HAL Id: jpa-00212466
https://hal.archives-ouvertes.fr/jpa-00212466
Submitted on 1 Jan 1990
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Critical mixing in monomolecular films :
pressure-composition phase diagram of a
two-dimensional binary mixture
C.L. Hirshfeld, M. Seul
To cite this version:
Critical
mixing
in
monomolecular films :
pressure-composition
phase diagram
of
atwo-dimensional
binary
mixture
C. L. Hirshfeld(1)
and M. Seul(2)
(1)
WilliamsCollege,
Williamstown, MA, 01267, U.S.A.(2)
AT&T Bell Laboratories,Murray
Hill, NJ 07974, U.S.A.(Reçu le 17 novembre 1989, révisé le 7
février
1990,accepté
le 2avril)
Résumé. - Nous
avons mesuré les isothermes
pression-surface
de monocouches mixtescomposées
dedimyristoyl
lécithine et de cholestérol à 23,5°C,
et nous les avonsanalysées
afin d’établir lediagramme
dephase pression-composition
de cesmélanges
bidimensionnels. Ces mesures sont confirmées par l’observation directe de laséparation
dephase.
Nous identifions un intervalle de non-miscibilité des états fluidesqui
se termine par unpoint critique,
accessible à latempérature
ambiante. Nous proposons que, dans lesphases
mixtes coexistantes, la lécithine setrouve dans des états distincts.
Abstract. - The
pressure-composition phase diagram
of mixedmonolayers
ofdimyristoyl
phosphatidylcholine (DMPC)
and cholesterol at a temperature of 23.5 °C is derivedby
numericalanalysis
of pressure-area isotherms and corroboratedby
direct fluorescencemicroscopic
observations. We
identify
a fluid-fluidmiscibility
gap, terminatedby
an upper criticalpoint
which is accessible near room temperature. We propose that thecoexisting
mixedphases
of cholesteroland DMPC contain the
phospholipid
in two distinct states.Classification
Physics
Abstracts68.10 - 64.70M - 64.75 - 87.15
,
1. Introduction.
The
discovery
of domain formationduring phase
coexistence in monomolecular films ofamphiphiles
confined to an air-water interface hasprovided
amajor
new stimulus to theinvestigation
of thesesystems.
A richvariety
of domainshapes
has been documented[1, 2].
Recently proposed
phenomenological
theories invoke apicture
ofcompeting
attractive vander Waals and
long-ranged, repulsive dipolar
or Coulomb interactions to account for the appearance of domains of finite size[3, 4].
At the level of the available mean field treatments,amphiphilic monolayers
are viewed to beequivalent
to uniaxialferromagnets [5, 6],
ferromagnetic
surfacelayers [7]
and thin ferrofluidic films[8].
Specifically,
the mean fieldphase diagram
contains a coexistenceregion
characterizedby
intralayer periodic
modulationsof the relevant order
parameter
[3].
Particularly pertinent
inascertaining
the rangeof validity
of these theoretical considerations is thestudy
of monomolecular films in thevicinity
of a criticalpoint.
The existence of criticalpoints
in severalsingle
component
monolayer
films at the air-water interface has beendemonstrated in film balance studies
[9,
10]
andcomputer
simulations[11].
In1538
component
monolayers
lateraldensity
andtemperature
must be varied to reach the criticalpoint.
As thequantitative analysis
ofconfigurations
anddynamics
of domain walls[12],
recordedby
optical
videomicroscopy, requires
extendedperiods
of observation of individual domains in films at the air-waterinterface,
it isimperative
to suppressmonolayer
flow. We have found this to be a task morereadily
accomplished
whentemperature
is eliminated as anexperimental
variable,
because even small thermalgradients generally
lead to flow. Theseconsiderations
suggest
the introduction ofcomposition
as a new variable and thus thestudy
ofmulticomponent
surfactantsystems.
The isotherms of monomolecular films
containing
two constituents have beeninvestigated
by
several authors. Mixed films of cholesterol andphospholipids
orfatty
acids have been ofparticular
interest,
due to the constituents’biological
andphysiological significance [13-18].
However,
it was theapplication
of fluorescencemicroscopy
whichrecently
enabled Subramaniam and McConnell[19]
todirectly
observephase separation phenomena
in mixed filmscontaining
thephospholipid
dimyristoylphosphatidylcholine (DMPC). They interpreted
theirintriguing
observations toimply
criticalmixing,
and the existence of an upper consolutepoint,
accessible at or near roomtemperature.
Two
important experimental
considerations render this mixed filmsystem
particularly
suitable for detailed
experiments.
Attemperatures
exceeding
thepostulated
criticaltempera-ture of the
liquid-condensed (LC)-liquid-expanded
(LE ;
fornomenclature,
see e.g.[20])
coexistence
of DMPC,
approximately
20 ° C[10],
DMPCsimply
behaves as a two-dimensionalfluid. This
greatly simplifies
the behavior of the cholesterol-DMPC mixture. Inaddition,
theexpected
absence of internal conformational order in DMPC facilitates dissolution of the fluorescentphosphatidylcholine analog
which is added as animpurity
to the hostphase
andvénérâtes
the contrastenabling
fluorescencemicroscopic
observation ofphase separation.
Inview of the
desirability
ofexperiments
in thevicinity
of a criticalpoint,
thissystem
thus offersitself as a candidate for closer
inspection.
We have
recently
applied techniques
ofdigital image analysis
to establish a detailedcharacterization of the domain
shapes
formed in this mixedmonolayer
system
by evaluating
powerspectra
of domain wallconfigurations.
Thisanalysis permits
the identification of several distinctregimes
of domainshape stability
as theputative
upper consolutepoint
isapproached.
As discussed elsewhere[12],
the appearance of different stableground
states of domain wallconfigurations
is ingeneral
accordance withstability analyses
based on the modelof
competing
interactions,
referred to at the outset.In the
present
article,
we derive thepressure-composition phase diagram
for thisbinary
mixed film at a
temperature
of 23.5° C,
based on theanalysis
ofthermodynamic
measure-ments, i.e. pressure-area isotherms and the bulk modulus curves derived from themby
numerical differentiation. We correlate the
thermodynamic
measurements with directoptical
observation ofphase separation, corroborating
the determination of the location of atwo-phase
coexistenceregion.
A rather detailedpicture
of theglobal phase
behavior may be deduced. Weidentify
an upper criticalpoint
in this mixturewhich,
at 23.5°C,
is located near amole fraction of cholesterol of 0.27 and a surface pressure of 11.5
dyne/cm.
Thecoexisting
phases
emerge at very low surface pressure(ir 5
0.5dynes/cm)
withrespective compositions
of
approximately
10 mole% and 55 mole% cholesterol.We propose a
simple
rationale for thequalitative
features of observedphase
behaviorby
noting
that,
to agood approximation,
the isobaric increase of cholesterol mole fraction in mixed films isequivalent
to an isothermalcompression
of thelipid
constituent,
DMPC.Specifically,
this leads us tosuggest
that,
in contrast tosimple
fluid-fluidimmiscibility,
themiscibility
gap in thephase diagram
studied here involves a second orderparameter,
relatedthe
distinctly
differentcompressibilities
of the two mixtures and relatedcharacteristics,
weargue that these
configurations
areanalogous
to thoseadopted, respectively,
in the LE and LCphases
ofsingle
component
monolayers
which have been shown to differprimarily
withrespect
to thedegree
of intramolecular chain order[21,
22].
In what
follows,
we firstdescribe,
in section2,
ourexperimental
methods and thedigital
filtersemployed
tocompute
the bulk modulus of the mixedmonolayer
films. Section 3 contains the resultsleading
to thephase-diagram
which we discuss andinterpret
in section 4.We summarize our conclusions in section 5.
2. Materials and
expérimental
methods.2.1 MATERIALS. - For the measurement of 7T-A isotherms it was crucial that the surfactants
and substrate be as pure as
possible ; impurities produced
substantial distortions in theisotherms. In
particular,
itproved
difficult to obtain cholesterol free of its oxidationproduct(s)
[23],
and difficult to prevent it fromoxidizing
at the air-water interface[18].
Cholesterol was
purchased
from bothSigma (Sigma
Grade 99+%,
as well as cell-culture-testedgrade
99+% ;
Sigma
Chem.Co.,
St.Louis,
MO)
and Serva(analytical grade,
99.9 %pure ;
Serva,
Westbury, NY),
the latteryielding
best results.Recrystallization
from ethanol removed a substantial fraction of oxidizedmaterial,
asjudged
from a characteristic inflectionin the isotherm
(compare
e.g.Fig. 2).
However,
even carefulrecrystallization
of cholesterol in an argonatmosphere
did notimprove
thepurity
of anewly opened supply.
DMPC(> 99
%pure)
was obtained from Avanti PolarLipids (Birmingham, AL)
and used without furtherpurification,
since its isotherms indicated sufficientpurity.
The fluorescentprobe
C6-NBD-PC(1-Palmitoyl-2-[6-[7-nitro-2-1,3-benzoxadiazol-4-yl)amino]caproyl]PC),
aphosphatidyl-choline
(PC) analog
with the NBDfluorophore
attached to onealiphatic
chain,
was alsoobtained from
Avanti,
and used without furtherpurification.
Stock solutions of DMPC and of cholesterol were made up in
spectroscopic grade
chloroform,
atapproximately
10 mM(8 mg/ml),
and diluted toapproximately
1 mM with chloroform to makespreading
solutions. 9: 1hexane/ethanol
was also tried as aspreading
solvent,
but seemed toaggravate
cholesterol oxidation(perhaps
due to water absorbedby
the ethanol or due to the process ofchanging solvents).
Mixtures of DMPC and cholesterol weremade
by
diluting appropriately
mixed amounts of the two stocks. A new cholesterol stock hadto be made up from a fresh bottle of cholesterol every two or three
days
because ofoxidation,
as
judged
by
the isotherms of cholesterol or cholesterol-rich mixtures.Those mixtures which were studied
by epifluorescence microscopy
wereprepared
with2 mole% C6-NBD-PC. This
dye
was alsokept
in chloroform stock solution. The inclusion ofthe
dye
caused a small(
5%)
shift of the isotherm of the 30 mole% cholesterolmixture,
leaving
its overallshape
unaltered.2.2 PRESSURE-AREA ISOTHERMS. - Pressure-area isotherms
were recorded under
computer
control on a
Langmuir-Blodgett trough
system
(KSV
2200LB ;
KSVChemicals, Helsinki,
Finland).
The entiretrough, kept
in a laminar flowhood,
wasplaced
in aplexiglass
enclosureto reduce
monolayer
flow. The area of the teflontrough
was varied betweenapproximately
15 x 45CM2
and 15 x 15CM2
by adjusting
theposition
of astepping
motor drivenbarrier,
machined from white delrin. To reduce thermal
drift,
thesubphase
temperature
was heldconstant at 23.5 ° C
by circulating
waterthrough glass tubing submerged
in thesubphase.
The surfacetemperature
was read with a thermistor(Yellow Springs
InstrumentCo.,
YellowSprings, OH),
while the surface pressure measurementemployed
aWilhelmy plate
ofroughened
platinum, supplied by
KSV.Isotherms,
usually containing
200-300points,
were1540
Fig.
1. - Schematicrepresentation
ofexperimental
set-up. Not shown is adipping
well, attached to the KSVLangmuir Blodgett trough, required
in thedeposition
ofmonolayer
films onto substrates.At the
beginning
of thisproject,
thetrough
wasthoroughly
cleanedby allowing purified
water to stand in it
overnight,
thenby rinsing
it with water andorganic
solvents,
andfinally by
running
many isotherms andrepeating
thecleaning operation,
until a standard isotherm ofthe
phospholipid dipalmitoyl phosphatidylcholine (DPPC)
at - 22 ° C was obtained[10].
Theglass
tube which carriedtemperature-controlled
waterthrough
the substrate was cleaned in a mixture of sulfuric acid andhydrogen peroxide.
The bottom of themoving
barrier wasgiven
a new surface finish to remove anyirregularities.
Thereafter,
theonly
cleaning
necessary for theapparatus
consisted ofrepeated sweeping
of the substrate surface after each isotherm. The entire volume ofsubphase (of approximately 1.51)
wasexchanged weekly ;
at that time thetrough
was cleaned with ethanol.The aqueous
subphase
wascomposed
of water of 18 Mfl cmresistivity,
obtained from awater
purification
unitcontaining ion-excange,
carbon and« Organex »
filters(Ultra
pureCartridge
Kit ;
Millipore
Co., Bedford,
MA).
Thesubphase pH
drifted to a value ofapproximately
5.5 within 2 hrs. Measurements were also made on asubphase containing
100 mM 1-ascorbic acid
(Fisher
Scientific,
reagent
grade),
added to minimize cholesterol oxidation at the air-water interface[23].
Here,
thetypical subphase pH
was found to be 3.5. The addition of ascorbic acid to thesubphase helped
to stabilizecholesterol,
especially
inmonolayers
ofXchol >-
0.5,
but had otherwise no effect on the isotherms.After a
monolayer
wasspread
at the air-waterinterface,
thespreading
solvent was allowedcompressed
at a rate ofapproximately
2Â2/Molecule.min.
Compressing
moreslowly
produced
nosignificant
differences in theisotherm ;
compressing
much faster introduced substantial noise into the measurement of surface pressure(see
Sect.2.4)
and may haveproduced
kinetic effects on the isotherms as well.2.3 EPIFLUORESCENCE MICROSCOPY. - A small
epifluorescence
microscope,
adapted
froma beam
splitter
tubeassembly (Rolyn Optics,
Covina,
CA)
in a waypreviously
described[24],
wasemployed
to monitormonolayer
films whilerecording
their 1T-A isotherms on the KSVtrough.
For these measurements, 2 mole% of a fluorescentlipid (see
Sect.2.1 )
were added tothe
monolayer.
The 457.9 nm line of a 5 W argon laser(Innova
90-5 ;
Coherent,
PaloAlto,
CA)
was used for illumination. Fluorescenceimages
were collectedthrough
a 25X(NA 0.45)
objective, passed through
a 495 nm cut-off filter and recordedby
a CCD camera(CCD72,
Dage-MTI,
Michigan City, IN).
The camera head’s smallweight
and size made itpossible
toattach it
directly
to themicroscope
barrel. A video processor with a gray scale stretch circuit(Dage-MTI)
was instrumental inobtaining images
ofacceptable signal-to-noise
ratio,
given
the low
input light
levels otherwise found to be insufficient togenerate
usableoutput.
Thephotographs
infigures
4 and 5 were taken on a smallertrough
with a SIT camera, as describedelsewhere
[12],
requiring
the addition ofonly
1 mole% offluorophore.
2.4 ISOTHERM ANALYSIS. - Pressure-area isothermswere
analyzed by application
of a sevenpoint least-squares quadratic
orquartic smoothing
filter(chapter
3.3 in[25]),
followedby
a sevenpoint
Lanczosdifferentiating
filter(chapter
6.4 in[25])
tocompute
the two-dimensional bulkmodulus,
i.e. the inverse of the isothermalcompressibility :
where à and TT denote mean molecular area and surface pressure,
respectively. Building
vibrations,
whichcoupled
into the surface pressurereading
via surface excitations of thesubphase represented
theprimary
source of noise in therecordings.
While of little concern inthe isotherms
themselves,
these vibrations necessitated the introduction of asmoothing
filterto minimize noise contamination of the
computed
1 / K
vs. à curves.Digital filtering
wasimplemented numerically by executing
theequivalent
convolutionoperation
as a matrixmultiplication (see
e.g.chapter
3 in[26])
on a 32 bitfloating point
array processor(DT7020,
DataTranslation,
Marlborough, MA) residing
in apersonal
computer.
3. Results.
A
typical
set of pressure-areaisotherms,
compiled
infigure
2,
reveals the absence of anyobvious characteristic features
(«
breaks»)
marking phase
transformations in many othermonolayer
systems
[9,
10, 14,
20].
It is this observation which motivated a more carefulanalysis
based on theinspection
of the isothermalcompressibility ( K ),
or itsinverse,
thetwo-dimensional bulk modulus.
Figure
3 contains arepresentative sample
of the results. Several features are nowreadily
identified. The first is a discontinuous increase in1 / K
indicating
adrop
of thecompressibility
to a finite value. Thecorresponding
mean molecular area whichwe refer to as
in.t
in what follows(see particularly Fig.
7below)
marks the termination ofliquid-vapor
coexistence[13].
Thesignificant
decrease ofa-onset
withincreasing
cholesterol1542
Fig.
2. - Set of pressure-area isotherms of mixedmonolayers
ofdimyristoylphosphatidylcholine
(DMPC)
and cholesterol, recorded at 23.5 ° C on an aqueoussubphase,
in some casescontaining
100 mM 1-ascorbic acid. The mole fractions are as indicated. The features visible near 20dynes/cm
on theXh.1
= 0.5 andXchol
= 0.6 isothermssignal
the presence of oxidized cholesterol.The second characteristic feature in the
plots of log
( 1 / K )
vs. li is a distinctbreak,
apparent
fromfigures
3b,
3c and3d,
corresponding
to cholesterol mole fractionsof 0.15,
0.30 and0.45,
respectively. Consistently,
this break occurs at thejunction
of twoapproximately
linearsegments
in thesemi-logarithmic representation
of the1 / K
profiles (see Fig. 3)
and thussignals
a discontinuous increase in theirslope.
At thecorresponding point
on theisotherm,
the mixed
monolayer
becomes lesscompressible. Epifluorescence microscopy
reveals that this «stiffening »
of the surfacelayer
coincides with the transition from aphase-separated
regime,
characterizedby
aheterogeneous
distribution of fluorescent label in themonolayers,
to a
homogeneously
fluorescent state which weidentify
below with ahomogeneous
mixture(see Fig. 8). Figure
4 illustrates this conclusion with a series ofepifluorescence micrographs,
taken
during compression
of a mixedmonolayer
ofXh.1
= 0.3 atincreasing
pressures.Figure
5presents
micrographs
ofmonolayers
in thetwo-phase region
withcompositions
fixedbelow
(Xchol
=0.1)
and above(Xchol
=0.45)
the critical value(see
below).
Theseimages
provide
direct evidence for the coexistence of twophases,
thebright
areascorresponding
tothe
phase predominantly containing
DMPC. We will see below that theseoptical
observations areentirely
consistent with theanalysis
of thethermodynamic
data.A third feature
characterizing
the mixedmonolayer
may be extracted from thelog
( 1 / K )
vs. i curves(of Fig. 3) by plotting
the value of1 / K ,
assumed at onset and at several fixed values of the surface pressure, as a function ofincreasing
cholesterol mole fraction. A selection of suchplots
is shown infigure
6.They permit
theimportant
observation that the addition of cholesterol to the mixedmonolayer
does notsubstantially
alter the value of1 / K
from that found for pure DMPC until a threshold value of the cholesterol mole fraction(Xchol)
is attained. This threshold value was estimatedby applying
a linearregression analysis
to the low
X,,h.1 portion
of eachplot,
as indicatedby
the solid lines infigure
6. As manypoints,
in order of
increasing Xchob
aspossible
wereincluded,
until this caused asignificant
deterioration of thegoodness
of fitparameter.
The first value ofXchol
definedby
thisprocedure
not to lie on the initial linearportion
was identified as the threshold value. Thesepoints
mark thephase boundary,
shown infigure
8(«
V»)
whichseparates
region
II from theFig.
3. - Inversecompressibility
1 / K - -
il2013 )
(bulk modulus)
as a function of mean moleculara
area, for
increasing
values ofXchol
= 0.05, 0.15, 0.30, 0.45 and 0.55. Also shown, as solid lines, are thecorresponding
isotherms from which the bulk modulus curves werecomputed by application
of a sevenpoint
Lanczosdifferentiating
filter(see
text, Sect.2.4). Straight
lines serve as aguide
to the eye. The lefthand ordinate shows the surface pressure, the
right
hand ordinate refers to the bulk modulus. The inversion in the1 /K
curves forXchol
= 0.45 and 0.55 centered at J m 47Â2/Mol ,
arises from cholesteroloxidation.
meaningful
test of the functional formobeyed
as1 / K
approaches
thelimiting
value measuredfor pure cholesterol.
To
identify
thepartial
molecular areas ofcoexisting phases
in the surfacemonolayer
wefollow
(in Fig. 7)
the classical scheme ofplotting
the mean molecular area, li, at onset(liquid-vapor
coexistence),
and at several fixed values of the surface pressure, as a function ofcholesterol mole fraction. It is
readily
apparent
that,
in contrast to several other1544
Fig.
4. -Fluorescence
micrographs
ofDMPC/cholesterol
mixedmonolayer
containing
30 mole%cholesterol, and 1 mole% of the fluorescent
lipid analog
C6-NBD-PC(see
Sect.2.1). a)
Fluorescent andprobe
excluding phases
occupyapproximately equal
area fractions. In theregime
of strongsegregation
shown here, domainsadopt
a circularshape.
Thephotograph
is taken at a surface pressure ofapproximately
5dynes/cm.
The bar marks 50 fJ.m.b) Upon
approaching
the upper consolutepoint
the domain wall energy softens andsignificant
domainshape
fluctuations lead to distortedshapes [12].
Thephotograph
is taken at a surface pressure ofapproximately
10dynes/cm.
Fluctuationsdecay
on time scales consistent withrapid,
i.e. fluid-likeintralayer
molecular diffusion in bothphases. c)
Furthercompression
yields
ahomogeneously
fluorescent mixture, shown here at a surface pressure ofapproximately
14dynes/cm.
The appearance of thishomogeneous phase
coincides with the «stiffening
»of the
monolayer
described in connection withfigure
3.constituents does not obtain : this would
imply
â =ADMPC( 7r )
XDMPC
+achol C Tr ) Xchob
andthus a
straight
lineconnecting
ii(Xchol
=0)
=aDMPC and
a(Xchol
=1.0)
=achol.
However,
allplots (certainly
those up to 7r = 15dynes/cm)
are wellapproximated
by
arepresentation
permitting
the identification of three distinctregimes,
each characterizedby
a linear decreaseof â with
increasing
mole fraction of cholesterol. Construction of theintercepts (see
e.g.chapter
7.4 in[27]) yields
thepartial
molecular areas of DMPC and cholesterol. Asimplied by
Fig.
5. - Fluorescencemicrographs
ofDMPC/cholesterol
mixedmonolayers containing
1 mole% of the fluorescentlipid analog
C6-NBD-PC(see Sect. 2.1 )
and 10 mole%(Fig. 5a)
and 45 mole%(Fig. 5b)
cholesterol,
respectively. Photographs
were taken at surface pressures ofapproximately
0.5dyne/cm
and 0.2
dyne/cm, respectively,
at molecular areas close to jonset- In combination withfigure
4a thesephotographs
demonstrate thatfluorophore excluding regions
in themonolayer
occupy anincreasingly
larger
area fraction asXchol
is increased. In section 4 we suggest theseregions
to be associated with aDMPC/cholesterol
mixturecontaining
DMPC in an ordered conformation. Aheterogeneous
distri-bution of labelled domainsgenerally prevails
atlength
scalesexceeding
500 jim. The bar infigure
5a marks 50 fJ-m.three
regimes.
To extract thepartial
molecular areas listed in table 1below,
a linear fit wasapplied
to the middleportion
of eachplot.
Thatis,
beginning
withXehol
= 0.1,
the number ofpoints,
in order ofincreasing Xch.1,
included in the linearregression
was increased until this led to a deterioration of the fit. The solid lines shown in the middleportion
of theplots
infigure
7 were so obtained.Approximate straight
lines,
indicated dashed in thefigure,
werethan drawn to obtain
intercepts
with the ordinatesXehol
= 0 andXchol
= 1.0. Thescarcity
ofpoints
forXchol --
0.6 is theprimary
source of theuncertainty
in the values for thepartial
molecular areas listed in the table below. It is also reflected in the error bars
given
for thepoints
markedby
« à » infigure
8.The existence of three distinct linear
regimes
in thedependence
of â onXchol
indicates theexistence of two
phase
boundaries,
shown infigure
8(«
à»).
The first of theseseparates
region
IV from the rest of thephase diagram.
As indicated(by « à »),
the secondphase
boundary
coincides,
withinexperimental
error, with thephase
boundary
derived on the basisof
figure
6,
whichseparates
region
II from the rest of thephase diagram.
The behavior documented in
figure
7 and in the table isanalogous
to that observed anddiscussed
by
de Bernard in his carefulearly
film balancestudy
of mixedmonolayers
of eggphosphatidylcholine
and cholesterol[13].
It may be understoodsimply by observing
that theaddition of
cholesterol,
itselfquite incompressible
as demonstratedby
theshape
of its isotherm infigure
2,
induces a « condensation » of DMPC whichconsequently
assumes amolecular area close to its
partial
molecular area of 51A2,
while thecorresponding partial
molecular
area of cholesterol coincides with its actual molecular area(=
38A2),
as extractedfrom the
pertinent
isotherm infigure
2. We return to thispoint
in section 4 below. Thefindings
described so far lead to thephase
diagram
ofDMPC/cholesterol
mixedmonolayers
at 23.5°C,
shown infigure
8. Itsglobal
features exhibit a remarkablesimilarity
to1546
Fig. 6. - Examples
ofplots showing
thedependence
of1 / K
on cholesterol mole fraction at fixed ’TTvalues. The threshold values of
Xchol
for a finite increase in1 / K
over its baseline value were extracted from suchplots by
aprocedure
described in the text,yielding
the solidstraight
lines on the basis of linear fits. These threshold values define the appearance of a state of lowcompressibility
of the mixture,separating
theregion
ofhigh X,,h.1 (region
II inFig. 8)
from the rest of thephase diagram
shown infigure
8.Fig.
7. -Plots of the mean molecular area as a function of mole fraction, at fixed values of
Fig.
8. -Pressure-composition phase diagram
of mixedmonolayer
of DMPC and cholesterol at 23.5 ° C. Thephase
boundaries shown here are derived on the basis offigure
3( 0 »), figure
6(« V »),
figure
7(«
à»),
as discussed in section 3. Where indicated,they
were confirmedby
fluorescencemicroscopic
observation ofphase separation («
X »,(this work),
and « + »[19]).
All lines serve asguides
to the eye. Dashed lines were drawnby
hand ; the solid line follows a standard functional form forphase
coexistence as discussed in section 3. The three distinct linesseparating regions
III and II areargued
to coincide withinexperimental
error(Sect. 3) ;
an indication oftypical
errormargins
isgiven by
horizontal delimiters( 1 - > 1 ).
The various lines delineate a fluid-fluidmiscibility
gap(region III)
boundedby
three distinctregions
further described in section 3. Thenumbering
is consistent withfigure 5
in[15].
Ahighly
compressible
vaporphase, existing
at all values ofXchol
fora > aonset and 7T 5 0.5
dyne/cm (see
text andFig. 3),
is not shown.Table I. - Partial Molecular Areas. Listed are
partial
molecular areas(in
Â2 )
derivedfrom
figure
7 as discussed in the text. The notations « low », « middle » and «high » refer
to the 3 linearregimes
of
theplots
infigure
7. Partial molecular areasof
DMPC andcholesterol,
aredenoted
by
âdmpc
andàchob respectively,
and are estimated to be accurate to within:t 3 Á 2.
1548
isotherm
(Fig. 5
in[10]), suggesting
that small amounts of added cholesterol(Xchol «
0.1 )
arandomly
distributedimpurity,
eliminate the orderedphases
of DPPC[29].
On the basis of our
analysis
weidentify
region
III as a coexistenceregime
of two immiscible fluidphases
with an upper consolutepoint
near(Xghol
=0.27,
ir’ =11.5).
The datapoints
( 0 ») delineating
thisregion
were obtained from the breaks in thelog ( 1 / K )
vs.à curves described in connection with
figure
3 and confirmed in several cases(« X ») by
epifluorescence
microscopy.
The datapoints reported
in the fluorescencemicroscopic study
of Subramaniam and McConnell
[19]
are also indicated(«
+»).
The solidline,
little morethan a
guide
to the eye at thisstage,
follows the standard functional formwith the critical values indicated
above ;
/3
was held fixed at a value of1/3, generally
observedfor fluid-fluid coexistence
(see
e.g.[30])
and favoredby
comparison
with coexistence curvescomputed four 6
=1/2,
the mean fieldvalue,
and forf3 = 1/8,
the value indicatedby
apostulated
2dIsing analogy applied
tomonolayer
films[31].
However,
our data set is toosmall to make any
meaningful
detailed test ; the value of1/3
iscertainly
not to be takenliterally.
We estimate the critical values to be accurate to withindXghol
= ± 0.05 andAir’ = ± 1
dyne/cm.
In
region
1 DMPC and cholesterol form amacroscopically homogeneous
mixture whosecholesterol content varies in the range 0.1 _
Xchol S 0.35.
This mixture appears 1homogeneously
fluorescent(see
Fig. 4).
Characteristic « breaks » in theplots
of â as afunction of
Xchol
infigure
7suggest
that to the left of the lineXchol S 0.1,
thatis,
inregion
IV,
pure DMPC coexists with aDMPC/cholesterol
mixture ofXchol
-- 0.1.Monolayers
in thisregion
of thephase diagram
also appearoptically homogeneous.
In
region
II,
pure cholesterol coexists with aDMPC/cholesterol
mixture. Whenentering
this
region,
mixedmonolayers undergo
a transition to a state of lowcompressibility.
Pointsmarking
thephase boundary
were extracted fromplots
of thetype
shown infigure
3(«
D»),
plots
oflog ( 1 / K )
vs.Xchol (Fig.
6 ;
« V »)
and from thoseshowing
thedependence
ofà on
Xchol (Fig.
7 ;
« à»).
These different determinations of the location of thephase
linediffer
by approximately
0.1Xchol.
Given theappreciable uncertainty
in the data in thisregion,
we cannot say whether it is this
uncertainty
which sets the width of the transitionregion,
orwhether the three
procedures
in factcapture
differing signatures
associated with thetransition. In any case, it appears
likely
that the coexistenceboundary
meets thephase
boundary
in atriple point
near(Xchol
=0.35,
’TT = 10dynes/cm).
This is the scenario favoredby
Albrecht et al. in theirstudy
of mixedmonolayers
of cholesterol anddipalmitoyl
phosphatidylcholine (DPPC)
at 24.9 °C[15].
Epifluorescence microscopy
reveals thatlayers
composed
of 40 mole% and 45 mole% cholesterol remaininhomogeneous
whencompressed
up to surface pressures of 30dynes/cm, suggesting
a directpath
betweenregions
III and II atthose values of the cholesterol mole fraction.
4. Discussion.
In what
follows,
wesuggest
a rationale to account for thequalitative
featuresdisplayed
in thephase diagram
and discuss the notion that DMPC in fact assumes distinct states in the twoimmiscible fluid
phases
which coexist within themiscibility
gap(region
III inFig. 8).
To set the
stage,
we note that thesharp
rise in the isotherm of the pure cholesterol(7T,
Xcho0 phase diagram
offigure
8 may in fact be understoodby explicitly making
this identification. Thatis,
one assumes that cholesterol-cholesterol interactions aregoverned
by
ahard-core
repulsive
potential
with a characteristic scale setby
the molecules’ van der Waalsradius,
and oneregards
the actual molecular area, ah.1(as opposed
to itspartial
moleculararea,
àchol),
as constant.An immediate consequence of this
assumption
becomesapparent
whenconsidering
the behavior of thebinary
mixedmonolayer
along
an isobarictrajectory through
the(w,
Xchol) diagram.
As the set of isotherms infigure
2demonstrates,
the mean molecular areaà at which a
given
value of ir is attained decreases withincreasing Xchol. Figure
7displays
thedependence
of â onXchol explicitly
for a number of fixed values of 7r. Three linearregimes,
readily
identified in theplots
at 7r x- 15dyne/cm,
indicate that thecorresponding partial
molecular areas
( â )
of both constituents remain constantthroughout
eachregime,
butundergo
anabrupt change
at the transitions between them(see
also Tab.I).
A lineardependence
of â onXch.1, in conjunction
with theassumption
of theincompressibility
ofcholesterol,
directly
indicates an effectivecompression
of DMPC[13].
Consider,
forexample,
the middleportion
of theplot
for ’TT = 0dynes/cm.
Here,
àdmpc
= 90Â2
andàchol
=20
Â2
« achobimplying
that for each molecule of cholesterol(of
actual molecular areaachol = 40
A 2)
added to themixture,
the total molecular area increasesby only
20Â2.
It is the concomitantcompression (or
« condensation»)
of DMPC which balances theequation.
We may thusregard
an increase inXchol
at constant pressure to beeffectively equivalent
to acompression
of(the
remaining)
DMPC.Consequently,
oneexpects
the conformations exhibitedby
DMPCalong
any isobar in the(ir, Xcho0 diagram
to reflect those of theequivalent
pure DMPCmonolayer subjected
to isothermalcompression.
Specifically,
thissuggests
to us that themiscibility
gap,corresponding
toregion
III in thephase diagram
depicted
infigure
8,
involvescoexisting
mixedphases
in which DMPCpreferentially
assumes two distinct states. One of the twomixtures,
containing
predominantly
DMPC,
exhibitsbright
fluorescence(Figs.
4,
5)
and acompressibility
which isessentially
that of a pure DMPCmonolayer (see Fig. 6).
In contrast, the second mixture excludes most of the fluorescentlipid analog, consequently appearing
dark(Figs.
4,
5),
anddevelops
amarkedly
lower
compressibility,
as discussed in connection withfigures
3 and 6.Furthermore,
as noted in section 3 in connection withfigure
7,
whenreaching
thephase boundary
ofregion
II,
thepartial
molecular area of DMPCapproaches
a valueof approximately
51Â2
(see
Tab.I).
All these features arestrongly
reminiscent of the coexistence of theliquid-expanded (LE)
andliquid-condensed (LC) phases featuring prominently
in thephase diagram
of the purephospholipid monolayer (see Figs.
2 and 6 in[10]
and inset toFig. 8).
It is thereforetempting
to
suggest
that DMPC in the twocoexisting
mixtures assumesconfigurations
similar to those itexhibits in the LE and LC
phases, respectively.
A recent
report
ofexperiments probing
molecularconfigurations
inmonolayers
at anair-water interface
[21]
]
states that the transition into the LEphase
is characterizedby
the excitation of disordered molecular chain conformational states.X-ray
measurements have beeninterpreted
to indicate the existence in theLC-phase
offully
extendedaliphatic
chainsadopting
asignificant
tiltangle [22].
Monte Carlo simulations based on this mechanism[11,
31]
]
reproduce
many of theexperimentally
observed features.Accepting
thisscenario,
onewould,
in thespirit
of Doniach’ssimple
two-state modelof chain-melting [31, 32], picture
thepredominantly
DMPCcontaining phase
as a mixture of cholesterol with DMPC in a disordered chainconfiguration,
characterizedby « gauche »
excitations,
while in thecoexisting
second mixture DMPCapproaches
anall-trans,
ordered conformation. A more1550
In support of such a model for
lipid/cholesterol
mixtures we observe that in thephase
diagram
of purephosphatidylcholine
thetypical
values of the molecular area in the LCphase
are indeed close to 50
Â2 [10]. Figure
7 and theaccompanying
table 1 ofpartial
molecularareas reveal that
àdmpc,
thepartial
molecular area ofDMPC,
in thehigh Xchol region
is in factof that
magnitude,
whileJchob
thepartial
molecular area ofcholesterol, approaches
achol, its actual molecular area of
approximately
40Â2.
To the extent thatà ch.1 me
achol, as assumed at the outset of the
discussion,
àdmpc
= ADMPOplacing
DMPCalong
thephase boundary separating region
II from the rest of thediagram,
in the range of densities characteristic of its LCphase.
Single
component
lipid monolayers
also exhibit a characteristicdrop
incompressibility
whenentering
their LCphase [10, 20]. By
analogy,
one would attribute the sudden increase of1 / K,
described in connection with thelog
( 1 / K )
vs.Xchol plots
ôffigure
6,
to the appearanceof the
equivalent
state of DMPC in the mixedmonolayers.
This statementimplies
that thephase
linedelineating
thehigh
cholesterolregion (region
II inFig. 8) signals
the condensation of thelipid
constituent into its all-transconfiguration.
This transitionpersists
to values of the surface pressures farexceeding
the critical pressure fordemixing.
Considerations in accordance with thosepertinent
to purelipid monolayers [11]
]
wouldsuggest
a continuoustransition to a mixture of cholesterol and all-trans DMPC above the upper consolute
point
(Fig. 8).
Within the context of the
proposed
model one would attribute theinhomogeneous
distribution of
fluorescent
label within themiscibility
gap to the fact that thetype
of fluorescentlipid analog employed
here islargely
excluded from thephase containing
DMPC in its chain-ordered state. It is this very feature which makespossible
the fluorescencemicroscopic investigations
of the LE - LCphase
coexistence insingle
component
monolayers
[1,
2, 33,
34].
For the same reason, one wouldexpect
mixedmonolayers
to exhibit aninhomogeneous
fluorescence distribution inregion
II,
as we have observed.As the
pertinent
isotherm infigure
2demonstrates,
theordered,
LC-equivalent
state is inaccessible to pure DMPCmonolayers
at thetemperature
of thepresent
experiments,
namely
23.5 ° C. This is thefrequently
notes «condensing »
effect of cholesterol[13,
20,
35] :
according
to ourphase diagram,
theLE-equivalent
conformation of thelipid
accommodatescholesterol up to
only
a modestcomposition,
in thepresent
caseapproximately
10 mole%.Further admixture of cholesterol stabilizes the conformation associated with the
lipid’s
LCphase.
The newphase
appears in thepresent system
with acomposition
ofapproximately
55 mole% cholesterol at 7T = 0
dyne/cm.
Compression
ofsingle
component
monolayers eventually
induces apositionally
ordered solid[36].
This appearsunlikely
in the presence of excess cholesterol. Onepossible
scenario in thehigh
cholesterolregion might
be the formation of aglassy DMPC/cholesterol
mixture,
coexisting
with pure cholesterol. The characteristic response of such a state to mechanicalperturbations (e.g.
in torsional oscillatormeasurements)
should be distinctive andinteresting
to pursue.
We believe the
preceeding
interpretation
of fluid-fluidimmiscibility
inphospholipid/choles-terol mixed
monolayers
to beplausible.
Itcertainly
lends itself to be testedby
anexperimental technique
which is sensitive to the state ofaliphatic
chainordering [21].
However,
irrespective
ofspecific microscopic
model one may wish to invoke to characterizethe different states of DMPC in the two
coexisting
fluidphases,
acomplete theory
of mixedmonolayers
would in any case have to consider thecoupling
between an internaldegree
offreedom and the
macroscopic
orderparameter,
e.g.Ycy)l - X(2)chol,
in contrast tosimple binary
[30].
Thisgeneric
situation is reminiscent of the nematic to smectic A transition inthermotropic liquid crystals
and theequivalent phenomenon
insuperconductors [37].
5. Conclusions.
We have
investigated
thephase
behavior of a two-dimensionalmixture,
a monomôlecularfilm confined to an air-water interface
containing
DMPC and cholesterol.By
inspection
of thedependencies
of bulk modulus andpartial
molecular areas on cholesterol molefraction,
wehave established a
phase diagram
whoseglobal
features include a fluid-fluidmiscibility
gap,terminated
by
an upper consolutepoint
near(X’ 01
=0.27,
’TTc =11. 5)
at 23. 5 ° C. Thelocation of the
phase boundary
was confirmedby
direct observation ofphase
separation
viaepifluorescence microscopy.
We propose that DMPC assumes states oi different chainordering
in thecoexisting
fluidphases, corresponding
to thosecharacterizing
liquid-condensed andliquid-expanded phase,
in the purelipid monolayer.
Thishypothesis
may bereadily
testedby experiments
sensitive to chain conformational order[21]
]
andperhaps by
those sensitive to chain tilt[22].
Weexpect
that thepossibility
ofcoupling
of the mean-fieldorder
parameter
to a seconddegree
of freedom must be examined to obtain a correctpicture
of the critical
mixing.
The
probed phase diagram
contains three furtherregions.
These areoccupied by : firstly,
coexisting
pure DMPC and ahomogeneous,
fluidDMPC/cholesterol
mixture ofAchoi ===
0.1 ;
secondly,
amixture,
also fluid andhomogeneous (0.1 Xchol S
0.35,
7r 2:Ir 5
butcharac-terized
by
a reducedpartial
molecular area ofDMPC ;
thirdly,
ahighly
incompressible
mixture,
coexisting
with pure cholesterol in which DMPC assumes apartial
molecular areacoinciding
with thatmarking
the appearance of the LCphase
insingle
component
layers.
The latterregion,
with similarproperties
has also been identified in thephase diagram
of DPPC and cholesterol[10].
The
accessibility
of a criticalpoint
in a two-dimensionalmonolayer
film in a convenientrange of
experimental
parameters
hasalready
beenexploited
in thestudy
of a series ofdomain
shape
instabilities in thepresent system
[12].
It offerspromising possibilities
for more detailedexperiments
in the criticalregion.
Acknowledgments.
We would like to thank E. Chin and S.
Stuczynski
for their advice and the use of their facilities in therecrystallization
of cholesterol under argon. CLHacknowledges
support
through
the Summer ResearchProgram
for Women andMinorities,
sponsored by
AT&T Bell Laboratories.References
[1]
MILLER A., KNOLL W. and MÖHWALD H.,Phys.
Rev. Lett. 56(1986)
2633-2636;HECKL W. M. and MÖHWALD H., Ber.
Bunsenges. Phys.
Chem. 90(1986)
1159-1163.[2]
MCCONNELL H. M., TAMM L. and WEIS R. M., Proc. Natl.Acad.
Sci. USA 81(1984)
3249-3253 ;WEIS R. M. and MCCONNELL H. M., Nature 310
(1984)
47-49 ; J.Phys.
Chem. 89(1985)
4453-4459 ;
GAUB H. E., MOY V. T. and MCCONNELL H. M., J.
Phys.
Chem. 90(1986)
1721-1725.[3]
ANDELMAN D., BROCHARD F., DE GENNES P. G. and JOANNY J.-F., C. R. 301(1985)
675-678 ;1552
[4]
KELLER D. J., MCCONNELL H. M. and MOY V. T., J.Phys.
Chem. 90(1986)
2311-2315.[5]
CAPE J. A. and LEHMAN G. W., J.Appl. Phys.
42(1972)
5732-5756.[6]
GAREL T. and DONIACH S.,Phys.
Rev. B 26(1982)
325-329.[7]
YAFET Y. and GYORGY E. M.,Phys.
Rev. B 38(1988)
9145-9151.[8]
ROSENSWEIG R., ZAHN M. and SHUMOVICH R., J.Magn. Magn.
Mat. 39(1983)
127-132.[9]
KIM M. W. and CANNELL D. S.,Phys.
Rev. A 13(1976)
411-416.[10]
ALBRECHT O., GRULER H. and SACKMANN E., J.Phys.
France 39(1978)
301-313.[11]
MOURITSEN O. G., IPSEN J. H. and ZUCKERMANN M. J., J. Coll.Interf.
Sci. 129(1989)
32-40.[12]
SEUL M. and SAMMON M. J.,Phys.
Rev. Lett. 64(1990)
1903-1906.[13]
DE BERNARD L., Bull. Soc. Chim. Biol. 40(1958)
161-170.[14]
MOTOMURA K., TERAZONO T., MATUO H. and MATUURA R., J. Coll.Interf.
Sci. 57(1976)
52-57.[15]
ALBRECHT O., GRULER H. and SACKMANN E., J. Coll.Interf.
Sci. 79(1981)
319-338.[16]
DEMEL R. A., VAN DEENEN L. L. M. and PETHICA B. M.,Biophys.
Biochim. Acta 135(1967)
11-19.[17]
GERSHFELD N. L., Ann. Rev.Phys.
Chem. 27(1976)
349-368.[18]
CADENHEAD D. A., KELLNER B. M. J. and PHILLIPS M. C., J. Coll.Interf.
Sci. 57(1976)
224-227.[19]
SUBRAMANIAM S. and MCCONNELL H. M., J.Phys.
Chem. 91(1987)
1715-1718.[20]
CADENHEAD D. A., MÜLLER-LANDAU F. and KELLNER B. M. J.,Ordering
in Two Dimensions,Ed. S. K. Sinha
(Elsevier
N.Holland)
1980, pp. 73-81.[21]
GUYOT-SIONNEST P., HUNT J. H. and SHEN Y. R.,Phys.
Rev. Lett. 59(1987)
1597-1600.[22]
KJAER K., ALS-NIELSEN J., HELM C. A., TIPPMAN-KRAGER P. and MÖHWALD H., J.Phys.
Chem.93