INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

The layer of polar lipid headgroups separates the apolar hydrocarbon core of the membrane from the surrounding aqueous phase and stabilizes its lamellar structure. Many cellular processes, such as binding and insertion of proteins, lateral diffusion, ligand-receptor recognition, and certain steps in membrane fusion, critically depend on the physical properties of this boundary layer. The intrinsic property of this layer is the presence of ionizable groups, which participate in the ionic equilibrium. Low-affinity cations bind to the negatively charged groups and neutralize a part of the surface charge. Cations with a higher affinity may cause profound effects on the packing of lipids, conformation of headgroups, and the phase state of hydrocarbon chains. Studies in model membrane systems illustrate the ability of mono- and multivalent ions to cause isothermal phase transition in pure lipids, phase separation, and clustering of individual components in mixtures (Boggs, 1987; Hauser, 1991; Graham et al., 1985). In native membranes, which exist in liquid crystalline state, such changes may potentially exert effects on the conformational dynamics of membrane-embedded proteins, and more specifically, on proteins that experience large conformational rearrangements in their transmembrane domains during their functional cycle (Cantor, 1999).

A nonspecific blockage of mechanosensitive channels by Gd³⁺ ions may be a conspicuous example of a lipid-borne effect. Gd³⁺, a small lanthanide, blocks many types of mechano-gated channels in sub-millimolar concentrations, irrespective of their origin, conductance, or selectivity (Yang and Sachs, 1989; Hamill and McBride, 1996). Early attempts to understand this property of Gd³⁺ implied the presence of specific motifs common to these proteins that may be targeted by the ion with high affinity. Another explanation to such generalized effect may be binding of Gd³⁺ to the lipid component of the cell membrane and alteration of the physical properties of the bilayer surrounding the channels. The high-affinity binding of lanthanides has been known to affect physical properties of phospholipid bilayers by causing phase transitions (Hammoudah et al., 1979; Li et al., 1994; Verstraeten et al., 1997), liposome fusion (Bentz et al., 1988), and pore formation in erythrocytes (Cheng et al., 1999).

Previous data (Ermakov et al., 1992) indicated that Be²⁺, another small cation, is characterized with a higher affinity to phosphatidylcholine membranes compared to other divalent cations. It has been known for toxicity, which is manifested primarily as abnormal immune reactions causing lesions in lungs in response to inhaled Beryllium dust. This cation was recently shown to induce apoptosis in several macrophage cell lines (Sawyer et al., 2000). It appeared timely to study the interaction of Be²⁺ with the negatively charged lipid phosphatiylserine, which has recently been shown to be an important marker on the surface of apoptotic cells, recognizable by macrophages (Fadok et al., 1998, 2000).

In the present work, we take electrostatic approach to study high-affinity interactions of Be²⁺ and Gd³⁺ ions with membrane surfaces. The traditional electrokinetic method widely used for determination of ion binding to membrane surfaces does not provide complete information on the structure of membrane-water interface because it "senses" only changes in the diffuse part of the double layer, outside of the membrane. The inner, dipole component of the boundary potential is a parameter directly related to the chemical structure of the interface, orientation, and hydration of the polar headgroups (Gawrisch et al., 1992). Measurements of the dipole potential may provide additional information complementary to "local" structure usually studied by the infrared or NMR spectroscopic techniques (Hauser, 1991). Here, we present a detailed description of experiments, procedures, and quantitative analysis that allow us to distinguish the effects of cations of different affinity, Mg²⁺, Be²⁺, and Gd³⁺, on the surface and dipole components of the boundary potential in membranes of different compositions. For this purpose, we combine electrokinetic measurements of the surface potential, with the intramembrane field compensation (IFC) method specially designed for monitoring changes of the total boundary potential, including its dipole component. We show the equivalence of the IFC and electrokinetic methods in detection of low-affinity interactions as illustrated by Mg²⁺ adsorption on phosphatidylserine (PS). We find that Be²⁺ and Gd³⁺ ions characterized with high affinity may exert significant effects of both signs on the dipole potential at the boundary, dependent on the lipid composition of the bilayer and pH. We observe the most profound effect of Gd³⁺ and Be²⁺ specifically on the dipole potential of PS membranes, which in the case of Gd³⁺ correlates with changes of mechanical properties of the lipid bilayer.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Principles of measurements

To study the electric field distribution across the membrane, we use the combination of the traditional electrokinetic technique applied to suspensions of liposomes with the intramembrane field compensation (IFC) method. The mobility of liposomes in the electric field provides information about the electric charge and potential drop in the diffuse part of the double layer termed surface potential, _s. The IFC technique, in contrast, permits measurements of the difference of total boundary potentials between the two sides of the planar bilayer lipid membrane (BLM). The changes of total boundary potential, _b, observed upon a unilateral ion adsorption are compared with changes of the surface potential _s obtained on liposomes of the same composition. The difference _b  s is attributed to the variation of the dipole component of the boundary potential, _d.

Measurement of the surface potential

Surface potentials, _s, were calculated from the electrophoretic mobilities of lipososmes, assuming the distance of the shear plane from the physical boundary,  = 0.2 nm (McLaughlin, 1989). The classical Gouy-Chapman-Stern (GCS) model was combined with the Langmuir isotherm (McLaughlin et al., 1981), which took into account the competition of Be²⁺ or Gd³⁺ with cations of the background electrolyte and, in some instances, with protons. The maximal density of negatively charged binding sites always corresponded to the packing of PS in the bilayer (0.6 nm² per lipid molecule, or charge density S = 0.2 C/m²), and the stoichiometry of ion-phospholipid binding was assumed 1:1. To describe quantitatively high-affinity interactions of the ions with negatively charged membrane surfaces and to account for the possible effect of bulk depletion, we included the condition of mass balance to the set of equations (see Appendix for details).

Boundary potential: the intramembrane field compensation method

The advantage of BLM as experimental system is the accessibility of both sides of membrane-to-surface modifications and changes of bath composition. The voltage drop across each membrane-water interface (total boundary potential, _b) is a sum of the diffuse part of the electric double layer (surface potential, _s) and the internal dipole component, _d. Under short-circuit conditions, when the electric potential in the bulk solutions on both sides of BLM is the same, the voltage drop across the membrane core (intramembrane field) is equal to the difference of boundary potentials on each side. It has been known that the membrane capacitance depends on the applied voltage and has a minimum when the intramembrane field is zero (Babakov et al., 1966; Alvarez and Latorre, 1978; Cherny et al., 1980). We use this property of the "zero-field point" to measure the difference of boundary potentials between the two sides of the membrane by compensating the intramembrane field with an externally applied voltage. The dependence of membrane capacitance on voltage was modeled either as electrostriction of an elastic capacitor (Passechnik and Hianik, 1978; Schoch et al., 1979; Shimane et al., 1984), or as change in membrane area due to increase of the amplitude of thermal undulations with voltage (Leikin, 1985). Irrespective of the physical model, the voltage-dependence of capacitance C may be approximated near its minimum, C₀, by a parabola,

(1)

where U is the voltage across the membrane core (intramembrane field),   C₀/2h₀k is the "compliance" of the elastic capacitor with the modulus k and the distance between the plates h₀. When the voltage U = U₀ + U₁sin t is applied to the elastic capacitor, the current j = dCU/dt has three harmonics (Carius, 1977; Hianik and Passechnik, 1995),

(2)

The equation shows that, at U₀ = 0, the capacitance is minimal and the amplitude of the second harmonic becomes zero. This corresponds to a symmetrical membrane, or to a state in which the intramembrane field is completely compensated. Note that the amplitude of the third harmonics is independent of U₀ and is proportional to , which is effective mechanical compliance of the membrane to electrostriction (Alvarez and Latorre, 1978; Schoch et al., 1979; Cherny et al., 1980; Leikin, 1985; Hianik and Passechnik, 1995). The experimental procedure of the IFC, with an automatic minimization of membrane capacitance, was first introduced by Sokolov and Kuzmin (1980). The method is implemented with a circuit containing a lock-in detector of the second harmonic. The dc output from the detector is continuously recorded and fed with a proper phase back to the membrane. The dc bias automatically minimizes the capacitance and thereby keeps the intramembrane field at zero. In practice, the procedure works well when the real (ohmic) component of the membrane current is small compared to the capacitive component. This condition is satisfied at sufficiently high frequency of the fundamental harmonic f = 1/2.

The IFC method detects the difference of boundary potentials, _b, between the two sides of the membrane. In contrast to the traditional method of _b measurement using permeant hydrophobic ions (Andersen et al., 1976), IFC is less sensitive to changes in membrane fluidity. Initially, in a symmetric membrane _b = 0. A unilateral introduction of ions changes the boundary potential at the cis-side only, producing _b. Because the boundary potential of the opposite (trans-) side remains constant, it can be used as reference, and the change of the boundary potential on the cis-side must be equal to the measured difference of boundary potentials between the two sides, _b.

The automatic IFC procedure described above is highly advantageous for continuous monitoring of adsorption and desorption of charged molecules on one side of a planar membrane. By keeping the intramembrane field close to zero, the experimenter may avoid electric breakdown of the BLM when the difference of boundary potentials between the sides is high. We should note that, in the case of unilateral adsorption of Gd³⁺ on PS membranes, the observed _b can be as high as 350 mV.

Experimental procedures

Multilayer liposomes were prepared by a conventional technique. Lipids in chloroform were dried under vacuum for 30 min in a round-bottom flask on a rotor evaporator. The lipid film was then rehydrated in the background electrolyte of a desired composition for 10 min, and than the flask was shaken by hand until a homogeneous suspension was obtained. The optimal concentration of lipids for electrokinetic measurements in our experiments was 1 mg/ml. Typically, the ions of interest were added to the prepared liposomes before measurements. The concentration of the introduced ion was varied from low to high with a series of sequential additions of concentrated stock solutions. The exact compositions of electrolytes in each experiment are listed in figure legends. The electrophoretic mobility of liposomes was measured on a photon correlation spectrometer Zetasizer-2 (Malvern Instruments, Worcestershire, UK). The output from a single run was the main frequency of the final spectrum, which is linearly related to the electrophoretic mobility µ and -potential (Uzgiris, 1978; Hunter, 1981).

Planar lipid membranes were formed on the aperture of 1 mm² in the septum of a Teflon chamber from lipid solutions in decane (15 mg/ml). All lipids (bovine PS and egg phosphatidylcholine, PC) were purchased from Avanti Polar Lipids Inc. (Alabaster, AL) or from Sigma (St Louis, MO) and used without further purification. All salts were of reagent grade (Aldrich, Milwaukee, WI) and the buffers (see legends) were prepared with double distilled water.

The experimental setup for IFC measurements includes a conventional sine-wave generator, a current-to-voltage converter (model 181, Princeton Applied Research, Princeton, NJ), a first harmonic rejection RC-filter, narrow-band amplifiers for the 2nd and 3d harmonics, a lock-in amplifier (model 126, Princeton Applied Research), digital voltmeters, and a chart recorder. The ac (20 mV, 272 Hz) and dc voltage components were summed with an operational amplifier and applied to the membrane with Ag/AgCl electrodes, either directly or via salt bridges. GdCl₃ or BeSO₄ were manually introduced to the cis-compartment of the chamber with continuous stirring of the buffer. In some experiments, the solution in the cis-compartment was exchanged with a two-channel peristaltic pump (Microperpex 2132, LKB, Sweden) allowing for a precise balance of the buffer inflow and outflow.

Membrane tension was assessed by applying gradients of hydrostatic pressure with simultaneous measurements of the capacitance, which is proportional to the area of the black part of the BLM. The membrane bulged under the pressure gradient was treated as a spherical cap, and the tension, , was calculated with the equation that includes the membrane capacitance C, its variation C, the specific value, C_s (0.5 µF/cm²) and the hydrostatic pressure across it, all related by the law of Laplace:

(3)

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Low-affinity ion-membrane interactions: a comparison of IFC with electrophoresis

After the planar membrane is formed in a symmetrical electrolyte and its capacitance stabilizes, the difference of boundary potentials between the two sides, _b, measured by IFC is usually close to zero (0 ± 3 mV). Introduction of inorganic multivalent cations to the cis compartment generates _b (further denoted as _b), with the sign corresponding to a positive shift of the surface potential on the cis side of BLM. An extensive perfusion of the cis compartment with the background electrolyte returns _b to zero. These experiments convinced us that interaction of ions with the membrane surface is completely reversible for all ionic species studied here. The boundary potential of the trans side, therefore, remains constant throughout the experiment (i.e., for 1-1.5 hr), suggesting that cations do not permeate through the membrane, and all changes of the potential measured by IFC are restricted to the cis side.

Interaction of Be²⁺ with uncharged lecithin membranes results in significant positive shifts of the - and boundary potentials, as shown by both electrokinetic and IFC measurements (Fig. 1). At low concentrations, -potentials essentially coincide with _b. Unlike _b, -potential exhibits a maximum in the range of Be²⁺ concentrations between 10 and 100 mM. Taking into account the position of the shear plane ( = 0.2 nm) in Eq. A2 (see Appendix), we calculated the surface potentials from -potentials. After such transform, we observe no difference between the data obtained with the two methods. The theoretical curves for  and surface potentials (solid lines), calculated according to the GCS model (Eqs. A2-A5 and A7) using the same parameters, correspond well to both sets of experimental points (Ermakov et al. 1994).

View larger version (20K): [in this window] [in a new window]   FIGURE 1   Electrostatic effects of Be2+ binding to phosphatidylcholine membranes. The surface potential (crosses) was calculated from -potentials (filled circles) according to Eq. A2 with the parameter x =  = 0.2 nm. The changes of the total boundary potential were measured with IFC (open symbols). Both types of measurements were done in 0.1 M KCl, 20 mM imidazole, pH 6.4. The theoretical curves for the surface (curves 1 and 2) and -potential (curves 3 and 4) were calculated according to the GCS model using the isotherm (Eq. A7) with parameters S = 0.2 C/m2, K2 = 400 M1 (curves 1 and 3); S = 2 C/m2, K2 = 40 M1 (curve 2); S = 0.02, C/m2, K2 = 4000 M1 (curve 4). The data set shown with open symbols is reproduced from Ermakov et al. (1994), with permission.

We presume that the maximum of - potential curve in Fig. 1 is a result of two processes, the surface potential increase due to the cation adsorption, followed by the decline due to self-screening by the electrolyte. A similar nonmonotonous character of -potential dependencies on divalent ion concentration has been reported previously (Tatulian, 1993). The GCS model predicts that, at a high ionic strength, the effect of screening is much more pronounced at a distance from the charged surface compared to that at the surface itself. Thus, the -potential shows the maximum at lower concentrations than the surface potential, _s. By fitting the  and _s curves to the model, we obtain  = 0.18 nm, which independently confirms previous estimates of the shear plane distance (for review see McLaughlin, 1989). A good correspondence of the data between the electrokinetic and IFC techniques strongly suggests that interaction of Be²⁺ with the surface of PC membranes induces changes in the diffuse part of the electric double layer only. On the assumption of the 1:1 binding stoichiometry for Be²⁺-PC, the data presented in Fig. 1 predict the binding constant of 400 M¹. The electrostatic effects reported for other divalent cations (Ca²⁺, Mg²⁺, Sr²⁺, Ba²⁺) do not exceed 15 mV (Tatulian, 1993, 1999). The binding constants were calculated in the referenced paper on the assumption of a different stoichiometry. When the 1:1 stoichiometry is applied to these data, the binding constants for these cations are found two orders of magnitude less than that for Be²⁺ (McLaughlin et al., 1978).

In this study, we use Mg²⁺ as a low-affinity ligand to compare its effects with those of high-affinity ions. The data illustrating the Mg²⁺ adsorption on negatively charged membranes made of PS obtained with the two methods are shown in Fig. 2. The curve for surface potentials is similar to those previously published (McLaughlin et al., 1981;Ermakov et al., 1992). The binding constant found from the zero charge point is 4 M¹, indicating very low affinity. The second set of data obtained with IFC shows the variation of the boundary potential, _b. To compare how the two potentials change with concentration, we represent _b in the same scale as _s. For this purpose, we offset the IFC data by the value of initial surface potential determined from the -potential of PS liposomes measured in the pure background electrolyte. Here, we assume that the surface component, _s, is the same for planar bilayers and liposomes of the same composition. As previously, _s is obtained from -potentials using Eq. A2 with the parameter  = 0.2 nm. Under experimental conditions given in Fig. 2, the initial value for _s is 85 mV. Both data sets coincide and demonstrate good correspondence with the GCS model, indicating that the changes of the boundary potential take place only in the diffuse part of the double layer, leaving the dipole potential unchanged. The theoretical curve (Fig. 2, solid line) calculated according to the adsorption isotherm (Eq. A6) has the predicted maximal slope of 28 mV per decade.

View larger version (12K): [in this window] [in a new window]   FIGURE 2   Binding of Mg2+ to PS membranes. The data for the surface (filled symbols) and boundary potentials were obtained in 0.1 M KCl, 20 mM imidazole, pH 6.4. The entire data measured with IFC (open symbols) were offset for 85 mV, which corresponds to the surface potential of liposomes without Mg2+. All surface potentials here and below were calculated from electrokinetic data with the account for the shear plane distance  = 0.2 nm. The theoretical curve (solid line) and its asymptote (dashed line) were calculated according to Eq. A6 with parameters S = 0.2 C/m2, K1 = 0.2 M1 for K+ and K2 = 4 M1 for Mg2+.

The dipole component in the presence of ions with high affinity

The data of the two methods illustrating the adsorption of Be²⁺ on PS membranes are represented in the same surface potential scale in Fig. 3. The curves have different shapes compared to that for Mg²⁺, and show a noticeable discrepancy between the two methods. The maximal magnitude of the boundary potential change is about 35 mV higher. Both curves (_b and _s) display the maximal slope, significantly exceeding the expected slope for divalent cations, apparently due to bulk depletion of Be²⁺. To describe this observation quantitatively, we supplemented the GCS model (Eqs. A2, A4-A6) with the condition of the material balance (Eqs. A8-A10). A theoretical curve (Fig. 3, dashed line) was generated with parameters c_lip = 0 (no depletion) and K₂ = 10³ M¹. Another curve (solid line), which illustrates depletion at low Be²⁺ concentrations, corresponds to c_lip = 1 mM, the amount of lipid used in the electrokinetic experiment. The latter is in good agreement with the initial part of the experimental curve. Note that the amount of lipid introduced into the bilayer chamber during membrane formation is typically less than that used in electrophoretic measurements, but is more difficult to control. At lower c_lip, the steepest region of the corresponding theoretical curve is expected to be shifted leftward. The shape of the _b curve suggests that the depletion also takes place in the IFC experiment. At higher Be²⁺ concentrations, the experimental curves for _s and _b have plateau-like regions, not predicted by the theory. The surface potential obtained from electrokinetic data levels at ~0 mV, whereas the boundary potential plateaus at +30-35 mV. This difference can be attributed to the increase in the dipole component of the boundary potential in PS membranes in the presence of Be²⁺.

View larger version (12K): [in this window] [in a new window]   FIGURE 3   Binding of Be2+ to PS membranes. Open symbols show IFC data for boundary potentials offset for 85 mV, which corresponds to the initial s in the background electrolyte; closed symbols represent surface potentials calculated from electrokinetic data. Experimental conditions were as in Fig. 2. Theoretical curves were calculated according to Eq. A6 with binding constants K1 = 0.6 M1 for K+, and K2 = 1000 M1 for Be2+ and with clip = 0 (dashed line) and clip = 1 mM (solid line).

Experiments with Gd³⁺ revealed much more pronounced effects on the boundary dipole component. Figure 4 A shows the IFC and electrokinetic data obtained with Gd³⁺ in membranes made of PS. As in the previous case with Be²⁺, the data of the two methods represented in the same scale are significantly different. All curves display a region of steep rise of the potential; in IFC experiments, it takes place at lower Gd³⁺ concentrations. The most drastic difference is observed in the magnitude of the effects; the maximal increase of _s is for 160 mV, whereas _b changes for up to 350 mV. Such large effects of Gd³⁺ on _b are observed at neutral pH on pure PS only. Mixing PS with PC (to 60-80 mol% PS in the mixture) changes _s slightly, but strongly diminishes the maximal _b. Note that, in membranes made of pure PC, the changes of _s and _b coincide in the entire range of concentrations (see Fig. 4 A, inset), similar to what was observed with Be²⁺ (Fig. 1). This suggests that the changes of the boundary potential in the latter case are restricted to the diffuse part of the double layer.

View larger version (24K): [in this window] [in a new window]   FIGURE 4   Binding of Gd3+ to membranes made of PS, PC and their mixtures. (A) Boundary and -potentials were measured with membranes made from 100% PS (1 and 3) and from the mixture PS:PC = 3:2 (2) (10 mM KCl, pH 7.1). The data are presented against the total concentration of GdCl3 in the chamber. The IFC data (open symbols) are offset by 114 mV for pure PS (squares) and for 101 mV for the mixture (triangles). The surface potentials obtained from electrokinetic data are shown by closed symbols. Inset: Gd3+ binding to membranes made of pure PC. The symbolic notation of is the same as previously. All theoretical curves are calculated in the framework of GCS model (Eqs. A2, A4, A5) taking into account the competitive binding of Gd3+ and K+ to PS (Eq. 6, K1 = 4, K2 = 5 · 104 M1, respectively) and Gd3+ to PC (Eq. A7, K2 = 103 M1). The parameter clip in the condition of mass balance (Eq. A8) is 0 (dashed line), 0.02, and 0.3 mM for curves 2 and 3, respectively. (B) The surface charge density, , and occupancy of binding sites, , calculated for pure PS (50 mM KCl, pH 7.0, K1 = 1 M1, K2 = 5 · 104 M1, clip = 0). The dotted line indicates the level of 33% occupancy corresponding to the zero charge point.

Theoretical curves in Fig. 4 A are generated using binding constants of Gd³⁺ to PS and PC of 5 · 10⁴ and 10³, respectively. The dashed curve in Fig. 4 A is built in the framework of the traditional GCS model (c_lip = 0); the two solid curves are computed with the same binding parameters, but account for Gd³⁺ depletion with parameters c_lip of 0.02 mM (curve 2) and 0.3 mM (curve 3), respectively. These curves well approximate the electrokinetic data (curve 3) and IFC data obtained in membranes made with 60% PS in mixture with PC (curve 2). In the latter case, the amplitude of the IFC signal does not exceed _s, and the right part of the _b curve coincides with the curve predicted according to the GCS theory, suggesting that adsorption of Gd³⁺ in this particular case affects only the diffuse component of the boundary potential. Large _b obtained in pure PS membranes that drastically exceed _s, cannot be explained in the framework of the GCS model and indicates a significant increase of the dipole component. To study the onset of this positive dipole effect and to avoid the effect of bulk depletion, we performed a series of experiments, in which the concentration of Gd³⁺ near the membrane was "clamped" at concentrations between 10⁸ to 10² M by perfusing large volumes of Gd³⁺-containing buffers (at least 10 chamber volumes). The results revealed a sudden change of the boundary potential at ~10⁶ M Gd³⁺, just below the zero charge point (data not shown).

Using the same set of parameters in the model, we calculated the variations of the surface charge density, , and the occupancy of binding sites by Gd³⁺, , on PS membranes as a function of bulk Gd³⁺ concentration (Fig. 4 B). At low concentrations (10¹¹-10⁷ M), a relatively steep increase of  reflects the substitution of K⁺ ions of background electrolyte at the binding sites by Gd³⁺. In the middle part (10⁶-10² M), the occupancy varies insignificantly, staying close to the zero charge point ( = 33%). Only at submolar concentrations, the range that is difficult for experimentation,  grows steeply and then saturates. The surface potential (Fig. 4 A, dashed line) changes almost linearly with the log of [Gd³⁺]. This asymptotic behavior with the slope of 20 mV/decade is dictated by the Boltzmann relationship (Eq. A5), which predicts the concentration of cations near the charged surface, c(0), practically independent on their bulk concentration and the slowly changing charge density at the surface. This phenomenon reflects merely the screening of surface charge by trivalent cations. The curves clearly illustrate the tendency of the system to stay close to the point of electroneutrality at intermediate concentrations of the adsorbed ion.

At higher concentrations of Gd³⁺, outside of the range of depletion, the experimental curves obtained with both methods have similar shape (Fig. 4 A). We used these data subsets to quantify the effects of Gd³⁺ on the dipole component. Two series of experiments were designed to study the Gd³⁺-induced dipole potential as a function of density of ionized PS in the membrane. One series resulted in pairs of _s and _b data taken for different PS/PC mixtures under equal experimental conditions. In another series, the carboxyl groups of PS were titrated with an acid, thus the measurements was taken on pure PS membranes at pH ranging from 2.8 to 7.0. In the latter case, the competitive binding of H⁺, K⁺, and Gd³⁺ was taken into account in calculations of surface potentials. The surface potentials determined from electrokinetic data and the parameters used in calculations are given in Table 1. As previously, we subtracted _s from _b at every Gd³⁺ concentration to determine the change of the dipole component. For this purpose, we used the surface potential, _s, after its correction for the proper value of c_lip, adjusted as a free parameter, such that the regions of the steepest slope for both _b and _s curves coincide. (Technically, we could use a theoretical approximation of _s [Fig. 4 A, dashed line], the position of which, in the _s  log C graph, is determined only by the binding constant for Gd³⁺ and density of binding sites.)

The changes of the dipole component _d = _b  s, obtained with membranes made from different PS/PC mixtures, are represented in Fig. 5 A. The right part of the curve for pure PS exhibits a shallow linear slope as the Gd³⁺ concentration increases. All other data show no systematic dependence on Gd³⁺ concentration. The points obtained with pure PC are scattered strictly around zero mV. A similar _d graph was built for PS membranes studied at different pH (Fig. 5 B). The mean values of _d found from the right sides of the _d  log[Gd³⁺] graphs are presented in Fig. 6 as a function of density of negatively charged binding sites relative to its maximal value in pure PS at pH 7.0 (S = 0.2 C/m²). Independent of the method of varying the density of the ionized form of PS, we observe the same changes of the dipole component of the boundary potential induced by Gd³⁺. At low densities (<60%), we see a negative change of the dipole potential of about 30 mV. At the highest density achieved with pure PS and neutral pH, the maximal magnitude of effects is +140 mV. The sign "+" means that the change of the dipole component augments the change of the surface component upon adsorption of positively charged ions and increases the absolute value of the outwardly directed dipole. Note that the two series of experiments were performed at different ionic strength of the background electrolyte (10 or 50 mM KCl), yet the magnitude of the dipole effect was the same. This indicates that the changes of electric potential take place in the unscreenable part of the double layer, i.e., inside the membrane. A large positive dipole effect of Gd³⁺ clearly correlates with a high contents of negatively charged PS in the bilayer.

View larger version (22K): [in this window] [in a new window]   FIGURE 5   Changes of the dipole potential induced by Gd3+ adsorption on membranes made from (A) different PS/PC mixtures (10 mM KCl, pH 7.1) and on membranes made of (B) pure PS at different pH (50 mM KCl). The boundary potential changes were measured by IFC method and offset by the value of surface potential calculated according to the GCS model with parameters listed in Table 1. For calculation of mean values of the dipole potential (dotted lines), we use data points to the right of the vertical dividing line. The position of this line corresponds approximately to the region of the steepest slope for most of the experimental b curves.

View larger version (13K): [in this window] [in a new window]   FIGURE 6   The mean values of dipole potentials induced by Gd3+ adsorption on membranes with different density of negatively charged binding sites, S, normalized to the maximal value Smax = 0.2 C/m2 corresponding to pure PS at pH 7.0. The squares represent the data for different PS/PC mixtures (Fig. 5 A), the triangles are the data for pure PS measured at different pH (Fig. 5 B).

One possible reason for the dipole potential change may be binding of cations at the plane beneath the layer of negatively charged headgroups. The sign of the effect in this case should be the same as observed in experiments, but the magnitude is expected to be proportional to the density of adsorbed Gd³⁺ ions, analogous to the density-dependent dipole effect of 1-anilino-8-naphtalenesulfonate (Ermakov et al., 1983). Figure 7 shows combined data for _d obtained for pure PS at two different ionic strengths and presented as a function of the occupancy, , calculated from the model that accounts for bulk depletion. The experimental points consistently deviate from the hypothetical linear dependence (dashed line); they show that _d undergoes stepwise increase near the zero charge point (  30%), remaining constant at higher occupancies.

View larger version (14K): [in this window] [in a new window]   FIGURE 7   The dipole potential versus the occupancy of binding sites by Gd3+ cations calculated for experiments with membranes made of pure PS. The data sets are the same as in Fig. 5. Background electrolyte was 50 mM KCl (filled circles) and 10 mM KCl in all other experiments, pH 7.0.

Mechanical properties of the bilayer

Our experimental setup permitted simultaneous recording of the first three harmonics of the capacitive current, which report on the capacitance itself, electrical asymmetry of the membrane, and its compliance to electrostriction, respectively (see Materials and Methods, Eq. 2). Adsorption of Gd³⁺ had no visible effect on the BLM capacitance (represented by the 1st harmonic), however we consistently observed a decrease of the amplitude of the third harmonic. The parameter  proportional to the ratio of the first and third harmonics, which reflects the compliance of the membrane to electrostriction, also declined. The major change in  was observed at Gd³⁺ concentrations of ~10⁷ M (i.e., about two orders of magnitude lower than the zero charge point), where the electrostatic effects are small. In the range of 10⁶-10⁵ M Gd³⁺, the membrane seems to be very rigid, such that the amplitude of the third harmonic becomes comparable to the noise level (data not shown). For this reason, we can correlate the changes of  neither with the density of absorbed ions, nor with the onset of the dipole effect.

In the presence of Gd³⁺, we observed a strong increase of membrane tension, , for BLMs made with PS, but not with PC (Fig. 8). Tensions were determined from relative expansions of the BLM measured as an increase of capacitance, C/C under different hydrostatic gradients (see Eq. 3). Note that solvent-containing bilayers can expand because of incorporation of an extra lipid material from the surrounding meniscus. The calculated tension remains practically constant as the relative membrane expansion increases (Fig. 8 A). The mean values for tensions obtained at different Gd³⁺ concentrations are plotted in Fig. 8 B. The PC membranes are characterized with the relatively low tension (~0.2 mN/m), essentially independent on the Gd³⁺ concentration. The same independence of Gd³⁺ exhibit membranes made of PS mixed with PC (60% PS). For membranes formed of pure PS, the tension rises at ~10⁵ M Gd³⁺, i.e., in the range of concentrations where the dipole potential steeply increases, which approximately corresponds to the zero charge point. The maximal increase of BLM tension is about 6 times over the control.

View larger version (14K): [in this window] [in a new window]   FIGURE 8   Surface tension of planar lipid membranes made from PC (open symbols) and PS (closed symbols) in the presence of Gd3+. The tension was determined by capacitance measurements under different hydrostatic pressure gradients between cis and trans compartments. (A) The tension of BLM plotted as a function of relative membrane expansion (measured as changes of BLM capacitance, C/Co). The concentrations of Gd3+ are 0, 5, 7.5 µM (membranes from PS, curves 1-3) and 0, 5, 10, 20, 120 µM (membranes from PC, curves 4-8), respectively. (B) The relative changes of BLM tension at different total Gd3+ concentrations on the cis-side of the chamber. The BLM tension in the background electrolyte (10 mM KCl, pH 7.1) was taken as unity. Triangles represent the data for 60% of PS in the PS/PC mixture.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

We demonstrate that Be²⁺ and Gd³⁺ ions exert dipole effects at the interfaces of PS-containing membrane. We come to this conclusion by comparing the changes of the total boundary potential with changes of its surface component, measured with the IFC method and electrophoretic technique, respectively. Although the physical principles of the measurements are completely different, the data obtained on Be²⁺ or Gd³⁺ with PC, and Mg²⁺ with PS, membranes (Figs. 1, 2, and 4 A, inset) show a remarkable agreement of the two methods. These examples of low-affinity interactions illustrate the equal ability of each technique to detect the changes of surface potential. The data imply no electrostatically detectable rearrangement of the interface in these instances. The discrepancy between the methods may occur at high ionic strength (Fig. 1) due to the inequality of the shear plane to the physical boundary. This deviation of the -potential from the boundary potential allowed for independent verification of the shear plane distance (parameter ), which is commonly used for quantitative analysis of electrophoretic data obtained in phospholipid liposomes (Eisenberg et al., 1979; McLaughlin, 1989).

The electrostatic effects were treated in the framework of the GCS model, with the assumptions of 1:1 binding stoichiometry and competition between the cations. All data were fit satisfactorily with the same set of principle parameters of the model, which includes binding constants for all cations and the surface density of binding sites. (The parameters K and S are not completely independent in the fitting procedure [see Fig. 1, curves 2-4]. Indeed, at low occupancies, the surface charge density, , in Eq. A7 is proportional to the product K × S. The problem of adequate fitting of data for monovalent cations is discussed in Ermakov (1990). For calculations, we choose S = 0.2 C/m², which corresponds to the maximal stoichiometry of binding, one ion per one lipid [McLaughlin et al., 1981; Westman et al., 1984]). Our data do not pose any reason to suspect that we deal with gadolinium complexes of variable valence or that the affinity of binding sites changes with their density or occupancy. We have explained quantitatively the high slope of _s isotherms observed in the micromolar range of concentrations as the effect of bulk depletion of the cations possessing exceptionally high-affinity to the negatively charged surface of PS. A similar effect of depletion was mentioned for electrokinetic measurements of PS liposomes in the presence of La³⁺ (Bentz et al., 1988). The quantitative description of depletion required only one adjustable parameter c_lip, which was not essential for the evaluation of the magnitude of dipole effects.

It is commonly accepted that, due to the intrinsic dipole potential, the core of phospholipid membranes is 200-300 mV more positive relative to the boundary layer of electrolyte (Andersen et al., 1976; Flewelling and Hubbell, 1986;Tocanne and Teissie, 1990). We could not find any reference in the literature on the magnitude of the dipole potential for membranes made specifically of PS. Our data indicate that Gd³⁺ bound to PS may either decrease or increase the intrinsic dipole potential. We consistently see a decrease of the dipole potential either at low PS content in the membrane-forming mixture, at low pH (Figs. 5 and 6), or at low occupancy of binding sites (Figs. 6 and 7). At high PS content and high occupancies by the ion, we observe an appreciable increase of the intrinsic _d in the presence of Be⁺² (+30 mV), and a very strong positive dipole effect of Gd⁺³ (+140 mV), indicative of restructuring of the surface. The strong dipole effect clearly correlates with the increase in membrane tension (Fig. 8) and transversal rigidity, suggesting the change in lipid packing confirmed in monolayer experiments (Yu. A. Ermakov, V. L. Shapovalov, and S. Sukharev, in preparation). Evidently, the altered bilayer structure may change other properties, such as viscosity and diffusion coefficients for lipids and other substances inside the membrane, which argues against the use of lipophilic ions for dipole potential measurements in these instances. Rigidification of the bilayer may lead to significant errors in _d estimation.

Our observations of negative dipole effects are consistent with the potency of many cations to dehydrate the membrane surface. Water molecules oriented by the phosphate and carbonyl groups of phospholipids augment the outwardly directed dipole contribution of carbonyls (Gawrisch et al., 1992). A removal of water, therefore, is predicted to diminish the intrinsic dipole potential. For instance, Li⁺ causes substantial dehydration of phosphate and carbonyl groups of PS headgroups, which, in pure synthetic lipids, is typically associated with isothermal liquid-gel phase transitions (Hauser, 1991). Peculiarly, Mg²⁺ and Ca²⁺, which also increase the melting temperature, cause less dehydration, reducing the amount of water near the phosphate groups of PS only (Hubner et al., 1994). Our measurements on bovine PS showed no dipole effect of Mg²⁺, in contrast to the high-affinity cations of Be²⁺ and Gd³⁺. Previously, using the same approach, we found that, during the phase transition in DPPC membranes in the presence of Be²⁺, _b and _s have opposite signs, indicating the decrease of the intrinsic dipole potential consistent with the predicted effect of dehydration. Noticeable isotope effects on the -potentials and phase transition temperature shifts induced by Be²⁺ observed on DPPC liposomes in D₂O support the same conclusion (Ermakov et al., 1994).

The difference in the behavior of PS and PC in the presence of ions can be accounted for by the chemistry of their headgroups. The dipole potential at the surface of PC membrane is attributed to the presence of two carbonyl groups, the P-N⁺ phosphocholine group and oriented water typically hydrating the sn-2 carbonyl and the phosphate group (Gawrisch et al., 1992). In PS, there is additional an negatively charged carboxyl group with its own dipole moment. We demonstrated the critical role of the ionized form of PS in the generation of the positive dipole effect, pointing to the primary role of these carboxyls. However, at the present stage, we cannot conclude whether the dipole moments of carboxyls only, or of other groups as well, contribute to the gross dipole potential changes resulted from the high-affinity ion adsorption. The high-magnitude positive dipole effect appears to result from concerted interactions of more than one adjacent PS headgroup. The onset of the effect occurs at Gd³⁺ concentration near the zero charge point, i.e., when three PS headgroups coordinate one Gd³⁺ ion.

Pettersheim and Sun (1989) have shown that lanthanide ions are coordinated within a layer of PS molecules by both phosphates and carboxyls. This makes possible a cross-linking of the PS sheet, because one lanthanide ion may coordinate three PS molecules, whereas each PS may potentially interact with two different ions. The authors observed a stronger conformational effect in the inner, more compressed leaflet of liposomes and proposed that the lanthanide-induced "conformational change is not a direct consequence of forming the cation-PS complex, but appears to be a more delocalized effect of the cation headgroup packing, i.e., electrostatic repulsion of the serine ammonium group." We infer that this delocalized effect may be due to a strong condensation of the lipid layer, a decrease of the area per headgroup and a resulting electrostatic or steric conflict between the headgroups, which forces the conformation to change. Importantly, in our experiments, the subsequent dipole rearrangement, not the binding parameters for Gd³⁺, critically depends on the density of ionized PS groups. Consistent with the inference that the dipole effect is a result of lipid condensation, a compression of dimirystoyil PS monolayers to ~30 mN/m (with no Gd³⁺ in the subphase) is found to increase the dipole potential an additional 100 mV, comparable to the dipole effect of Gd³⁺ (Yu. A. Ermakov, V. L. Shapovalov, and S. Sukharev, in preparation). The observed electrostatic effects do not provide information on the exact conformation assumed by the PS headgroup in the presence of Gd³⁺, but the sign and magnitude of the total dipole potential change (+170 mV, accounting for 30 mV due to dehydration) suggest that positively charged amino groups move to a certain depth toward the membrane interior, and stay behind the layer of phosphates, possibly forming hydrogen bonds with dehydrated carbonyls. The negatively charged carboxyls in this conformation are likely to remain on the periphery.

The observed interaction of Be²⁺ (K = 10³ M¹) and Gd³⁺ (K = 5 · 10⁴ M¹) with PS correlates with the broad spectrum of membranotropic effects of these ions in model systems, and in vivo. Be²⁺ is the smallest of all divalent ions with the charge density (z/r = 5.66) similar to that of some trivalents. It induces membrane rigidification and phase separation with a potency comparable to that of Al³⁺, Ga³⁺, In³⁺, and Sc³⁺ (Verstraeten et al., 1997). Long known for its high toxicity, the inhaled beryllium (or BeO) dust causes immune-mediated lesions in lungs (chronic beryllium disease), associated with lymphocyte infiltration and aggregation of macrophages (Finch et al., 1998). It has also been established recently that macrophages specifically recognize PS in the outer leaflet of the cells undergoing apoptosis, and the lipid in these cases is the signal that triggers phagocytosis (Fadok et al., 1998, 2000). It seems logical to propose that Be²⁺ somehow promotes the accumulation of PS on the outer surface of a normal epithelium, which is then attacked by phahgocytes. The questions of whether the putative redistribution of PS takes place in the presence of Be²⁺ and whether it is mediated by physical lipid clusterization by the ion, by promoting nonspecific flip-flop, activation of scramblase, or by blockage of specific aminophospholipid translocase (Bratton et al., 1997), are pertinent in this context.

Gd³⁺, a "small" lanthanide, actively displaces Ca²⁺ from the surface of neutral PC membranes and causes fatty acid chain ordering and phase separation (Li et al., 1994). Its effects range from inhibition of gravitropism in plant roots (Millet and Pickard, 1988), blockage of various mechano-gated channels (Yang and Sachs, 1989; Hamill and McBride, 1996; Oliet and Bourque, 1996), to inhibition of hemagglutinin-mediated cell fusion (L. Chernomordik, personal communication) and lipid clusterization and pore formation in erythrocytes (Cheng et al., 1999).

This work and the preceding studies (Bentz et al., 1988) show that PS polar moeities may act as ubiquitous nonspecific receptors for lanthanide ions in native membranes. The lipid ordering, rigidification, and phase separation induced by the ions may directly effect membrane-embedded proteins. The increase of membrane tension and decrease of transversal compliance reported here suggest that the change of mechanical properties of the bilayer does take place in the presence of Gd³⁺. Mechanosensitive channels are the most likely proteins that are sensitive to such perturbations. In the event of phase separation, mechanosensitive channels embedded in "frozen" lipid domains may be mechanically isolated from the tension-transducing fluid part of the membrane and, thus, may turn insensitive to stretch. Under normal conditions, when such channels are activated by tension, their proteins expand in the plane of the membrane (Sukharev et al., 1999). The condensation of surrounding lipids in the presence of Gd³⁺ may exert positive pressure on the proteins, which would oppose the conformational transition favored by tension. Another attractive hypothesis is that Gd³⁺-induced lipid condensation may change the pressure profile across the lipid bilayer (Cantor, 1999), thus biasing the conformation of the transmembrane domains in certain proteins toward the closed state. The questions on the magnitude of pressure that can be generated by ion binding, the conformation of polar headgroups, the state of hydrocarbon chains under such conditions, and the effects on lipid-embedded proteins outline the scope of problems for future research.