Article pubs.acs.org/Langmuir
Formation of Inverse Topology Lyotropic Phases in Dioleoylphosphatidylcholine/Oleic Acid and Dioleoylphosphatidylethanolamine/Oleic Acid Binary Mixtures Richard J. Gillams,† Tommy Nylander,‡ Tomás S. Plivelic,§ Marcus K. Dymond,∥ and George S. Attard*,† †
Faculty of Natural & Environmental Sciences, University of Southampton, Southampton SO17 1BJ, U.K. Physical Chemistry, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden § MAX IV Laboratory, Lund University, P.O. Box 118, SE-221 00 Lund, Sweden ∥ School of Pharmacy and Biomolecular Sciences, University of Brighton, Lewes Road, Brighton BN2 4GJ, U.K. ‡
S Supporting Information *
ABSTRACT: The addition of saturated fatty acids (FA) to phosphatidylcholine lipids (PC) that have saturated acyl chains has been shown to promote the formation of lyotropic liquidcrystalline phases with negative mean curvature. PC/FA mixtures may exhibit inverse bicontinuous cubic phases (Im3m, Pn3m) or inverse topology hexagonal phases (HII), depending on the length of the acyl chains/fatty acid. Here we report a detailed study of the phase behavior of binary mixtures of dioleoylphosphatidylcholine (DOPC)/oleic acid (OA) and dioleoylphosphatidylethanolamine (DOPE)/oleic acid at limiting hydration, constructed using small-angle X-ray diffraction (SAXD) data. The phase diagrams of both systems show a succession of phases with increasing negative mean curvature with increasing OA content. At high OA concentrations, we have observed the occurrence of an inverse micellar Fd3m phase in both systems. Hitherto, this phase had not been reported for phosphatidylethanolamine/fatty acid mixtures, and as such it highlights an additional route through which fatty acids may increase the propensity of bilayer lipid membranes to curve. We also propose a method that uses the temperature dependence of the lattice parameters of the HII phases to estimate the spontaneous radii of curvature (R0) of the binary mixtures and of the component lipids. Using this method, we calculated the R0 values of the complexes comprising one phospholipid molecule and two fatty acid molecules, which have been postulated to drive the formation of inverse phases in PL/FA mixtures. These are −1.8 nm (±0.4 nm) for DOPC(OA)2 and −1.1 nm (±0.1 nm) for DOPE(OA)2. R0 values estimated in this way allow the quantification of the contribution that different lipid species make to membrane curvature elastic properties and hence of their effect on the function of membrane-bound proteins.
■
INTRODUCTION Over the past two decades there has been growing interest in the effect that free fatty acids (FFAs) may have on the organization and local structure of lipid membranes. This interest stems from an increasing recognition of the multiple biological effects that FFAs are able to elicit as well as the fact that lipids are susceptible to hydrolysis. Early studies established that unsaturated fatty acids (FAs), such as oleic acid (OA) and linoleic acid enhanced the growth rate of both normal and neoplastic rat mammary epithelial cells in culture by a factor of 2.1 Other studies on primary cell cultures also showed the ability of FFAs to modulate cell function. For example, Pérez et al.2 reported that cis-unsaturated fatty acids, such as OA, block the secretion of growth hormone and prolactin in GH3 cells stimulated by thyrotropin-releasing hormone. The wider significance of this observation is that FFAs have been implicated in the pathogenesis of the altered patterns of the secretion of growth hormone that are associated © 2014 American Chemical Society
with obesity and Itsenko−Cushing syndrome (a hormonal disorder). The blocking of growth hormone secretion was attributed to the FAs perturbing the function of integral membrane proteins, such as enzymes, channels, and pumps. More direct evidence of the effect of cis-fatty acids, in particular, OA, on proton flux across lipid bilayers has been reported3 and appears to support this view. More recently, evidence has begun to emerge on the possible links among dietary FFAs, changes in cellular function, and conditions such as inflammation-related diseases and prostate cancer.4−6 Although a number of biological effects resulting from the interaction of FFAs with cells and cell membranes have been documented, the detailed mechanisms through which the FFAs affect membrane proteins and how these effects are related to Received: November 6, 2013 Revised: March 4, 2014 Published: March 7, 2014 3337
dx.doi.org/10.1021/la404275u | Langmuir 2014, 30, 3337−3344
Langmuir
Article
aggregate structures that have positive curvatures, unlike the case of protonated acids. It is well known that the dissociation constants, pKa, for long-chain fatty acids are much higher than for soluble carboxylic acids (pKa values are on the order of 5.0).26 For the oleic acid in an aggregate, pKa values of as high as 9.0 have been reported.27−29 As a consequence, we would expect that in water and at high lipid concentration most of the fatty acids will be fully protonated. In this work, we aim to extend our understanding of the effects of different fatty acid concentrations on the phase behavior of phospholipid−fatty acid binary mixtures at excess hydration. We rationalize these effects in terms of the curvature of the lipid/aqueous interface using lipids found in biological membranes, namely, dioleoylphosphatidylcholine (DOPC)/ oleic acid (OA) and dioleoylphosphatidylethanolamime (DOPE)/oleic acid. DOPC and DOPE were chosen because PC and PE lipids account for the majority of lipid species within most eukaryotic cells (typically >60%). Oleic acid was chosen as the fatty acid/radyl component because it promotes the formation of fluid phases at room temperature and above. Furthermore, in lipidomic studies30 it appears that the cis unsaturation of OA provides a closer approximation to the fluid hydrophobic interior of biological membranes than do saturated species.
changes in cellular function remain unclear. Because the partitioning of FFAs into cell membranes is a prerequisite for their biological activity, understanding how FFAs affect lipid membrane properties may provide insights into how simple physical interactions determine biological activity and influence the control of phospholipid homeostasis.7−9 For example, it is now well established that local changes in membrane curvature and membrane topology, such as may occur because of local variations in lipid composition, can have significant effects on the activity of membrane-associated proteins10 (e.g., ion channels,11 phospholipases,12 and various lipid biosynthesis enzymes13,14). Of particular relevance to this work is the observation of a synergistic activation of CTP/phosphocholine cytidylyltransferase by mixtures of phosphatidylethanolamine and oleic acid and how this is related to the curvaturedependent elastic properties of a membrane.15 In recent years, an increasing body of evidence has emerged implicating membrane curvature elasticity in the function of membraneassociated proteins and in controlling lipid homeostasis.7−9 FFAs are known to be potent modulators of the curvature of phospholipid (PL) assemblies. Evidence for this comes from studies of the lyotropic liquid-crystalline behavior of phosphatidylcholine (PC)/fatty acid mixtures at limiting hydration. For example, Seddon et al.16 and Templer et al.17 reported the phase behavior of mixtures of a range of saturated PCs and saturated FAs in a molar ratio of 1:2. This molar ratio of PC to FA was chosen because of evidence of the formation of a distinct molecular complex, which we denote as PC(FA)2.18,19 The interactions that stabilize the PC(FA)2 complexes result in species that have very small headgroups and large hydrocarbon cross sections. As a consequence, the lyotropic phase behavior of most PC(FA)2 complexes that have been reported is dominated by inverse topology phases (i.e., phases that have polar/apolar interfaces with a negative mean curvature). For mixtures containing short-chained fatty acids, such as dilauroylphoshatidylcholine (DLPC)/lauric acid (LA) and dimyristoylphoshatidylcholine (DMPC)/myristic acid (MA), Im3m and Pn3m inverse bicontinuous cubic phases are observed, as well as the inverse hexagonal phase (HII) and Fd3m phases.20 For mixtures containing longer fatty acids (e.g., dipalmitoylphoshatidylcholine (DPPC)/palmitic acid (PA)) HII phases are formed. Although PC/FA mixtures with a 1:2 molar ratio have been studied extensively,16,17,21−23 the phase behavior of mixtures away from this composition regime has received relatively little attention.19,24,25 Phase diagrams for binary mixtures of DPPC and a range of fatty acids have been reported.19 On the basis of differential scanning calorimetry and Fourier transform infrared spectroscopy observations, it has been suggested that the solidstate and gel-phase diagrams of DPPC/FA mixtures can be considered to be a juxtaposition of two pseudobinary phase diagrams: one for FA mole fractions in the mole fraction range of 0 < x < 0.67, corresponding to a mixture of DPPC and DPPC(FA)2, and one in the composition range of 0.67< x < 1, corresponding to a mixture of DPPC(FA)2 and FA. The formation of the DPPC(FA)2 complex even at low FA concentrations, with its large negative spontaneous curvature, introduces a strong drive toward the formation of phases with negative curvature. This was observed in the case of DPPC/PA mixtures.19,25 We also note that a complication that arises from studies of mixtures of fatty acids with phospholipids is the fact that the deprotonated form of the fatty acids (i.e., soaps) tends to form
■
EXPERIMENTAL SECTION
1,2-Dioleyl-sn-glycero-3-phosphocholine and 1,2-dioleyl-sn-glycero-3phosphoethanolamine were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Oleic acid was purchased from Sigma-Aldrich (UK). Ultrapure water of 18.2 mΩ conductivity (Barnstead Nanopure Diamond) was used for the preparation of all samples. Preparation of Lyotropic Liquid-Crystal Phases. Samples for X-ray measurements were prepared by weighing dry quantities of each lipid (used as received) into 500 μL microcentrifuge tubes. Typically, the total mass of dry lipid samples was in the range of 50 to 100 mg. The moisture content of the lipids was estimated to be ≤0.1% w/w, leading to an uncertainty in the lipid composition of the mixtures of up to 2 mol %. Lyotropic liquid-crystal preparations were made in excess of limiting hydration by adding 100 μL of pure water to each sample. All samples were mixed manually using a small spatula for several minutes prior to centrifugation at 17 000 g (Heraeus Pico 17 Centrifuge) for 5 min. Manual mixing and centrifugation cycles were performed three times before samples were incubated at 37 °C for 2 to 3 days. Prior to data collection, further manual mixing was carried out. Small-Angle X-ray Diffraction (SAXD) Studies. All studies were performed at SAXS station I911-4 of the MAX IV Synchrotron, Lund, Sweden.31 The diffractograms were recorded using a wavelength λ of 0.91 Å, and a bidimensional hybrid pixel X-ray detector (Pilatus 1M, Dectris) ensured a scattering vector q range of 0.08−5.20 nm−1 (q = (4π/λ)sin(θ), where 2θ is the scattering angle). The q calibration was done using a silver behenate standard sample. Data were reduced using the bli911-4 software31 and analyzed with PRIMUS software.32 Prior to measurement, samples were sandwiched between ultralene films and loaded into custom-built sample holders. Temperature scans were made in 5 to 10 °C step increments, and diffraction patterns were measured after a 60 s temperature equilibration and when the temperature fluctuations were less than 0.1 °C. Lattice parameters, which refer to the length scales of the repeat units of the phase, were obtained by fitting the observed diffraction peaks to the Miller index function appropriate to the phase (Supporting Information Figure S1a−S1d,S2). Errors in the lattice parameters estimated in this way are on the order of ±0.2 Å.
■
RESULTS AND DISCUSSION Phase Behavior. The phase behavior of binary mixtures DOPC/OA and DOPE/OA as determined from our SAXD
3338
dx.doi.org/10.1021/la404275u | Langmuir 2014, 30, 3337−3344
Langmuir
Article
Figure 1. Temperature−composition phase diagram of binary mixtures of (a) DOPC + OA and (b) DOPE + OA at limiting hydration. The gray points indicate where observations were made. Gray areas indicate the extent of homogeneous phases (as judged from the SAXD data). White areas indicate biphasic regimes, and the dashed line in plot (a) indicates the likely existence of a coexistence region comprising the two cubic phases, over a very narrow range of compositions. Compositional uncertainties are smaller than the sizes of the plot symbols.
Figure 2. Dependence of the lattice parameter on the mole fraction of oleic acid at 37 °C for (a) DOPC/OA and (b) DOPE/OA mixtures. Phases: Lα (●), HII (◊), Im3m (*), Pn3m (▲), andFd3m (○). The lattice parameter of each phase constitutes the fundamental length scale that defines the translational periodicity of that phase.
DOPE system is that the starting point is a curved phase (HII) as opposed to a lamellar phase. Above 0.4 OA mole fraction, the HII phase transforms to an Fd3m phase. Thus, in DOPE/ OA mixtures the composition at which the transition from HII to Fd3m occurs (xHII−Fd3m) is shifted to lower OA mole fractions (0.4 < xHII−Fd3m < 0.5) than for DOPC/OA mixtures (0.6 < xHII−Fd3m < 0.7). We note that within the composition resolution of our experiments (x = ± 0.01) the positions of the phase boundaries are essentially unchanged with temperature. Structural Changes. The changes in the lattice parameters of the phases formed by DOPC/OA and DOPE/OA mixtures at limiting hydration are illustrated by the data obtained at the physiologically relevant temperature of 37 °C, shown in Figure 2. DOPC/OA Mixtures at 37 °C. At low OA concentrations, DOPC forms an Lα phase with an interlamellar spacing of 6 nm, which does not change as further OA is added to the mixture, even when the lamellar phase coexists with a putative Im3m bicontinuous cubic phase (x = 0.2). The uncertainty in the identification of this phase is due to the overlapping diffraction peaks arising from the coexisting phases. At this composition, the lattice parameter of the cubic phase is 40 nm.
data and supported by polarized light microscopy observations (Figure S3) is mapped in Figure 1. As expected from previous studies on PC/saturated FA mixtures, the addition of oleic acid to DOPC even at low concentration drives the phase behavior toward the formation of phases with increasing negative curvature (Figure 1a).16,19,21,25 Thus, a mixture containing 0.2 mole fraction of OA forms coexisting fluid lamellar (Lα) and inverse cubic phases. Indexing of the peaks suggests an Im3m cubic phase. At 0.3 mole fraction of OA, there is a different coexistence region comprising an additional cubic phase with different topology, probably Pn3m, and an inverse hexagonal phase (HII). This tentative identification of the two QII phases indicated in Figure 1a, which is due to the uncertainty in assigning overlapping peaks from the two phases, leads to a phase sequence that is consistent with that reported for a range of PC/FA mixtures.25 Between ∼0.35 and ∼0.65 OA mole fraction, the system forms a homogeneous hexagonal phase, which by a mole fraction of 0.7 gives way to an Fd3m inverse micellar cubic phase. At OA mole fractions >0.8, the system forms a fluid inverse micellar phase (L2). A similar progression to phases with increasing negative curvature of the lipid aqueous interface is also observed for DOPE/OA mixtures (Figure 1b). The main difference in the 3339
dx.doi.org/10.1021/la404275u | Langmuir 2014, 30, 3337−3344
Langmuir
Article
Figure 3. Effect of temperature on lattice parameters of the different phases formed by (a) DOPC/OA and (b) DOPE/OA mixtures. The OA mole fraction is indicated on each label after the phase name.
curvatures of the assemblies as temperature is increased. These findings are comparable to observations of temperature effects on HII phases of PC/FA mixtures reported in the literature (e.g., −2 × 10−2 nm K−1 for the HII phase of (1:2) diarachidoylphosphatidylcholine (DAPC)/arachidic acid (AA)16 and −2 × 10−1 nm K−1 for (1:5.7) DMPC/MA25). Our experimental findings for the effect of OA on the HII phase regarding the increase in curvature suggest that the OA is mostly located in the acyl chain region and hence protonated. In the case of the Lα phase that occurs only in DOPC and in the DOPC/OA mixture with a 0.2 mole fraction of OA, the increase in the d spacing of the phase as a function of temperature is 0.75 × 10−2 nm K−1 for DOPC and 1 × 10−2 nm K−1 for DOPC/OA. These values are comparable to the value of 1 × 10−2 nm K−1 that has been reported for the Lβ phase of (1:2) DAPC/AA and 0.5 × 10−2 nm K−1 for the Lα phase of (2:1) DMPC/MA.25 We finally note that the observed swelling of Lα in the presence of OA suggests that, in contrast to the case for the inverse phases, the OA is located with the carboxyl group fully in the lipid headgroup region and is likely to be partially deprotonated and charged. We have previously observed that an increase in curvature of a liquid-crystalline phase might cause an increase in the OA pKa to the larger value observed by pH-stat titration of a cubic phase exposed to lipolytic enzymes.33,34 Spontaneous Curvatures of Binary Fatty Acid− Phospholipid Mixtures from Hexagonal Phase Data. The temperature dependence of the lattice parameters of lyotropic phases over compositional regimes where they are not in coexistence with other phases, as observed by SAXD, can be used to obtain estimates of the spontaneous curvature of the different PL/OA mixtures and also of their component species. Helfrich introduced and refined the concept of membrane rigidity and the bending modulus, which we apply in this study.35 The Helfrich free energy (per unit area) for an elastic deformation or bending of the lipid monolayer (or bilayer) with principal curvatures c1 and c2 away from the spontaneous curvature c0 for lipid species k is
Further addition of OA to DOPC leads to a different biphasic regime in which an HII phase coexists with a putative Pn3m phase. The lattice parameter of the HII phase in this coexistence region is 8 nm, and that of the Pn3m phase is 29 nm. The lattice parameter of the HII phase decreases slightly with increasing OA concentration over the stability range of the HII phase (0.3 ≤ x ≤ 0.7), going from 8 to 6 nm. When the OA mole fraction is 0.7, the DOPC/OA system at 37 °C is biphasic, with the HII phase coexisting with an Fd3m inverse micellar cubic phase having a lattice parameter of 17 nm. Over the relatively narrow range of compositions over which the Fd3m phase occurs, its lattice parameter decreases and is 15 nm when x = 0.8. DOPE/OA Mixtures at 37 °C. DOPE/OA mixtures with OA mole fractions up to 0.6 form an inverse hexagonal phase with a lattice parameter that decreases slightly with increasing OA concentrations (from 7.0 nm when x = 0 to 5.0 nm when x = 0.6). The Fd3m phase that occurs in the mole fraction range of 0.5 ≤ x ≤ 0.7 has a lattice parameter of 16 nm that does not change when the phase coexists with the HII phase (x = 0.5, 0.6) but falls to 14.5 nm when only the Fd3m phase is present (x = 0.7). A comparison of the HII and Fd3m lattice parameters for DOPC/OA mixtures with those of DOPE/OA mixtures indicates that, as expected, the latter systems have the more highly curved interfaces (i.e., they have smaller lattice parameters, corresponding to smaller mean curvatures). Temperature Dependence of Phase Structural Parameters. With the exception of the Lα phase, the lattice parameters of all of the lyotropic phases observed in the DOPC/OA and DOPE/OA mixtures studied decrease approximately linearly with increasing temperature, as shown in Figure 3. Linear fits of the data for the HII phase of the DOPC/OA mixtures give a slope of −0.9(±0.1) × 10−2 nm K−1, whereas for the DOPE/ OA mixtures the slope is −1.3(±0.3) × 10−2 nm K−1. The resolution of the data is not sufficient to determine whether there is any correlation between the rate of lattice parameter change with temperature and the mole fraction of OA in the mixture, as is the case with DMPC/MA mixtures.25 The trends we observe correspond to an increase in the negative mean 3340
dx.doi.org/10.1021/la404275u | Langmuir 2014, 30, 3337−3344
Langmuir
Article
Figure 4. Graphs obtained by plotting eq 5 using curvature values of the lattice parameters of the HII phases of (a) DOPC/OA and (b) DOPE/OA. Lines are linear fits to the data. DOPC/OA mixtures: crosses, filled triangles, and empty inverted triangles show data obtained at mole ratios of 6:4, 5:5, and 4:6, respectively. DOPE/OA mixtures: crosses, filled triangles, inverted empty triangles, and empty diamonds show data obtained at mole ratios of 1:0, 8:2, 7:3, and 6:4, respectively.
κ (c1, k + c 2, k − c0, k)2 + κGc1, kc 2, k 2
Ek =
1 2
(1)
where κ is the mean curvature elastic modulus (for the lipid layer) and κG is the Gaussian curvature elastic modulus, which is a consequence of deformation relative to a saddle point or a surface that is as convex as it is concave (zero mean curvature). For an ideal binary mixture of amphiphiles (species i and j), the total curvature elastic energy is assumed to be a linear combination of the curvature energies of the individual components such that E T = xEi + (1 − x)Ej
xκi 2
{(
2
2
(
)
+ c 2,(A) i
2
− c 2,(B) i
dl(A) = 2R w(A) + 2l(A)
and l
1
}
(1 − x)κj 2
{(c
(
2
2
2
)
(
)
}
{
}
(A) (B) (B) (A) (B) (B) + xκG, i c1,(A) + (1 − x)κG, i c1,(A) i c 2, i − c1, i c 2, i j c 2, j − c1, j c 2, j
(3)
For hexagonal phases c2 = 0 and making the assumption that κi ≈ κj,14 eq 3 is reduced to 2ΔE T =x κ
{(c
2
2
2
2
) − (c1,(B)i ) } + (1 − x){(c1,(A)j ) − (c1,(B)j ) } (B) (A) (B) − 2(1 − x)(c1,(A) j − c1, j )c 0, j − 2x(c1, i − c1, i )c 0, i (A) 1, i
= R p(A) = R w(A) + lx(A)
(7)
(8)
and lx is the distance from the pivotal plane (R0), which is the plane on which the area per lipid molecule is not changed upon applying a bending moment (Supporting Information, Figure S1d). Spontaneous curvatures are conventionally measured at the edge of the water pore.36 Note that R0, the radius of curvature of the pivotal plane, is R0 = 1/c0. The expectation is that, for small deviations of Rp away from R0, lx does not change significantly relative to Rw and thus Rp, at different temperatures, can be estimated from eq 6.35 Evidence supporting the assumption that changes in lx are small compared to Rw comes from studies determining the spontaneous curvature of DOPC in the presence of DOG.38 Over the composition range of 0.2 to 0.35 mole fraction of
}
(A) (B) (B) (B) (A) (B) + 2 c1,(A) − 2 c1,(A) j c 2, j − c1, j c 2, j j − c1, j + c 2, j − c 2, j c 0, j
{
(6)
is the radius of the water cylinder in the HII phase is the lipid length, both at temperature A, and
l(A) = lx(A) + l(A) y
2
) − (c1,(B)j ) + (c2,(A)j ) − (c2,(B)j )
(A) 1, j
}
where
(A) (B) (B) (B) (A) (B) + 2 c1,(A) − 2 c1,(A) i c 2, i − c1, i c 2, i i − c1, i + c 2, i − c 2, i c 0, i
+
{
where Rw(A) (A) c1(A)
)
(5)
for pairs of temperatures (A and B) gives a line with an intercept of ΔET/κ and a slope equal to the spontaneous curvature of the mixture, c0. c1(A) and c1(B) can be obtained from the lattice parameters of our binary lipid system data because the inter pore spacing dl at temperature A is related to c1(A) as follows in eqs 6−8
2
(
}
{
(2)
− c1,(B) i
ΔE T κ
1 (c1(A))2 − (c1(B))2 against c1(A) − c1(B) 2
) ( ) ( ) ( )
c1,(A) i
2
( ) } = c0{c1(A) − c1(B)} +
− c1(B)
which is in the form of a linear equation. Thus, for a given binary mixture, a plot of
where Ei is the curvature energy of component i and x is its mole fraction. Taking the difference in the curvature energy (ΔET) between temperatures A and B and assuming that for each species of the mixture c0 is independent of temperature (i.e., for component i, (B) c(A) 0,i ≅ c0,i ≡ c0,i) gives ΔE T =
2
{(c1(A))
(4)
This equation can be rearranged in terms of the mixture parameters to give 3341
dx.doi.org/10.1021/la404275u | Langmuir 2014, 30, 3337−3344
Langmuir
Article
DOG in DOPC, the average value of lx is 0.55 ± 0.05 nm, whereas Rw changes from 4.3 to 2.6 nm so the variance in lx with respect to Rw is less than 2%. Similar estimates can be made for mixtures that contain DOPE.39 A test of the reliability of this approach can be made using our data for pure DOPE (x = 0) in Figure 3b to determine R0 from eq 5. The gradient of this plot is −0.42 nm−1, which means that R0 is equal to −2.38 ± 0.12 nm. This estimate is in excellent agreement with values reported in the literature. Gruner et al. report a value of −2.38 ± 0.24 nm for DOPE in water,40 Fuller et al.41 determine R0 = −3.0 nm in pure water from DOPE/DOPS HII phases, and Leikin et al.39 quote a value of −2.85 nm from studies of DOPE/DOG HII phases. Figure 4 shows plots of eq 5 for the DOPC/OA and DOPE/ OA mixtures that form pure HII phases (i.e., HII phases that are not in coexistence with other phases). In all cases, the data points are in excellent agreement with the linear fits. The slopes of the regression curves were used to estimate R0 for the two sets of mixtures, and these are shown in Table 1. Because c0 for
increases in the HII phase, the assumption that lx is constant becomes less reasonable. Therefore, this method of estimating spontaneous curvature will be most reliable when applied to data obtained at temperatures
