E.
Drolle
ab,
W. F. D.
Bennett
c,
K.
Hammond
d,
E.
Lyman
e,
M.
Karttunen
f and
Z.
Leonenko
*abd
aDepartment of Biology, University of Waterloo, Canada. E-mail: zleonenk@uwaterloo.ca
bWaterloo Institute of Nanotechnology, University of Waterloo, Canada
cDepartment of Physics and Astronomy and Department of Chemistry and Biochemistry, University of California, Santa Barbara, USA
dDepartment of Physics and Astronomy, University of Waterloo, Canada
eDepartment of Physics and Astronomy, 217 Sharp Lab, Newark, USA
fDepartment of Mathematics and Computer Science & Institute for Complex Molecular Systems, Eindhoven University of Technology, MetaForum, 5600 MB Eindhoven, the Netherlands
First published on 22nd November 2016
The molecular arrangement of lipids and proteins within biomembranes and monolayers gives rise to complex film morphologies as well as regions of distinct electrical surface potential, topographical and electrostatic nanoscale domains. To probe these nanodomains in soft matter is a challenging task both experimentally and theoretically. This work addresses the effects of cholesterol, lipid composition, lipid charge, and lipid phase on the monolayer structure and the electrical surface potential distribution. Atomic force microscopy (AFM) was used to resolve topographical nanodomains and Kelvin probe force microscopy (KPFM) to resolve electrical surface potential of these nanodomains in lipid monolayers. Model monolayers composed of dipalmitoylphosphatidylcholine (DPPC), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1,2-dioleoyl-sn-glycero-3-[phospho-rac-(3-lysyl(1-glycerol))] (DOPG), and cholesterol were studied. It is shown that cholesterol changes nanoscale domain formation, affecting both topography and electrical surface potential. The molecular basis for differences in electrical surface potential was addressed with atomistic molecular dynamics (MD). MD simulations are compared the experimental results, with 100 s of mV difference in electrostatic potential between liquid-disordered bilayer (Ld, less cholesterol and lower chain order) and a liquid-ordered bilayer (Lo, more cholesterol and higher chain order). Importantly, the difference in electrostatic properties between Lo and Ld phases suggests a new mechanism by which membrane composition couples to membrane function.
Though the plasma membrane contains many different lipid species, the largest fraction is cholesterol, which typically comprises 20–30% by mole. Cholesterol is known to broaden the liquid to gel phase transition, forming an intermediate “liquid-ordered” (Lo) phase at sufficiently high concentration.9 Cholesterol tends to order saturated chains, reducing the area per phospholipid and thickening the membrane while maintaining the fluid phase.10 Addition of a lower melting temperature lipid introduces a region of Lo/Ld coexistence.11
Ternary mixtures have been studied extensively over the past decade. The most well-characterized is DPPC/DOPC/Chol, as this mixture forms micron scale domains and can be imaged directly by light microscopy.12,13 MD simulations support a model in which cholesterol associates preferentially with saturated acyl chains over unsaturated ones.14–19 Atomistic MD simulations of the DPPC–DOPC–Chol system have found cholesterol to arrange itself normal to the bilayer surface (upright) at sufficiently high concentrations.14 More recently, long timescale simulations of the same mixture revealed nanoscale structure within the Lo phase.20 A coarse-grained MD simulation of DPPC–DOPC–Chol bilayer showed fast partitioning of the mixture into raft-like domains, with the liquid-ordered domains (DPPC) being highly enriched in cholesterol and thicker than the surrounding DOPC regions.16,21 Monolayer simulations have shown phase separation similar to the bilayer model, again with cholesterol enrichment in the DPPC phase, at surface pressures from 1–30 mN m−1.17 However, electrostatic effects have been comparatively understudied, and yet offer an obvious mechanism for targeting proteins to specific regions of membranes. Many membrane proteins are directed to the membrane by nonspecific electrostatic interactions.22
Direct imaging of lipid bilayers on the nanoscale is difficult due to their soft nature—they are held together by hydrophobic interactions within a 2 molecule thick membrane (∼5 nm) solvated by two chemically diverse water interfaces. Atomic force microscopy (AFM) is one of the few methods which provides direct, nanoscale information on membrane organization.23–27 Of particular interest to the present work is Kelvin probe force microscopy (KPFM), a high-resolution, sensitive method for direct measurement of surface potential, recently used to map surface potentials in model lipid systems.28,29 For example, the presence of cholesterol causes electrostatic domain formation in a pulmonary surfactant lipid mixture containing DPPC and anionic lipids.28,29
In the following, KPFM is used to characterize cholesterol induced electrostatic domains in model systems. Combination of AFM with KPFM allows for measuring of both topographical and electrostatic domains. Atomistic MD simulations used to model these domains, support the experimental results, demonstrate that the results extend to bilayer systems, and provide insight into the mechanistic details responsible for the electrostatic domains.
Lipids | Ratio (w/w) |
---|---|
DPPC–DOPC | 607![]() ![]() |
DPPC–DOPC–Chol | 560![]() ![]() ![]() ![]() |
DPPC–DOPG | 50![]() ![]() |
DPPC–DOPG–Chol | 40![]() ![]() ![]() ![]() |
We used small lipid bilayers as mimics for the bulk phases of the AFM experiments. The gel-phase bilayer was 72 DPPC lipids (only the last 50 ns were used for analysis; after it reached equilibrium in the gel-phase). The DOPC:
DPPC
:
Chol ratio for the Ld (42
:
20
:
6) and Lo (10
:
38
:
20) phase was taken from the ternary phase diagram. Simulations were also performed with DOPG lipids in compositions of 44
:
20
:
6 and 10
:
38
:
20 DOPG
:
DPPC
:
Chol in order to investigate the influence of headgroup charge on the potential, with Na+ to neutralize the charge of the PG head groups.
The electrostatic potential (Ψ) across the bilayer was determined by integrating the charge distribution (ρ) twice according to Poisson's equation:
Boundary conditions were chosen such that the electric field and potential are zero at the center of the bilayer.42 After computing the potential, the curves were shifted to be equal to zero in bulk water, as we are interested in the difference in potential between water the membrane center. For the electrostatic potential decompositions, the charge distribution is calculated for a particular chemical group (e.g., water) which is then numerically integrated to obtain the potential.43
Fig. 2A presents results for a DPPC–DOPC mixture with and without cholesterol. In the absence of cholesterol, domains are observed both in AFM and KPFM. The difference in height between these domains (Δh, visualized in Fig. 1) averaged to 1.22 ± 0.03 nm. It is likely that higher domains correspond to DPPC molecules, as in their gel phase, the tail groups are more ordered and thus are slightly thicker than the disordered tail groups of fluid phase (also known as liquid disordered, Ld, phase) DOPC.44 These higher domains were much larger in size than the lower domains, reaching lateral dimensions of up to 570 nm in length (i.e. X) and 775 nm in width (i.e. Y); the lower domains were less prevalent in the monolayer, with average X dimensions of 150 nm and Y dimensions of up to 350 nm. The domains also have distinct electrical surface potentials. We analyzed the difference in the electrical surface potential (ΔV) between the higher and lower domains (visualized in Fig. 1). The average ΔV of the electrostatic domains was determined to be 41.6 ± 5.39 mV (Fig. 2B).
With the addition of cholesterol to the DPPC–DOPC system, smaller topographical domains on the monolayer samples (Fig. 2C) with an average Δh of the domains of 0.97 ± 0.06 nm were observed. This average Δh is slightly lower than that observed in the DOPC–DPPC system without cholesterol (1.22 nm). This is consistent with the idea that cholesterol causes slight disorder in the already ordered tail groups of gel phase lipid molecules such as DPPC by creating an intermediate Lo phase.16,44,45 The idea that cholesterol molecules are clustered in higher concentration near the DPPC molecules than the DOPC molecules agrees with the idea that cholesterol prefers to interact with saturated rather than unsaturated lipids.44 It also agrees with many simulation studies that show cholesterol molecules organized in higher concentration with DPPC molecules rather than the DOPC molecules.15
The corresponding surface potential image (Fig. 2D) shows distinct electrostatic domains where higher topographical domains correspond to areas of higher electrical surface potential. These electrostatic domains had an average ΔV of 67.25 ± 7.03 mV. This difference is larger than that observed in the DOPC–DPPC–Chol system, which had an average ΔV of 41.6 mV, supporting the idea that cholesterol has a measurable effect on the V of the mixed lipid system.
Fig. 3 shows snapshots of the model bilayer systems that were simulated. To mimic the DPPC–DOPC, gel–liquid phase separated monolayer system, a DPPC bilayer simulation was run at 298 K for 300 ns, equilibrating to a gel-phase. The straight and aligned tails in the gel phase are clearly visible in Fig. 3. A DOPC bilayer also at 298 K was run for 100 ns, equilibrating to a fluid phase. The compositions of the ternary mixtures in the presence of cholesterol are described in Methods. The Na+ ions, rendered as green spheres in Fig. 3, bind to the negatively charged DOPC head groups.
Fig. 4 compares the electrostatic potential and mass density profile for the gel DPPC and liquid DOPC bilayer. For the gel bilayer, due to tight packing and slow dynamics, two distinct density peaks are observed for the gel–DPPC bilayer head group region, whereas for the liquid DOPC there is a smooth distribution for the head groups. Qualitatively the simulations match the AFM results; the gel bilayer is thicker than the liquid DOPC phase. Simulations show a sharp drop in density near the bilayer center, because the two gel leaflet tails cannot interdigitate. This is, however, somewhat complicated by the tilt of the DPPC hydrocarbon chains in the gel phase, Fig. 3.
The electrostatic profile (Fig. 4C) is shifted to longer distances for the gel bilayer, due to the increased bilayer thickness. The potential is higher for the gel bilayer, until right at the bilayer center, where the density drops, and the potential for the liquid DOPC becomes slightly larger. This is different than the KPFM results on the DPPC–DOPC phase separated monolayer systems, where the gel DPPC system had a larger potential compared to DOPC. We speculate that the discrepancy could be due to differences between a bilayer and monolayer system, given that the coupling of the gel leaflets is not present in a monolayer. The gel DPPC electrostatic potential is larger than DOPC, until right at the low density bilayer center. Despite similar total electrostatic potentials, the components are quite different, with ∼1 V difference (although we note that these contributions are not ‘felt’ by the system due to explicit screening of the charges, to yield the total electrostatic potential). Water at the interface is ordered resulting in a positive dipole potential that offsets the large negative dipole potential at the bilayer center from the negative phosphate group's position relative to the positive choline group.
Fig. 5 shows electrostatic and density profiles across cholesterol mixtures, a 10:
38
:
20 (DOPC
:
DPPC
:
Chol) for the Lo phase, and 42
:
20
:
6 at 298 K for the Ld phase, both at 298 K. Cholesterol is known to order saturated lipids, resulting in the Lo phase, which has structural order similar to the gel phase, but is still liquid with relatively fast lateral diffusion of lipids compared to the gel phase. Density profiles show that the bilayer is thicker for the Lo bilayer, but the Lo bilayer's density is more similar to the Ld bilayer than the gel system, without distinct phosphate and choline peaks. Due to the thicker density, the potential profile is shifted out towards water for the Lo and gel phase. There is also a small trough in the electrostatic potential near the center of the Lo similar to the gel phase systems. The electrostatic potential at the center of the Lo phase (1.11 V) is higher compared to the Ld phase (1.03 V), by a difference of 80 mV. Considering the limitations of fixed charge models for bilayer electrostatics, this difference is very similar to the measurements of differences in electrostatic surface potential of 35 to 70 mV obtained by KPFM (Table 2). Because these are bilayers compared to supported monolayers for the KPFM results, we would not expect perfect agreement. Of note, the electrostatic contributions from water and the lipid bilayer are very similar for the Lo and Ld bilayers, as opposed to the gel bilayer (Fig. 4).
Average difference in height Δh (nm) | Average difference in electrical surface potential ΔV (mV) | |
---|---|---|
DPPC–DOPC | 1.22 ± 0.03 | 41.60 ± 5.39 |
DPPC–DOPC–Chol | 0.97 ± 0.06 | 67.25 ± 7.03 |
Effect of cholesterol = decrease in Δh, increase in ΔV | ||
DPPC–DOPG | 0.80 ± 0.04 | 138.67 ± 6.79 |
DPPC–DOPG–Chol | 1.18 ± 0.06 | 117.55 ± 8.65 |
Effect of cholesterol = increase in Δh, decrease in ΔV |
Comparison between the monolayer and bilayer systems is not, however, straightforward.46 Since the monolayer DPPC–DOPC–Chol systems were prepared at a relatively high surface pressure of 35 mN m−1 and low cholesterol concentration corresponding to the so-called α-region,46,47 it is likely that they are in the gel, or liquid-condensed phase. This does not directly correspond to the Lo phase. Although beyond the scope of the current study, direct comparison (to Lo) would require surface pressures around or below 15 mN m−1.47 We used the compression of 35 to mimic physiological compressions reported for the membrane.48–50 We can, however, make the following conclusions: consistent with experiments, the lipid tail in the gel phase of the pure DPPC display the characteristic tilt51 which is absent in the Lo phase mixtures when cholesterol is interacting with them; addition of cholesterol or DOPC leads to straightening of the DPPC hydrocarbon chains even in the gel phase.52
Fig. 6A shows the topography for the pure DPPC–DOPG system. Distinct differences in topography are observed that likely arise due to the phase separation of gel phase DPPC and fluid phase DOPG. We assume that higher domains are saturated with gel phase DPPC and lower domains are saturated with fluid phase DOPG. The average difference in height, Δh, between these domains was determined to be 0.80 nm ± 0.04 nm. The corresponding KPFM image shows differences in potential (Fig. 6B), average ΔV is −138.67 ± 6.79, Table 2.
In the DPPC–DOPG–Chol image larger domains are observed than in the cholesterol-free sample (Fig. 6C). These surface features have an average Δh of 1.18 ± 0.06 nm. There also appear to be areas of the sample consistent with multilayer formation. This feature is normally seen at higher pressures, when the lateral pressure exerted on the lipid monolayer causes the monolayer to buckle and fold, alleviating the high pressure.
The corresponding FM-KPFM image shows electrostatic domains in areas that correlate with topographical domains (Fig. 6D). Though cholesterol's effect on the ΔV – an average of −117.55 ± 8.65 mV for this sample – is only slightly smaller than the −138.67 mV for the DPPC–DOPG sample, Table 2, it does show that cholesterol influences surface potential in lipid systems containing charged head groups.
In order to make comparison with simulations we need to compare the electrical surface potential between DPPC/Chol phase and DOPG/Chol phase, assuming that domains we observed are the result of DPPC–DOPG demixing and that cholesterol is present in each lipid phase. The electrostatic profile for the PG bilayers are slightly different than the other systems, with a small negative potential near the water interface (Fig. 5B), and a significantly lower potential at the center of the bilayer. For pure DPPC (gel) and DOPG (fluid) bilayers, the electrostatic potential difference is 40 mV. This result is different from experiments, most likely due to effect of counter-ions that were accounted for in bilayer system, in MD simulations. In contrast, KPFM experiments are done on supported lipid monolayer, which was deposited on mica by Langmuir–Blodgett deposition from the water–air interface, therefore there are likely less counter-ions present in this system.
Comparing the electrostatic potential between the PG-Ld bilayer (0.89 V) and the PG-Lo phase (0.96 V), we find a difference of 70 mV. Because the PG lipid head group has a net charge, and Na+ ions were present to keep the system neutral, the electrostatic contributions from these groups are large. The contribution from water is similar for the PG-Lo and PG-Ld bilayers (∼11 V), which is considerably larger than the Lo (∼4 V), indicating strong ordering of water at the interface of the charged lipid system. Despite the large negative surface charges from PG, there is only a small negative total electrostatic potential at the interface, due to the accumulation of Na+ on the surface to shield the head group charges; very recent studies indicate the need to improve the ionic force fields used in lipid simulations.53 Similar concerns regarding the lipid phases of the monolayers and bilayers as discussed in connection with the mixed DPPC–DOPC–Chol exist. When PG lipids are present, uncertainties in the ion–PG interactions contribute to the quantitative discrepancies.53
The results of MD simulations are summarized in Table 3. Despite the discrepancies, both results are valuable in assessment of complex matter of understanding of electrical surface potential in lipid monolayers and bilayers.
Lipid composition | Calculated potential (V) |
---|---|
DOPC–DPPC–Chol (Ld) | −1.03 |
DOPC–DPPC–Chol (Lo) | −1.11 |
DOPG–DPPC–Chol (Ld) | −0.89 |
DOPG–DPPC–Chol (Lo) | −0.96 |
DPPC (gel) | −0.89 |
DOPG | −0.85 |
DOPC | −1.03 |
Using MD simulations, it is possible to measure the effect that cholesterol has on a single molecule's electrical surface potential or on a single component lipid system's membrane dipole potential, as seen in Starke-Peterkovic et al.'s work;54 however, when it comes to multiple molecules, it becomes a much more complex issue. This is the case with domains within the membranes, which could contain thousands of interacting molecules. This work presents a first approach where experimental AFM and KPFM imaging were combined with MD simulations to address both topographical and electrostatic nanodomains in the multicomponent lipid systems. Extending this approach to more complex mixtures should help determine whether such electrostatic domains are a feature of the plasma membrane of intact cells.
This journal is © The Royal Society of Chemistry 2017 |