTAL PROCEDURES

Cell and virus

HEL cells were supplied by Dr John Docherty (Department of Microbiology, The Pennsyl-vania State University). The viruses used were HSV-i, strain KOS, and a mutant of this strainisolated in our laboratory, syn 20, which causes extensive cell fusion in an otherwise normalinfection. The culture media and reagents, and cell and virus growth procedures have beendescribed previously (Person et al. 1976).

Membrane probes

The structures and syntheses of membrane probes are given below.

F18 NH R18

Fluorescence and cell fusion 169

.F18: 0-5 g of fluorescein isothiocyanate (Sigma Chemical Co., St Louis, MO) was reactedwith 0-5 g of octadecylamine (Aldrich Chemical Co., Milwaukee, WI) in 10 ml dry pyridine for4 days. The product, F18, was purified by 2-dimensional thin layer chromatography on anumber of 2-mm thick, 20 x 20 cm Silica Gel G plates (Analabs, Inc., North Haven, CT).After development in the first solvent the plates were rotated 180° and developed in the secondsolvent. The developers were chloroform/acetone/methanol/acetic acid/water (50/20/10/10/5)and water/pyridine (4/1) for the first and second dimension, respectively. Octadecylamine hadthe lowest mobility in the first solvent (verified by ninhydrin staining), and F18 had a lowermobility than fluorescein isothiocyanate in the second solvent. The F18 streak was scrapedfrom the plate, dissolved in ethanol and the chromatography steps repeated. The final solutionwas centrifuged at 7500 g 3 times to remove Silica Gel G, the ethanol allowed to evaporate, andthe product stored as a dry powder at 20 °C.

i?i8: A suspension of i-7 g of Rhodamine B (Aldrich Chemical) in 10 ml dry benzene wastreated with 0-3 ml of dry pyridine and 0-27 ml of thionyl chloride was added dropwise withstirring and cooling. The reaction mixture was stirred at room temperature for 12 h. Onegramme of octadecanol was then added and allowed to react for an additional 12 h with con-tinued stirring. The benzene was allowed to evaporate, the powder dissolved in a small volumeof ethanol and the solution streaked on to chromatography plates which were then developedin the 2 solvent systems as described above, followed by development in ether to move theoctadecanol away from the product, R18. R18 was resuspended in ethanol and the chromato-graphic separation repeated twice. The final R18 ethanol solution was centrifuged as notedabove, the alcohol allowed to evaporate, and the product stored as a powder at 20 °C.

Preparation of F18- or Ri8-labelled cells

Standard labelling media containing less than 0-2 % ethanol were prepared by dissolving theprobes in alcohol and adding sufficient growth medium to produce absorbances of 0085 at506 nm (F18) or 0-260 at 565 nm (R18). Cells in late logarithmic phase of growth were har-vested and seeded into 16-oz. (434-ml) prescription bottles at 7-5 x io' cells per bottle. After6-h incubation at 37 °C the medium was decanted and replaced by 10 ml labelling medium perbottle; 12-14 h later the labelling medium was decanted and the cells washed twice with 10 mlof TBS. If the cells were to be infected, virus was added in o-8 ml TBS and allowed to attachfor 1 h. A multiplicity of infection (MOI) of 10 adsorbed plaque-forming units (PFU) per cellwas used (Person et al. 1976). Cells were then harvested and the 2 labelled populations weremixed in equal proportions and seeded into 30-mm diameter sealable glass culture dishes(stock #1934-12030, Bellco Glass, Vineland, N.J.). Generally 1 x io8 Fi8-dyed cells and1 x io6 Ri8-dyed cells in a total of 2 ml of growth medium were used. While being flushedwith 5 % CO, the Petri dishes were sealed with a 5 % solution of warmed agarose and incubatedat 38 °C.

Preparation of FiS-labelled virus

A virus stock (approximately 1 x 10' PFU/ml) was layered on a 30-ml, 5-30 % Dextran T-10(Pharmacia Fine Chemicals, Upsala, Sweden) gradient made with growth medium. The gradientwas centrifuged at 4 °C for 50 min at 22000 rev/min in a Beckman SW25.1 rotor. The virusband was collected, diluted with 20 ml TBS and pelleted. The pellet was resuspended in 2 mlof 8 X the standard F18 labelling solution. The probes were allowed to partition into the viralenvelope for 12 h at room temperature. The virus was then pelleted to remove unadsorbedprobe molecules.

Description of fluorometer

The 457-459 nm lines from an Osram caesium lamp (Spectronics, Westbury, N.Y.) werefiltered (460-nm band pass filter, 500-nm cut-off filter, Ditric Optics, Marlboro, MA; CorningGlass 5-61 coloured glass filter, Corning, N.Y.) and allowed to shine on the bottom of a 30-mmglass culture dish. Light emitted at right angles was filtered (Ditric Optics 600-nm cut-oninterference filter; Ditric 590 cut-on coloured glass filter) and detected by a water-cooled photo-

170 P. M. Keller, S. Person and W. Snipes

multiplier tube (Emitronics, Inc., Plainview, N.Y. #95588). A manually operated shutterallowed one to determine background readings with the culture dish in place. The photo-multiplier tube was operated with 1 kV across the dynode chain of a high-voltage power supply(Model 3oooR, Gencom Div., Emitronics). The output was measured by a digital voltmeter(Weston # 1292, Weston Instruments, Inc., Newark, N.J.).

RESULTS

The fluorescent probes used in the experiments reported here were synthesized bythe covalent linkage of fluorescein or rhodamine to a saturated hydrocarbon chain, 18carbon atoms long. The hydrocarbon chains cause the probes to partition into cellmembranes; more than 95% of the fluorescent material recovered from F18- andRi8-labelled cells was in the organic phase of a Folch extraction (Folch, Lees &Sloane-Stanley, 1957). Thin layer chromatography of Folch extracts prepared afterovernight exposure of cells to F18 or R18 showed that the extracted material wasessentially all in the form of F18 and R18 (Keller, 1976).

6 8 10Time after infection, h

12 14

Fig. 1. R18 fluorescence of mock infected, wild type infected and syn 20 infected cells.R18 fluorescence was measured using culture dishes containing 10' F18- and io8 R18-labelled cells for (O) mock infected, (A) wild type infected, and (A) syn 20 infectedcells. Wild type infected cells produced about 20 % of the fluorescence increase re-lative to syn 20 infected cells. The increase in fluorescence for syn 20 infected cellsbegan at a time when fusion was observed to occur in parallel experiments using theCoulter counter assay to monitor fusion. One relative quantum transferred representsan increase of o-i mV. Each point represents the average relative quanta transferredusing 3 culture dishes.

The absorption and fluorescence spectra of F18 and R18 are very similar to those offluorescein and rhodamine, respectively. The emission spectrum of F18 overlaps theabsorption spectrum of R18 such that resonance energy transfer can be detected bypreferentially exciting F18 at 460 run and recording the fluorescence, mainly due toR18, at wavelengths greater than 600 nm. Cell populations labelled with either F18

Fluorescence and cell fusion 171

or R18 were infected with virus, harvested, mixed in equal proportions, and seededinto culture dishes at a concentration sufficient to form a monolayer upon cell attach-ment. At various times after infection F18 was excited and the R18 fluorescence wasmeasured (see Experimental procedures). An increase in fluorescence was observedbeginning approximately 4 h after infection of HEL cells with a syn mutant virus,syn 20 (Fig. 1). If all of the syn 20 infected cells contained either F18 or R18, nochange in fluorescence occurred over the course of the experiment (Keller, 1976).There was no change in the fluorescence of mock infected cells (treated the same asvirus infected cells except that no virus was present in the adsorption medium) anda small increase in fluorescence of wild type infected cells (see Discussion) (Fig. 1).

1 0

0 8

0-6

0-4

0-2

I

8-

-

-

i

D D D o ^ \

\

•

1 1

1 1

\s "

1 12 3 4 5 6

Time after infection, h

Fig. 2. The effect of fluorescent probes on the kinetics of cell fusion. Cells were labelledand infected with syn 20 using the standard procedures. For F18- plus Ri8-labelledcells one-half of the standard probe concentration was used for each probe. Infectedcells were harvested at various times after infection and the number of single cells re-maining determined by a Coulter counter assay (Person et al. 1976). Zero time corre-sponds to the time when the cells were incubated following adsorption. The numberof small single cells, N, was measured and plotted as the ratio N/No, where No was theaverage number of small single cells found during the first 5 h after infection. The ratioN/No represents the fraction of cells remaining unfused. Data are shown for ( • ) F18-labelled cells, (O) Ri8-labelled cells, (A) F18- plus Ri8-labelled cells, and (•)unlabelled cells.

The probes at the concentrations routinely used do not perturb the experimentalsystem. Coulter counter fusion kinetics curves were determined for unlabelled cells,cells containing either F18 or R18, and for cells containing F18 and R18. We found nodifference in the time of onset or rate of fusion in any of the cases (Fig. 2). In additionthe probes did not alter the cell or virus growth rates (Figs. 3, 4) or the cell-to-glassor virus-to-cell attachment rates (data not shown). However, these biological functionsare altered if higher probe concentrations are used (Keller, 1976).

172 P.M.Keller, S. Person and W. Snipes

To determine how the increase of fluorescence depended upon the distance ofseparation of probe molecules, the relative number of probe molecules per cell wasvaried. Both probes were always present in the same proportions. Labelled, syn 20 infec-ted HEL monolayers were prepared in the usual way and the increase in fluorescencemeasured 8 h after infection. The percent increase in fluorescence was plotted as a

30

'/bottle

o

X

6f 5

Cel

2

1 i

/

[/

i i

i

_

-

1

20 40 60Time after seeding, h

80

Fig. 3. The effect of the fluorescent probes on cell growth. Previously labelled HELcells were seeded into 2-oz. (57-ml) bottles at 3 x 10s cells per bottle at timeo and in-cubated at 37 °C in complete growth medium. (5 ml per bottle). Cells were labelled asnoted in the Experimental procedures except that when both probes were used theconcentration of each probe was one-half of that normally used. At the indicated timesthe cells were harvested as indicated in the Experimental procedures and counted usinga Coulter counter. The number of cells per bottle is plotted as a function of time afterseeding for ( • ) Fi8-labelled cells, (O) Ri8-labelled cells, (A) F18- plus Ri8-labelledcells, and (Q) cells containing no label.

function of the relative probe concentration (Fig. 5). 100% of relative quanta trans-ferred was taken as the amount observed when the concentration was equal to thestandard concentration. Data are plotted on a log-log plot and the straight lines givenare calculated for the expected fluorescence increase due to photon emission andreabsorption (1/r2) and for resonance energy transfer (1/r6). Lines are given for both2-dimensional and 3-dimensional cases, that is when the increase in fluorescence isdue to the interaction of probe molecules along a surface or due to the interaction ofprobe molecules distributed throughout a volume. The data give a reasonable fit toeither of the resonance energy transfer lines, and surprisingly give an especially goodfit for the interaction of probe molecules in a 3-dimensional distribution (seeDiscussion).

Fluorescence and cell fusion 173

In related experiments the total number of cells per Petri dish was kept constant andthe ratio of F18 to R18 dyed cells was varied. Cells were virus-infected and treated inthe standard manner and the fluorescence increase measured 8 h after infection. Maxi-mum increase of fluorescence was observed when 40-50 % of the cells were labelledwith R18 (Fig. 6).

Du.Q.

107

108

1[

105

I

-

9

I

O j

o /

1A

i

i i

>

o

1 1

1

1

1 O

8 -1

— ic

8-

-

i8 12 16 20Time after infection, h

24 28

Fig. 4. The effect of fluorescent probes on virus growth. Previously labelled cells (seeFig. 3) were seeded into 2-oz. (57-mJ) bottles at 27 x io" cells per bottle to give a mono-layer upon cell attachment. One hour later the cells were infected with syn 20 virus atan MOI of 10 adsorbed PFU/cell. The infected cells were incubated at 34 °C incomplete growth medium (5 ml/bottle). At the times indicated the monolayer wasfreeze-thawed 3 times to release intracellular virus and the number of PFU/mlmeasured by a plaque assay (Person et al. 1976). The number of PFU/ml is plotted asa function of time after infection for (#) Fi8-labelled cells, (O) Ri8-labelled cells,(A) F18- plus Ri8-labelled cells, and (•) cells containing no label.

The interaction of Fi8-labelled HSV-i virions with Ri8-labelled cells was alsoobserved. Labelled virus was added to cells and the R18 fluorescence measured atvarious times thereafter. There is an immediate and rapid increase in fluorescence asa function of time after adding virus. The increase is similar to the rate of virus attach-ment. When the labelled virus was incubated with HSV-i antibodies for 30 min priorto infection no increase in fluorescence was observed when the virus was added tocells (Fig. 7). Similar data were obtained using labelled wild type virus instead ofsyn 20 (data not shown).

To demonstrate visually that the probes partition into cells and mix only as a resultof fusion, cells were labelled and fluorescent micrographs were taken. Fi8-labelledcells fluoresced green and Ri8-labelled cells fluoresced yellow-orange (Fig. 8). Atearly times in infection for mock infected, wild type infected or syn 20 infected cells,the cells are distinct and either green or yellow-orange (Fig. 8A). The same was true

12 CEL 28

174 P- M. Keller, S. Person and W. Snipes

later in infection for mock infected and most wild type infected cells. However, forsyn 20 infected cells 8-5 h after infection, cells which were fused had an intermediatecolour (Fig. 8 B).

100

80

60 -

oc

fe 20 -

ro•3cr

ID

cr

4 -

0-2 0-4 0-6 0-8 1 0Relative concentration

Fig. 5. R18 fluorescence as a function of probe concentration. Labelled HEL cellswere infected with syn 20 and the cells prepared for fluorescence measurements by theusual procedures. However, in these experiments, the concentrations of F18 and R18,although kept in the same proportions, were varied from o to 1 o, 1 -o representing theusual F18 and R18 dye concentrations. Fluorescence was measured 8 h after infection,at a time when the fluorescence increase had normally plateaued; 100% fluorescencewas taken as the fluorescence at the 1 -o dye concentration. The data points are indicatedas ( • ) and the lines were calculated as follows. Let iV represent the number of probemolecules incorporated into cellular membranes and assume it is proportional to thenumber added to the growth medium. If the cellular probes are distributed in 3dimensions (3D) the average separation between probes (r), r oc (1/iV)1'* or N oc i/r*.In 2 dimensions (2D), N <xi/rV Photon emission and reabsorption (PER) has a 1/r1

distance dependence and resonance energy transfer (RET) has a i/r9 distance depen-dence. Fluorescence enhancement (FE) oc donor absorption (proportional to N) xenergy transfer from donor to acceptor (proportional to i/ra or i/r'). For the 2D case(FE)PEE oc N* and ( F E W °C N*. For the 3D case (FE)PEtt oc JV»/» and (FE)REToc JV*. The slopes of the lines in the figure are equal to the exponents of N. A, PER,3D; B, PER, 2D; C, RET, 3D; D, RET, 2D.

DISCUSSION

Two species of fluorescent membrane probes and a fluorometer to measure theirinteractions in cells have been developed. The probes, at the concentrations used, donot perturb the experimental system. Probes were not metabolically altered and appear

Fluorescence and cell fusion

6 -

S 4 -

2 -

.a

/

[/' O

1

i

i

\

i

-

0-2 0-4 0-6 0-8Fraction of F18 labelled cells

i | | [ _ _

0-8 0-6 0-4 0-2Fraction of R18 labelled cells

Fig. 6

20 40 60

Time after infections , minFig. 7

Fig. 6. The maximum fluorescence enhancement as a function of the fraction of Fi8-or Ri8-labelled cells. Labelled cells were infected with syn 20 and the cells preparedfor fluorescence measurements by the usual procedures. However, in these experimentsthe fraction of Fi8-labelled cells to Ri8-labelled cells was varied. The fraction ofFi8-labelled cells increases from left to right; the fraction of Ri8-labelled cells in-creases from right to left. Fluorescence was measured 8 h after infection, at a timewhen the fluorescence increase had normally plateaued.

Fig. 7. R18 fluorescence in a virus-cell system and the attachment of virus to host cells.Fi8-labelled syn 20 virus was added to 3 x ioe Ri8-labelled cells containing twice theusual probe concentration and the fluorescence increase measured as a function of time.Sufficient virus was added in 02 ml of growth medium to give about 10 PFU/cell.Data are shown for syn 20 virus labelled with F18 ( # ) and for the same preparationsof virus incubated with HSV-i antiaera for 30 min prior to mixing the virus withlabelled cells (O). Data for the attachment of syn 20 to HEL cells are also given ( x ).

to partition exclusively into membranes. Experiments also indicate that there is nodetectable loss of probe molecules into the medium during cell fusion, and the fluor-escence micrographs are a visual indication that the probes remain in separate cells untilcell fusion occurs. We conclude that the mixing of the probes occurs only in themembranes of fused cells, and that the intermixing is a valid assay of cell fusion. Thefact that the fluorescence increases as a result of fusion shows that at least some of thelipid regions of the cell membranes are free to diffuse and mix in the plane of themembrane.

At 38 °C for syn 20 infected cells an increase in fluorescence begins 4-5 h after in-fection and increases until about 7 h after infection. For wild type infected cells therewas about a 20 % increase in fluorescence relative to syn 20 infected cells. Fusionbegins in the wild type infected cells but shortly thereafter the rate of fusion decreasessharply as has been noted previously (Person et al. 1976). A model to account for thisfinding postulates that a fusion factor is present in wild type infected cells but only asmall amount of fusion is produced, perhaps due to the subsequent appearance on thecell surface of glycoproteins that inhibit fusion (Knowles, 1976).

176 P. M. Keller, S. Person and W. Snipes

The studies on the maximum increase in fluorescence as a function of probe con-centration indicate that the fluorescence enhancement is due to resonance energytransfer between F18 and R18. We estimate that cells labelled under standard con-ditions contain approximately 3 x 1 o8 probe molecules per cell after fusion has occurred.Microscopic observation at large concentrations reveal that the probes rapidly diffusethroughout the membranes of the cell. Taking 24 x io12 nm3 (2-4 x io15 A3) as the cellvolume, and assuming a random distribution of probe molecules within this volume, wecalculate an average separation between probes of approximately 20 nm (200 A). Anysequestering or localization of probes in cellular structures would result in a smallerseparation between probe molecules.

Since the probes are localized in planar membrane structures, it might be expectedthat their interactions would follow the kinetics of a 2-dimensional array, providedthat the separation between probes is small compared to the distance between layersof membranes. Our data best fit the kinetics derived for 3-dimensional interaction ofprobe molecules via the process of resonance energy transfer. Since the probes appearto be spread throughout all of the cellular membranes they are presumably present onboth sides of a given membrane and some energy transfer may be occurring betweenprobes on opposite sides of the bilayer. Alternatively, the 3-dimensional effect couldresult from a large contribution from those probes located in intracellular membranestructures where many layers of closely packed membranes are prominent. The goodfit to 3-dimensional kinetics may be fortuitous, with a small contribution from photonemission and reabsorption reducing the slope from that expected for resonance energytransfer in 2 dimensions.

It was also possible to observe the interaction between the Fi8-labelled virus andRi8-labelled cells. An increase in fluorescence was observed and the kinetics wereconsistent with the adsorption of virus to the host cell. It is not clear whether theincrease in fluorescence is due to virus adsorption, phagocytosis or fusion of viral andcell membranes. These data are presented only to show that the probe assay involvingan increase in fluorescence has potential wide applications beyond those of cell fusionmeasurements. In this regard a very recent report used resonance energy transfer tomeasure the distribution of concanavalin A on the cell surface of normal and trans-formed cells (Fernandez & Berlin, 1976). Although the result was largely known, the

Fig. 8. Fluorescence micrographs of mock-infected and syn 20 infected labelled cells.Cells were labelled using 1-5 times the normal probe concentration of F18 and R18.Cells were harvested, mixed in equal proportions and about 6 x io5 cells seeded into30-mm culture dishes containing coverslips. They were examined under a Leitz fluor-escence microscope using excitation and emission filters that allowed the simultaneousobservation of both probes. Daylight high-speed Ektachrome (Kodak, Rochester, NY)film was used and developed at an ASA of 1000. The exposure times were approxi-mately 2 min. Photographs are shown for mock infected cells, 1-5 h (A), and for syn 20infected cells, 8-5 h (B) after infection. Fi8-labelled cells are green, Ri8-labelled cellsare yellow-orange and fused cells appear as a composite of these colours, x 300approx.

P. M. Keller, S. Person and W. Snipes

8A

B

(Facing p. 176)

Fluorescence and cell fusion 177

experiments definitely demonstrate that resonance energy transfer can be used tomeasure the change in distribution of a molecular species on the cell surface.

We acknowledge discussion of data with Robert W. Knowles, Thomas C. Holland andG. Sullivan Read. The initial work was stimulated by a discussion with Dr Alec D. Keith.The excellent technical assistance of Susan C. Warner and V. Craig Bond is acknowledged aswell as help by David Spencer in the construction of the fiuorometer. The research was sup-ported by grants from the Public Health Service (AI 11513) and the U.S. Energy Researchand Development Administration [E(n-i)-34i9]. P.M.K. was a PHS predoctoral trainee.

REFERENCESBECKER, J. S., OLIVER, J. M. & BERLIN, R. D. (1975). Fluorescent techniques for following inter-

actions of microtubule subunits and membranes. Nature, Lond. 254, 152-154.DEVAUX, P. & MCCONNELL, H. M. (1972). Lateral diffusion in spin-labeled phosphatidyl-

choline multilayers. J'. Am. Chem. Soc. 94, 4475-4481.FERNANDEZ, S. M. & BERLIN, R. D. (1976). Cell surface distribution of lectin receptors deter-

mined by resonance energy transfer. Nature, Lond. 264, 411-415.FOLCH, J., LEES, M. & SLOANE-STANLEY, G. (1957). A simple method for the isolation and

purification of total lipids from animal tissues. J. biol. Chem. 226, 497-509.FORSTER, T. (1948). Zwischenmolekulare Energie Wanderung und Fluoreszenz. Anvln Phys.

2. 55-75-FRYE, L. D. & EDIDIN, M. (1970). The rapid intermixing of cell surface antigens after form-

ation of mouse-human heterokaryons. J. Cell Set. 7, 319-335.GABOR, G. (1968). Radiationless energy transfer through a polypeptide chain. Biopolymers 6,

809-816.HAUGLAND, R. P., YGUERABIDE, J. & STRYER, L. (1969). Dependence of the kinetics of singlet-

singlet energy transfer on spectral overlap. Proc. natn. Acad. Sci. U.S.A. 63, 23-30.HUBBELL, W. L. & MCCONNELL, H. M. (1969). Motion of steroid spin labels in membranes.

Proc. natn. Acad. Set. U.S.A. 63, 16-22.KELLER, P. M. (1976). A Fluorescence Enhancement Assay of Membrane Interactions. Ph.D.

dissertation. The Pennsylvania State University.KNOWLES, R. W. (1976). Cell Fusion and the Glycoproteins of Herpes Simplex Virus. Ph.D

dissertation. The Pennsylvania State University.KORNBERG, R. D. & MCCONNELL, H. M. (1971). Inside-outside transitions of phospholipids

in vesicle membranes. Biochemistry, N.Y. 10, 1111-1120.LATT, S. A., CHEUNG, H. T. & BLOUT, E. R. (1965). Energy transfer. A system with relatively

fixed donor acceptor separation. J. Am. Chem. Soc. 87, 995-1003.PERSON, S., KNOWLES, R. W., READ, G. S., WARNER, S. C. & BOND, V. C. (1976). The kinetics

of cell fusion induced by a syncytia-producing mutant of herpes simplex virus type I.J. Virol. 17, 183-190.

SINGER, S. J. & NICOLSON, G. L. (1972). The fluid mosaic model of the structure of cell mem-branes. Science, N.Y. 175, 720-731.

STRYER, L. & HAUGLAND, R. P. (1967). Energy transfer: A spectroscopic ruler. Proc. natn.Acad. Set. U.S.A. 58, 719-726.

(Received 21 March 1977)