Parameters Affecting the Fusion of Unilamellar

Phospholipid Vesicles with Planar Bilayer Membranes

FREDRIC S. COHEN,* MYLES H. AKABAS, JOSHUA ZIMMERBERG, and

ALAN FINKELSTEIN
*Department of Physiology, Rush Medical College, Chicago, Illinois 61612; and Departments of
Physiology/Biophysics and Neuroscience, Albert Einstein College of Medicine, Bronx, New York 10461 .
Dr. Zimmerberg's present address is National Institutes of Health Physical Science Laboratory,
Bethesda, Maryland 20014.

ABSTRACT It was previously shown (Cohen, F. S., J . Zimmerberg, and A. Finkelstein, 1980, /.
Gen . Physiol ., 75 :251-270) that multilamellar phospholipid vesicles can fuse with decane-
containing phospholipid bilayer membranes . An essential requirement for fusion was an
osmotic gradient across the planar membrane, with the vesicle-containing (cis) side hyper-
osmotic with respect to the opposite (trans) side . We now report that unilamellar vesicles will
fuse with "hydrocarbon-free" membranes subject to these same osmotic conditions . Thus the
same conditions that apply to fusion of multilamellar vesicles with planar bilayer membranes
also apply to fusion of unilamellar vesicles with these membranes, and hydrocarbon is not
required for the fusion process . If the vesicles and/or planar membrane contain negatively
charged lipids, divalent cation (-15 mM Ca") is required in the cis compartment (in addition
to the osmotic gradient across the membrane) to obtain substantial fusion rates. On the other
hand, vesicles made from uncharged lipids readily fuse with planar phosphatidylethanolamine
planar membranes in the near absence of divalent cation with just an osmotic gradient .
Vesicles fuse much more readily with phosphatidylethanolamine-containing than with phos-
phatidylcholine-containing planar membranes . Although hydrocarbon (decane) is not required
in the planar membrane for fusion, it does affect the rate of fusion and causes the fusion
process to be dependent on stirring in the cis compartment .

The mechanisms causing and controlling fusion of biological branes on the basis oftwo criteria : (a) the transfer of vesicular
membranes are of interest to scientists from many diverse contents across the planar membrane ("exocytosis") (22), and
disciplines . Of the many possible components in the fusion (b) the incorporation into the planar membrane of a mem-
process, the one certainty is that ultimately fusion occurs brane-associated protein originally present in the vesicular
between two lipid bilayers-one from each membrane. Con- membranes (4) . The essential condition for obtaining fusion
sequently, we have been investigating the fusion of phospho- is an osmotic gradient across the planar membrane, with the
lipid vesicles to planar phospholipid bilayer membranes. We cis side (the side containing the vesicles) hyperosmotic with
have tried to answer several questions with this approach: respect to the opposite, trans, side (4) . Although divalent
first, can fusion occur in this system? Second, if it can, what cations at millimolar concentrations on the cis side greatly
conditions are necessary to obtain it and what parameters increase the rate offusion, they are not essential to the process.
control its rate? Third, what are the forces and energetics in We proposed that divalent cations increase the frequency and
the process (that is, what mechanisms underly fusion in this duration of close contact ofvesicles with the planar membrane
system)? Finally, to what extent is the fusion process in this and that fusion results from osmotic swelling ofthese vesicles
model system relevant to biological fusion events, particularly followed by rupture of vesicular and planar membranes in
exocytosis? the region ofcontact (4). In support of this view, Zimmerberg
We previously demonstrated that fusion can occur between et al. (23) reported that micromolar amounts specifically of
phospholipid vesicles and planar phospholipid bilayer mem- Ca" (rather than millimolar amounts of any divalent cation)
THE JOURNAL OF CELL BIOLOGY " VOLUME 98 MARCH 1984 1054-1062
1054 ® The Rockefeller University Press - 0021-9525/84/03/1054/09 $1 .00
stimulated fusion if the planar membrane contained a cal- conductance from nonspecific leaks in the planar membrane . Fig. I illustrates

cium-binding protein, but an osmotic gradient across the the voltage-dependent properties of porin inserted from a detergent solution
into a planar membrane . Channels are open at voltages below ±100 mV and
planar membrane was still essential. begin to close at larger voltages of either polarity. The rate of decline of the
Two features detracted from the possible biological rele- macroscopic conductance induced by many porin channels increases with
vance of these previous studies : first, the vesicles were multi- voltage. Porin channels are similar in many respects to the voltage-dependent

lamellar, rather than unilamellar; second, the planar mem- anion channel obtained from the outer membrane of mitochondria (5, 17) and

branes contained large amounts of hydrocarbon solvent

used as a membrane marker in earlier fusion experiments (4) . We used porin,
rather than voltage-dependent anion channel, in the present experiments be-
(decane) . This paper addresses both issues. We find that the cause of the ease in preparing large, purified quantities.
same conditions required for obtaining fusion of multilamel- Porin was initially obtained as a generous gift from Dr. Carl Schnaitman

lar vesicles to planar membranes also hold for unilamellar (University of Virginia). This, and two subsequent preparations of porin, were

vesicles. We also show that fusion readily occurs with planar

prepared by a modification of the procedure of Rosenbusch (15) (C . Schnait-
man, personal communication) ; the details of the preparation are given in
membranes largely devoid of hydrocarbon solvent ; i .e., hy- reference 1 . Porin was stored dry as a lyophilized powder in the refrigerator or
drocarbon is not required for fusion in this system. Hydro- freezer for at least two years without significant loss of activity . Porin stock

carbon in the planar membrane, however, does modify the suspensions were stored in the refrigerator and showed no loss of activity over

fusion process, which we describe and discuss . In the accom- several months.

panying paper (2) we examine the nature of the interaction Planar Membranes: Two types of planar phospholipid bilayer mem-
branes were used in the present study : (a) "hydrocarbon-free" membranes
and attachment of vesicles with the planar membrane, and formed by folding together two monolayers (10), and (b) decane-containing
demonstrate a "pre-fusion" state that is separable from the membranes formed by the brush technique (11) . Experiments were performed
osmotically driven fusion event. at room temperature in a chamber that consisted of a 125 um thick Teflon
partition separating two lucite compartments, each containing 2 to 3 ml of the
desired salt solution. Each compartment contained a small magnetic stirring
MATERIALS AND METHODS bar which permitted vigorous mixing of the compartment's contents. When

Preparation of Vesicles required, the contents of either compartment could be intermittently or contin-
uously perfused with a fresh solution via Teflon tubes connected to a peristaltic
CHOLATE DIALYSIS PROCEDURE: 4-6 mg of egg phosphatidy1choline pump; the total compartmental volume could be exchanged several times in I
(PC)' in chloroform were dried under nitrogen in a glass tube and then in a min .
vacuum desiccator for 30 min. 0 .1 ml of 50 mg/ml cholate (Aldrich Chemical "Hydrocarbon-free" membranes were formed across a 0.36-mm diameter
Co ., Milwaukee, WI) in "buffer (500 mM KCI or NaCl, 10 mM HEPES, 1 hole that was pre-coated with squalene (14) purified on an alumina column ;
MM MgC12, 0.1 mM EDTA, pH 8 .0), 0 .025 ml of porin stock suspension (I decane-containing membranes were formed across either a 0 .36- or a 1 .0-mm
mg/ml in "buffer") and 0 .375 ml of "buffer" were then added to the lipid- diameter hole from 2 to 5% lipid solutions in ndecane. The salt solutions in
containing tube . (The exact lipid to cholate ratio required to reconstitute porin the two compartments generally consisted of 100 mM KCl or NaCl, 10 mM
into the vesicular membrane varied somewhat among batches of porin and was MES, l mM EDTA, pH 6 .0; for the hydrocarbon-free membranes, 2 mM
empirically determined for each batch.) The mixture was vortexed for l min, MgC1 2 was often included in the salt solution to aid in membrane formation.
sonicated briefly in a cylindrical bath sonicator (Lab Supplies Company, Inc., Asolectin (lecithin type II from Sigma Chemical Co. [St. Louis, MO]) was
Hicksville, NY) to clarity, transferred to a 12,000-mol-wt cutoff dialysis tube purified by the method of Kagawa and Racker (6) to remove neutral lipids; its
(Spectra/Poi 2 membranes, Fischer Scientific Co ., Pittsburgh, PA), and dialyzed composition after this procedure is 40% PC, 33% phosphatidylethanolamine
overnight at 4°C against 200 ml of "buffer." Vesicles were removed from the (PE), 14% phosphatidylinositol, and smaller percentages of lysophosphatidyl-
dialysis tube and stored on ice. Egg PC was from Avanti Biochemical Co . choline and cardiolipin (9) . All other lipids used to form planar membranes
(Birmingham, AL) (see reference 13) . were purchased from Avanti. Salts were reagent grade and used as obtained ;
SON ICATION-FREEZE-THAW PROCEDURE: 5mgofthedesired lipid water was deionized and then distilled in a glass still.
in either chloroform or chloroform:methanol (2:1) was dried under nitrogen in Electrical measurements were made under voltage clamp conditions through
a glass tube and then in a vacuum desiccator for 30 min. 0.475 ml of "buffer" a single pair of saturated calomel electrodes contacting the aqueous solutions
and 0 .025 ml of porin stock suspension (see above) were added to the lipid- through saturated KCI bridges. Generally, a 20 mV potential (cis side positive)
containing tube; the mixture was vortexed for 30 s, transferred to a polycarbon- was applied across the planar membrane and the resulting transmembrane
ate screw-top tube, and sealed under nitrogen . The cloudy suspension was current monitored on a chart recorder and an oscilloscope screen ; the trans-
clarified by sonication for -10 min at 4°C in a Branson Sonifier Model W 185 membrane current is directly proportional to membrane conductance (which
with Cup Horn attachment (Branson Sonic Power Co., Danbury, Conn .). It equals current/voltage), since the voltage is held constant. The compartment
was then rapidly frozen in a dry ice/ethanol bath for 70 s and allowed to thaw to which vesicles were added is called the cis compartment; the opposite
at room temperature . This freeze-thaw cycle was repeated three to five times, compartment is referred to as the trans compartment .
during which time the suspension became cloudy . The cloudy vesicle suspen-
sion was then sonicated in a cylindrical bath sonicator for 15 s to disperse the
vesicles . These vesicles, now ready for use, were prepared fresh daily and stored RESULTS AND DISCUSSION
on ice . All lipids were from Avanti Biochemical Co. (see reference 7).
Reconstitution of porin by both vesicle-formation techniques was assayed Fusion of Unilamellar Vesicles to "Hydrocarbon-
by the vesicles' inability to retain K* when dialyzed against an isosmotic Li' Free" Planar Membranes
solution. (The K* initially inside a vesicle with at least one porin channel in its
membrane will exchange for Li' during the dialysis .) After dialysis, the amount CONDITIONS FOR AND CHARACTERISTICS OF FU-
of K* that remained trapped inside porin-free vesicles was determined, using a SION : When cholate-dialysis, porin-containing unilamellar
K'-selective electrode, from the amount of K* released from the vesicle. by vesicles are added to the aqueous phase on one side (the cis
gramicidin A . This technique was also used to determine the average diameter
of vesicles prepared in the absence of porin. The volume-average diameter of
side) of a "hydrocarbon-free" planar bilayer membrane, no
both cholate dialysis and freeze-thaw vesicles was -700 .4 . incorporation of porin channels occurs in the planar mem-
Porin : Porin is a large (110,000 mol wt) intrinsic membrane protein that brane, regardless of the incubation time. Thus, porin does not
forms channels in the outer membrane of gram-negative bacteria and is spontaneously transfer from vesicular to planar membranes.
responsible for the permeability of that membrane to small solutes (12). It is If an osmotic gradient of -200 mosM or higher is now
established across the planar membrane by addition of urea
easily purified in large quantities, and remains stable even under such harsh
conditions as sonication and exposure to concentrated detergent solutions. It is
a particularly good marker for fusion because of its characteristic voltage- or KCI to the cis compartment, a few porin channels may
dependent properties (18, 19), which make it easy to distinguish porin-induced insert over the next several minutes (Fig. 2). Under voltage-
clamp conditions, these insertions appear as discrete current
'Abbreviations used in this paper. PC, phosphatidylcholine; PE, (conductance) jumps. If Ca" is now added to the cis com-
phosphatidylethanolamine ; PS, phosphatidylserine. partment to a concentration of 5 to 15 mM, the rate of
COHEN ET AL . UnilamellarVesiclesandBilayerMembranes 105 5
 v
N

FIGURE 1 The insertion of porin channels and the voltage-dependent behavior of porin-induced conductance in planar lipid
bilayers . An asolectin membrane clamped at V = 20 mV separates symmetric solutions (100 mM NaCl, 10 mM MES, 1 mM EDTA,
pH 6.0) . At the start of the record, porin in cholate buffer (0 .1 mg/ml porin, 1 mg/ml cholate, 500 mM KCI, 10 mM HEPES, 1 mM
M9Cl 2 , 0 .1 mM EDTA, pH 8 .0) is added to the cis compartment to a concentration of 1 .7 x 10' mg/ml . Shortly after this addition,
the transmembrane current begins to increase in a step-like manner . 31/2 min later an osmotic gradient is formed by addition of
NaCl to the cis side to a concentration of 50 mM . Aside from the instantaneous increase in the membrane conductance due to
the increase in the salt concentration, this does not affect the rate of channel insertion into the planar bilayer . The addition to
the cis side of CaCl2 to a concentration of 10 mM stops the step increases in conductance . Ca*` seems to inhibit the insertion
into the planar membrane of porin in detergent solution, perhaps by chelating the cholate in the compartment . In the inset, the
voltage dependence of the porin-induced conductance is shown . The transmembrane voltage is clamped to the voltages
indicated . At voltages >100 mV the conductance declines; the larger the voltage the faster the decline . The decline in the
conductance is indicative of the closing of the porin channels .

channel insertion is greatly stimulated (Fig. 2). This slows membrane is caused by the osmotic gradient (cis side hyper-
steadily with time, and insertions eventually cease to occur osmotic with respect to the trans side) and not by a direct
after several minutes to more than an hour. If, while channel effect ofosmoticants on the vesicles. (We have never observed
insertion is occurring, the osmotic gradient is abolished by porin insertion when the trans side was hyperosmotic with
addition of osmoticant to the trans compartment, channel respect to the cis side) .
insertion ceases ; it resumes when the osmotic gradient is re- Insertion rates can be so high (as in Figs. 2-4) that individ-
established (Fig. 3) . ual events cannot be resolved. If, however, fewer vesicles are
The osmotic gradient can be created by making either the added to the cis compartment, one can clearly resolve the
cis side hyperosmotic (Fig. 2) or the trans side hypeosmotic individual current jumps and show that each represents the
(Fig. 4). Thus, insertion of porin channels into the planar simultaneous insertion ofmany porin channels into the planar
1056 THE JOURNAL OF CELL BIOLOGY - VOLUME 98, 1984
 particularly if the planar and/or vesicular membrane contain
negatively charged lipids. Ca++ is not unique in this respect .
10 nA Mg++ and Ba++ are also effective, although about twofold
I Min higher concentrations of Mg++ are required to achieve similar
rates .
The osmotic requirement and Ca++ stimulation of channel
insertion into planar membranes exist only if porin has been
incorporated into vesicles. Porin-free vesicles added to the cis
compartment, whether followed or preceeded by the addition
ÉÉ É of an osmoticant and Ca++, induce no change in the con-
c) o N ductance of planar membranes. Porin added directly from its
stock suspension, with or without an osmotic gradient and
Ca++, produces no change in planar membrane conductance,
nor does the addition of porin-free vesicles alter this result.
When porin solubilized in cholate is added to the cis com-
partment, channels insert singly, or infrequently as small
multiples, rather than as large multiples (see Fig. 1), and the
insertion rate is not increased by an osmotic gradient across
the membrane. Ca` added to the cis compartment actually
FIGURE 2 Fusion of unilamellar vesicles to a "hydrocarbon-free" inhibits, rather than stimulates, channel insertion, possibly
planar membrane . An asolectin membrane clamped at V = 20 mV because Ca++ chelates with cholate. Thus, only ifporin is first
separates symmetric solutions (100 mM NaCl, 10 mM MES, 3 mM reconstituted in vesicles does an osmotic gradient and Ca++
MgCl 2 , 0 .1 mM EDTA, pH 6 .0) . At the start of the record, cholate promote incorporation of porin channels into planar mem-
dialysis PC vesicles (-10"/ml) are added to the cis compartment. branes.
No fusion events (current jumps) occur during the next 3 '/2 min . SIMILARITY OF RESULTS WITH UNILAMELLAR AND
Urea is then added to the cis compartment to a concentration of
MULTILAMELLAR VESICLES : All of the features of porin
400 mM . Two or three fusion events (current jumps) occur over the
next 4 min . After the addition of CaCl2 to the cis compartment to a channel insertion into planar membranes from unilarnellar
concentration of 15 mM, the rate of fusion (current jumps) dramat-
ically increases. (Inset) Demonstration of the voltage dependence
of the fusion-induced conductance . After fusion has proceeded for
12 min (during which time CaCl 2 was also added to the trans
compartment to a concentration of 15 mM), the voltage across the
planar membrane is stepped from 0 to 20 mV and back to 0, and
then stepped from 0 to 120 mV . Note that in the latter case, the
current (conductance) declines in a manner characteristic of porin
channels closing, thus demonstrating that the current jumps seen
in the main record resulted from the insertion of porin channels
into the planar membrane .

membrane. ("Simultaneous" here means "within 200 gs",

which is our time resolution). In Fig. 5, for example, only one
current jump has occurred. When the voltage across the planar
membrane was increased from 20 to 100 mV, the current
decreased stepwise, each step corresponding to the turn off of
one to three porin channels. In this record, about 15 porin
channels were associated with the current jump. Ifthe voltage
is increased from 20 mV to >_ 100 mV across a membrane in
which there have been numerous insertions, a macroscopic
decline of current occurs, characteristic of many porin chan-
nels turning off (see Fig. 1 and inset to Fig. 2). We routinely
checked in this way that the large conductances obtained in FIGURE 3 Demonstration that an osmotic gradient is required
many of our experiments resulted from insertions of porin across the planar membrane for fusion to occur . A 'hydrocarbon-
channels, and not from nonspecific "leaks" induced in the free" asolectin membrane clamped at V = 20 mV separates the
planar membranes. same symmetric solutions as in Fig . 2 . At the start of the record,
unilamellar cholate dialysis PC vesicles (-10"/ml) are added to the
An osmotic gradient across the planar membrane is abso-
cis compartment, and no fusion events (current jumps) occur even
lutely essential to the above phenomena; Ca" by itself does after the addition of CaCl2 (15 mM) to the cis side . However, when
not induce channel insertion, even at symmetrical concentra- urea is added to the cis side to a concentration of 400 mM, fusion
tions of 100 mM. In Fig. 3, for example, we see that 15 mM events (current jumps) occur at a rapid rate . These cease to occur
Ca" alone is ineffective, but the subsequent application of shortly after the abolishment of the osmotic gradient across the
an osmotic gradient induces massive incorporation of chan- planar membrane by the addition of urea to the trans side to a
nels into the membrane. 5-15 mM Ca++ in the cis compart- concentration of 400 mM ; they resume, when the osmotic gradient
ment does, however, greatly increase the rate of channel is re-established by a second addition of urea (increment of 400
insertion when an osmotic gradient already exists (see Fig. 2), mM) to the cis side .

COHEN ET AL. Unilamellar Vesicles and Bilayer Membranes 105 7

 in 100 mM KCL is - 1 .5 x 10 - ' ° mho [ 18; personal observa-
tions of the authors].) This mean may be artifactually high,
as it was often difficult to resolve small current steps from the
background noise . The large range may reflect the heteroge-
neous size distribution of sonicated-freeze-thaw vesicles. This
was very apparent in electron micrographs of negatively
stained preparations, with vesicle diameters ranging from 200
to several thousand angstroms . When assayed by the K+
release method (see Materials and Methods), porin-free vesi-
cles had an average diameter of -700 Á.
Cholate dialysis and sonication-freeze-thaw vesicles offer a
significant advantage over large multilamellar vesicles for
Li
U studying fusion. Because much greater numbers ofthe smaller
N
U,
N U vesicles can be added to the cis compartment, much higher
t.)
N NO rates offusion are obtainable . In typical experiments, between
10' and 10 4 fusion events occur in 10 min, whereas fewer
N
U
than 100 events occur with large multilamellar vesicles (4).
1 (Normalizing for the number of vesicles added to the cis side,
approximately the same number of fusion events are seen
FIGURE 4 Demonstration that fusion is induced when the osmotic over a wide variety of vesicle sizes.) This permits a more
gradient across the planar membrane is created by making the trans detailed study ofthe parameters affecting the fusion process.
side hypoosmotic . A "hydrocarbon-free" asolectin membrane sep-
arates symmetric solutions (100 mM NaCl, 300 mM urea, 10 mM "Hydrocarbon-Free" vs. Decane-
MES, 3 mM CaCl 2 , 0 .1 mM EDTA, pH 6 .0) . At the start of the record, containing Planar Membranes
cholate dialysis PC vesicles (-10"/ml) are added to the cis side
followed 2 min later by the addition of CaCl2 (15 mM) to the same The earlier study by Cohen et al. (4) left unresolved the role
side . No fusion events (current jumps) occur until the trans side is of hydrocarbon in their planar membranes: was it essential to
perfused with urea-free buffer (100 mM NaCl, 10 mM MES, 3 mM
CaC1 2 , 0 .1 mM EDTA, pH 6 .0), thus establishing an osmotic gradient
across the planar membrane .

vesicles-the simultaneous insertion of many channels, the

requirement for an osmotic gradient across the planar mem-
brane (cis side hyperosmotic), the stimulation of insertion

L
rate by addition ofdivalent cations to the cis compartment-
pertain to insertion of porin channels from multilamellar
vesicles. More importantly, these are the same features pre-
viously described for the fusion of multilamellar vesicles with 100 pA

decane-containing planar bilayer membranes (4, 22). We 0 .5 min

therefore conclude that the insertion of porin channels into
planar bilayers described in Figs. 2-5 results from fusion of
unilamellar vesicles with planar membranes. Furthermore,
the fusion process is the same for unilamellar and multila-
mellar vesicles.
This last point is of considerable practical importance . The
unilamellar vesicles described in this section were made by
the cholate dialysis procedure . Although we readily reconsti- 0
E

tuted porin into vesicles made from PC, we were not as 0

successful with other lipids; the titration of lipid, porin, and E

0
cholate proved rather critical and difficult to reproduce . On
the other hand, there was no difficulty reconstituting porin FIGURE 5 Demonstration of simultaneous insertion of many chan-
into any desired lipid mixture by the sonication-freeze-thaw nels into a planar membrane. An asolectin membrane clamped at
technique . Although vesicles formed this way are said to be V = +20 mV separates assymmetric solutions . The cis side contains
unilamellar (7), a significant fraction are undoubtedly not (C. cholate dialysis PC vesicles (-10 9/ml) and 200 mM urea, 100 mM
Miller, personal communication) . Since, however, the fusion NaCl, 10 mM MOPS, 10 mM CaCl 2 , 0 .1 mM EDTA, pH 7 .0 . The
process is apparently similar for both unilamellar and multi- trans side contains 100 mM NaCl, 10 mM MOPS, 0 .1 mM EDTA,
pH 7 .0 . No change in the membrane conductance has occurred
lamellar vesicles, we could study it using sonicated-freeze-
before the start of the record . About 1 min after the start of the
thaw vesicles without qualms . Most of our experiments were record, at point a, there is a single step jump of 160 pA in the
with vesicles prepared in this way. transmembrane current . Half a minute later, the voltage is switched
The number of channels incorporated into a planar mem- from +20 to +100 mV . The current begins to decline immediately
brane by a single fusion event ranged from one to several in step-like fashion, each step indicative of the closing of one to
hundred, with a mean of about twenty . (The number of three porin channels . This shows that the single current jump at
channels in a fusion event was determined from the size of point a resulted from the simultaneous insertion of many porin
the current jump. The single channel conductance of porin channels into the planar membrane (in this case about 15) .

058 THE JOURNAL OF CELL BIOLOGY " VOLUME 98, 1984

or did it significantly affect the fusion process? Low fusion reduced the number of fusion events in an experiment to a
rates with large multilamellar vesicles precluded the use of mean of one or less. In the present system, with many more
"hydrocarbon-free" membranes, since such membranes can fusion events, we could investigate the effect of hydrocarbon
only be formed across holes significantly smaller than 1 mm on fusion. By forming decane-containing and "hydrocarbon-
in diameter-the size oftheir decane-containing membranes. free" bilayers of essentially the same area, we could directly
The decrease in membrane surface area alone would have compare the two types of membranes.
Fusion of vesicles to "hydrocarbon-free" and decane-con-
H
taining planar membranes occurs under the same set of
N
Ul conditions. Thus, an osmotic gradient across the planar mem-
_U
a
O L
brane (cis side hyperosmotic) is essential to the process, and
(n N y~ the rate of fusion is dramatically increased by millimolar
a)
> a m concentrations of Ca" in the cis compartment, when the
planar and/or vesicular membrane contain negatively charged
lipids. (The concentration of Ca" that stimulates fusion in
"hydrocarbon-free" membranes (- 15 mM) is slightly higher
than that which stimulates fusion in decane-containing mem-
branes (5-10 mM) . Also, stimulation by Ca" is usually
delayed -30-60 s in "hydrocarbon-free" membranes but not
in decane-containing membranes.) Osmotic gradients about
twice as large as those applied to decane-containing mem-
branes are required for comparable rates of fusion to "hydro-
carbon-free" membranes ; this may simply reflect a smaller
water permeability of the latter . Thus, the presence of large
amounts of hydrocarbon in the planar membrane is not a
requirement for fusion ; fusion occurs between unilamellar
Stirring dependence of fusion to a decane-containing
FIGURE 6
vesicles and "hydrocarbon-free" membranes. (Because the
membrane . An asolectin membrane clamped at V = 20 mV sepa-
hole in the partition across which "hydrocarbon-free" mem-
rates symmetric solutions (100 mM KCI, 10 mM MES, 1 mM EDTA,
pH 6 .0) . Before the start of this record, 200 mM urea and 10 mM branes were formed was pre-coated with squalene, we cannot
CaC1 2 were added to the cis side . With both compartments stirred, be certain that these membranes are absolutely hydrocarbon-
PC vesicles (-10"/ml) are added to the cis compartment . About 20 free, although the amount of squalene in the bilayer is mini-
s later fusion begins . Stirring is stopped at point A, and within 1 min mal (20, 21). We have also obtained fusion with membranes
fusion ceases . When stirring is resumed at point B, fusion resumes. formed across a hole that was pre-coated with vaseline . Even

00

_y y Ü
l
- OV N+
O
7Y 7 U

FIGURE 7 The effect of sonicated decane on fusion to a "hydrocarbon-free" membrane . An asolectin "hydrocarbon-free"
membrane clamped at V = 20 mV separates symmetric solutions (100 mM KCI, 10mM MES, 2 MM M9Cl 2, 0 .1 mM EDTA, pH
6 .0) . Vesicles (egg PC ; --10"/ml) are added to the cis compartment, followed by 450 mM urea and 15 mM CaCl 2 . After the
addition of calcium, fusion begins and is not affected when stirring ceases 8 min later . Stirring is resumed in the intervening 12
min, at the end of which fusion has nearly plateaued . 2 min after the addition to the trans compartment of 50 ul of sonicated
decane in buffer (1 ml decane in 15 ml buffer), the fusion rate accelerates . When stirring is now stopped, the fusion rate slows
and eventually ceases after -4 min; fusion resumes when stirring is begun again . This decane-accelerated fusion is still dependent
on the existence of the osmotic gradient . The reduction of the osmotic gradient by the addition of 300 mM urea to the trans
compartment stops fusion .

COHEN ET AL . Unilamellar Vesicles and Bilayer Membranes 105 9

in these membranes there may be small amounts of hydro-
carbon in the membrane .) A
Fusion with decane-containing membranes differs from
that with "hydrocarbon-free" membranes in one significant
respect : the former is sustained only while the aqueous phases
are stirred. Fusion, which is proceeding normally, ceases
within 1 min after stirring is stopped and commences again
shortly after it is resumed (Fig. 6). This phenomenon is
repeatable several times on the same membrane. In contrast,
fusion of vesicles to "hydrocarbon-free" membranes is totally
unaffected by stirring. It is only stirring in the vesicle-contain-
ing (cis) compartment that influences fusion to decane-con-
taining membranes; fusion rate is unaffected by stirring in the
trans compartment .
What is the possible meaning of these differences between
"hydrocarbon-free" and decane-containing membranes? We
believe that there are two steps in the fusion process: first,
attachment of vesicles to the planar membrane leading to a 1 .5 nA
"pre-fusion" state, and second, osmotic swelling of these
attached vesicles that can ultimately result in fusion. In the I min
accompanying paper (2) we show that these two steps are
experimentally separable and that it is the formation of the
pre-fusion state that is stirring dependent in decane-contain-
ing membranes . A possible interpretation ofthe stirring effect,
therefore, is that the pre-fusion state cannot form in the bilayer
proper of decane-containing membranes, but can occur in
the toroidal region. Stirring in the cis compartment may then
sweep these vesicles into the bilayer region where they can
osmotically swell and fuse. The differences between PE and B
PC, which we attribute to the pre-fusion process (2), are
masked in decane-containing membranes, possibly because
attachment occurs in the toroidal region.
If the pre-fusion state is formed in the toroidal region of N_
decane-containing membranes, where is it formed in "hydro- U
Co
carbon-free" membranes? At present we cannot say with N U
certainty that this occurs in the bilayer proper. The energy U
to
O
N
t
N
barrier to vesicle-planar membrane attachment, which we > Û
j-

discuss in the accompanying paper, may be fortuitously low

in the boundary region between monolayer and bilayer. If so,
other factors must exist in biological systems that reduce the
Demonstration that vesicles fuse much more readily to
energy barrier for vesicle attachment to the plasma mem- FIGURE 8
a PE-containing, negatively charged "hydrocarbon-free" membrane
brane. than to a PC-containing, negatively charged "hydrocarbon-free"
A "hydrocarbon-free" membrane can be converted to a membrane . (A) Fusion to a plant PEPS membrane. A plant PEPS
decane-containing membrane by adding a suspension of (4 :1) membrane clamped at V = 20 mV separates symmetric solu-
decane droplets to the trans compartment (Fig. 7). The ca- tions (100 mM KCI, 10 mM MES, 2 mM MgCl2 , 0 .1 mM EDTA, pH
pacitance of the planar membrane declines to -60% of its 6 .0) . Vesicles (egg PC :PS, 4 :1 ; -10"/ml) are added to the cis
previous magnitude, presumably because of decane entry into compartment, and in the subsequent 2 min no fusion occurs . When
the membrane, and fusion is affected in two ways: first, the an osmotic gradient is formed by addition of urea to the cis side to
rate increases dramatically, and second, it is now stirring- a concentration of 450 mM, fusion occurs . (The reason for the
dependent . exceptionally large size of two of the fusion events is unclear .) CaC1 2
is then added to the cis side to a concentration of 15 mM . This
Effect of Lipid Composition on induces a high rate of fusion . (The reason for the absence of the
large events after the addition of Ca" is also unclear .) Fusion
Fusion ("Hydrocarbon-Free" Bilayers) continues at a high rate for over 12 min . (B) Fusion to a plant PC :PS
NEGATIVELY CHARGED LIPIDS: An osmotic gradient membrane . A plant PC :PS (4 :1) membrane clamped at V = 20 mV
and 15 mM Ca" are both necessary to obtain fusion when separates the same symmetric solutions as in A . Vesicles (egg PC :PS,
vesicles and/or planar membranes contain negatively charged 4 :1 [the same vesicle preparation as used in A] ; ---10"/ml) are added
lipids (e.g., Figs. 2 and 8). (Occasionally low rates of fusion to the cis compartment, and in the subsequent 2 min no fusion
occurs . When urea is added to the cis side to a final concentration
occur with just an osmotic gradient, as in Fig. 2 and 8).
of 450 mM, there is only one fusion event during the next 2 min
Examples include fusion ofPC and PC:PE (1 :1) vesicles with (in contrast to the many seen in A under the same circumstances) .
either asolectin, PC:PS (4:1), or PEPS (4 :1) membranes; CaCl 2 is then added to the cis side to a concentration of 15 mM .
fusion of PC:PS (4:1) and PEPS (4:1) vesicles with PC, PE, Fusion begins -1 min later and continues at a slow rate . This slow
and PC:PE (1 :1) membranes, and fusion of PC:PS (4:1) and rate contrasts with high fusion rates to plant PEPS (seen in A) and
PEPS (4:1) vesicles with asolectin, PC:PS (4:1) and PEPS asolectin membranes .

06 0 THE JOURNAL OF CELL BIOLOGY " VOLUME 98, 1984

 process does not require negatively charged lipids in both
membranes. In fact, it appears that it does not require them
in either membrane. Vesicles made from uncharged lipids
(PC :PE [1 :11 or PC) readily fuse with planar PE membranes
with just an osmotic gradient (Fig. 9A) . (For reasons of
membrane stability, we always included small concentrations
ofdivalent cation (generally 2 mM Mg") when working with
"hydrocarbon-free" membranes. We therefore cannot defini-
tively declare that fusion can occur with "hydrocarbon-free"
membranes in the complete absence of divalent cations. We
can say, however, that fusion does occur with decane-contain-
ing membranes in the complete absence ofdivalent cations .)
(15 mM Ca" increases the fusion rate, possibly because of
negatively charged contaminants in the lipids. The significant
rates of fusion that occur with no Ca" present illustrate that
these presumed negative contaminants are not necessary for
fusion . When known amounts of negatively charged lipids
are present in the vesicular or planar membrane, such high
fusion rates do not occur in the absence of Ca" .) Thus,
neither Ca" nor negatively charged lipids appear to be nec-
essary for fusion to occur .
PE V S . PC: Vesicles fuse much more readily with PE-
containing than with PC-containing planar membranes. This
is apparent in three ways. First, good fusion rates occur with
PE membranes (see Fig. 9A), but not with PC membranes
(Fig. 9B). Second, an osmotic gradient alone will induce
reasonable numbers of fusion events (which quickly cease)
with PE:PS (4:1) planar membranes but not with PC:PS (4:1)
membranes (Fig. 8). Third, after Ca" addition, fusion rates
are some 30-fold higher with PE:PS (4:1) membranes than
with PC:PS (4:1) membranes (Fig. 8). (An osmotic gradient
alone will not induce significant fusion with a negatively
charged planar membrane containing both PE and PC [e.g.,
asolectin], but after Ca" addition, the fusion rate is the same

N N
V û
0 t
d N
O
U

FIGURE 9 Fusion of uncharged vesicles to uncharged "hydrocar-

bon-free" membranes . (A) Fusion to a PE membrane . A membrane
(E. coli PE) clamped at V = 20 mV separates symmetric solutions
(100 mM KCL, 10 mM MES, 2 MM M9CI Z , 0 .1 mM EDTA, pH 6 .0) .
When egg PC :egg PE (1 :1) vesicles (-10"/ml) are added to the cis
compartment, no fusion occurs within the next 2 min . After urea
(450 mM) is added to the cis side, fusion begins . 4 min later, CaCI Z
(15 mM) is added to the cis side, and the rate of fusion increases .
The rate slows steadily over the next 10 min . (B) Fusion to a PC
membrane . A diphytanoyl PC membrane clamped at V = 20 mV
separates symmetric solutions (100 mM KCI, 10 mM MES, 2mM
M9CI Z , 0 .1 mM EDTA, pH 6 .0) . Vesicles (egg PC :egg PE,1 :1 [same
as in A] ; -10"/ml) are added to the cis side . There is no fusion
during the next 2 min . When urea is added to the cis side to a FIGURE 10 Fusion of PC vesicles to a decane-containing PC mem-
concentration of 450 mM, there is still no fusion . When, in addition, brane . A diphytanoyl PC membrane clamped at V = 20 mV sepa-
CaCI Z (15 mM) is added to the cis side, a very low rate of fusion rates symmetric solutions (100 mM KCI, 10 mM MES, 1 mM EDTA,
results . Thus, although PC :PE vesicles readily fuse with a "hydro- pH 6 .0) . Egg PC vesicles (---10"/ml) are added to the cis side with
carbon-free" PE membrane, they fuse poorly with a "hydrocarbon- continuous stirring . There is no fusion during the subsequent 4
free" PC membrane . minutes . Urea (200 mM) is then added to the cis side, and fusion
begins immediately . The rate of fusion is fairly steady for the next
membranes. Although the propensity of lipids to transfer
(4 :1) 4 min . The addition of CaCI Z (10 MM) to the cis side appears to
from one membrane to another (3, 8, 16) complicates the slow the rate of fusion . Thus, although uncharged vesicles fuse
interpretation of those experiments with charged lipids in poorly to a hydrocarbon-free PC membrane (Fig . 9B), they fuse
only one membrane, our results do suggest that the fusion readily to a decane-containing PC membrane (this figure) .

COHEN ET AL . UnilamellarVesicles and Bilayer Membranes 106 1

as for a PEPS [4:11 membrane. PC in the planar membrane 4. Cohen, F. S., J . Zimmerberg, and A. Finkelstein. 1980. Fusion of phospholipid vesicles
with planar phospholipid bilayermembranes. II. Incorporation of a vesicularmembrane
appears to inhibit fusion ifjust an osmotic gradient is present, marker into the planar membrane. J Gen. Physiol. 75:251-270 .
but has little effect on fusion after Ca" is added.) 5. Colombini, M. 1979 . A candidate forthe permeability pathway of theouter mitochon-
drial membrane . Nature (Lond). 279:643-645.
Distinctions between PE and PC are obscured in decane- 6. Kagawa, Y., and E. Racket. 1971 . Partial resolution of the enzymes catalyzing oxidative
containing membranes. Here, presumably because ofan effect phosphorylation. J. Biol. Chem. 246:5477-5487.
7. Kasahara, M., and P. C. Hinkle. 1977. Reconstitution and purification of D-glucose
of decane, an osmotic gradient alone produces high rates of transporter from human erythrocytes. J. Biol. Chem. 252:7384-7390 .
fusion between PC vesicles and PC planar membranes (Fig. 8. Martin, F. J., and R. C. MacDonald. 1976 . Phospholipid exchange between bilayer
membrane vesicles. Biochemistry. 15 :321-327.
10) . This rate of fusion is not increased by the addition of 9. Miller, C., and E. Racket. 1976. Fusion of phospholipid vesicles reconstituted with
cytochrome c oxidase and mitochondrial hydrophobic protein. J Membr. Biol. 26 :319-
Ca". (In the experiment shown in Fig. 10, the fusion rate 333.
actually appears to be slowed by Ca" .) For the decane- 10. Montal, M. 1974 . Formation of bimolecular membranes from lipid monolayers. Meth-
ods Enzymol. 32:545-554.
containing membranes it is thus unambiguous that negatively 11 . Mueller, P., D. O. Rudin, H. T. Tien, and W. C. Wescott. 1963 . Methods for the
charged lipids are not required for fusion ; only an osmotic formation of single bimolecular lipid membranes in aqueous solution. Journal of
Physical Chemistry. 67 :534-535 .
gradient is required. Similarly, whereas fusion rates are higher 12 . Nikaido, H. 1979 . Permeability of the outer membrane of bacteria . Angew. Chem . Int.
in negatively charged PE-containing "hydrocarbon-free" Ed. Eng. 18 :337-350 .
13 . Racket, E. 1979 . Reconstitution of membrane processes. Methods Enzymol. 55 :699-
membranes than in negatively charged PC-containing "hydro- 711.
14. Reyes, J., and R. Latorre. 1979 . Effect of theanesthetics benzyl alcohol and chloroform
carbon-free" membranes, no systematic differential effects on bilayers made from monolayers. Biophys. J. 28:259-280.
exist between negatively charged PE-containing and PC-con- 15 . Rosenbusch, J. P. 1974 . Characterization ofthemajor envelope protein from Escherichia
coli. Regular arrangement on the peptidoglycan and unusual dodecyl sulfate binding. J.
taining decane-containing membranes. Biol. Chem. 249:8019-8029 .
16 . Sandra, A., and R. E. Pagano . 1979 . Liposome-cel l interactions. Studiesof lipid transfer
using isotopically asymmetric vesicles. J Biol. Chem. 254:2244-2249 .
This work was supported by National Institutes of Health grants GM 17 . Schein, S. J., M. Colombini, and A. Finkelstein . 1976. Reconstitution in planar lipid
29210-06, 5T32GM7288, and GM 27367-03. bilayers of a voltage-dependent anion-selective channel obtained from Paramecium
mitochondria . J Membr. Biol. 30:99-120.
18 . Schindler, H., and J. P. Rosenbusch . 1978. Matrix protein from Escherichia coli outer
Receivedfor publication 15 June 1983, and in revisedform 28 October membranes forms voltage-controlled channels in lipid bilayers . Proc. Nail. Acad. Sci.
1983. USA. 75:3751-3755 .
19 . Schindler, H., and J. P. Rosenbusch . 1981 . Matrixprotein in planar membranes : clusters
of channels in a native environment and their functional reassembly . Proc. Nail Acad
Sci. USA. 78 :2302-2306.
REFERENCES 20 . Simon, S. A., L. J. Lis, R. C. MacDonald, and J. W. Kaufman. 1977 . The noneffect of
a large linear hydrocarbon, squalene, on the phosphatidylcholine packing structure.
1. Akabas, M. H. 1983 . Fusion of unilamellar vesicles to planar lipid bilayers: a model for Biophys. J 19 :83-90 .
exocytosis. PhD Thesis. Albert Einstein College of Medicine, Yeshiva University, New 21 . White, S. H. 1978 . Formation of "solvent-free" black lipid bilayer membranes from
York. glyceryl monooleate dispersed in squalene. Biophys. J 23 :337-347.
2. Akabas, M. H., F. S. Cohen, and A. Finkelstein . 1984 . Separation of the osmotically 22 . Zimmerberg, J., F. S. Cohen, and A. Finkelstein . 1980 . Fusion of phospholipid vesicles
driven fusion event from vesicle-planar membrane attachment in a model system for with planar phospholipid bilayer membranes. I. Discharge of vesicular contents across
exocytosis. J. Cell Biol. 98:1063-1071 . the planar membrane . J. Gen. Physiol. 75 :241-250.
3. Cohen, J. A., and M. M. Moronne. 1976 . Interaction of charged lipid vesicles with 23 . Zimmerberg, J., F. S. Cohen, and A. Finkelstein . 1980 . Micromolar Ca=* stimulates
planar bilayer lipid membranes : detection by antibiotic membrane probes. J. Supramol. fusion of lipid vesicles with planar bilayers containing a calcium-binding protein. Science
Struct . 5:409-416 . (Wash. DC. 210:906-908.

1062 THE JOURNAL OF CELL BIOLOGY " VOLUME 98, 1984