Investigation of metal ion interaction with a lipid cubic phase using electrochemical impedance spectroscopy

Accepted Manuscript Investigation of metal ion interaction with a lipid cubic phase using electrochemical impedance spectroscopy Mohammad Yaser Khani Meynaq, Kenichi Shimizu, Mahdi Shahmohammadi Aghbolagh, Solomon Tesfalidet, Britta Lindholm-Sethson PII: DOI: Reference:

S0021-9797(16)30512-4 http://dx.doi.org/10.1016/j.jcis.2016.07.053 YJCIS 21439

To appear in:

Journal of Colloid and Interface Science

Received Date: Revised Date:

12 May 2016 20 July 2016

Please cite this article as: M.Y.K. Meynaq, K. Shimizu, M.S. Aghbolagh, S. Tesfalidet, B. Lindholm-Sethson, Investigation of metal ion interaction with a lipid cubic phase using electrochemical impedance spectroscopy, Journal of Colloid and Interface Science (2016), doi: http://dx.doi.org/10.1016/j.jcis.2016.07.053

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Investigation of metal ion interaction with a lipid cubic phase using electrochemical impedance spectroscopy Mohammad Yaser Khani Meynaqa, Kenichi Shimizua,b, Mahdi Shahmohammadi Aghbolagha, Solomon Tesfalideta , Britta Lindholm-Sethsona,* a

Department of Chemistry, Umeå University, SE 90187 Umeå Sweden Physical and Theoretical Chemistry Laboratory, Department of Chemistry, Oxford University, South Parks Road, Oxford, OX1 3QZ, United Kingdom b

8

Abbreviations: MO1, LCP2, NMR3, SAX4, DMPC5, EIS6, CPE7, FT-IR .

Abstract Hypothesis: Electrochemical impedance spectroscopy, EIS, can be used as a complementary technique to investigate ion interaction with the headgroup region in the aqueous channels of a lipid cubic phase, LCP.

Experiments: A freestanding membrane made of monoolein LCP was formed by filling a small aperture that separates two cell compartments. The cell compartments were filled with electrolyte solutions at two different ionic strengths: i.e: 10 and 100 mM, of KCl, CsBr and CaCl2. Electrochemical impedance spectroscopy was recorded between two platinum electrodes that were present at each side of the membrane.

1 Monoolein 2 Lipid Cubic Phace 3 Nuclear Magnetic Resonance 4 Small Angle X-ray 5 Dimyristoylphosphatidylcholine 6 Electrochemical impedance spectroscopy 7 Constant-phase element 8 Fourier Transform Infrared Spectroscopy

1

Findings:

The membrane resistance and capacitance were estimated from

equivalent circuit fitting of the impedance data. It was confirmed that calcium ions interacts strongly with the headgroup region in the aqueous channels giving significantly higher membrane resistances compared to monovalent alkali metal ions. The membrane capacitance with Ca 2+(aq) in solution was concentration dependent, which for the first time indicates formation of two different cubic phases at these conditions. lipid cubic phase; electrochemical impedance spectroscopy, monoolein, lipid–ion interactions, calcium, cesium, potassium, ionic permeability, membrane resistance, membrane capacitance. Keywords:

1. Introduction The interaction of biologically relevant cations and anions with lipid membrane surfaces has been of large research interest for a long period, as for example metal cations regulate membrane related physiological process. Especially important is calcium ion interaction, that for instance affects membrane stability and structure [1] and play an important role in signal transduction in the interaction with neuron membranes [2]. The interaction of cations with lipid headgroup can be studied using monoolein (MO) lipid cubic phases (LCP) with or without addition of phospholipids. In pure MO the two hydroxyl groups in the glycerol give polar characteristics to the headgroup and form hydrogen bonds with the hydration shell of solvated cations. Ions with different sizes and charges have different corresponding hydration numbers in bulk, suggesting a competition between ion binding to water and lipids [3, 4]. Ionic interactions have been investigated using molecular dynamic and metadynamic simulations [5], fluorescence measurements [6], nuclear magnetic 2

resonance (NMR) [7], Fourier Transform Infrared Spectroscopic (FT-IR) [8], and small angle X-ray (SAX) [9]. Some of these findings are summarized below: From molecular dynamic simulations and fluorescence measurements it is suggested that alkali metal ions predominantly interact with carbonyl groups and phosphates in the headgroup region of a phospholipid membrane [5, 10]. Moreover potassium ion adsorption to the membrane was found to be weak especially when chloride is the counterion. Metadynamic simulations were used to estimate the free energy landscape for some metal cations at dimyristoylphosphatidylcholine (DMPC) membrane surfaces [11]. It was suggested that the most favourable state for sodium and potassium ions is the fully hydrated state and therefore both these ions prefer to remain in the aqueous phase. In the case of calcium ions the most stable state is bound to the lipid headgroups coordinating four lipid oxygens and two water molecules. It was early recognized that calcium ions are responsible for adhesion of cells to different substrates and to each other. Manery et al reports on calcium bridging between carboxyl groups in the headgroup region and also that alkaline earth metals easily form stable chelates. The chelation induces a displacement of hydration water from the metal ion, and thereby causing dehydration of the membrane [12]. Infrared spectroscopy showed that divalent ions such as magnesium and calcium penetrate deeply into the polar headgroup region and cause partial dehydration. A complex rearrangement of the carbonyl region is following calcium ion binding involving both hydration and conformational changes. The hydration shell becomes less structured than the pure hydrated lipid [13]. In accordance to this Sinn et al found that calcium binding to the lipid membrane headgroup is 3

entropy driven because of the dehydration. Moreover, the dehydration following calcium ion binding leads to a tighter packing of the headgroups and the hydrocarbon chains [14]. Cook et al observed that each calcium ion is coordinated to the hydroxyl groups of lipid bilayer which can lead to an arrangement of hydroxyl groups suitable for binding calcium ions. Calcium ions are most stable when it is bound to the lipid oxygens in the membrane, and the corresponding bond thereby inducing a condensing effect on the membrane [3]. The condensing effect of calcium might induce pore formation in a planar lipid membrane of fixed area. This is probably not the case in the LCP. The radii of the hydrated potassium were reported by different authors giving values that vary between 0.201 and 0.331 nm [15-18]. The corresponding values for cesium ions were 0.295 and 0.329[19, 20]. Thus, the hydrated potassium ion is somewhat smaller than the hydrated cesium ion at infinite dilution. The aqueous ionic radii of the Br -(aq) and Cl-(aq) anions are both 0.30 nm [21]. The inner diameter of the water channels within the monoolein LCP saturated with pure water is estimated to 5-10 nm [4, 22, 23]. Therefore, no physical hindrance is expected for hydrated anions and cations to move in the aqueous channels in the MO-LCP. The differences in ionic mobility can therefore be explained by interactions with the membrane headgroup region in the LCP channels [24]].

Electrochemical impedance spectroscopy (EIS) was first introduced to investigate dielectric properties of tissues and other biological samples for more than a century ago [25, 26] and references cited therein. Thus one of the first estimations of the molecular thickness of cell membranes, was performed by Fricke [27] with dielectric relaxation studies in cell suspensions. Later, EIS was used at black lipid membranes for investigation of antigen-antibody and

4

enzyme-substrate reactions by de Castillo [28] and some similar early work were presented by de Levie et al and Coster et al. [29-31]. Since black lipid membranes are very fragile, stabilisation of artificial lipid membranes on solid support was suggested. Hence, polymer cushioned electrodes were used by for instance, i.e.: [32, 33] or lipid bilayers on thiolipid support by Vogel et al [34]. Supported monolayers on mercury provide a nice and reproducible model that was employed for metal ion interaction investigations with the lipid headgroup region [35]. In the present work we extend the use of EIS to investigation of metal ion interaction within the ion channels of a free-standing membrane of a LCP. A monoolein/water LCP was covering a small circular aperture that separates two cell compartments containing one platinum electrode each. The compartments were filled with different electrolyte solutions at two ionic strengths, 10 mM and 100mM. Electrochemical impedance spectra were recorded continuously with this two-electrode setup to monitor changes in membrane properties as the electrolyte ions diffuse into the aqueous channels. The dielectric properties of the LCP membrane were estimated from circuit fitting, i.e.: membrane resistance and membrane capacitance. EIS is a new method for detection and tracking ionic interactions with the membrane surfaces in the aqueous channels inside LCP. EIS is inexpensive and simple to use as compared to existing techniques. The hypothesis is that EIS can enable disclosure of novel findings on lipid head group interactions with ions. The condensing effect of calcium ions was observed from an increased membrane resistance, which confirms earlier findings with other techniques. The two concentrations of CaCl2 gave significantly different membrane capacitances, which might suggest the prevalence of two different cubic phases.

5

2. Experimental Materials Monoolein (1-oleoly-rac-glycerol, ≥99%, Sigma), calcium chloride, potassium chloride, and Cesium bromide (99%, Sigma-Aldrich) were used as received. Water was obtained from Milli-Q water purification system; Millipore Corp. 18 MΩ cm. Instrumentation Modulab (Solartron Analytical) was used for Electrochemical Impedance Spectroscopy measurements (EIS); amplitude 10 mV, frequency range 0.1-100 000 Hz. A two-electrode arrangement was used and the platinum electrodes were at open circuit potential with a bias of Zero Volt between them. All electrochemical measurements were carried out at room temperature. Zview version 3.4 was used for quantitative evaluation of the data. Preparation of Cubic Phases Dry solid monoolein was weighed in a vial and melted at 40° C. Then, water was added to the vial at room temperature in the ratio: 36 w% of water and 64 w% of monoolein and mechanically mixed [36] according to the phase diagram in Fig. 2. The vial was centrifuged for 10 min at 4000 rpm, until the viscous and transparent cubic phase was formed. Note that the phase diagram reflects a situation with pure water in the aqueous channels, whereas the present paper investigates the effect of electrolyte ions entering the channels. In the final equilibrium state the channels contain large concentrations of ions and therefore the phase diagram is probably altered somewhat. A transparent LCP at the end of the experiments indicates a stable phase. 6

Electrochemical cell and preparation before measurements A homebuilt two-compartment electrochemical cell made of teflon was used for all experiments, Fig. 3. The two compartments contain one platinum electrode each ( = 0.5 cm, A = 0.07 cm2). A four electrode set up was also investigated as suggested by Coster et al [37] with two Ag/AgCl electrodes inserted in each compartments close to the membrane. However, at high resistance membranes large measurement errors were obtained, especially at the high frequency limit with negative RealZ values and at the low frequency limit the measurements became noisy. With membranes of lower resistances the results were similar to the two – electrode setup except for the low frequency part where the capacitive charging of the platinum electrodes was absent as expected. Therefore all results in this paper are based on the stable two-electrode setup. The time constant for capacitive charging of the platinum electrodes is well separated from the charging of the LCP membrane and therefore it was possible to compensate for the metal/solution impedance in data analysis. A cylindrical teflon brick with dimensions  = 12 mm, thickness 3 mm and with a cylindrical fully penetrating hole in the centre,  = 1 mm, was fixed between the two compartments. Before mounting the cell the aperture was filled with the monoolein cubic phase. The amount of cubic phase was estimated from weighing before and after addition of LCP to the aperture and was around 4.0 ± 0,1 mg. The cylindrical aperture was unfilled for blank measurements. Electrolyte solutions were prepared from KCl, CsCl and CaCl2 at two different ionic strengths, i.e.: 100 mM and 10 mM respectively. Thus, the concentration of CaCl2 was 33.3 and 3.33 mM respectively. Two openings at the top of the

7

two compartments were used for a simple addition of electrolyte solution. Measurements started immediately after the cell was assembled and the compartments filled with electrolyte solution.

3. Results and discussion 3.1 Blank measurements: Three scans were performed for each electrolyte with the connecting cylindrical aperture without LCP. The blank measurements give straight capacitive lines in the complex impedance plane, and reflect the capacitive charging of the two platinum electrodes. The solution resistance between them can be estimated from the intersection with the real axis. A simple equivalent circuit is suggested for these blank measurements Fig 4, where the double layer capacitance of the two platinum electrodes, Cdl, is in series with the electrolyte resistance between the two electrodes R s.

The obtained Cdl is

composed of two double layer capacitances in series. The geometrical capacitance of the measurement cell is modelled as a constant phase element, CPEcell in parallel with Rs and Cdl according to for instance JR Macdonald et al. [38] The values from data fitting are found in Table 1. It is clear that the constant phase elements obtained at an ionic strength of 100 mM are not the expected non-ideal capacitances. The reason for this is that the time constant for charging the cell is outside the time window of the impedance measurements. Therefore, a mean value was calculated from the values obtained from blank measurements with the three electrolyte solutions at I = 10 mM. The obtained mean CPE was 1/1,29 10-11 * (jω)0,96 (F cm−2 s0,96 ) and used as a fixed element in the equivalent circuits, Fig. 4, 6a and 9a, when analyzing data from measurement on the freestanding membrane. Constant-phase elements, CPE, are frequently used as circuit elements in equivalent circuits for fitting impedance data, when pure capacitances or 8

resistances are not appropriate. This gives a possibility to mirror distributed reactivities and inhomogeneities in the surface, for instance. A constant phase element has two parameters: the pre-factor Q and the exponent n and is given by the formula: CPE = 1 /Q * (jω)n [39]. When n = 0, the constant phase element is transformed into a pure resistance = 1/Q, whereas n = 1 indicates a pure capacitance equal to the prefactor Q. With n = 0.5 the constant phase element, mirrors transport properties such as in the Warburg impedance. 3.2 Measurement on freestanding LCP membrane: The cylindrical aperture connecting the two cell compartments was filled with LCP. The permeability of the freestanding LCP membrane with different electrolytes in the cell compartment was investigated with measurements of impedance spectroscopy in a two-electrode setup. Initially the channels in the LCP are filled with pure water so the measurements reflect also the diffusion of ions into the aqueous channels in the LCP membrane and the subsequent change in dielectric properties of the film. The cubic phase remains transparent indicating stability, however some minor changes in the unite cell parameters of the cubic phase cannot be excluded as a result of the metal ion interactions. It is demonstrated in several papers that the channel widths in LCP under certain circumstances, can be varied in a rather wide range, without loosing the cubic phase structure. Thus, Mezzenga et al [40] used SAXS measurements, to show that the channel diameter in a fully hydrated bicontinuous cubic phase based on the monolinolein (MLO)/water system was varied continuously from 4 -12 nm simply by increasing the amount of sugar ester content from 0 to 25 wt% in the initial lipid blend. Furthermore, Angelov et al [41] used time resolved X-ray diffraction to investigate the MO/octylglucoside/water LCP phase, Pn3m, and found a sharp transition from large channels to small channels as the temperature increased above 44°C with with the LCP phase intact. The results from the electrochemical impedance

9

spectroscopy in this investigation indicates that the metal cations can have an effect on the ion channel diameter, where a decreased diameter is reflected as an increase in membrane resistance. Twenty-four impedance scans were done consecutively taking almost three hours for all electrolyte solutions except for 10 mM CaCl2. In the latter case the equilibration time was much longer and therefore a measurement time of 22 hours was used. The lipid phase is still transparent after exposure to all different electrolytes, which indicates that all measurements are performed on a stable LCP. 3.2.1 Effect of anions: Lipid membrane interaction with several lyotropic anions was investigated from the red shift in fluorescence excitation spectra in the presence of two dyes [42]. The strength of interaction of a particular anion with the membrane was shown to depend on its free energy of hydration. Chloride and bromide ions are both hydrophilic anions and also close to each other in the lyotropic series, indicating that both interact only weakly with the lipid headgroups in phosphatidylcholine vesicles. Ionic mobility in water at 25 °C for K+(aq), Cs+(aq) and Ca++(aq) ions in the limiting case of infinitely diluted solutions are 7.12 ± 0.51, 7.32 ± 0.66 and 2.06 ± 0.12 10-8 m2 V-1 s-1 respectively and for Cl-(aq) and Br-(aq) anions they are 6.88 ± 0.31 and 7.20 ± 0.70 10 -8 m2 V-1 s-1 [24]. Thus, the ionic mobilities of bromide and chloride ions are comparable. Taken together, it is concluded that observed changes in dielectric properties of the cubic phase in the present investigation is not caused by differences in anion mobility within the aqueous channels. The differences in membrane resistance and capacitances will be discussed as an effect of cationic interactions with the membrane headgroup region.

10

3. 3 LCP membrane in contact with the electrolytes: I = 10 mM KCl and CsBr. In Fig. 5 results are shown from 24 successive impedance scans with I = 10 mM KCl. A blank measurement is also shown. In the high to medium frequency range, large and depressed semicircles are observed and at low frequencies a capacitive line is seen. The impedance spectra changes with time: i.e.: the high frequency resistance decreases continuously and also the size of the semicircle. This smooth change is attributed to diffusion of electrolyte ions into the LCP. The final scans are almost overlapping which indicates that the cubic phase becomes saturated with the electrolyte solution at the end of the experiment. In the transfer from the depressed semicircle to the capacitive line something that might be mistaken for an inductive loop is observed. In the control measurements with a four – electrode setup an identical semicircle was observed followed by a decrease in resistance at low frequencies, but the capacitive charging of platinum electrodes was absent as expected. This indicates that the observed loop in the two-electrode setup is only a reflection of the continuous change in resistance of the LCP due to diffusion of the electrolyte ions into the membrane. The impedance spectra for 10 mM CsBr show similar characteristics as in the case of 10 mM KCl.: that is depressed semicircles are observed for high and medium frequencies followed by a loop. At low frequencies the impedance response is again dominated by the double layer capacitances of the platinum electrodes. The semicircles contain information on the dielectric properties of the LCP. An equivalent circuit is suggested for extraction of numerical information in Fig. 6a. In series with Cdl, and Rs a parallel circuit is inserted that models the free standing membrane. Thus, RCP is the membrane resistance and CCP the membrane capacitance. This parallel circuit is the origin of the semicircle and 11

the characteristic time constant for charging the membrane can be estimated from the maximum value of the semicircle from τc = RCP x CCP. The total cell capacitance is modelled as a constant phase element, CPEcell, in parallel with the whole circuit. Results from circuit fitting show that RCP decreases from 716 kOhm to an almost constant value after 2.5 hours where the final value is 352 kOhm. The membrane capacitance is almost constant during the 2.5 hours i.e.: 1.53 ± 0.02 pF. Similar results were obtained when fitting experimental data from measurements in CsBr at ionic strength 10 mM. The initial and final membrane resistances were 676 kOhm and 277 kOhm respectively. The membrane capacitance is again almost constant and similar to the previous case, i.e.: 1.57 ± 0,03 pF It should be noted that both initial and final membrane resistances are lower in the case of cesium ions in electrolyte than with potassium. It is reported that potassium ions interact with membrane to a somewhat larger extent than cesium ions, which might be the reason [43]. All initial and final membrane capacitances and resistances are summarized in Table 2. 3.4 LCP membrane in contact with the electrolytes I = 10 and 100 mM CaCl2 In Fig. 7, 24 successive impedance scans in both I = 10 mM and 100 mM CaCl2 are shown with the corresponding two blank measurements. At the lower concentration, I = 10 mM, a depressed semicircle is observed at high and medium frequencies with a subsequent transition to a capacitive line. The semicircle is decreasing slightly with time also at the end of the 24 scans, which indicates that longer time is needed to saturate the membrane with electrolyte ions. Therefore, the measurements were repeated 190 times taking

12

approximately 22 hours, data not shown. Obviously the diffusion of divalent Ca2+-ions into the aqueous channels of the LCP is much slower than in the case of monovalent alkaline cations. Data fitting of data from measurements in 10 mM CaCl2 gives a continuously decreasing RCP from initially 1.5 MOhm to a final value of 0.97 MOhm after 22 hours. The membrane capacitance was not constant during the experiment and decreased from 0.44 pF to 0.33 pF. At the higher concentration, I = 100 mM, the features of the impedance spectra are similar to the results obtained from 10 mM monovalent electrolyte solutions, with a large depressed semicircle at medium to high frequencies followed by a loop and then a capacitive line at low frequencies. No saturation of the membrane with electrolyte ions is observed at the final scans. The results from data fitting show a decreasing R CP from initially 621 kOhm to a final value of 86 kOhm. In this case the membrane capacitance is larger than in the case of monovalent ions and decreased from 2.3 to 1.9 pF during the experimental time. The membrane resistance at low ionic strength, i.e.: 10 mM, is significantly larger than that for 100 mM. Moreover, the difference in membrane capacitance is five-fold between the high and low ionic strength, which indicates that two different types of cubic phases are prevalent at the two concentrations. The fitting parameters to the model in Fig 6a display high errors for the membrane capacitance; 6 - 10 %. This can be explained with a system that is not in equilibrium during measurements and also that the time-constant is high in the system so the whole relaxation process is not covered within the time window of the measurements. Anyhow, data fitting gives a good indication on

13

the size of both membrane resistance and capacitance. All initial and final membrane capacitances and resistances are summarized in Table 2. 3.5 LCP membrane in contact with the electrolytes I = 100 mM KCl, CsBr Measurement results with 100 mM KCl and CsBr are different from those at 10 mM. Only a fraction of the depressed high frequency semicircle is seen in the first scans and it is disappearing with time so that the last scans only show a capacitive line. In Fig. 8 results from measurements in 100 mM KCl are shown. Measurements in 100 mM CsBr have similar characteristics and resistances and also in this case the high frequency resistance is decreasing with time. It was found that the equivalent circuit in Fig. 6a was not appropriate to get an acceptable fit to the experimental data. The characteristic time-constant τc = RC for charging the cell membrane becomes small since the membrane resistance is small. Therefore, there are not enough data points from the high-frequency regime where this relaxation occurs, and only a fraction of the semicircle reflecting the dielectric properties of the membrane is within the frequency regime of the experiment. However, a similar equivalent circuit as the one used for blank measurements gave good results and was used in this case, Fig. 9a. It follows that LCP membrane resistance and solution resistance are not separated in the timescale of the experiment. The resistance RTOT is the sum of RCP and RS. Therefore, only RCP can be determined for these experimental conditions and not CCP. A typical EIS spectrum for 100 mM KCl is shown in Fig. 9b, together with the corresponding fitting results. RCP is decreasing with time from 103 kOhm to a stable value of 32 kOhm. Similar results were obtained when fitting experimental data from the ionic strength 100 mM CsBr. Initial resistance for 100 mM CsBr is 174 kOhm and 14

the final was 31 kOhm, which is the same as in 100 mM KCl within the experimental error. All initial and final membrane capacitances and resistances are summarized in Table 2. In summary: It was illustrated that electrochemical impedance measurements can be used for investigation of the permeability and interaction of metal cations, i.e.: K+(aq), Cs+(aq) and Ca2+(aq), in LCP. When the freestanding membrane was exposed to either 100 mM KCl or CsBr, the membrane resistance was larger initially in the case of CsBr whereas both systems gave the same final RCP. The system changes very fast in the beginning of electrolyte exposure, which might be a consequence of difficulties to start exactly at the same time in each experiment. The initial values of R CP are therefore not comparable between the systems and are therefore discarded. RCP was always larger in the case of KCl when the surrounding electrolyte was either 10 mM KCl or CsBr. The final RCP was 352 kOhm with KCl and 277 kOhm with CsBr. This supports the findings of Robert Va´cha, et al who reported that the larger cesium ions, do not preferentially adsorb and only weakly penetrate into the headgroup region, whereas the smaller potassium ions have a slightly larger tendency to interaction [43]. This can explain the higher membrane resistances in the presence of potassium ions, when the LCP membrane was equilibrated with low concentration alkali metal electrolytes. Both K+(aq) and Cs+(aq) prefer to remain in the aqueous region in the channels and bind only weakly to hydroxyl groups through their hydrated shell. The results indicate that impedance measurements can discriminate between these monovalent cations, where the ionic radii for six- and seven-coordinate K+(aq) is smaller than eight-coordinate Cs+(aq). The almost identical membrane capacitances in the two systems, is also a clear indication that the monovalent cations are not bound to the headgroup region in the aqueous channels.

15

In 100 mM monovalent electrolyte solution the hydration number of the cations will decrease [44] and so also the ionic radii. This can explain why the membrane resistances are the same after 3 hours for the two cases. The membrane resistances were significantly higher when the freestanding membrane was exposed to CaCl2 as compared to exposure to monovalent electrolytes with the corresponding ionic strengths. This is a consequence of the strong interaction of Ca2+(aq) with hydroxyl groups in the monoolein headgroup region. The Ca2+(aq) bridging between headgroups and the dehydration in the headgroup leads to condensation and a probable decrease of the radius of the aqueous channels [14, 45]. All changes in membrane resistances are shown in Fig. 10 where RCP were obtained as described above. Interestingly the membrane capacitance in CaCl2 electrolyte is larger than in the monovalent case for high ionic strength, and for the low ionic strength the membrane capacitance is significantly lower. We speculate that this is a consequence of two different cubic phases, similar to what was observed by Awad and coworkers [46]. They showed that two different cubic phases consisting of 30%-DOPG/70%-MO was obtained when large unilamellar vesicles, were exposed to different concentrations of calcium electrolyte using SAX measurements. In their case the two different cubic phases had different lattice constants with 15 nm for concentrations below 25 mM and 10 nm at higher concentrations. The lattice constant for the low concentration case indicates a more open structure than for the high concentration case. However the DOPG/MO cubic phase carries negative charges in the headgroup region. The observation of a high resistance and low capacitance in the low concentration system in the present work indicates that the cubic phase is more condensed than in the high concentration case. This is not surprising considering that binding of calcium

16

ions to the walls in the aqueous channels implies addition of charges to the MO headgroups. The imposed electrostatic repulsion in the headgroup region will increase at higher concentrations inducing a phase transition to another cubic phase that is more open. We speculate that this is the reason for the lower membrane resistance and higher membrane capacitance at the higher ionic strength. Conclusions and Future aspects It is demonstrated that Electrochemical Impedance Spectroscopy is a versatile tool for investigation of ionic interaction with lipid membrane, organized in a lipid cubic phase. Firstly: the difference between monovalent, K+(aq) Cs+(aq)

and divalent, Ca2+(aq)

and

cations was clearly demonstrated from

observations of differences in membrane resistance and membrane capacitance. This is in good accord with earlier reported findings using metadynamics simulations [11, 43]. Secondly: an indication of two different MO cubic phases at two different ionic strengths of CaCl2 was observed. This is a new finding but similar observations has been reported by Awad et al [46]. Thirdly: A higher membrane resistance was observed when the LCP membrane was exposed to 10 mM KCl compared to the same ionic strength of CsBr. This indicates that K+(aq) penetrates more into the MO headgroup region than Cs +(aq) as discussed by Vácha et al in ref [43]. These results support our hypothesis that EIS can be used for investigation of ionic interactions with lipid headgroups. EIS has the capability to confirm earlier findings such as calcium condensing effect on membranes and also to detect unknown phenomena in lipid systems, such as MO LCP phase transitions at different CaCl2 ionic strengths.

17

The results from this investigation are certainly promising and the methodology will be further investigated with a DOPC/MO cubic phase using the same metal cations in the electrolyte solution. The permeability of LCP for control of different ionic drugs will be studied to increase the knowledge in this emerging research field. The proposed phase transition from one cubic phase to another should be further investigated by successive changes in the ionic strength with small increments between the ionic strengths 10 mM and 100 mM of CaCl2 and the introduction of for instance SAX measurements for confirmation of our findings.

Acknowledgement The authors thank prof Göran Lindblom for valuable discussions. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. References: 1. 2. 3. 4. 5. 6. 7.

Lipowsky R., Sackmann E. Structure and Dynamics of Membranes: I. From Cells to Vesicles / II. Generic and Specific Interactions. 1995: Elsevier Science. Alberts B., Johnson A., Lewis J., Walter P., Raff M., Roberts K. Molecular Biology of the Cell 4th Edition: International Student Edition. 2002: Routledge. Cook W.J., Thewalt U., Bugg C.E. Calcium-binding to non-ionic lipids: Crystal structure of a calcium chloride complex of geraniol. Biochem Biophys Res Commun, 1976, 68(1),143-148. Ganem-Quintanar A., Quintanar-Guerrero D., Buri P. Monoolein: a review of the pharmaceutical applications. Drug Dev Ind Pharm, 2000, 26(8), 809-20. Berkowitz M.L., Vácha R. Aqueous Solutions at the Interface with Phospholipid Bilayers. Acc Chem Res, 2012, 45(1),74-82. Kalinin S.V., Molotkovsky J.G. Anion binding to lipid bilayers: determination using fluorescent membrane probe by direct quenching or by competitive displacement approaches. J. Biochem. Biophys. Methods, 2000, 46(1–2), 39-51. Boettcher J.M., Davis-Harrison R.L., Clay M.C., Nieuwkoop A.J., Ohkubo Y.Z., Tajkhorshid E., Morrissey J.H., Rienstra C.M. Atomic View of Calcium-Induced Clustering of Phosphatidylserine in Mixed Lipid Bilayers. Biochemistry, 2011, 50(12), 2264-2273.

18

8. 9. 10.

11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

Dluhy R., Cameron D.G., Mantsch H.H., Mendelsohn R. Fourier transform infrared spectroscopic studies of the effect of calcium ions on phosphatidylserine. Biochemistry, 1983, 22(26), 6318-6325. Lotan O., Fink L., Shemesh A., Tamburu C., Raviv U. Critical Conditions for Adsorption of Calcium Ions onto Dipolar Lipid Membranes. J. Phys. Chem. A, 2016, 120 (19),3390–3396 Vácha R., Jurkiewicz P., Petrov M., Berkowitz M.L., Böckmann R.A., BaruchaKraszewska J., Hof M., Jungwirth P. Mechanism of Interaction of Monovalent Ions with Phosphatidylcholine Lipid Membranes. J Phys Chem B, 2010, 114(29), 95049509. Yang J., Calero C., Bonomi M., Martí J. Specific Ion Binding at Phospholipid Membrane Surfaces. J Chem Theory Comput, 2015, 11(9), 4495-4499. Manery J. Effects of Ca ions on membranes. Fed Proc, 1966, 25(6P1), 1804-&. Martín-Molina A., Rodríguez-Beas C., and Faraudo J. Effect of Calcium and Magnesium on Phosphatidylserine Membranes: Experiments and All-Atomic Simulations. Biophys J, 2012, 102(9), 2095-2103. Sinn C.G., Antonietti M., Dimova R. Binding of calcium to phosphatidylcholine– phosphatidylserine membranes. Colloids Surf A Physicochem Eng Asp, 2006, 282–283, 410-419. Kiriukhin M.Y., Collins K.D. Dynamic hydration numbers for biologically important ions. Biophys Chem, 2002, 99(2), 155-168. Rogers H., Spector R.G. An Introduction to Mechanisms in Pharmacology and Therapeutics, 2013, Elsevier Science. Zhou J., Lu X., Wang Y., Shi J. Molecular dynamics study on ionic hydration. Fluid Phase Equilibria, 2002, 194–197, 257-270. Volkov A.G., Paula S., Deamer D.W. Two mechanisms of permeation of small neutral molecules and hydrated ions across phospholipid bilayers, Bioelectrochem Bioenerg, 1997, 42(2), 153-160. Nightingale E.R. Phenomenological Theory of Ion Solvation. Effective Radii of Hydrated Ions. J. Phys. Chem, 1959, 63(9), 1381-1387. Ohtomo N., Arakawa K. Neutron diffraction study of aqueous ionic solutions. I. Aqueous solutions of lithium chloride and cesium chloride. Bull Chem Soc Jpn, 1979, 52(10), 2755-2759. Kielland J. Individual Activity Coefficients of Ions in Aqueous Solutions. J Am Chem Soc, 1937, 59(9), 1675-1678. Li D., Caffrey M. Lipid cubic phase as a membrane mimetic for integral membrane protein enzymes. Proc Natl Acad Sci U S A, 2011, 108(21), 8639-8644. Jeong S.W., Orädd G., Lindblom G. Encapsulation and Diffusion of Water-Soluble Dendrimers in a Bicontinuous Cubic Phase. Langmuir, 2002, 18(4), 1073-1076. Koneshan S., Rasaiah J.C., Lynden-Bell R.M., Lee S.H. Solvent Structure, Dynamics, and Ion Mobility in Aqueous Solutions at 25 °C. J. Phys. Chem. B, 1998, 102(21), 4193-4204. Foster K.R., Schwan H.P. Dielectric properties of tissues and biological materials: a critical review. Crit Rev Biomed Eng, 1989, 17(1) 25-104. Kell D.B., Davey C.L. In: Cass AEG (ed) Conductimetric and impedimetric devices in biosensors, a practical approach. 1990: IRL Press, Oxford University. Fricke H. THE ELECTRIC CAPACITY OF SUSPENSIONS WITH SPECIAL REFERENCE TO BLOOD. J. Gen. Physiol, 1925, 9(2) 137-152.

19

28. 29. 30. 31.

32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45.

del Castillo J., Rodriguez A., Romero C.A., Sanchez V. Lipid Films as Transducers for Detection of Antigen-Antibody and Enzyme-Substrate Reactions. Science, 1966, 153(3732) 185-188. de Levie R., Vukadin D. Dipicrylamine transport across an ultrathin phosphatidylethanolamine membrane. J. electroanal. chem. interfacial electrochem, 1975, 62(1) 95-109. de Levie, R., Thomas J.W., Abbey K.M. Membrane admittance measurements under computer control. J. electroanal. chem. interfacial electrochem, 1975, 62(1) 111-125. Coster H.G.L., Smith J.R. The molecular organization of bimolecular lipid membranes. A study of the low frequency Maxwell-Wagner impedance dispersion. Biochimica et Biophysica Acta (BBA) - Biomembranes, 1974, 373(2) 151-164. Sackmann E. Supported Membranes: Scientific and Practical Applications. Science, 1996, 271(5245) 43-48. Lindholm-Sethson B. Electrochemistry at Ultrathin Organic Films at Planar Gold Electrodes. Langmuir, 1996, 12(13) 3305-3314. Lang H., Duschl C., Vogel H. A new class of thiolipids for the attachment of lipid bilayers on gold surfaces. Langmuir, 1994, 10(1) 197-210. Lindholm-Sethson B., Geladi P., Nelson A. Interaction with a phospholipid monolayer on a mercury electrode: Multivariate analysis of impedance data. Analytica Chimica Acta, 2001, 446(1–2) 121-130. Rowiński P., Korytkowska A., Bilewicz R. Diffusion of hydrophilic probes in bicontinuous lipidic cubic phase. Chem. Phys. Lipids, 2003, 124(2), 147-156 Coster H.G.L., Chilcott T.C., Coster A.C.F. Impedance spectroscopy of interfaces, membranes and ultrastructures. Bioelectrochem Bioenerg, 1996, 40(2) 79-98. MacDonald J.R., Hurt R.L. Some simple equivalent circuits for ionic conductors. J. electroanal. chem. interfacial electrochem, 1986, 200(1) 69-82. Jorcin J.-B., Orazem M.E., Pébère N., Tribollet B. CPE analysis by local electrochemical impedance spectroscopy. Electrochim Acta, 2006, 51(8–9), 14731479. Negrini R., Sánchez-Ferrer A., Mezzenga R. Influence of Electrostatic Interactions on the Release of Charged Molecules from Lipid Cubic Phases. Langmuir, 2014, 30(15) 4280-4288. Angelov B., Angelova A., Ollivon M., Bourgaux C., Campitelli A. Diamond-Type Lipid Cubic Phase with Large Water Channels. J. Am. Chem. Soc, 2003, 125(24) 7188-7189. Clarke R.J., Lüpfert C. Influence of Anions and Cations on the Dipole Potential of Phosphatidylcholine Vesicles: A Basis for the Hofmeister Effect, Biophys J, 1999, 76(5), 2614-2624. Vácha R., Siu S.W.I., Petrov M., Böckmann R.A., Barucha-Kraszewska J., Jurkiewicz P., Hof M., Berkowitz M.L., Jungwirth P. Effects of Alkali Cations and Halide Anions on the DOPC Lipid Membrane. J Phys Chem A, 2009, 113(26), 7235-7243. Marcus Y. Concentration Dependence of Ionic Hydration Numbers. J Phys Chem B, 2014, 118(35), 10471-10476. Yaghmur A., Sartori B., Rappolt M. The role of calcium in membrane condensation and spontaneous curvature variations in model lipidic systems. Phys Chem Chem Phys, 2011, 13(8), 3115-3125.

20

46.

Awad T.S., Okamoto Y., Masum S., Yamazaki M. Formation of Cubic Phases from Large Unilamellar Vesicles of Dioleoylphosphatidylglycerol/Monoolein Membranes Induced by Low Concentrations of Ca2+. Langmuir, 2005. 21(25): p. 11556-11561.

21

Figure legends

1a) Molecular structure of Mono-Olein (1-0leoyl-rac-glycerol) (MO) b) Phase diagram of the monoolein/water system (Fig 2 in Biomaterials, 2000. 21: 223-34 with kind permission from the publisher) 2) Cartoon representation of cubic-Pn3m phase (Fig 1 in Biochim. Biophys. Acta, 1989. 988, 221-256 with kind permission from the publisher) 3) Schematic picture of the two-compartment electrochemical cell. 4) Equivalent Circuit: series combination of double layer capacitances of the platinum electrodes, Cdl, solution resistance, Rs., cell capacitance modeled as a constant phase element, CPE cell 5) Electrochemical impedance scans with MO cubic phase in the aperture including the corresponding background measurement. 1-16th scan and the 24th scan are displayed. Electrolyte solution = 10 mM KCl, Rs = solution resistance is obtained from the interception of the background measurement with the real axis, RCP = membrane resistance 6a) Equivalent Circuit; See Fig 4 with addition of a parallel circuit that models the membrane capacitance and resistance i.e.: CCP and RCP respectively. b) Raw data from 24th impedance scan in 10 mM M KCl and MO in aperture and the corresponding results from fitting to Fig 6a. The corresponding background measurement is included. 7) Electrochemical impedance scans with MO cubic phase in the aperture and two ionic strengths of CaCl2: I = th

th

10 mM and 100 mM. 1-16 scan and the 24 scan are displayed for both ionic strengths. Rs = solution resistance, RCP = membrane resistance as in fig 5. The corresponding background measurements are included. 8) Electrochemical impedance scans with MO cubic phase in the aperture; KCl: I = 100 mM. 1-16th scan and the th

24 scan are displayed. RS = solution resistance as in fig 5, RCP = membrane resistance. The corresponding background measurement is included. 9a) Equivalent Circuit: series combination of double layer capacitances of the platinum electrodes, C dl, solution + membrane resistance, RTOT, cell capacitance modeled as a constant phase element, CPE cell 9b) Raw data from 24th scan in 100 mM KCl and MO in the aperture and the corresponding fitting results to equivalent circuit in 9a 10) Time dependence of membrane resistances for all systems. The electrolyte solutions are indicated with the ionic strength in mM.

22

a. b.

23

24

25

Cdl

Rs

CPEcell

Element Cdl Rs CPEcell-T CPEcell-P

Freedom Free(+) Free(+) Fixed(X) Fixed(X)

Data File: Circuit Model File: Mode: Maximum Iterations: Optimization Iterations: Type of Fitting: Type of Weighting:

Value 3,4213E-05 37954 0 1

Error N/A N/A N/A N/A

Error % N/A N/A N/A N/A

C:\Users\yakh0005\Desktop\blank.mdl Run Fitting / All Data Points (1 - 61) 100 0 Complex Calc-Modulus

26

27

28

29

30

31

1600 1400

Rcp/KOhm

1200 KCl-100

1000

CsBr-100

800

CaCl2-100

600

KCl-10

400

CsBr-10

200

CaCl2-10

0 0

2000

4000

6000 t/s

8000

10000

12000

32

Table 1 Result from Zview fitting for all blank measurements: Cdl = double layer capacitance of two platinum electrodes in series, Rs = solution resistance; Q and n are parameters in the constant phase element that models a non-ideal cell capacitance. Electrolyte/mM KCl / 100 KCl / 10 CsBr / 100 CsBr / 10 CaCl2 /100 CaCl2 /10

Cdl/µF 28,6 ±0,2 27,5±0,1 26,7±0,2 23,24±0,02 25,79±0,03 25,7±0,1

Rs/kOhm 3,94±0,02 37,53±0,035 4,103±0,008 36,30±0,02 8,11±0,03 67,8±0,2

Q*10-10 670±36 0,133±0.009 3340±130 0,138±0,002 3,06±0,06 0,115±0,009

n 0,1173±0,0006 0,958±0,005 0,165±0,006 0,955±0,001 0,7±0,1 0,965±0,006

33

Table 2: Initial and final membrane resistances and capacitances for all systems as obtained from circuit fitting. Corresponding error percentages are shown in parentheses. I.S = ionic strength. The final scan for 10mM CaCl2 was obtained after 22 h whereas for all other systems it was 2,5 h. I.S scan KCl CsBr CaCl2

10 mM Rcp/kOhm First Last 716 352 (0,46) (0,10) 676 277 (0,76) (0,20) 1500 979 (0,19) (0,16)

Ccp/pF First Last 1,53 1,51 (8,02) (2,38) 1,53 1,57 (13,5) (5,41) 0,44 0,33 (4,08) (5,98)

100 mM Rcp/kOhm First Last 103 32 (1,23) (0,21) 174 31 (2,30) (0,27) 621 86 (1,02) (0,14)

Ccp/pF First Last -

-

2,28 (12,76)

1,88 (7,05)

34

Graphical abstract

35