Energetics of the Mixing of Phospholipids in Bilayers Determined

Nov 17, 2016 - ... of the Mixing of Phospholipids in Bilayers Determined Using Vesicle Solubilization. Keisuke Ikeda and Minoru Nakano. Graduate Schoo...
0 downloads 8 Views 1MB Size
Download PDF
Article pubs.acs.org/Langmuir

Energetics of the Mixing of Phospholipids in Bilayers Determined Using Vesicle Solubilization Keisuke Ikeda* and Minoru Nakano Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama 930-0194, Japan S Supporting Information *

ABSTRACT: Here, we report an experimental approach for determining the change in the free energy and the enthalpy that accompanies the mixing of the anionic phosphatidylglycerol and the zwitterionic phosphatidylcholine. The enthalpy change originates in the thermal changes of disrupting lipid bilayer vesicles titrated into a surfactant micelle solution and is monitored using isothermal titration calorimetry. The difference in the solubilization enthalpies between pure and mixed lipid vesicles yields the lipid mixing enthalpy. The Gibbs free energy changes are estimated by determining the thermodynamic equilibrium constants of forming a molecular complex between phospholipids and methyl-βcyclodextrin. We provide direct experimental evidence that mixing of the anionic lipid and the zwitterionic lipid is explained well by the entropic term of the electrostatic free energy of a charged surface in the Gouy− Chapman model. The present strategy enables us to determine the precise energetics of lipid−lipid interactions in near-native environments such as liposomes without any chemical modification to lipid molecules.

■

systems,10,11 determining the thermodynamic parameters of lipid−lipid interactions in bilayers is still challenging. Here, we determined the enthalpy and free energy changes that accompany the mixing of two miscible phospholipids in lipid bilayers. The mixing of the zwitterionic 1-palmitoyl-2oleoylphosphatidylcholine (POPC) and the anionic 1-palmitoyl-2-oleoylphosphatidylglycerol (POPG) was analyzed as an example. Lipid vesicles of the mixed lipid compositions were solubilized by the Triton-X100 surfactant micelles and monitored using isothermal titration calorimetry (ITC), which provided the standard enthalpy changes of vesicle solubilization, ΔH0sol. On the other hand, the vesicles were solubilized using methyl-β-cyclodextrin (MβCD) and monitored by the turbidity of the solution. Analysis of the lipid− MβCD complexation equilibrium provides the standard Gibbs free energy changes of vesicle solubilization, ΔG0sol. ΔH0sol and ΔG0sol are used to calculate the enthalpy change of bilayer mixing, ΔH0mix, and excess free energy, ΔGEmix, at the molar ratio of POPG (X) in the binary vesicles, which are defined assuming an additivity of the parameters

INTRODUCTION Lipid−lipid interactions in biomembranes play key roles in a variety of biological processes in living cells, such as signal transduction, transport of molecules, and metabolism. Biomembranes of eukaryotic and bacterial cells are composed of more than a thousand different lipid species, controlling a variety of structures and functions.1 The coupling of lipid molecules induces the formation of microdomains called lipid rafts, which regulate the biomembrane actions by recruiting or excluding specific membrane proteins in clusters.2,3 A recent model of raft domains proposes that transient homodimers of glycosylphosphatidylinositol (GPI)-anchored proteins are the basic units of raft organization.4 Lipid−lipid interactions between cholesterol, the saturated chains of GPI-anchored proteins and glycosphingolipids, stabilize the homodimers of GPI-anchored proteins.5 Membrane protein functions are also regulated by the binding of specific lipids and/or the physicochemical parameters of the membrane environment, such as phase, mobility, and bilayer thickness.6,7 The lipid−lipid interactions that determine the domain formation and the physical properties of lipid bilayers have been investigated by extensive studies of model membrane systems.8,9 It is essential to elucidate the changes in the thermodynamic parameters such as free energy, enthalpy, and entropy accompanying the lipid− lipid interactions in order to understand the molecular mechanisms of biomembrane structures and functions. Although the excess free energy of lipid mixing can be conventionally determined using the Goodrich−Pagano− Gershfeld approach using a π−A isotherm of monolayer © 2016 American Chemical Society

0 0 0 ΔHmix (X ) = −ΔHsol (X ) + X ·ΔHsol,PG + (1 − X ) · 0 ΔHsol,PC

(1)

Received: September 9, 2016 Revised: November 17, 2016 Published: November 17, 2016 13270

DOI: 10.1021/acs.langmuir.6b03333 Langmuir 2016, 32, 13270−13275

Article

Langmuir

of the reaction mixture was transferred to a 10K Nanosep ultrafiltration device (PALL Corporation, Port Washington, NY) and the filtrate was collected by centrifugation (10 000 × g, 10 min). The concentrations of POPC and total phospholipids (POPC + POPG) in the filtrate were determined.

E 0 0 ΔGmix (X ) = −ΔGsol (X ) + X ·ΔGsol,PG + (1 − X ) · 0 ΔGsol,PC

(2)

■

Here, the subscripts PC and PG represent the thermodynamic parameters for pure bilayer systems of POPC and POPG, respectively. We also assume that the solubilization of the lipids in the micelles completely suppresses the phospholipid− phospholipid interactions at a low lipid-to-surfactant ratio. Note that ΔGEmix represents the nonideal part of the mixing Gibbs energy change (see details in the Theory section).

■

THEORY Thermodynamic Model of MβCD-Phospholipid Complexation and Lipid Mixing. The complex formation of MβCD and POPC was formulated by Anderson et al.16 They demonstrated that the POPC−MβCD complexation is described by an equilibrium of [lipid in bilayer] + n[free MβCD] ⇄ [lipid−MβCDn], with an equilibrium constant K.16 They assumed that every complex has the same 1:n stoichiometry between the phospholipid and MβCD. We applied their approach to binary lipid systems of lipid 1 and lipid 2. The thermal equilibrium of the formation of the MβCD−lipid complex in the presence of lipid vesicles is described as Y1 K1 = γ1X1YDn1 (5)

EXPERIMENTAL SECTION

Materials. POPC with >99% purity and POPG, sodium salt with >98% purity were purchased from NOF Corporation (Tokyo, Japan). 1-Palmitoyl-2-oleoyl-sn-glycero-3-ethylphosphocholine (ethyl-POPC), chloride salt was purchased from Avanti Polar Lipids (Alabaster, AL). MβCD (average Mn 1310) was purchased from Sigma-Aldrich (St. Louis, MO). Triton X-100 was purchased from Nacalai Tesque (Kyoto, Japan). Preparation of Lipid Vesicles.12 Aliquots of POPC and POPG in chloroform/methanol (1:2) were transferred to a round-bottom flask, and the solvent was removed using a rotary evaporator. After drying under vacuum overnight, the residual lipid film was hydrated in a buffer [20 mM 3-morpholinopropane-1-sulfonic acid (MOPS)/130 mM NaCl/1 mM EDTA-2Na, pH 7] to produce a multilamellar vesicle (MLV) suspension. Large unilamellar vesicles (LUVs) were prepared as follows. The MLV suspension was subjected to five cycles of freezing and thawing for reducing the lamellarity of the vesicles13 and filtered 31 times through a polycarbonate filter with 100 nm pores using a LiposoFast extruder (Avestin, Ottawa, Canada). The concentration of POPC in the vesicles was determined using a Phospholipids C test kit (Wako, Tokyo, Japan). The total phospholipid concentration (POPC and POPG) was determined using the phosphorus analysis.14 ITC.15 ITC was carried out using a MicroCal iTC200 isothermal titration calorimeter (Malvern Instruments, Worcestershire, UK). Triton X-100 (200 μL, 1% v/v) in MOPS buffer was titrated with LUVs (2 mM total lipid) at 37 °C. The titration was performed with a first injection of 0.4 μL and seven subsequent injections of 2 μL at 150 s intervals. The heat generated by the injection of LUVs into the buffer was subtracted as a blank. Solubilization of MLVs by MβCD. A 100 μL reaction mixture of the MLV suspension (0−3 mM total lipid) and MβCD (0−70 mM) in MOPS buffer was prepared in a 96-well microplate and incubated for 24 h at 37 °C to attain equilibrium. The absorbance at 450 nm was measured using an iMark microplate absorbance reader (Bio-Rad, Hercules, CA), and the absorbance of the buffer was subtracted. The boundary lipid molar concentration, C*L , just above which the lipid vesicles begin to coexist with the dissolved species, at each MβCD concentration was determined from the absorbance A as a function of the molar concentration of the lipids. We ignored the low values of absorbance (A < 0.01) indicating the complete solubilization of MLVs by MβCD. Then, the following function was used for estimating CL* as an x-intercept

A = a([lipid] − C L*)

K2 =

(6)

Here, subscripts 1 and 2 represent lipids 1 and 2, respectively. K denotes the equilibrium constant. Y and YD are the mole fractions of the lipid−MβCD complex and free MβCD in solution, respectively. To account for the deviations from ideal mixtures, we introduced an activity of phospholipid, γX, where γ is an activity coefficient at a molar fraction of X (X1 = 1 − X2) in the binary bilayers. The total mole fraction of the complexes in solution, Y is given by Y = Y1 + Y2 = K1γ1X1YDn1 + K 2γ2X 2YDn2

(7)

When we assume the identical binding stoichiometry between the lipids and MβCD (n = n1 = n2), it becomes Y = K′YDn, where K′ = K1γ1X1 + K2γ2X2. The total mole fraction of MβCD, YT = YD + n1Y1 + n2Y2 = YD + nY is given by YT = nY + (Y /K ′)1/ n

(8)

At the boundary concentration of the total lipid, C*L , Y is identical to YL*. The above equation then assumes the same form as eq 4. The constant, K′, is exactly equal to the binding constant, K, for pure lipid bilayers with γ = 1 and X = 1. In fact, C*L agreed well with the concentration of the complex or the dissolved lipid in coexistence with the lipid vesicles for a pure POPC system (Figure S1). Now, we assume that the lipid composition does not change during the solubilization of the mixed vesicles, Y1 = X1Y and Y2 = X2Y, yielding K′ = K1γ1 = K2γ2. In this regime, the standard Gibbs free energy change of dissolving the vesicles is given by ΔG0sol = −RT ln K′ = −RT ln K1γ1 = −RT ln K2γ2. The excess Gibbs free energy change of lipid mixing is given by eq 2

(3)

Boundary concentrations were determined from at least three data points. A plot of the molar concentration of MβCD, [MβCD] as a function of CL* is analyzed by the following equation16

YT = nY L* + (Y L*/K )1/ n

Y2 γ2X 2YDn2

E 0 0 0 ΔGmix (X1) = −ΔGsol (X1) + X1·ΔGsol,1 + X 2 ·ΔGsol,2

(4)

= RT (X1 ln γ1 + X 2 ln γ2)

Here, YT and Y*L are mole fractions: YT = [MβCD]/(55.5 M) and Y*L = CL*/(55.5 M), respectively. Data fitting was performed using the IGOR Pro software (WaveMetrics, Lake Oswego, OR). Ultrafiltration of the reaction mixture was also carried out to determine the concentration of the dissolved lipids by separating the lipid molecules bound to MβCD and MLVs. For this purpose, 300 μL

(9)

ΔGEmix

Here, in the present analysis does not include the contribution of the ideal mixing entropy term of the two phospholipids in the binary bilayer, −TΔSideal. Note that the assumption of a constant lipid composition during the 13271

DOI: 10.1021/acs.langmuir.6b03333 Langmuir 2016, 32, 13270−13275

Article

Langmuir solubilization is valid at C*L , although not always true in general. To confirm that the lipid composition was unchanged during solubilization, we carried out ultrafiltration of the POPC/ POPG (1/1) binary vesicles (3 mM) in the presence of MβCD (60 mM). A POPC ratio of 0.52 ± 0.04 to the total lipid concentration dissolved in the filtrate (1.03 mM) was observed. Gouy−Chapman Model for Charged Lipid Bilayer Solubilization and Lipid Mixing. The electrostatic free energy of a planar charged surface can be described by the Gouy−Chapman model.17 The following theoretical treatment has already been reported by Bergström and Eriksson for a mixture of ionic and nonionic surfactants.18 The Helmholtz free energy of the surface per charge, Ae,c in a 1:1 electrolyte solution is given by ⎧ ⎪ Ae,c /kT = 2⎨ ln(S + ⎪ ⎩

S2 + 1 ) −

⎪ ( S2 + 1 − 1) ⎫ ⎬ ⎪ S ⎭

(10)

Here, S is a dimensionless parameter with the electrolyte concentration of C0 S = σ / 8000kTNAC0ε0εr

(11)

The surface charge density, σ, is written by σ (X ) = X e / a l

(12)

where X is the molar fraction of the charged lipid. The area of cross section of the lipid head group, al, is assumed to be constant at 0.6 nm2. Then, the free energy per lipid, Ae,l is given by ⎧ ⎪ Ae,l /kT = 2X ⎨ ln(S + ⎪ ⎩

S2 + 1 ) −

Figure 1. Titration of POPC/POPG LUVs into a Triton X-100 micelle solution monitored using ITC. (a) Thermal changes during eight injections (0.4 μL × 1, and 2 μL × 7) of lipid vesicles (2 mM) into a buffer containing Triton X-100 (1% v/v, 200 μL). (b) Changes in the enthalpy of solubilization of LUVs, ΔH0sol (kcal/mol), as a function of the molar ratio of POPG in LUVs, XPOPG. The error bars represent ±SD (n = 7). The solid blue line shows the theoretical electrostatic enthalpy changes, ΔH0e,sol, calculated from the Gouy− Chapman model of vesicle solubilization. All measurements were performed in 20 mM MOPS/130 mM NaCl/1 mM EDTA-2Na buffer (pH 7) at 37 °C.

⎪ ( S2 + 1 − 1) ⎫ ⎬ ⎪ S ⎭

(13)

Similarly, the internal energy per lipid, Ue,l is Ue,l /kT = X( S2 + 1 − 1)/S

(14)

shows the thermograms of ITC when the Triton X-100 solution (1% v/v, ∼15 mM) was titrated using lipid vesicles. After eight injections, the final concentration of lipids in the ITC cell reached 134 μM. Under these conditions, the added lipids were completely dissolved in the micelles. The observed peak areas represent the enthalpy changes of solubilization, ΔH0sol, for the transfer of lipid molecules from the lipid bilayers to the surfactant micelles. The contribution of monomeric lipids in solution can be neglected because the concentration of the phospholipid monomer in solution coexisting with the lipid vesicles is low (10−5 to 10−10 M).17,19 LUVs with higher molar ratios of POPG exhibited larger exothermic peaks, indicating that ΔH0sol became more negative as XPOPG increased (Figure 1). This is probably because the electrostatic repulsion between the negatively charged POPG head groups in the lipid bilayers is relieved by the dilution in nonionic micelles, whereas zwitterionic POPC had a small ΔH0sol during the solubilization. A standard enthalpy change of mixing, ΔH0mix, can then be derived from eq 1 and plotted as a function of XPOPG (Figure 3a). The free energy change of lipid mixing can be derived from the free energy changes of transferring a lipid molecule from pure and mixed bilayers into a solution, which is related to the monomer concentrations of the phospholipids coexisting with the lipid vesicles. Unfortunately, determining the transfer

When the charged surface is dissolved into a solution by the addition of surfactant micelles or MβCD, the electrostatic free energy change and the internal energy change are given by 0 ΔAe,sol (X ) = −Ae,l (X )

(15)

0 ΔUe,sol (X ) = −Ue,l(X )

(16)

ΔH0e,sol

ΔU0e,sol

ΔG0e,sol

For incompressible systems, = and = ΔA0e,sol. Finally, the electrostatic Gibbs free energy change and the enthalpy change in mixing charged and neutral lipids are given by 0 0 0 ΔGe,mix (X ) = −ΔGe,sol (X ) + X ·ΔGe,sol (1) 0 + (1 − X ) ·ΔGe,sol (0)

(17)

0 0 0 ΔHe,mix (X ) = −ΔHe,sol (X ) + X ·ΔHe,sol (1)

■

0 + (1 − X ) ·ΔHe,sol (0)

(18)

RESULTS AND DISCUSSION We determined the enthalpy changes that occur during the solubilization of the phospholipid vesicles in the presence of an excess amount of Triton X-100 surfactant micelles.15 Figure 1a 13272

DOI: 10.1021/acs.langmuir.6b03333 Langmuir 2016, 32, 13270−13275

Article

Langmuir

Figure 2. Solubilization of the phospholipid vesicles in buffers containing MβCD. (a) Absorbance of the POPC vesicle suspension as a function of the lipid concentration in the presence of MβCD at 450 nm. The molar concentrations of MβCD in solution, [MβCD] for each measurement are indicated. (b) Plot of [MβCD] vs C*L for vesicles of different XPOPG values. The solid lines are the fitting curves (eq 4), where n = 4. (c) Changes in 0 = −RT ln K as a function of XPOPG. The error bars represent ±SD (n = 2). The solid blue line shows the standard free energy of solubilization ΔGsol the theoretical electrostatic free energy changes, ΔG0e,sol, calculated from the Gouy−Chapman model of vesicle solubilization. Offset of −10 kcal/mol was added. All reactions were performed in 20 mM MOPS/130 mM NaCl/1 mM EDTA-2Na buffer (pH 7) at 37 °C.

decreased with an increase in XPOPG, suggesting that the solubilization of the charged bilayers reduces the unfavorable electrostatic interactions, as seen in the case of ΔH0sol. The excess free energy mixing of POPC and POPG in bilayers, ΔGEmix, is given by eq 2 (Figure 3b).

energy of the phospholipids is hampered by the small monomer concentrations.17,19 Instead, we analyzed the solubility curve of the lipid vesicles in the presence of MβCD. A number of studies have revealed that MβCD forms a complex with phospholipids by incorporating the hydrophobic acyl chains of phospholipids into its hydrophobic cavity and dissolves the bilayer vesicles.16,20−22 Anderson et al. demonstrated that the POPC−MβCD complexation is described by an equilibrium of [lipid in bilayer] + n[free MβCD] ⇄ [lipid−MβCDn], with an equilibrium constant K.16 They determined the boundary concentration of the phospholipid (CL*) at a given MβCD concentration, below which the lipid vesicles were completely dissolved. C*L is a function of the total MβCD concentration (eq 4). Thus, the standard free energy change of vesicle solubilization by MβCD is ΔG0sol = −RT ln K. Equation 4 also holds for mixed bilayers by simply assuming that the lipid composition of the bilayer does not change during the solubilization process. Identical 1:n stoichiometry for the lipid−MβCD complexes is also required. We investigated the solubilization of POPC, POPG, and binary vesicles in the presence of MβCD on the basis of their optical densities (Figures 2a and S2). The optical density of the suspension linearly increased with the lipid concentration at a low MβCD concentration of 10 mM. However, at higher MβCD concentrations, two distinct regions are detected: a flat region with no signals, suggesting a complete solubilization of the vesicles below C*L , which is followed by a region of linearly increasing intensity corresponding to an increase in the vesicle concentration. The value of CL* at each MβCD concentration can be determined from the x-intercept of the least-squares fit of the increasing signals (Figure 2a). Figure 2b represents the boundary curves of vesicle solubilization, above which the vesicles are completely dissolved by a complexation with MβCD. The curve fittings of eq 4 yield n = 4.18 ± 0.12 and 3.75 ± 0.63 for pure POPC and POPG vesicles, respectively. This observation agrees with the previous study of n = 4 for POPC, determined by the light scattering of the vesicles.16 A global fitting for all of the data of XPOPG from 0 to 1 gives n = 4.21 ± 0.13. These results suggest that the stoichiometry of complexation of POPG with MβCD is almost the same as POPC. The change in the standard free energy of solubilization of the vesicles ΔG0sol = −RT ln K as a function of XPOPG, assuming n = 4 is plotted in Figure 2c. The energy change

Figure 3. Thermodynamic parameters of mixing of POPC and POPG to produce a mixed bilayer with a POPG molar ratio of XPOPG. (a) Enthalpy change of mixing, ΔH0mix, determined by the solubilization of the lipid vesicles by Triton X-100 micelles (Figure 1). (b) Excess free energy change of mixing, ΔGEmix, determined by the solubilization of the lipid vesicles by MβCD (Figure 2). The solid blue lines show the theoretical electrostatic enthalpy changes, ΔH0e,mix (a) and free energy changes, ΔG0e,mix (b) from the Gouy−Chapman model of charged surfaces. 13273

DOI: 10.1021/acs.langmuir.6b03333 Langmuir 2016, 32, 13270−13275

Article

Langmuir Both the ΔH0mix and ΔGEmix of the binary vesicles of POPC and POPG show slightly negative deviations from zero over the entire range of XPOPG (Figure 3). These observations are in line with the differential scanning calorimetry study of the phosphatidylcholine−phosphatidylglycerol mixture.23 These parameters are compared with the calculated values ΔH0e,mix 0 and ΔGe,mix from a theoretical model for electrostatic interactions of charged surfaces based on the Gouy−Chapman theory (Figure 3, solid blue lines).17,18 It is known that the Gouy−Chapman model well describes a charged lipid membrane in an ionic solution.24 However, experimental evidence is lacking as to whether the theory is applicable to the coupling of charged phospholipids in mixed bilayer vesicles. We observed an almost perfect correlation between the experimental and theoretical parameters. In addition, the model also satisfactorily describes the changes in ΔH0sol and ΔG0sol (Figures 1b and 2c, solid blue lines). Note that there is a slight difference between the experimental solubilization free energy changes, ΔG0sol, and the theoretical ones, ΔG0e,sol, at a large XPOPG (Figure 2c). This difference is probably due to the simplified assumptions for the calculations. For example, we ignored a direct binding of counter cations to the negatively charged POPG surfaces. Assuming a binding constant of 1 M−1 for POPG−Na+ binding,25 approximately 10% of the negative charges of POPG are neutralized under our experimental conditions. This causes a slight shift of +0.1 kcal/mol in ΔG0e,sol. We also assumed a constant cross section area of 0.6 nm2 for the lipid headgroup. A variation of ±0.1 nm2 for the area causes a shift of ±0.2 kcal/mol in ΔG0e,sol. However, we confirmed that these variations have only a slight effect on ΔH0e,mix and ΔG0e,mix. The above results indicate that the electrostatic contribution of the entropy changes mainly drives the mixing of POPC and POPG, whereas the enthalpic contribution of ≤−0.1 kcal/mol is still favorable. This large entropy change originates from the changes in the distribution of counterions around the charged surfaces. A decrease in the surface potential and the surface charge density of the negatively charged POPG bilayer by mixing with the neutral POPC releases the ions constrained near the surface, resulting in an increase in the entropy. It should be noted that the above electrostatic contributions are comparable or smaller than the contribution of the ideal mixing entropy, −TΔSideal = RT{X ln X + (1 − X) ln(1 − X)}, corresponding to −0.4 kcal/mol at X = 0.5 and T = 310 K, which is not included in the experimental ΔGEmix (Figure 3b), as described in the Theory section. A higher bilayer stabilization was observed in a 1:1 mixture of cationic ethyl-POPC and anionic POPG than in the pure bilayers, as demonstrated by the large negative ΔH0mix of −1.12 ± 0.08 and ΔGEmix of −1.76 ± 0.25 kcal/mol at a low salt concentration, indicating strong electrostatic interactions (Figure 4). The thermodynamic parameter changes determined by the vesicle solubilization method may include some errors. For example, we assumed the complete separation of phospholipids in the Triton-X100 micelles, although attractive lipid−lipid interactions would inhibit the dissociation. In general, the lipid−lipid interactions in bilayers have been reported to be weak up to ∼1 kcal/mol.9 Therefore, we concluded that the contribution of the residual lipid−lipid interactions in the micelles is negligible at a low lipid-to-surfactant ratio of