Sub-ten-nanometer heterogeneity of solid supported lipid membranes determined by solution atomic force microscopy

Biochimica et Biophysica Acta 1858 (2016) 181–188

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbamem

Sub-ten-nanometer heterogeneity of solid supported lipid membranes determined by solution atomic force microscopy Chian Sing Ho, Nawal K. Khadka, Jianjun Pan ⁎ Department of Physics, University of South Florida, Tampa, FL 33620, USA

a r t i c l e

i n f o

Article history: Received 25 August 2015 Received in revised form 19 October 2015 Accepted 5 November 2015 Available online 6 November 2015 Keywords: Membrane rafts Nanodomain Critical fluctuations Nanoscopic Phase coexistence Nanoheterogeneity

a b s t r a c t Visually detecting nanoscopic structures in lipid membranes is important for elucidating lipid–lipid interactions, which are suggested to play a role in mediating membrane rafts. We use solution atomic force microscopy (AFM) to study lateral and normal organization in multicomponent lipid membranes supported by mica substrate. Nanoscopic heterogeneity is observed in a three-component system composed of 1-palmitoyl-2-oleoyl-snglycero-3-phosphocholine (POPC)/brain-sphingomyelin (bSM)/cholesterol (Chol). We find sub-ten-nanometer correlation lengths that are used to describe membrane lateral organization. In addition, we find that the correlation length is independent on cholesterol concentration, while the height fluctuation (variation) is not. To explore the mechanism that controls the size of membrane heterogeneity, we extend our study to a fourcomponent system composed of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC)/POPC/bSM/Chol. By systematically adjusting the relative amount of DOPC and POPC, we obtain macroscopic-to-nanoscopic size transition of membrane heterogeneity. In contrast to the results from vesicle based fluorescence microscopy, we find that the structural transition is continuous both in the lateral and normal directions. We compare our nanoscopic structures to two theoretical models, and find that both the critical fluctuations and the nanodomain models are not sufficient to account for our solution AFM data. Finally, we propose a nanoheterogeneity model that could serve as the organization principle of the observed nanoscopic structures in multicomponent lipid membranes. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Nonrandom lateral organization is a hallmark of the plasma membrane of eukaryotic cells [1]. Accumulating evidences point to the existence of nanoscopic spatial heterogeneity, or the so-called membrane rafts, in living cells. Since the conceptualization, membrane rafts have been proposed to play an array of biological functions, including signaling, membrane trafficking, viral entry and budding, and amyloid catalysis, to name a few. Most of the evidence supporting the existence of membrane rafts is based on indirect spectroscopy techniques. More recently, super-resolution microscopy revealed hindered diffusion [2] and transient trapping [3] of fluorophore-tagged molecules in living cells. Despite the compelling evidences, controversies remain pertaining to membrane rafts size, shape, and dynamics, or even their definitive existence has been challenged [4]. For example, a recent experiment employing super-resolution stimulated emission depletion microscopy and scanning fluorescence correlation spectroscopy (STED–FCS) showed that the diffusion behavior of fluorescent phospholipid and cholesterol analogues is homogeneous in living cells [5]. It is clear that further elucidation of membrane rafts and their exact nature requires direct visual evidence. ⁎ Corresponding author. E-mail address: [email protected] (J. Pan).

http://dx.doi.org/10.1016/j.bbamem.2015.11.001 0005-2736/© 2015 Elsevier B.V. All rights reserved.

Simplified models employing a few lipid species have proven valuable to elucidate key properties associated with the heterogeneous organization in cell plasma membranes [6–8]. These models often contain a high-melting (high-Tm) lipid, a low melting (low-Tm) lipid, and cholesterol (Chol). Indeed, micron-scale liquid-ordered (Lo) and liquid-disordered (Ld) phase coexistence has been observed in a variety of ternary mixtures containing 1,2-dioleoyl-sn-glycero-3phosphocholine (DOPC) as the low-Tm lipid, see Refs. [9,10]. The popularity of these mixtures is highlighted by the readily observable micronscale domains using a light microscope. A more biologically relevant system is obtained by substituting DOPC with 1-palmitoyl-2-oleoylsn-glycero-3-phosphocholine (POPC) as the low-Tm lipid. Although optically homogeneous, mixtures containing POPC are often heterogeneous at the nanometer-scale as indicated by many spectroscopy measurements. To translate the nanoscopic heterogeneity inferred from model membranes to rafts in living cells, several models have been proposed to describe the nanoscopic structures present in lipid membranes. Among them the critical fluctuations and the nanodomain models have gained particular interest. Both models can successfully explain size transition from a few nanometers to microns. However, the critical fluctuations model cannot account for nanoscopic heterogeneity implied in a large compositional space of POPC containing membranes, while the nanodomain model suffers from the lack of visual evidence. In addition, it is known that nanomaterials often exhibit distinct properties

182

C.S. Ho et al. / Biochimica et Biophysica Acta 1858 (2016) 181–188

from their bulk counterparts. This raises the question as to whether micron-sized domains can be faithfully translated into nanodomains. Liquid compatible atomic force microscopy (AFM) is uniquely suited to explore lateral organization of planar lipid membranes [11–20]. Height contrast down to sub-angstrom enables AFM to detect structural features ranging from micron to nanometer sizes. In this paper we first use AFM to study composition dependent nanoscopic structures in a three-component system composed of POPC/brain-sphingomyelin (bSM)/Chol. Nanoscopic compositional heterogeneity is directly seen for the first time. Calculations of the height weighted pair distribution functions indicate that the heterogeneous structures have a correlation length of ~5–10 nm. To explore the mechanism of size transition in heterogeneous lipid membranes, we then investigate a four-component system composed of DOPC/POPC/bSM/Chol. Macroscopic-to-nanoscopic size transition is identified by gradually substituting DOPC with POPC. Based on the obtained nanoscopic structures in two multicomponent lipid systems, we propose a new model, the nanoheterogeneity model, to act as the organization principle of the nanoscopic structures in multicomponent lipid membranes. 2. Materials and methods All lipid compositions are presented in mole ratio, fraction, or percentage. POPC, DOPC, bSM, egg sphingomyelin (eSM), rhodamineDPPE (1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine), and cholesterol were purchased as lyophilized powders from Avanti Polar Lipids (Alabaster, AL). 2.1. Atomic force microscopy Stock solutions were prepared by dissolving lipids in organic solvents (chloroform for POPC, DOPC, and cholesterol, and chloroform/ methanol 3:1 for bSM and eSM). Lipid mixtures were prepared by mixing appropriate ratios of stock solutions in 13 ml glass test tubes. Organic solvents were removed by a gentle stream of argon gas using a 12-position N-EVAP evaporator (Organomation Associates, Inc.), and then vacuumed for N2 h. Lipid dry films were hydrated in ultrapure water and ultrasonicated using a Sonic Dismembrator at 40 W for 12 min. The obtained small unilamellar vesicles (SUVs) were centrifuged for 15 min at 13,000 rpm. Dynamic light scattering measurement using a Dynapro Nanostar (Wyatt Technology, Santa Barbara, CA) indicates that the SUVs have an average diameter of ~30 nm. Solution AFM height images (at room temperature) were acquired using a Multimode 8 AFM (Bruker, Santa Barbara, CA) coupled with a Nanoscope V controller. A bungee cord supported platform sitting on a vibration isolated optical table was used to enhance system resolution (signal-to-noise ratio). After incubating SUVs (~ 0.2 mg/ml of total lipids) in the AFM liquid cell for N30 min, either at room temperature or at 50 °C using a heating accessory (Bruker model: MMHC-A60), a solid supported planar bilayer was formed by vesicle fusion onto a freshly cleaved mica substrate [21–23]. A unique mode referred to as the PeakForce quantitative nanomechanics (QNM) in liquid was used for data collection. A special Si3N4 cantilever designed to work with the PeakForce QNM mode (Bruker model: ScanAsyst-Fluid +) was used for bilayer scanning. The sensitivity of the cantilever (i.e., deflection versus applied voltage) was determined by deflection measurement on a mica film; the spring constant of the cantilever was determined using thermal oscillation (the method is built into the AFM software). Multiple square images (≥ 3) at different locations of the bilayer surface were acquired at a scan rate of 0.5–1.0 Hz. For control, we also scanned samples prepared at different days (e.g., POPC/ bSM 3:2 + 20% Chol). The resulting bilayer structures are very similar for the same lipid composition. The peak force of each scan was set at ~300–600 pN depending on bilayer stiffness. (Note that the peak force does not affect bilayer topology.) AFM height images were leveled by subtracting a linear or second order polynomial background. The

leveled images were used to calculate the height weighted pair distribution function [24,25]: GðRÞ ¼ hhðrÞ  hðr þ RÞi  hhðrÞihhðr þ RÞi

ð1Þ

where brackets denote ensemble average over the radial distance r. Correlation length ξ corresponds to the exponential decay length of G(R). 2.2. Fluorescence microscopy DOPC/eSM/Chol mixtures were prepared from stock solutions in the same manner as in AFM experiment. For fluorescence imaging, each lipid mixture contains 0.2% of the Ld phase marker, rhodamine-DPPE. Lipid mixtures were deposited onto ITO-coated glass slides and dried under vacuum for N2 h. Giant unilamellar vesicles (GUVs) in 100 mM sucrose were generated by swelling under an AC field of 10 Hz and 2.0 V (60 °C). Electroformed GUVs were dispersed in 100 mM glucose. After settling for N1 h, GUVs were transferred into a silicone gel well framed by a cover slip and a glass slide. Fluorescence images (at room temperature) were collected using an inverted microscope (Nikon Eclipse Ti-U), a CFI Super Fluor ELWD 60× objective, and an EM-CCD camera (Andor iXon 897). 3. Results and discussion 3.1. Three-component lipid membranes Many indirect measurements have suggested the existence of nanoscopic structures in POPC/sphingomyelin (SM)/Chol. Domain sizes were proposed to range from hundreds to a few nanometers [26–31]. We use AFM to directly visualize solid supported planar bilayer structures composed of POPC/bSM/Chol. We first fix cholesterol concentration at 20% while varying the ratio of POPC/bSM. To span a large compositional space, eight ratios of POPC/bSM are chosen (Supporting Information, Fig. S1); the corresponding AFM topographic images are shown in Fig. 1. Heterogeneous structures with enhanced height variations (amplitude of ~0.2 nm) are observed when the ratio of POPC/bSM is near 1:1. Bilayers become exceedingly smooth at POPC/bSM ratios of 5:1 and 1:3. To quantitatively characterize the structural transition as a function of POPC/bSM ratio, we calculate height probability distributions (Supporting Information, Fig. S2). By fitting to Gaussian functions, we find that the full width half maximum (FWHM) is 0.09, 0.10, 0.11, 0.16, 0.16, 0.09, 0.08, and 0.08 nm for POPC/bSM ratio at 5:1, 3:1, 2:1, 4:3, 1:1, 3:4, 1:2, and 1:3, respectively. The obtained FWHM values are consistent with our qualitative description of bilayer heterogeneity based on the height images and profiles (Fig. 1). In addition, the sub-angstrom values of FWHM highlight the good resolution of our experimental setup in the normal direction. Overall, our AFM data indicate that bilayers are more heterogeneous when the composition of POPC/bSM + 20% Chol is near the center of the Gibbs triangle phase diagram. Moving toward binary axes of POPC/Chol or bSM/Chol results in bilayers with less heterogeneity. Such a trend is compatible with several of the reported phase diagrams [31–33], and is against the prediction of phase coexistence near the POPC/Chol binary axis [34,35]. We next explore bilayer structures of POPC/bSM 1:1 with different cholesterol concentrations. The results are shown in Fig. 2. Dispersed macroscopic domains – presumably corresponding to the solid phase [9] – with irregular boundaries are observed at 0% Chol. Addition of 5% Chol disintegrates the solid domains into smaller sizes with smeared boundaries. Domain sizes remain similar at 8% Chol. However, the area fraction of the solid domains becomes noticeably larger. The bilayers at 10 and 12% Chol exhibit a remarkable structural feature with no distinct domains and surround. It seems that the transition

C.S. Ho et al. / Biochimica et Biophysica Acta 1858 (2016) 181–188

183

Fig. 1. Planar bilayer structures of POPC/bSM at different ratios +20% Chol. (A) AFM height images. Z-scale indicated by the color bar is 0.5 nm. Scale bars are 200 nm. (B) Height profiles (in nm) along horizontal lines crossing the center of the corresponding images in (A).

from 8 to 10% Chol is achieved by eroding the protruded domains. At 14% Chol, regions with larger heights get dissolved, resulting in structures that are best described by compositional heterogeneity at 16% Chol. The heterogeneous structures are continuously smoothed out at higher cholesterol concentrations as indicated by the height profiles (Fig. 2B). The bilayer becomes very homogeneous at 45% Chol. To characterize bilayer lateral heterogeneity, we calculate height weighed pair distribution functions G(R) for POPC/bSM 1:1 with different cholesterol concentrations (Fig. 3). Characteristic exponential decay is observed at short distances [24,25]. Linear fits to the logarithm of G(R) give rise to the correlation length ξ of 7, 5, 8, and 6 nm at 16, 20, 22, and 28% Chol, respectively. The correlation lengths are essentially the same considering the statistical uncertainties (± 3 nm based on sample to sample variation at fixed lipid composition). This result is rather surprising since the bilayer height variation σh characterized by the FWHM decreases from 0.22 to 0.08 nm when cholesterol concentration increases from 16 to 28% (Supporting Information, Fig. S3). It seems that increasing cholesterol concentration only suppresses bilayer height variation σh, but leaving the correlation length that describes the length scale of bilayer lateral heterogeneity unchanged. The uncoupled transition of the height variation σh and the correlation length ξ cannot be accounted for by conventional line tension theories [36–38], which predict that the size of an isolated domain is governed by the thickness difference between the domain and the surround. On the other hand, the inadequacy of the line tension theories is not so surprising since our observed sub-ten-nanometer heterogeneity does not correspond to isolated nanodomains with well-defined interior that differs significantly from the surround. Instead, the sub-ten-nanometer structures are better described by compositional heterogeneity.

3.2. Four-component lipid membranes. Since POPC and DOPC induce, respectively, nanoscopic and macroscopic heterogeneous structures in multicomponent lipid membranes [9], macroscopic-to-nanoscopic size transition can be obtained by altering the parameter ρ ≡ DOPC / (DOPC + POPC), while keeping the mole fractions of bSM and cholesterol fixed. Bilayer heterogeneous structures of DOPC + POPC/bSM/Chol 2:2:1 (the total mole fraction of DOPC + POPC is fixed at 40%) as a function of ρ are shown in Fig. 4. At ρ = 1.0, stripe-like Lo domains with large connectivity are observed. The height difference between coexisting phases is ~ 0.8 nm. We note that the morphology of Lo domains is dependent on their surface area fraction. Round domains are preferred at smaller area fractions, whereas complex domain morphologies emerge at larger area fractions [20,39]. Similar stripe-like Lo domains are observed at ρ = 0.9. Further replacement of DOPC with POPC roughens domain boundaries and perforates domain interiors (ρ = 0.8–0.4). Moreover, the area fraction of the Lo domains becomes larger and the height difference between coexisting phases decreases from ~0.8 to ~0.3 nm as indicated by the height profiles (Fig. 4B). The overall transition can be understood by considering that the partitioning coefficient of DOPC into the Ld phase is larger than that of POPC. Therefore, larger content of POPC (smaller ρ) yields smaller fraction of the Ld phase. Further decrease of ρ (i.e., 0.3–0.0) results in nanoscopic heterogeneity that are observed in POPC/bSM/Chol. Height analysis indicates that the height variation σh represented by FWHM decreases from 0.24 to 0.16 nm when ρ decreases from 0.3 to 0.0 (Supporting Information, Fig. S4). Calculation of the height weighted pair distribution functions reveals that the correlation length ξ is 11 nm at ρ = 0.3, and 5 nm at ρ = 0.0. It seems that in the four-component system with

184

C.S. Ho et al. / Biochimica et Biophysica Acta 1858 (2016) 181–188

Fig. 2. Bilayer heterogeneous structures of POPC/bSM 1:1 as a function of cholesterol content. (A) AFM height images. Z-scale indicated by the color bar at the bottom is 1.2 nm for 0 to 14% Chol, 1.0 nm for 16 and 18% Chol, 0.8 nm for 20 and 22% Chol, and 0.5 nm for 25 to 45% Chol. Scale bars are 200 nm for all images. (B) Height profiles (in nm) along horizontal lines crossing the center of the corresponding images in (A).

Fig. 3. Height weighted pair distribution functions G(R) for POPC/bSM 1:1 with different cholesterol contents. Linear fits to the logarithm of G(R) give rise to the correlation length ξ, which is 7, 5, 8, and 6 nm for 16, 20, 22, and 28% Chol, respectively. Data are shifted vertically for better viewing.

fixed bSM and cholesterol fractions, the correlation length ξ is proportionally correlated to the height variation σh. This is in contrast to the three-component system with fixed ratio of POPC/bSM (Figs. 2 and 3). Together, our AFM data show that both the lateral and normal dimensions of heterogeneous bilayers can be tuned by adjusting the relative content of DOPC and POPC. Following the same principle, a different four-component system (e.g., DOPC/POPC/DSPC (1,2-distearoyl-sn-glycero-3-phosphocholine)/ Chol) has been studied before [40–43]. The authors employed fluorescence microscopy and FRET/SANS (Föster resonance energy transfer and small angle neutron scattering) to exploit bilayer macroscopic and nanoscopic structures, respectively. Similar size transition was reported as a function of ρ. One difference is that a discontinuous region with patterned structures referred to as the “modulated phase” was reported when ρ is in a narrow range (i.e., 0.15–0.25). This is different from our AFM data, which show that the structural transition from macroscopic to nanoscopic heterogeneity is continuous as a function of ρ. Compared to planar bilayers used in our studies, curvature effect of GUVs might play a role in the formation of the “modulated phase” [40,41,43]. Alternatively, coupling to solid substrate could arrest the formation of modulated phase in our study.

C.S. Ho et al. / Biochimica et Biophysica Acta 1858 (2016) 181–188

185

Fig. 4. Bilayer heterogeneous structures of DOPC + POPC/bSM/Chol 2:2:1 as a function of ρ ≡ DOPC / (DOPC + POPC). (A) AFM height images. Z-scale indicated by the color bar is 1.2 nm for ρ = 1.0 to 0.7, 0.8 nm for ρ = 0.6 and 0.5, and 0.6 nm for ρ = 0.4 to 0.0. (B) Height profiles (in nm) along horizontal lines crossing the center of the corresponding images in (A). The correlation lengths derived from the height weighted pair distribution functions are 5, 6, 8, and 11 nm for ρ at 0.0, 0.1, 0.2, and 0.3, respectively.

3.3. Models for nanoscopic structures in lipid membranes Currently there are two theoretical models describing nanoscopic structures present in lipid-only membranes: (1) Structural fluctuations associated with critical behavior. Fluctuations have been observed under microscope by tuning lipid composition and/or temperature in model GUVs [24,44–47] and in giant plasma membrane vesicles [48]. Fluctuations have also been inferred from nuclear magnetic resonance (NMR) measurements [49,50]. The size of the fluctuating structures is described by the correlation length, which can range from a molecular diameter to infinitely large, depending on the proximity to the critical point. This model predicts that plasma membranes are poised near (above) the critical point. External stimulus triggers the membrane to cross the phase boundary either by altering membrane composition or temperature, thus coarsening nanoscopic fluctuations into micron size. It is noteworthy that submicron fluctuations above the critical point have not been observed experimentally, but only been postulated based on temperature dependent fluctuations behavior. (2) Nanodomains with distinct structural and mechanical properties from the surround [12,28,51–53]. This model is built on the observation that some ternary mixtures (e.g., POPC/SM/Chol) are optically homogeneous. However, techniques that are sensitive to nanoscopic structures suggest a non-uniform mixing property. In parallel with the macroscopic Lo + Ld phase coexistence, it was suggested that nanometer-scale domains (i.e., nanodomains) with well-defined interior and boundary are present in lipid mixtures containing POPC. Evidences for the existence of stable nanodomains are primarily based on indirect spectroscopy and scattering measurements. Similar to the critical fluctuations model, the nanodomain model predicts that coalescence of nanodomains can be triggered by moving to a different location in the temperature dependent phase diagram. A variation of the nanodomain model is the 2-dimensional

microemulsion model, which postulates that isolated nanoscopic domains can be stabilized by interfacial surfactant-like molecules [24,54–56].

Our AFM based studies of the three- and four-component systems indicate that the sub-ten-nanometer heterogeneous structures do no correspond to isolated nanodomains. Topological structures with variable heights are mixed with no distinct boundary or interior. The overall organization is best described by compositional heterogeneity. We note that although non-ideal mixing is inevitable in multicomponent systems [24,57], our obtained nanoscopic structures do not correspond to non-ideal mixing either. This is evident that extremely homogeneous (flat) structures (candidates for non-ideal mixing in planar bilayers) are detected in POPC/bSM/Chol when the compositions are far away from the center of the phase diagram (Figs. 1 and 2). We also note that our observed nanoscopic structures are not due to temporal trapping caused by the mica substrate. This is supported by the observation that macroscopic domains are mobile in planar bilayers and can fuse together in less than 30 min (unpublished data). Similarly, lateral diffusion measurements indicate that fluorescent lipids in solid supported bilayers are mobile [58–60]. In particular, STED–FCS measurements showed that in mica supported phase coexisting bilayers, fluorescent lipids diffuse freely on spatial scales of 40–250 nm [61]. Since the free diffusing length scale is larger than the correlation length of our nanoscopic heterogeneity, it is reasonable to argue that mica substrate does not play a role in generating the sub-ten-nanometer structures observed in our study. Finally, the negligible influence of mica substrate on sub-ten-nanometer structures is also supported by the observation that preparation of planar bilayers at elevated or room temperatures results in the same nanoscopic structures. Our calculation of the height weighted pair distribution functions indicates that the correlation length of the nanoscopic heterogeneity is ~5–10 nm. Similar length scales have been estimated based on indirect measurements [27,30,31], although a different form of the nanoscopic entities was assumed (i.e., isolated nanodomains). We note that the

186

C.S. Ho et al. / Biochimica et Biophysica Acta 1858 (2016) 181–188

nanoscopic structures obtained in our studies are compatible with almost all of the experimental data (e.g., FRET, NMR, and SANS), from which the nanodomain model was derived, since these measurements only imply compositional/structural heterogeneity. Membrane rafts hypothesize that lipids and proteins form nanoscopic assemblies that can be stabilized to coalesce [62]. Both the critical fluctuations and nanodomain models are able to describe size transition of membrane assemblies, although the driving forces are different. The nanoscopic structures observed in our study do not conform to the nanodomain model, since they do not contain well-defined boundary and interior. Similarly, our nanoscopic structures cannot be entirely explained by the critical fluctuations model. This is based on the consideration that critical fluctuations only emerge at a localized region near the critical point [24], while our nanoscopic structures are observed in a broad compositional space. Based on these observations, we argue that alternative model(s) is required to describe nanoscopic structures in lipid membranes containing POPC or equivalent lipids. To this end, we are tempted to propose a nanoheterogeneity model, which integrates the following elements: (1) nanoscopic heterogeneous structures (~10 nm) do not contain well-defined boundary, and are better described by compositional heterogeneity; (2) the length scale of compositional heterogeneity can be described by the correlation length, which varies from a few to tens of nanometers; (3) the nanoheterogeneity can be obtained in a broad compositional space by recruiting POPC or equivalent membrane components; and (4) continuous nanoscopic-to-macroscopic size transition can be induced by incorporating DOPC or equivalent membrane components. Admittedly, our nanoheterogeneity model is a mere reflection of the collective behavior of nanoscopic structures observed in our solid supported bilayers. It will be optimistic to assume that this model can be

applied in a universal context. Nevertheless, the nanoscopic fluctuation-like characteristic (to a broader extent than the critical fluctuations model) is consistent with a recent experimental observation using STED–FCS measurements [5]. The authors found that fluorescent lipid with enhanced phase partitioning property, as well as cholesterol analogues, exhibit uniform diffusion behavior in living cells (within the instrument resolution limit of ~ 60–80 nm). This observation can be well accounted for by our nanoheterogeneity model, which predicts the absence of distinct boundary/barrier (unlike the nanodomain model) to slow down lipid diffusion at a particular region. Do our heterogeneous sub-ten-nanometer structures contradict lipid facilitated membrane raft model? The answer is maybe. This is seen by the contemporary membrane raft definition: “Membrane rafts are small (10–200 nm), heterogeneous, highly dynamic, sterol- and sphingolipid-enriched domains that compartmentalize cellular processes. Small rafts can sometimes be stabilized to form larger platforms through protein–protein and protein–lipid interactions” [62]. The main inconsistence is the concept of “domain” in the nanometer regime, which is not supported by our data. Similar inconsistence is seen in the critical fluctuations model, which postulates the absence of stable or dynamic nanodomains. In fact, our sub-ten-nanometer structures resemble compositional fluctuations observed near the critical point of the three-component lipid bilayer system containing DOPC/eSM/Chol (Fig. 5). It is clearly that when lipid composition is far away from the critical point (e.g., 20% Chol), both the GUV and solid supported bilayer exhibit macroscopic phase separation; when lipid composition migrates to the vicinity of the critical point (e.g., 38% Chol), critical fluctuations and fluctuation-like heterogeneity are observed in the GUV and the planar bilayer, respectively. Similar fluctuation-like structures in solid supported bilayers have been reported [20]. Compared to critical

Fig. 5. Fluorescence (GUVs) and AFM (planar bilayers) images of DOPC/eSM 3:2 at two cholesterol concentrations. At 20% Chol, both the GUV (A) and the solid supported planar bilayer (B) exhibit macroscopic phase separation; increasing cholesterol concentration to 38% yields critical fluctuations in the GUV (C) and fluctuation-like heterogeneity in the planar bilayer (D).

C.S. Ho et al. / Biochimica et Biophysica Acta 1858 (2016) 181–188

fluctuations localized near the critical point, our sub-ten-nanometer structures in POPC-containing bilayers are observed in a broad compositional space. Since only one or a few critical points can exist within the Gibbs phase diagram, our data cannot be solely accounted for by the critical fluctuations model. Instead, we suggest that fluctuation-like sub-ten-nanometer compositional heterogeneity is a more general organization principle that does not require membranes to be in the vicinity of the critical point when POPC-like lipids are recruited. 4. Conclusions AFM is a powerful technique that has been broadly used to explore heterogeneous topographical structures of lipid membranes. Domains larger than ~ 50 nm can be easily visualized by differentiating 2dimensional height profiles. Visualizing structures with length scales b20 nm demands high-resolution instrumental setup. Empowered by the high-resolution liquid-compatible Multimode 8 AFM, we present the first nanoscopic heterogeneous structures of mica supported POPC-containing multicomponent lipid bilayers that exhibit sub-tennanometer correlation lengths. Our obtained sub-ten-nanometer structures indicate that isolated disk-like domains in the nanometer regime do not exist. This observation is supported by the analysis of height fluctuations (variations) across the membrane surface, which shows that the height contrast varies from ~0.2 nm to smaller than 0.1 nm when the sub-ten-nanometer structures emerge. Due to entropic penalty, the small height contrast results in unfavorable association of “raft lipids”; therefore, sub-ten-nanometer compositional heterogeneity dominates. We also study a four-component system containing DOPC/ POPC/bSM/Chol. By fixing the total ratio of DOPC and POPC while varying the parameter ρ ≡ DOPC / (DOPC + POPC), we obtain bilayer structural transition from micron- to nanometer-scale. This observation confirms previous finding that by tailoring lipid composition, the degree (or size) of membrane heterogeneity can be actively regulated. Transparency document This Transparency document associated with this article can be found, in the online version. Acknowledgments J. Pan acknowledges the startup package from the University of South Florida. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbamem.2015.11.001. References [1] D. Lingwood, K. Simons, Lipid rafts as a membrane-organizing principle, Science 327 (2010) 46–50. [2] C. Eggeling, C. Ringemann, R. Medda, G. Schwarzmann, K. Sandhoff, S. Polyakova, V.N. Belov, B. Hein, C. von Middendorff, A. Schonle, S.W. Hell, Direct observation of the nanoscale dynamics of membrane lipids in a living cell, Nature 457 (2009) 1159. [3] S.J. Sahl, M. Leutenegger, M. Hilbert, S.W. Hell, C. Eggeling, Fast Molecular Tracking Maps Nanoscale Dynamics of Plasma Membrane Lipids, 107P Natl Acad Sci USA, 2010 6829–6834. [4] S. Munro, Lipid rafts: elusive or illusive? Cell 115 (2003) 377–388. [5] A. Honigmann, V. Mueller, H. Ta, A. Schoenle, E. Sezgin, S.W. Hell, C. Eggeling, Scanning STED–FCS reveals spatiotemporal heterogeneity of lipid interaction in the plasma membrane of living cells, Nat. Commun. 5 (2014) 5412. [6] K. Simons, W.L.C. Vaz, Model systems, lipid rafts, and cell membranes, Annu. Rev. Biophys. Biomol. 33 (2004) 269–295. [7] E. London, How principles of domain formation in model membranes may explain ambiguities concerning lipid raft formation in cells, Biochim. Biophys. Acta, Mol. Cell Res. 1746 (2005) 203–220. [8] K. Jacobson, O.G. Mouritsen, R.G.W. Anderson, Lipid rafts: at a crossroad between cell biology and physics, Nat. Cell Biol. 9 (2007) 7–14.

187

[9] G.W. Feigenson, Phase boundaries and biological membranes, Annu. Rev. Biophys. Biomol. 36 (2007) 63–77. [10] D. Marsh, Cholesterol-induced fluid membrane domains: a compendium of lipidraft ternary phase diagrams, Biochim. Biophys. Acta Biomembr. 1788 (2009) 2114–2123. [11] Y.F. Dufrene, G.U. Lee, Advances in the characterization of supported lipid films with the atomic force microscope, Biochim. Biophys. Acta Biomembr. 1509 (2000) 14–41. [12] G.W. Feigenson, J.T. Buboltz, Ternary phase diagram of dipalmitoyl-PC/dilauroyl-PC/ cholesterol: nanoscopic domain formation driven by cholesterol, Biophys. J. 80 (2001) 2775–2788. [13] S. Chiantia, J. Ries, N. Kahya, P. Schwille, Combined AFM and two-focus SFCS study of raft-exhibiting model membranes, ChemPhysChem 7 (2006) 2409–2418. [14] S.D. Connell, D.A. Smith, The atomic force microscope as a tool for studying phase separation in lipid membranes (review), Mol. Membr. Biol. 23 (2006) 17–28. [15] A.J. Garcia-Saez, S. Chiantia, P. Schwille, Effect of line tension on the lateral organization of lipid membranes, J. Biol. Chem. 282 (2007) 33537–33544. [16] L.J. Johnston, Nanoscale imaging of domains in supported lipid membranes, Langmuir 23 (2007) 5886–5895. [17] M.P. Mingeot-Leclercq, M. Deleu, R. Brasseur, Y.F. Dufrene, Atomic force microscopy of supported lipid bilayers, Nat. Protoc. 3 (2008) 1654–1659. [18] T. Wang, H. Shogomori, M. Hara, T. Yamada, T. Kobayashi, Nanomechanical recognition of sphingomyelin-rich membrane domains by atomic force microscopy, Biochem. U.S 51 (2012) 74–82. [19] L. Picas, P.E. Milhiet, J. Hernandez-Borrell, Atomic force microscopy: a versatile tool to probe the physical and chemical properties of supported membranes at the nanoscale, Chem. Phys. Lipids 165 (2012) 845–860. [20] S.D. Connell, G. Heath, P.D. Olmsted, A. Kisil, Critical point fluctuations in supported lipid membranes, Faraday Discuss. 161 (2013) 91–111. [21] Z.V. Leonenko, A. Carnini, D.T. Cramb, Supported planar bilayer formation by vesicle fusion: the interaction of phospholipid vesicles with surfaces and the effect of gramicidin on bilayer properties using atomic force microscopy, Biochim. Biophys. Acta Biomembr. 1509 (2000) 131–147. [22] J. Jass, T. Tjarnhage, G. Puu, From liposomes to supported, planar bilayer structures on hydrophilic and hydrophobic surfaces: an atomic force microscopy study, Biophys. J. 79 (2000) 3153–3163. [23] I. Reviakine, A. Brisson, Formation of supported phospholipid bilayers from unilamellar vesicles investigated by atomic force microscopy, Langmuir 16 (2000) 1806–1815. [24] A.R. Honerkamp-Smith, S.L. Veatch, S.L. Keller, An introduction to critical points for biophysicists; observations of compositional heterogeneity in lipid membranes, Biochim. Biophys. Acta Biomembr. 1788 (2009) 53–63. [25] J.J. Amazon, G.W. Feigenson, Lattice simulations of phase morphology on lipid bilayers: renormalization, membrane shape, and electrostatic dipole interactions, Phys. Rev. E 89 (2014) 022702. [26] R.F.M. de Almeida, L.M.S. Loura, A. Fedorov, M. Prieto, Lipid rafts have different sizes depending on membrane composition: a time-resolved fluorescence resonance energy transfer study, J. Mol. Biol. 346 (2005) 1109–1120. [27] A. Pokorny, L.E. Yandek, A.I. Elegbede, A. Hinderliter, P.F.F. Almeida, Temperature and composition dependence of the interaction of delta-lysin with ternary mixtures of sphingomyelin/cholesterol/POPC, Biophys. J. 91 (2006) 2184–2197. [28] A. Bunge, P. Muller, M. Stockl, A. Herrmann, D. Huster, Characterization of the ternary mixture of sphingomyelin, POPC, and cholesterol: support for an inhomogeneous lipid distribution at high temperatures, Biophys. J. 94 (2008) 2680–2690. [29] T. Bartels, R.S. Lankalapalli, R. Bittman, K. Beyer, M.F. Brown, Raftlike mixtures of sphingomyelin and cholesterol investigated by solid-state H-2 NMR spectroscopy, J. Am. Chem. Soc. 130 (2008) 14521–14532. [30] P. Pathak, E. London, Measurement of lipid nanodomain (raft) formation and size in sphingomyelin/POPC/cholesterol vesicles shows TX-100 and transmembrane helices increase domain size by coalescing preexisting nanodomains but do not induce domain formation, Biophys. J. 101 (2011) 2417–2425. [31] R.S. Petruzielo, F.A. Heberle, P. Drazba, J. Katsaras, G.W. Feigenson, Phase behavior and domain size in sphingomyelin-containing lipid bilayers, Biochim. Biophys. Acta Biomembr. 1828 (2013) 1302–1313. [32] S.L. Veatch, S.L. Keller, Miscibility phase diagrams of giant vesicles containing sphingomyelin, Phys. Rev. Lett. 94 (2005) 148101. [33] I.V. Ionova, V.A. Livshits, D. Marsh, Phase diagram of ternary cholesterol/ palmitoylsphingomyelin/palmitoyloleoyl-phosphatidylcholine mixtures: spin-label EPR study of lipid-raft formation, Biophys. J. 102 (2012) 1856–1865. [34] R.F.M. de Almeida, A. Fedorov, M. Prieto, Sphingomyelin/phosphatidylcholine/cholesterol phase diagram: boundaries and composition of lipid rafts, Biophys. J. 85 (2003) 2406–2416. [35] K.K. Halling, B. Ramstedt, J.H. Nystrom, J.P. Slotte, T.K.M. Nyholm, Cholesterol interactions with fluid-phase phospholipids: effect on the lateral organization of the bilayer, Biophys. J. 95 (2008) 3861–3871. [36] P.I. Kuzmin, S.A. Akimov, Y.A. Chizmadzhev, J. Zimmerberg, F.S. Cohen, Line tension and interaction energies of membrane rafts calculated from lipid splay and tilt, Biophys. J. 88 (2005) 1120–1133. [37] V.A.J. Frolov, Y.A. Chizmadzhev, F.S. Cohen, J. Zimmerberg, “Entropic traps” in the kinetics of phase separation in multicomponent membranes stabilize nanodomains, Biophys. J. 91 (2006) 189–205. [38] K.B. Towles, N. Dan, Line tension and coalescence in heterogeneous membranes, Langmuir 23 (2007) 13053–13058. [39] R.M.A. Sullan, J.K. Li, C.C. Hao, G.C. Walker, S. Zou, Cholesterol-dependent nanomechanical stability of phase-segregated multicomponent lipid bilayers, Biophys. J. 99 (2010) 507–516.

188

C.S. Ho et al. / Biochimica et Biophysica Acta 1858 (2016) 181–188

[40] T.M. Konyakhina, S.L. Goh, J. Amazon, F.A. Heberle, J. Wu, G.W. Feigenson, Control of a nanoscopic-to-macroscopic transition: modulated phases in four-component DSPC/DOPC/POPC/Chol giant unilamellar vesicles, Biophys. J. 101 (2011) L8–L10. [41] S.L. Goh, J.J. Amazon, G.W. Feigenson, Toward a better raft model: modulated phases in the four-component bilayer, DSPC/DOPC/POPC/CHOL, Biophys. J. 104 (2013) 853–862. [42] F.A. Heberle, M. Doktorova, S.L. Goh, R.F. Standaert, J. Katsaras, G.W. Feigenson, Hybrid and nonhybrid lipids exert common effects on membrane raft size and morphology, J. Am. Chem. Soc. 135 (2013) 14932–14935. [43] T.M. Konyakhina, J. Wu, J.D. Mastroianni, F.A. Heberle, G.W. Feigenson, Phase diagram of a 4-component lipid mixture: DSPC/DOPC/POPC/Chol, Biochim. Biophys. Acta Biomembr. 1828 (2013) 2204–2214. [44] S.L. Veatch, S.L. Keller, Seeing spots: complex phase behavior in simple membranes, Biochim. Biophys. Acta, Mol. Cell Res. 1746 (2005) 172–185. [45] S.L. Veatch, K. Gawrisch, S.L. Keller, Closed-loop miscibility gap and quantitative tielines in ternary membranes containing diphytanoyl PC, Biophys. J. 90 (2006) 4428–4436. [46] A.R. Honerkamp-Smith, P. Cicuta, M.D. Collins, S.L. Veatch, M. den Nijs, M. Schick, S.L. Keller, Line tensions, correlation lengths, and critical exponents in lipid membranes near critical points, Biophys. J. 95 (2008) 236–246. [47] A.R. Honerkamp-Smith, B.B. Machta, S.L. Keller, Experimental observations of dynamic critical phenomena in a lipid membrane, Phys. Rev. Lett. 108 (2012) 265702. [48] S.L. Veatch, P. Cicuta, P. Sengupta, A. Honerkamp-Smith, D. Holowka, B. Baird, Critical fluctuations in plasma membrane vesicles, ACS Chem. Biol. 3 (2008) 287–293. [49] S.L. Veatch, O. Soubias, S.L. Keller, K. Gawrisch, Critical Fluctuations in DomainForming Lipid Mixtures, 104P Natl Acad Sci USA, 2007 17650–17655. [50] J.H. Davis, M.L. Schmidt, Critical behaviour in DOPC/DPPC/cholesterol mixtures: static H-2 NMR line shapes near the critical point, Biophys. J. 106 (2014) 1970–1978. [51] J.R. Silvius, Fluorescence energy transfer reveals microdomain formation at physiological temperatures in lipid mixtures modeling the outer leaflet of the plasma membrane, Biophys. J. 85 (2003) 1034–1045.

[52] F.A. Heberle, J. Wu, S.L. Goh, R.S. Petruzielo, G.W. Feigenson, Comparison of three ternary lipid bilayer mixtures: FRET and ESR reveal nanodomains, Biophys. J. 99 (2010) 3309–3318. [53] F.A. Heberle, R.S. Petruzielo, J. Pan, P. Drazba, N. Kucerka, R.F. Standaert, G.W. Feigenson, J. Katsaras, Bilayer thickness mismatch controls domain size in model membranes, J. Am. Chem. Soc. 135 (2013) 6853–6859. [54] R. Brewster, P.A. Pincus, S.A. Safran, Hybrid lipids as a biological surface-active component, Biophys. J. 97 (2009) 1087–1094. [55] M. Schick, Membrane heterogeneity: manifestation of a curvature-induced microemulsion, Phys. Rev. E 85 (2012) 031902. [56] B. Palmieri, M. Grant, S.A. Safran, Prediction of the dependence of the line tension on the composition of linactants and the temperature in phase separated membranes, Langmuir 30 (2014) 11734–11745. [57] D.G. Ackerman, G.W. Feigenson, Lipid bilayers: clusters, domains and phases, Essays Biochem. 57 (2015) 33–42. [58] J.M. Crane, L.K. Tamm, Role of cholesterol in the formation and nature of lipid rafts in planar and spherical model membranes, Biophys. J. 86 (2004) 2965–2979. [59] C. Scomparin, S. Lecuyer, M. Ferreira, T. Charitat, B. Tinland, Diffusion in supported lipid bilayers: influence of substrate and preparation technique on the internal dynamics, Eur. Phys. J. E 28 (2009) 211–220. [60] S.M. Sterling, R. Dawes, E.S. Allgeyer, S.L. Ashworth, D.J. Neivandt, Comparison actinand glass-supported phospholipid bilayer diffusion coefficients, Biophys. J. 108 (2015) 1946–1953. [61] A. Honigmann, V. Mueller, S.W. Hell, C. Eggeling, STED microscopy detects and quantifies liquid phase separation in lipid membranes using a new far-red emitting fluorescent phosphoglycerolipid analogue, Faraday Discuss. 161 (2013) 77–89. [62] L.J. Pike, Rafts defined: a report on the Keystone Symposium on Lipid Rafts and Cell Function, J. Lipid Res. 47 (2006) 1597–1598.