Two cations, two mechanisms: interactions ofsodium and calcium with zwitterionic lipid

membranes†

Matti Javanainen,†,‡,⊥ Adéla Melcrová,¶,⊥ Aniket Magarkar,§,‖ Piotr Jurkiewicz,∗,¶

Martin Hof,¶ Pavel Jungwirth,∗,§,† and Hector Martinez-Seara∗,§,†

†Laboratory of Physics, Tampere University of Technology, P.O. Box 692, FI-33101Tampere, Finland

‡Department of Physics, University of Helsinki, P.O.Box 64, FI-00014 Helsinki, Finland¶J. Heyrovský Institute of Physical Chemistry, Czech Academy of Sciences, Dolejskova 3,182 23 Prague 8, Czech Republic., Academy of Sciences of the Czech Republic, Prague,

Czech Republic§Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech

Republic, Prague, Czech Republic‖Faculty of Pharmacy, University of Helsinki, Viikinkaari 5E, FI-00014 Helsinki, Finland

⊥These authors contributed equally to this work.

E-mail: piotr.jurkiewicz@jh-inst.cas.cz; pavel.jungwirth@uochb.cas.cz; hseara@gmail.com

1

Abstract

Adsorption of metal cations to a cellular membrane changes its properties, such asinteractions with charged moieties or the propensity for membrane fusion. It is, however,unclear whether cells can regulate ion adsorption and the related functions via locallyadjusting their membrane composition. We employed fluorescence techniques and com-puter simulations to determine how the presence of cholesterol – a key molecule inducingmembrane heterogeneity – affects the adsorption of sodium and calcium to zwitterionicphosphatidylcholine bilayers. We found that the transient adsorption of sodium is depen-dent on the number of phosphatidylcholine head groups, while the strong surface bindingof calcium is determined by the available surface area of the membrane. Cholesterolthus does not affect sodium adsorption and only plays an indirect role in modulatingthe adsorption of calcium by increasing the total surface area of the membrane. Theseobservations also indicate how lateral lipid heterogeneity can regulate various ion-inducedprocesses including adsorption of peripheral proteins, nanoparticles, and other moleculesto membranes.

Graphical TOC Entry

Increase in surface areaincreases Ca2+ binding

Increase in PC numberincreases Na+ binding

2

Metal cations play a central role in cell func-tion and regulation.1,2 In resting mammaliancells, the most common monovalent cationsNa+ and K+ (at concentrations of 100–150 mM)are segregated between the extracellular and cy-tosolic compartments, respectively. Free diva-lent cations are more scarce. For example, Ca2+exhibits a significantly lower concentration onthe cytosolic side (1–100 nM) than on the ex-tracellular one (1–2 mM).3 Global changes inthe distribution of these cations due to cellhomeostasis trigger changes in cell volume, os-motic balance regulation, and electrostatic po-tential.1,2 For these reasons, cationic concentra-tions and their fluxes in and out from the cellare tightly regulated.1 More localized changesof the cation distributions play an importantrole in, e.g., membrane fusion4,5 and signal-ing6,7. Furthermore, the differential accumu-lation of cations at the surface of a membranecan ultimately modulate its interaction with thesurrounding macromolecules or locally alter thephysical properties of the membrane.8–10Over the past decades, considerable progress

has been made in unraveling the role of ions insignaling6 as well as the regulation of cell vol-ume1 and transmembrane potential2. However,it is only recently that advances have been madein the experimental description of the directinteractions between ions and membranes11,12,the quantitative aspects of which are still highlycontroversial13. Moreover, it remains to be de-termined whether cells possess mechanisms toselectively modulate their membrane affinity forcertain ions.There is clear evidence that ions interact in a

specific way with membranes12,13. While mono-valent alkali cations do not adsorb efficientlyto membranes unless negatively charged lipidsare present11, divalent ions such as Ca2+ doso more readily even if no attractive net neg-ative charge is present12. The extracellularleaflet of the cell membrane is mainly composedof zwitterionic phosphatidylcholine and sphin-gomyelin together with cholesterol, all of whichare overall neutral lipids. Therefore, any se-lective adsorption of cations to the surface ofsuch a membrane results in positive net chargeat the membrane–water interface.14,15 Cells can

very precisely control and regulate membranelipid compositions both globally and locally.16If such changes in composition could also altermembrane–cation interactions, the cell wouldpossess a powerful tool to induce and controla plethora of functions. As the outer layer ofthe plasma membrane is mostly constituted byzwitterionic and neutral lipids, there is no evi-dent mechanism for this regulation. Motivatedby several studies which proposed that choles-terol which is a neutral lipid and one of the maincomponents of the plasma membrane, may af-fect Na+ binding,17,18 we have studied in detailthe underlying mechanisms determining Ca2+and Na+ interaction at the membrane surface.We find that the number of bound calcium ionsdepends on the total membrane area, whereasthe number of phospholipid head groups deter-mine the number of bound sodium.To unravel the mechanisms governing Ca2+

and Na+ interactions at the membrane surface,we have characterized the properties of lipo-somal membranes with fluorescence measure-ments in membranes with and without choles-terol, at different temperatures, and at the pres-ence of varying levels of Na+ and Ca2+. Fur-thermore, we have performed molecular dynam-ics simulations to investigate how increasingcholesterol levels and temperature affect thebinding of Ca2+ and Na+ to the membrane. Re-markably, results from both of these methodolo-gies show excellent agreement with each other,with the latter one providing a detailed insightinto the microscopic origins of our experimentalobservations.To compare the impacts of the binding of

sodium and calcium ions on model membranes,we performed fluorescent measurements withthe fluorophore Laurdan. Laurdan probes thehydration and packing around the lipid sn-1carbonyl region.19 We have performed two setsof experiments – time-dependent fluorescenceshift (TDFS) and excitation generalized polar-ization (GPEX) – the latter serving as a refer-ence methodology. TDFS presents several ad-vantages with respect to GPEX; While GPEX isa ratiometic phenomenological experiment thatallows to roughly distinguish between lyotropiclipid phases,20 TDFS allows to deconvolute hy-

3

dration (i.e. polarization) and mobility (i.e.packing) at the carbonyl region into two pa-rameters – the emission shift (∆ν) and the re-laxation time (τr).21No significant differences in the emission

shifts (∆ν) of Laurdan were detected for theherein studied systems. This means that theprobe does not detect any changes in mem-brane hydration, which we further discuss inSection S3.2 in the ESI†. The relaxation times(τr) measured for Laurdan embedded in largeunilamellar vesicles composed of either purepalmitoyloleoylphosphatidylcholine (POPC) orits 4:1 mixture with cholesterol (POPC/Chol)are shown in Fig. 1. The influence of sodiumand calcium are studied at three different ionconcentrations (0, 150 mM and 1 M), all at310 K. Additional measurements at 288 K and298 K provide similar trends (see Section S3.2,ESI†).

0 150 10000

1

2

3

4

+62%

+11%

+34%

+5%

Ion concentration (mM)

Rel

axat

ion

time

τ r(n

s) POPC/Chol (20%) + CaCl2POPC/Chol (20%) + NaClPOPC + CaCl2POPC + NaCl

Figure 1: Relaxation times from time-dependent fluorescence shift measurements at310 K. The percentage values represent therelative increase of Laurdan relaxation timesupon the addition of 1 M of salt. Whereasthe methodology itself is quite precise (error of∼0.05 ns), we estimate the total error associ-ated with the values for the relative increaseto be ∼10 percentage points, estimated fromcomparing values obtained using different fit-ting strategies.

As shown in Fig. 1, we observe significantchanges in relaxation time τr as a function ofcholesterol content and salt concentration. Fur-ther data measured at different temperatures isshown in Fig. S7. Decreasing temperature or in-creasing cholesterol concentration increases the

time required for the Laurdan environment torelax. The former leads to a decrease in the areaper lipid, whereas the latter results in a moreordered and rigid membrane. This means thatin both cases the mobility in the carbonyl re-gion – where Laurdan resides – decreases, whichresults in the slowdown of Laurdan rotationaldynamics. Decrease in the temperature by 22 Khas a larger effect than the addition of 20%cholesterol to the system (see Fig. S7). How-ever, the addition of Na+ at any investigatedconcentration does not have a significant effecton either POPC (∆τPurer = 5%) or POPC/Chol(∆τChol

r = 11%) membranes, both values fallingwithin the error estimate. This suggests thatsodium cations do not significantly disturb thelipid surface structure. Therefore, the lipid bulkstructural and dynamic properties remain unal-tered in the region where Laurdan resides.In contrast, the addition of Ca2+ affects the

measured relaxation times severely. This showsthat calcium adsorption is sensed at the car-bonyl level, while no direct probe–ion interac-tions are present (see Section S3.2, ESI†). Thisis caused by the bridging of lipid molecules bycalcium ions and the subsequent lateral mem-brane compression which results in the hin-drance of the phospholipid mobility.11,22 In-terestingly, it is clearly visible that the ad-dition of cholesterol at 310 K not only in-creases the relaxation times, but also signifi-cantly amplifies the effects of calcium, i.e., from∆τPurer = 34% to ∆τChol

r = 62%. This am-plified effect of calcium due to the presence ofcholesterol is consistent with an increased num-ber of Ca2+ per phospholipid adsorbed in themembrane. These additional ions cause fur-ther phospholipid bridging, lower mobility ofthe lipids around Laurdan, and consequentlylarger relaxation times.Motivated by our unexpected finding that

cholesterol modulates the binding of Na+ andCa2+ in different ways, and aiming to under-stand the molecular mechanisms behind thisbehavior, we simulated several phosphatidyl-choline (PC) and PC/Chol membranes undervarious ionic environments (no ions, 130 mM or1 M of NaCl, or 450 mM of CaCl2, see Table S1,ESI†). Upon addition of cholesterol to a given

4

PC membrane, the total membrane area is in-creased (see Section S5.1, ESI†). However, dueto the condensing effect of cholesterol,23 thisincrease is less than in the case of ideal mix-ing. In systems with NaCl, the total number ofadsorbed sodium ions remains almost constantwith cholesterol content (see Fig. 2) leading toa decrease in the surface charge density of thebilayer. We also observe that at a higher NaClconcentration the number of bound cations in-creases. These findings indicate that the bind-ing of sodium depends solely on its bulk concen-tration and on the number of PC head groupsavailable at the surface. Recent studies on theeffect of cholesterol on the binding of sodiumto PC membranes17,18 report seemingly contra-dicting results, which we found to be explain-able by factors related to normalization (seeSection S6, ESI†).

0 10 20 300.020.050.08

Cholesterol content (%)

0.250.3

0.350.4

Cat

ions

per

PC

1M NaCl 450 mM CaCl2130 mM NaCl

Figure 2: Average number of bound cations perPOPC head group as a function of cholesterolconcentration. Error bars show Std. Dev.

Next, the same set of POPC bilayers withvarying amounts of cholesterol were simulatedin the presence of nominally 450 mM of CaCl2.This amount of CaCl2 was selected so that af-ter adsorption to the membrane free cationswould always remain in the aqueous phase forthe present unit cell size. Unlike for NaCl, thenumber of bound calcium ions increases as afunction of cholesterol content (see Fig. 2). Theobserved trends of the effect of cholesterol onboth sodium and calcium binding are sharedwith nuances across several lipid and ion forcefields (see Section S5.2, ESI†).The tendency of sodium and calcium to ad-

sorb to bilayers is plotted as a function of areaper PC in Fig. 3. Considering that salt and

0.060.090.120.15

Na+

Cha

rge

dens

ity(e

/nm

2 ) POPC/chol DOPC (Ld)DPPC (gel)

0.45 0.5 0.55 0.6 0.65 0.7 0.751.001.051.101.15 Ca2+ B

Area per PC (nm2)

0.25

0.30

0.35 Ca2+

0.45 0.5 0.55 0.6 0.65 0.7 0.75

0.040.050.060.07

Na+A

Boun

dio

nspe

rPC

Figure 3: (A) The dependence of cation adsorp-tion (130 mM of NaCl or 450 mM of CaCl2) onthe area per PC. (B) The dependence of sur-face charge density on the area per PC in mem-branes. Red dots are POPC with 0,10, 20, or30% cholesterol at 310 K (smallest to highestarea). Black dot is DOPC at 310 K. Green dotis DPPC at 295 K. Dashed lines show fits to thedata, see text for details. Error bars show Std.Dev.

cholesterol can disturb the lateral packing ofthe bilayer, we observe two distinct trends forNa+ and Ca2+. The number of bound Na+ perPC is constant regardless of the area per lipid(Fig. 3A). The independence of ion adsorptionon the total membrane area results in varia-tions of surface charge densities in the rangeof ∼0.07–0.12 e/nm2 in the simulated systems(Fig. 3B). To summarize, Na+–membrane in-teraction is mostly governed by the total num-ber of PC head groups present. In the caseof calcium, however, the number of adsorbedcations per PC depends roughly linearly onthe available surface area (Fig. 3A) produc-ing an approximately constant surface chargedensity of ∼1.06 e/nm2 in the present sim-ulations (Fig. 3B). Notably, even a dipalmi-toylphosphatidylcholine (DPPC) bilayer in the

5

gel-phase follows the same observed trend forboth cations. The effect of increased tempera-ture on both sodium and calcium adsorption issimilar but somewhat more complex (see Sec-tion S4.2, ESI†).The differences between Na+ and Ca2+ re-

ported by the present experiments and simu-lations are explained by the different natureof their binding to the membrane. Whereassodium has a low binding energy, calcium bind-ing is nearly irreversible. Our simulationsalso suggest that while sodium shows rapidexchange between the membrane surface andthe bulk solvent, calcium saturates at the sur-face relatively slowly (see Section S5.3, ESI†),and desorption events are rare. In both cases,the amount of phosphatidylcholine lipids de-termines the number of possible binding sites,however, real competition for them only ex-ists for Ca2+. With Na+, most binding sitesare essentially free or only transiently occupied.At these conditions, only the bulk Na+ den-sity and the rather constant surface PC den-sity determine the number of surface boundNa+. This finding indicates that the Na+ con-centration would be enhanced in ordered mem-brane domains with higher density of PC headgroups, i.e. binding sites. This is indeed ob-served in heterogeneous ternary systems (seeSection S4.1, ESI†).In contrast to sodium, the strong binding of

calcium results in a practically complete cov-erage of the membrane surface area at the rel-atively high calcium concentrations employedhere. In our simulations, we find that eachcalcium is coordinated on average to three PCphosphate groups (Section S5.4, ESI†). Thisstrong coordination at the head group levelslows down lipid rotation (Section S4.3, ESI†)and self-diffusion (Section S5.1, ESI†).An increase in cholesterol concentration leads

to higher calcium adsorption even thoughcholesterol itself neither interacts with theions nor does it affect the penetration depth.Cholesterol thus does not seem to take part inthe calcium coordination and the number of co-ordinated phosphates is unaffected by the pres-ence of cholesterol. These results indicate thatcholesterol simply allows more calcium adsorp-

tion by increasing the membrane area maintain-ing a fairly constant critical surface charge den-sity. This unspecific role is emphasized by thefact that other ways of affecting the membranearea (e.g. by increasing lipid unsaturation orby taking the membrane to a gel phase) leadto similar observations. This suggests that inthis case the charge–charge repulsion betweenCa2+ is the limiting factor of adsorption. A de-tailed explanation of all these observations canbe found in the ESI† (Section S5).

0.63 0.66 0.69 0.72

Cholesterol as aspacer

σ+

N+Ca2+ Na+

Area per PC (nm2)

Are

apr

obab

ility

Num

ber

|Den

sity

Figure 4: Schematic picture describing the ef-fect of cholesterol on cation adsorption. Blackand red curves show the distribution of the areaper phospholipid in POPC membranes with 0%and 30% cholesterol, respectively. Green andblue lines show data for the effect of cholesterol-induced area increase on calcium and sodiumadsorption respectively; solid lines show thenumber of adsorbed ions (N+), while the dashedlines show the resulting surface charge density(σ+). The number of bound calcium ions is pro-portional to the total area, resulting in a con-stant surface charge density. The number ofsodium ions, on the other hand, is proportionalto the number of PC head groups, resulting ina surface charge density that is proportional tothe area per PC lipid. Note that this figure de-picts trends, whereas the data is properly shownin Figs. 2 and 3.

In conclusion, we have performed fluorescentmeasurements in POPC membranes with in-creasing cholesterol and sodium or calcium con-centrations and complemented these results bysimulating analogous systems using different

6

lipid and ion force fields with the aim to un-ravel the effect of cholesterol on Na+ and Ca2+cation adsorption onto lipid bilayers. We findthat the interaction of Na+ with the membranedepends on the total number of PC headgroupspresent, which determines the number of bind-ing sites. In contrast, the level of adsorbed cal-cium is determined primarily by the accessiblesurface area of the membrane.This significantly different behavior of the

two cations results from the weak and tran-sient sodium binding with the lipid head groupsthat contrasts with the stronger calcium ad-sorption that is capable of inducing the cluster-ing of phosphatidylcholine head groups. Thereis quantitative agreement between simulationsand experiments concerning these observations,which are graphically summarized in Fig. 4.The present findings indicate that a cell,

which can control its global and local mem-brane lipid composition, may use this abilityto selectively modulate the membrane interac-tion with Na+ and Ca2+ by inducing lateral het-erogeneity (regions of varying lipid packing) orcontrolling its volume. Taking into considera-tion the important role of these two cations inthe regulation of cellular functions, the mech-anisms explained here could be exploited bythe cell and may constitute a hitherto unrecog-nized regulation mechanism. The present find-ings also help us to understand how cations in-teract with cholesterol-rich ordered regions inheterogeneous membranes. This informationmay prove crucial for understanding the roleof ions in modulating the adsorption of chargedmolecules onto the membrane surface.We thank CSC – IT Center for Science (Es-

poo, Finland) for computing resources. HMSand MJ thank the Academy of Finland (Centreof Excellence program) and the European Re-search Council (Advanced Grant CROWDED-PRO-LIPIDS) for financial support. PJun ac-knowledges support from the Czech ScienceFoundation (grant no. 16–01074S) and theAcademy of Finland via the FiDiPro program.AM, PJur and MH acknowledge the Czech Sci-ence Foundation (grant no. 17-03160S) and theCzech Academy of Sciences for the PraemiumAcademiae award.

References(1) Jentsch, T. J. VRACs and Other Ion Channels and Trans-

porters in the Regulation of Cell Volume and Beyond. Na-ture Reviews Molecular Cell Biology 2016, 17, 293–307.

(2) Sten-Knudsen, O. Biological Membranes: Theory of Trans-port, Potentials and Electric Impulses; Cambridge Univer-sity Press, New York, 2002; ISBN: 9780521036351.

(3) Andersen, O. S. Encyclopedia of Metalloproteins; SpringerScience + Business Media, 2013; pp 580–587.

(4) Hu, K.; Carroll, J.; Fedorovich, S.; Rickman, C.; Sukho-dub, A.; Davletov, B. Vesicular Restriction of Synapto-brevin Suggests a Role for Calcium in Membrane Fusion.Nature 2002, 415, 646–650.

(5) Tarafdar, P. K.; Chakraborty, H.; Dennison, S. M.;Lentz, B. R. Phosphatidylserine Inhibits and Calcium Pro-motes Model Membrane Fusion. Biophysical Journal 2012,103, 1880–1889.

(6) Berridge, M. J.; Bootman, M. D.; Roderick, H. L. CalciumSignalling: Dynamics, Homeostasis and Remodelling. Na-ture Reviews Molecular Cell Biology 2003, 4, 517–529.

(7) Orrenius, S.; Zhivotovsky, B.; Nicotera, P. Regulation ofCell Death: The Calcium–Apoptosis Link. Nature ReviewsMolecular Cell Biology 2003, 4, 552–565.

(8) Szekely, O.; Steiner, A.; Szekely, P.; Amit, E.; Asor, R.;Tamburu, C.; Raviv, U. The Structure of Ions and Zwitte-rionic Lipids Regulates the Charge of Dipolar Membranes.Langmuir 2011, 27, 7419–7438.

(9) Antipina, A. Y.; Gurtovenko, A. A. Molecular mechanismof calcium-induced adsorption of DNA on zwitterionic phos-pholipid membranes. The Journal of Physical Chemistry B2015, 119, 6638–6645.

(10) McManus, J. J.; Rädler, J. O.; Dawson, K. A. Does Cal-cium Turn a Zwitterionic Lipid Cationic? The Journal ofPhysical Chemistry B 2003, 107, 9869–9875.

(11) Jurkiewicz, P.; Cwiklik, L.; Vojtíšková, A.; Jung-wirth, P.; Hof, M. Structure, Dynamics, and Hydration ofPOPC/POPS Bilayers Suspended in NaCl, KCl, and CsClSolutions. Biochimica et Biophysica Acta (BBA) - Biomem-branes 2012, 1818, 609–616.

(12) Melcrová, A.; Pokorna, S.; Pullanchery, S.; Kohagen, M.;Jurkiewicz, P.; Hof, M.; Jungwirth, P.; Cremer, P. S.; Cwik-lik, L. The Complex Nature of Calcium Cation Interac-tions With Phospholipid Bilayers. Scientific Reports 2016,6, 38035.

(13) Catte, A.; Girych, M.; Javanainen, M.; Loison, C.;Melcr, J.; Miettinen, M. S.; Monticelli, L.; Määttä, J.;Oganesyan, V. S.; Ollila, O. S. et al. Molecular Electrometerand Binding of Cations to Phospholipid Bilayers. PhysicalChemistry Chemical Physics 2016, 18, 32560–32569.

(14) Jones, M. N. The Surface Properties of Phospholipid Li-posome Systems and Their Characterization. Advances inColloid and Interface Science 1995, 54, 93–128.

(15) Melcr, J.; Bonhenry, D.; Timr, Š.; Jungwirth, P. Transmem-brane Potential Modeling: Comparison Between Methodsof Constant Electric Field and Ion Imbalance. Journal ofChemical Theory and Computation 2016, 12, 2418–2425.

(16) Saheki, Y.; Bian, X.; Schauder, C. M.; Sawaki, Y.;Surma, M. A.; Klose, C.; Pincet, F.; Reinisch, K. M.;Camilli, P. D. Control of Plasma Membrane Lipid Home-ostasis by the Extended Synaptotagmins. Nature Cell Biol-ogy 2016, 18, 504–515.

(17) Yang, J.; Bonomi, M.; Calero, C.; Martí, J. Free EnergyLandscapes of Sodium Ions Bound to DMPC-CholesterolMembrane Surfaces at Infinite Dilution. Physical ChemistryChemical Physics 2016, 18, 9036–9041.

(18) Magarkar, A.; Dhawan, V.; Kallinteri, P.; Viitala, T.; El-mowafy, M.; Róg, T.; Bunker, A. Cholesterol Level AffectsSurface Charge of Lipid Membranes in Saline Solution. Sci-entific Reports 2014, 4, 1–5.

(19) Jurkiewicz, P.; Cwiklik, L.; Jungwirth, P.; Hof, M. Lipid Hy-dration and Mobility: An Interplay Between FluorescenceSolvent Relaxation Experiments and Molecular DynamicsSimulations. Biochimie 2012, 94, 26–32.

7

(20) Parasassi, T.; De Stasio, G.; d’Ubaldo, A.; Gratton, E.Phase Fluctuation in Phospholipid Membranes Revealed byLaurdan Fluorescence. Biophysical Journal 1990, 57, 1179–1186.

(21) Amaro, M.; Sachl, R.; Jurkiewicz, P.; Coutinho, A.; Pri-eto, M., M.and Hof Time-Resolved Fluorescence in Lipid Bi-layers: Selected Applications and Advantages Over SteadyState. Biophysical Journal 2014, 107, 2751–2760.

(22) Böckmann, R. A.; Grubmüller, H. Multistep Binding of Di-valent Cations to Phospholipid Bilayers: A Molecular Dy-namics Study. Angewandte Chemie International Edition2004, 43, 1021–1024.

(23) Róg, T.; Pasenkiewicz-Gierula, M.; Vattulainen, I.; Kart-tunen, M. Ordering effects of cholesterol and its analogues.Biochimica et Biophysica Acta (BBA)-Biomembranes2009, 1788, 97–121.

8