HAL Id: jpa-00210018
https://hal.archives-ouvertes.fr/jpa-00210018
Submitted on 1 Jan 1985
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.
Lyotropic effects of alkanes and headgroup composition on the lα -Hii lipid liquid crystal phase transition :
hydrocarbon packing versus intrinsic curvature
G.L. Kirk, S.M. Gruner
To cite this version:
G.L. Kirk, S.M. Gruner. Lyotropic effects of alkanes and headgroup composition on the lα -Hii lipid liquid crystal phase transition : hydrocarbon packing versus intrinsic curvature. Journal de Physique, 1985, 46 (5), pp.761-769. �10.1051/jphys:01985004605076100�. �jpa-00210018�
761
Lyotropic effects of alkanes and headgroup composition on the L03B1 -HII lipid liquid crystal phase transition : hydrocarbon packing versus intrinsic curvature
G. L. Kirk (*) and S. M. Gruner (**)
Dept. of Physics, Princeton University, Princeton, N.J. 08544, U.S.A.
(Reçu le 29 octobre 1984, accepté le 21 janvier 1985)
Résumé. 2014 Les effets liés à l’addition d’une petite quantité d’hydrocarbure et à la composition de groupements polaires de mélanges de phospholipides ont été examinés relativement à la transition de phase lamellaire (L03B1) hexa- gonale inverse (HII) dans des cristaux liquides eau-phospholipide. L’étude du réseau cristallin et la détermination des
phases ont été faites aux rayons X par diffraction. Kirk, Gruner, Stein (Biochem. 23 (1984) 1093) ont postulé que la
compétition entre la répulsion d’hydratation, l’élasticité due à la courbure d’une monocouche de lipide et l’empile-
ment d’hydrocarbones détermine la transition L03B1-HII. Dans cet article, nous décrivons les expériences qui per- mettent d’explorer les effets de courbure et d’empilement Nous montrons que le rayon de courbure des tubes de
lipides de la phase HII peut être ajusté en changeant les proportions de lipides de haute et basse courbure. Cepen- dant, pour de grandes courbures, les contraintes d’empilement lipide-hydrocarbone doivent être diminuées par
l’adjonction d’hydrocarbure. Sans hydrocarbure, la phase L03B1 persiste à haute température; la simple adjonction
de 5 % d’hydrocarbure réduit de façon spectaculaire l’intervalle de température à l’intérieur duquel la phase L03B1 existe; on observe une réduction de la température de transition L03B1 HII et une expansion spectaculaire du réseau HII. La mesure des dimensions internes du réseau HII montre que l’épaisseur de la couche du lipide est à peu près
constante à une température donnée et que la presque totalité de l’expansion du réseau est due à une augmenta- tion du rayon des groupes formés de molécules d’eau. Les implications biologiques sont discutées.
Abstract. 2014 The effects of mixed phospholipid headgroup composition and the addition of small amounts of alkane were examined with respect to the lamellar (L03B1) to inverse hexagonal (HII) phase transition in phospholipid-
water liquid crystals. X-ray diffraction was used to probe the lattices and determine the phases. It has been postulat-
ed that competition between hydration repulsion, lipid monolayer curvature elasticity, and hydrocarbon packing
determine the L03B1-HIItransition, [Kirk, Gruner and Stein, Biochem. 23(1984) 1093]. Here, experiments which explor-
ed the effects of curvature and packing are described. It is shown that the radius of curvature of the lipid tubes
of the HII phase could be adjusted by mixing high and low curvature lipids. However, large curvatures could not be expressed unless lipid hydrocarbon packing constraints were relieved, for instance, by the addition of alkane.
Without alkane, the L03B1 phase extended to high temperatures; adding just 5 % alkane significantly reduced the temperature span of the L03B1 phase, by lowering the L03B1 to HII transition temperature, and dramatically expanded
the HII lattice. Measurements of the internal dimensions of the HII lattice showed the lipid layer thickness to be nearly constant at a given temperature and that almost all of the lattice expansion was due to an increase in the radius of the water cores. Biological implications are discussed.
J. Physique 46 (1985) 761-769 MAI 1985,
Classification
Physics Abstracts
64.70M
1. IntroductioiL
A rigorous physical description of inverse phase for-
mation has yet to be developed for a wide class of
lyotropic liquid crystals [1]. For example, there is only a rudimentary understanding of the molecular forces which drive the L.-HI, phase transition [2]
(*) Current address : RESONEX, Inc., 610 Palomer
Ave., Sunnyvale, CA 94086, U.S.A.
(**) To whom reprint requests should be addressed.
that is observed with neutral (zwitterionic) phospho- lipid-water liquid crystals. In the La phase the amphi- philic lipid molecules form parallel bilayer sheets separated by layers of water (Fig. la). For certain
phospholipids, such as phosphatidylethanolamines,
either raising the system temperature or lowering
the water content initiates a first order transition to the HI, phase [3]. In the H,I phase, the monolayers
of lipid curl into extended cylinders with water cores
and with the hydrophobic lipid tails projecting out-
wards. These tubes stack on a 2-dimensional hexago-
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01985004605076100
762
Fig. 1. - Schematic cross-sections through L.(a) and H,ib) phospholipid phases. Lipid headgroups are represented by black circles attached to two melted hydrocarbon chains.
Water is shown as cross-hatched areas. The distance, d, is the ordinate for figures 3-10. In the Hn phase, R. is the radius of the water cylinder and d,. is the minimally extend- ed mean length of a lipid molecule. The lipid tails are at their shortest extent at point C and at their longest at point D in the lattice. Thus, lipids must stretch to fill point D.
nal lattice (Fig. 1 b). Lyotropic liquid crystal transi- tions, such as the Li-H,,, are very complicated to deal with theoretically because of the number of interac- tions involved and because so many system para- meters change abruptly at the transition. The amphi- philic nature of the lipids and the presence of water introduces many poorly understood polar and sur- face interactions. The lipid surfaces undergo a topologi- cally discontinuous change at the transition, i.e., there is no way to continuously deform the lipid monolayers of the bilayer phase into the tubes of the His phase without tearing the lipid-water interface.
Both the curvature of the lipid-water interface and the shape of the mean volume per lipid molecule change dramatically at the transition.
The first task confronting a quantitative theory of the transition is to decide which contributions to the system Hamiltonian are important. Recently, Kirk
et al. [4] have presented a theory of the LIX-HII transi- tion in which three contributions were postulated
as driving the transition : elastic bending of the lipid monolayers, hydration repulsion of the headgroup surfaces, and hydrocarbon packing energies.
Elastic contributions stem from the fact that the lipid monolayers (which form bilayers when placed
tails-to-tails) are practically flat in the La phase but rolled into cylinders of - 30 A radius in the Hn phase. Presumably there exists a minimum free energy of curvature associated with an equilibrium or intrinsic curvature of the lipid monolayers [4] at a given temperature. Enforced departures from the minimum energy curvature stores energy elastically.
Similar ideas have been applied to lipid bilayers (e.g.,
see [5] and the discussion in [4]).
Lipid headgroup also repel one another very
strongly (up to 109 dyne/cm2 ; see Parsegian et al. [6]) when opposed across thin water layers [7]. The repul- sive force is observed to fall off exponentially with bilayer separation with a universal decay length
- 2-3 A [7]. The repulsive force cannot be accounted for by standard DLVO theory [8, 9]. It has been suggested as arising from a surface-induced polariza- tion density in the water [10, 11]. ].
Hydrocarbon packing energies stem from the rela- tive inaccessibility of the lipid tails to certain volumes in the hydrocarbon zone of the H" phase. Consider points « C » and « D » of figure lb : in so far as the lipid headgroup surface is cylindrical, point « D » is further from the headgroups than point « C ».
This suggests that the hydrocarbon tails, which contain a fixed number (- 16) of carbon atoms, must stretch further to occupy the volume at « D ». Because the tails are melted, they contain many gauche rotamers (« kinks ») at the carbon-carbon bonds. This flexibi- lity endows the tails with polymer-like properties.
The many kinks yield an average tail length consi- derably shorter than that of a fully extended, all
trans tail. In LQ bilayers, this length decreases with temperature [2]. We postulate that there is an opti-
mum tail length at a given temperature and either stretching or compressing a tail from the optimum length costs energy. Thus, considering points C and D of figure 1 b, all tails cannot be at the optimum length in the H" phase. This is not necessarily the case for La bilayers.
Note, incidentally, that point D of figure 1 b would be more accessible, for large water cores, if the water core cross-section were not cylindrical but, for ins- tance, rounded hexagonal. However, this would cost bending energy of the lipid layer.
Kirk et al. [4] computed explicit functional forms for the curvature, hydration repulsion and hydro- carbon packing terms used to compute the total free energy of the lipid-water system. Because of the sys- tem complexity and an inadequate knowledge of the forces involved, these forms were necessarily based on various approximations and phenomenological parameters. Even so the resulting total free energy curves were reasonable in view of the observed beha- viour of lipids undergoing the transition. Moreover, the theory suggested experiments which could be used to test if curvature, hydration repulsion and hydrocarbon packing are, indeed, the dominant energies in the transition.
763
This article reports on experiments which primarily probe the hydrocarbon packing and curvature con-
tributions. It should be noted that it is difficult to
design experiments which independently examine
the surface curvature, hydration repulsion, and hydro-
carbon packing because each of these is coupled to
the liquid crystal geometry. Modification of any of the three interactions inevitably changes the geometry.
For example, in this paper we examine the conse-
quences of introducing small amounts of free alkanes
(oil) to the lipid-water mixtures. The fully hydro- phobic alkanes partition into the lipid hydrocarbon and, because they have no polar groups to anchor them at the water interface, they can migrate to
relax stress in the lipid hydrophobic zone. One con-
sequence of adding alkane is that the lipid cylinders expand in radius, thereby, changing the magnitudes
of the curvature and hydration forces. Another conse-
quence is a decrease in the HIj free energy resulting
in a lower temperature for the Lex to Hii transition.
Although we are unable to eliminate geometry- dependent hydration energies, by using an excess
of free water, hydration constraints can be relaxed.
The surface curvature can then be varied by using
mixtures of two lipids which have different polar head-
groups but similar lipid tails. In this way it is shown that when the mixture is such as would be expected
to produce cylinders of very large radius, a small
amount of free alkane can have enormous effects, i.e.,
curvature is otherwise constrained by hydrocarbon packing. Contrawise, a headgroup composition that produces a small cylinder radius is relatively insen-
sitive to the addition of free alkane. In this case, hydro-
carbon packing is constrained by surface curvature.
Thus, by balancing one contribution against another,
a qualitative picture of the relative importance of the
three forces can be developed.
2. Experimental methods.
X-ray diffraction was used to monitor the phase of the lipid liquid crystals. The X-ray apparatus has been described in detail elsewhere; only a brief descrip-
tion will be given here. CuKa X-rays were generated
on a Rigaku RU-200 microfocus rotating anode
machine which was equipped with two independent small-angle X-ray cameras. Both cameras were used in this study. One consisted of single mirror using
Franks X-ray optics [12] coupled to the Princeton
SIV TV area detector [13]. The other used orthogonal,
bent trapezoidal mirrors [14] and the Princeton SIT TV
area detector [15]. Lipid specimens were held in
1.5 mm glass X-ray capillaries sealed with epoxy
plugs. The specimen stage consisted of a thermo-
electrically servoed copper block whose temperature
was computer controlled [16]. The block had a hole drilled through it to accept the glass capillary and
another orthogonally directed hole to allow X-rays
to pass through the capillary.
Pure, powdered dioleoylphosphatidylethanolamine
(DOPE) and dioleoylphosphatidylcholine (DOPC) (Fig. 2) were obtained from Avanti Polar Lipids (Bir- mingham, Alabama). Pure lipid-aqueous solution dispersions of known concentration were made by weighing a capillary, adding dry lipid, weighing again, adding the aqueous solution and weighing again. The lipid was - 98 % anhydrous; an estimate
of this « bound » water was factored into the total water content. In the case of DOPE-DOPC-aqueous dispersions, lipids were mixed in known ratios by dissolving both lipids in chloroform, which was sub- sequently evaporated. Two procedures were used
to mix lipids with alkane. In the first, dodecane was
dissolved in the highly volatile solvent methyl-
butane. This was mixed with the dry lipid powder (approx. 1:1) and the solvent was then evaporated
under a gentle stream of nitrogen. A second procedure
involved injecting a known concentration of lipid
dissolved in chloroform into the capillary via a micro-pipette. The chloroform was then evaporated
under vacuum and the capillary was reweighed to precisely determine the amount of lipid clinging to the
inside capillary walls. Tetradecane dissolved in diethyl
ether was then added and the lipid was solubilized in the capillary, after which the ether was removed under vacuum. Another weighing determined the amount of added tetradecane. Tetradecane was used in the second procedure because it has a higher vapour pressure than dodecane. This procedure was used only on the 95 % and 63 % DOPE curves of figure 7.
Note, however, that specimens prepared by the dode-
cane and tetradecane procedure were not equivalent
in that the tetradecane procedure specimens were
more prone to exhibit co-existing Lex and HII phases.
The aqueous solution consisted of 10 mM HEPES buffer and 100 mM NaCI, adjusted to pH 7.4. (This
buffer solution was used to be consistent with the buffer used in a parallel NMR study. We have observ- ed that dilute monovalent salt effects are small.)
The lipid was mixed with the buffer in the capillary
and .allowed to equilibrate for at least a night.
Fig. 2. - The chemical formulas of DOPE and DOPC are shown.
764
The internal dimensions of the lamellar and H,I phases were determined by standard procedures [2].
To do this it was necessary to know the mass-fractions of lipid, oil, and aqueous solution and the partial specific volumes of the lipid. Values for the latter were
obtained from the literature [17, 18]. The partial spe- cific volumes of the alkanes were assumed to be the
same as in the bulk alkane. Since the alkane mass
never exceeded 10 % that of the lipid, any error intro- duced by this assumption would be small.
X-ray repeat distances were calibrated against
lead-nitrate powder. Reproducibility of the liquid crystal repeat for a given specimen was better than 0.5 A in the small-angle regime and better than 0.1 A
for repeats in the 4-5 A range. The estimated total
error in the computed internal lattice dimensions
was ± 0.5 A. As demonstrated below, the Hjj cylin-
der radius is a function of the DOPE : DOPC ratio.
Lipid fractions were accurate to several percent, by weight. This corresponds to a vertical uncertainty
of the H,I repeat curves of roughly ± 1.5 A for the
DOPE-DOPC mixtures. Thermal accuracy was better than + 0.3 °C.
The X-ray apparatus is highly automated. A typical
data taking strategy was to program a sequence of temperature steps, each of which was followed by a specimen equilibration time and a X-ray exposure.
Typically, temperature was stepped from - 40 to
+ 80 OC in 5 OC steps; thermal settling times were
under 1 min; specimen equilibration times were usually 15-60 min; X-ray exposures were generally
60-400 s. Diffraction patterns consisted of concentric
rings which were integrated to intensity vs. scattering angle. Peak positions and lattice assignments were least-squares determined as noted elsewhere [19, 20].
Several comments are pertinent to the interpretation
of the phase behaviour by thermal scans. Although
many specimens contained excess bulk water, the bulk water was frozen below 0 OC. This sometimes lead to anomalously small values of the repeat spac-
ing below 0 OC. For example, the L, to L, phase tran-
sition is normally accompanied by a thinning of the bilayer and an uptake of water by the lamellar lattice;
the water uptake is inhibited if the excess water pool
is frozen, leading to low repeat values. This also leads to variability in the lattice repeat below 00 because the amount of water left in the lattice depends
on the thermal history of the specimen since the last time it was brought above 00. These considerations do not affect the conclusions of this paper. Also, it is known that the L,,-HI, transition temperature exhibits hysteresis [21] depending on whether the specimen
is being heated or cooled and, thus, the La to H"
temperature, tends to be higher than the Hu to L., temperature. For consistency, all scans in this study,
unless noted otherwise, were from cold to hot. Gene- rally, there is also a temperature span over which the
La and HI, phases co-exist for a given experimental arrangement. In the data plots, phase co-existence
is indicated by the presence of two lattices at one
temperature.
Data is presented as graphs of the repeat dis- tance, d and phase vs. temperature, T. For lamellar patterns (L. and L phases), the distance is the total
X-ray repeat distance; for hexagonal lattices, it is
basis vector length, i.e., twice the distance of R,, +
dUll in figure 1.
Curves were fit to the data points by eye.
3. Experimental results.
The critical questions with which we are concerned
are what factors determine the radius of the HI,
tubes and the temperature ( TBH) at which the La Hn transition occurs. Prior to examining the behaviour
of DOPE when mixed with DOPC and alkanes, it is
useful to review the behaviour of pure DOPE in
excess water (Fig. 3). At - 10 °C and below, the small- angle X-ray scatter (SAXS) consisted of equally spac- ed peaks, indicating a lamellar phase, and the wide-
angle X-ray scatter (WAXS) exhibited a sharp, nearly symmetric spike at 4.2 A, indicating frozen hydro-
carbon [2]. We identify this as the frozen-chain multi-
bilayer L, phase [22]. At - 5 °C there was a sudden
shift in the repeat spacings and relative intensities of the SAXS peaks, but they remained equally spaced.
The WAXS peak was now a broad hump at - 4.6 A.
This indicated the bilayers now had melted hydro-
carbon and were in the Lex phase. The bilayer melt tem- perature was, therefore, Tm 5 °C. At + 5 °C, a
new set of weak SAXS peaks were superimposed
on the Lex peaks; the new set had spacings in the ratios
of 1 : qfl : 2, indicating the (1, 0), (1, 1) and (2, 0)
Fig. 3. - Lattice spacing, d (see Fig. 1) vs. temperature is shown for DOPE in excess water. The symbols for figures 3-
10 : Circles represent the HII phase, squares represent the L.
phase, and triangles represent the L, phase. The occasional open circle represents a data point acquired on a cooling
scan.
765
orders of a hexagonal lattice. We identify this as co- existing La and Hu phases. By + 10 OC, the La peaks
were essentially gone, and the Hu peaks had grown
in intensity. Thus, TBH was in the range of + 5 to + 10°. These results are entirely consistent with prior
work on DOPE. (See [21] and references therein).
The effect of adding 5 % alkane is shown in figure 4.
At - 9 °C the WAXS indicated the chains to still be frozen and the SAXS pattern indicated a single
lamellar phase. Thus, the specimen appeared to be
in the L, phase at - 9 °C. By - 5 OC, the WAXS
indicated the chains had melted and the SAXS
yielded a phase-separated pattern consistent with a
combination of lamellar and hexagonal lattices. At 0 OC, only the hexagonal pattern was apparent. Thus,
the addition of 5 % alkane appeared to dramatically
contract the temperature width of the La phase. Our temperature steps were too large to determine if the
specimen was ever in a pure La phase in the - 9 to
- 5 °C gap.
Our expectation was that alkane could directly
relieve hydrocarbon packing stress by partitioning
into relatively inaccessible regions of the Hu lattice (i.e., point D in Fig. 1) and that the alkane would have
only marginal effects on the membrane curvature.
In so far as it relieves packing stress, it removes a positive contribution to the Hu free energy and would be expected to shift the Hu free energy down with res- pect to the La state [4], thereby promoting the H,I phase at a lower temperature. The lipid monolayer
curvature may be more directly affected by introduc- ing lipid which, by itself, in water, remains in the La phase and only assumes the Hu phase at very high temperatures. DOPC was chosen for this purpose
Fig. 4. - The addition of alkane to the DOPE-excess water
system dramatically contracts the temperature span of the
Lex phase and allows the Hn phase to form just above the melting temperature of the lipid hydrocarbon. Percentages
refer to the non-water mass. Unless otherwise stated, all specimens involving alkane used the dodecane procedure.
because it contains hydrocarbon chains of the same
composition as DOPE and does not go into the HII phase below 100 °C [23]. As seen in figure 5, this has small effect on Tm but dramatically extends the temperature span of the La phase. When 25 % of
the lipid is DOPC, the bilayer to hexagonal transi-
tion temperature is TBH = 50 to 55 °C. If, in addition,
5 % alkane (referred to the total non-water mass)
is added (Fig. 6), the temperature span of the La phase
is dramatically reduced with TBH = - 5 to 0 °C.
Note the extraordinarily large Hu lattice size at 0 °C, corresponding to lipid-water tubes of 46 A radius.
We define the positive equilibrium radius of curva-
ture as the radius of the inverted (i.e., HII-like) water cylinder which minimizes the elastic and hydration
free energies (see [4] for a discussion). Assume that DOPE has a small equilibrium radius of curvature
and this is the fundamental reason DOPE is in the
Hjj phase at room temperature. Likewise, assume that DOPC has a very large equilibrium radius of
curvature and this is why pure DOPC dispersions
are in the La form at room temperature. Since the
two lipids appear to be well mixed, the equilibrium
curvature of a DOPE-DOPC monolayer may be
expected to be of a value between the extremes of the two pure constituents. This is suggested by figure 7,
in which the Hjj size is shown vs. temperature for various DOPE : DOPC ratios. The addition of 5 %
alkane helped to relieve the hydrocarbon packing
constraint and allowed the underlying equilibrium
curvature to be expressed. Note that at a given tem-
Fig. 5. - DOPE has a small equilibrium radius of cur-
vature and DOPC has a very large radius. In excess water,
a lipid mixture of DOPE and DOPC seeks to express a curvature intermediate between the two extremes. Since the equilibrium radius of curvature shrinks as the tempera-
ture is raised, the net result is to cause the Hu phase of the
mixture to appear at a higher temperature than with DOPE alone. Compare to figure 3. Percentages refer to the non-
water mass.