This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys.

Cite this:DOI: 10.1039/c6cp04631b

Nanoscale dynamics of phospholipids reveals anoptimal assembly mechanism of pore-formingproteins in bilayer membranes†

Nirod Kumar Sarangi,a K. G. Ayappa,*bc Sandhya. S. Visweswariahcd andJaydeep Kumar Basu*a

Cell membranes are believed to be highly complex dynamical systems having compositional hetero-

geneity involving several types of lipids and proteins as the major constituents. This dynamical and

compositional heterogeneity is suggested to be critical to the maintenance of active functionality and

response to chemical, mechanical, electrical and thermal stresses. However, delineating the various

factors responsible for the spatio-temporal response of actual cell membranes to stresses can be quite

challenging. In this work we show how biomimetic phospholipid bilayer membranes with variable lipid

fluidity determine the optimal assembly mechanism of the pore-forming protein, listeriolysin O (LLO),

belonging to the class of cholesterol dependent cytolysins (CDCs). By combining atomic force microscopy

(AFM) and super-resolution stimulated emission depletion (STED) microscopy imaging on model

membranes, we show that pores formed by LLO in supported lipid bilayers can have variable conformation

and morphology depending on the fluidity of the bilayer. At a fixed cholesterol concentration, pores

formed in 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) membranes were larger, flexible and more

prone to coalescence when compared with the smaller and more compact pores formed in the lower

fluidity 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) membranes. In contrast, 1,2-dimyristoyl-

sn-glycero-3-phosphocholine (DMPC) membranes did not show any evidence of pore formation.

Fluorescence correlation spectroscopy (FCS) in STED mode revealed the appearance of a length scale,

x, below which lipid dynamics, under the influence of LLO protein binding and assembly, becomes

anomalous. Interestingly, the magnitude of x is found to correlate with both lipid fluidity and pore

dimensions (and flexibility) in DOPC and POPC bilayers. However this length scale dependent crossover,

signalling the onset of anomalous diffusion, was not observed in DMPC bilayers. Our study highlights the

subtle interplay of lipid membrane mediated protein assembly and lipid fluidity in determining proteo-

lipidic complexes formed in biomembranes and the significant insight that STED microscopy provides in

unraveling critical aspects of nanoscale membrane biophysics.

1 Introduction

Lipid–protein interactions are fundamental to the organizationand function of biomembranes.1,2 However, the mechanismsby which proteins can modify membrane structure,3 or the

dependence of membrane lipid composition on the nature ofinteractions4 between intrinsic membrane proteins, are stillpoorly understood. Aggregation of proteins into larger proteo-lipidic complexes can occur through a process of self-assembly,which follows from an initial step of protein binding to thecellular membrane. This phenomenon of membrane mediatedself-assembly has widespread implications, ranging from acting asthe primary virulent pathway for pore forming toxins (PFTs) to theformation of protein aggregates in Alzheimer’s and Parkinson’sdisease.5 Pore-forming toxins are expressed by several strains ofbacteria such as Staphylococcus aureus, Escherichia coli, andListeria monocytogenes to name a few.6–9 Monomeric proteinsare expressed in a water soluble form which seek out the targetmembrane. Upon encountering the target cell membrane,rapid binding and oligomerization lead to the formation of a

a Department of Physics, Indian Institute of Science, Bangalore 560 012, India.

E-mail: basu@physics.iisc.ernet.in; Fax: +91 80 2360 2602; Tel: +91 80 2293 3281b Department of Chemical Engineering, Indian Institute of Science,

Bangalore 560 012, India. E-mail: ayappa@chemeng.iisc.ernet.in;

Fax: +91 80 2360 8121; Tel: +91 80 2293 2769c Centre for Biosystems Science and Engineering, Indian Institute of Science,

Bangalore 560012, Indiad Department of Molecular Reproduction, Development and Genetics,

Indian Institute of Science, Bangalore 560012, India

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6cp04631b

Received 3rd July 2016,Accepted 5th October 2016

DOI: 10.1039/c6cp04631b

www.rsc.org/pccp

PCCP

PAPER

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article OnlineView Journal

Phys. Chem. Chem. Phys. This journal is© the Owner Societies 2016

transmembrane pore which eventually causes cell lysis.7 Oligo-merization is unregulated, limited only by the concentration ofthe bound toxin, and binding is usually receptor mediated. Theprocess of binding and oligomerization in the membrane leadsto disruption in the surrounding lipid structure, alters lipiddynamics and induces local compositional heterogeneity aroundthe transmembrane pore complex.10,11 Pore aggregation canfurther extend the spatial influence of disruption in the lipidenvironment. Since transmembrane pores typically range from10 to 30 nm in diameter,12,13 capturing the induced perturbationin lipid structure and dynamics at these length scales has beena challenge. Unraveling these nanoscopic changes in themembrane environment has fundamental implications onour understanding of transmembrane mediated signaling andother biophysical cellular events.

Cholesterol dependent cytolysins (CDCs) are a large class ofcholesterol dependent PFTs which have been widely studied interms of the pore formation efficiency, kinetics and morphology.14,15

LLO expressed as a water-soluble monomer binds to cholesterolwhich acts as a receptor in the host cell membrane. Subsequentoligomerization leads to the formation of prepore complexes(30–50 subunits), which undergo a conformational change formingtransmembrane pores.16,17 While it is desirable to study poreformation in plasma membranes of cells, the chemical com-plexity of typical plasma membranes makes identification ofthe dominant processes determining pore formation difficult.Hence, the use of artificial lipid bilayer membranes with con-trolled complexity and organization is more useful as a bottom-upplatform on which protein–lipid interactions, in general, andPFT–lipid interactions, in particular, can be conveniently studied.Model biological membranes have been widely used to under-stand the organizational and dynamical aspects of real cellmembranes as well as their interactions with proteins, polymersand nanoparticles.18–28 In this regard, several recent studieswith supported lipid bilayer (SLB) membranes and LLO PFTshave revealed the subtle dependence of the nature of proteinassemblies and pores formed on the composition of lipidmembranes.29,30 Obtaining microscopic insight into the causalconnection between lipid composition and pore formationefficacy has proven to be difficult, largely due to both the smalllength scale involved for both pore formation (10–100 nm) aswell as the dynamic and co-operative nature of the potentialinteraction and response.

The advent of super-resolution microscopy techniques, espe-cially the stimulated emission depletion based technique(STED)31,32 in combination with well established fluorescencecorrelation spectroscopy (STED-FCS),33–38 has paved the way forexploration of fundamental biological and especially protein–membrane interactions with the high spatio-temporal resolutionrequired to probe the phenomenon at the nanoscale.39 Here,using supported phospholipid bilayer platforms with variablelipid composition/fluidity and STED-FCS, we have studied theprocess of lipid membrane bound assembly and pore formationof LLO and their connection to the nanoscale lipid dynamicalheterogeneities (DH) induced upon LLO binding to these modelbiological membranes. Cholesterol (25%) was included in all

lipid mixtures such as DOPC, POPC and DMPC, since it serves asthe receptor for LLO in biological membranes. By employingSTED-FCS, we show how the fluidity of the membrane lipids iscritical for the membrane-bound oligomerization process of LLO.While pore formation is associated with the onset of nanoscalelipid DH for the fluid phase lipid membranes composed of DOPCand POPC, neither pore formation nor DH is observed for DMPCmembranes. Interestingly, POPC with intermediate fluidityseems to form more homogeneous and smaller, compact pores,compared to DOPC which seems to form less compact andlarger as well as more heterogeneous pores. This also seems tobe consistent with the relatively larger length scale for the onsetof anomalous lipid diffusion for DOPC when compared withPOPC as revealed in STED-FCS measurements. Our results shedlight not only on the mechanistic pathway for induced nano-sized proteolipid domains directed by the host cell membranelipid composition and fluidity but also opens a new methodologyto obtain insight into various biological and bio-membranemediated processes occurring at the nanoscale.

2 Experimental2.1 Materials and methods

Phospholipids such as DOPC (499% purity), POPC (499% purity)and DMPC (499% purity) and cholesterol (Chl, 498% purity)were obtained from Avanti polar lipids. 1,2-Dimyristoyl-sn-gly-cero-3-phosphoethanolamine labeled with Atto 488 was obtainedfrom ATTO-TEC GmbH, Germany. Supported lipid bilayers areprepared by the self-assembly of lipids into bilayers onto a glasssubstrate (20 � 20 mm, Glaswarenfabrik Karl Hecht GmbH &Co KG, Germany); this planar configuration is ideally suited forobservation using fluorescence microscopy.40 The SLBs stainedwith Atto 488-PE (0.001 mol%) dye were prepared on pre-treatedglass slides by using the Langmuir–Blodgett (LB) method. Thelipid solutions were prepared in chloroform (HPLC grade, SigmaAldrich) and sprayed on the water subphase for the formationof interfacial monolayers. Ultrapure water with a resistivity of18.2 MO cm was used as the subphase for all monolayer studiesand was produced using a two-stage Elix-3 and Milli-Q (MilliporeAcademic) system. Prior to the experiment, the mini trough wascleaned with ethanol (extra pure AR grade, Fine Chemicals, India)several times and finally rinsed with ultra pure water. In addition,bilayers were also prepared using the vesicle (B100 nm) fusionmethod. The detailed description of the preparation of the SLBscan be found in the ESI.† Listeriolysin O (LLO) was expressed inE. coli and purified by Ni-NTA affinity chromatography asdescribed earlier.41 LLO (0.08 mM) was incubated with the SLBsfor 30 min at 37 1C and at pH 5.4. The plasmid encoding LLOwas obtained from Dr Gregor Anderluh.

2.2 Atomic force microscopy

Atomic force microscopy (AFM) measurements were carriedout using a Park System (NX 100, Korea) AFM under bufferusing contact mode with a Pyrex-Nitride-Probes–Silicon Nitride(PNP–SiN) cantilever with force constant 0.08 Nm�1. The SLBs of

Paper PCCP

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys.

pristine bilayers and LLO incubated bilayers are transferredinto the liquid cell. The images were acquired at the scanningrates of 0.65 Hz, operated at room temperature 22–23 1C using a50 mm2 scan head. Images obtained were processed and analyzedusing the NX 100 XEI 1.8.0 program.

2.3 STED-FCS nanoscope

For imaging and FCS, we applied STED-FCS nanoscopy using acommercial CW-STED setup (SP5x, Leica Microsystems GmbH,Mannheim, Germany) with a 592 nm fiber laser for STED. TheAPD signal intensity as well as fluorescence correlation spectro-scopy (FCS) from the SLBs was recorded using the microscopesoftware SymPhoTime, PicoQuant coupled with Leica LAS AFsoftware, which communicates with the PicoQuant SymPhoTimesoftware as an already integrated FCS package in LAS AF. Thedetailed description for the measurement parameters is pre-sented in the ESI.†

2.3.1 STED-FCS measurements and analysis. The SLBs (forthe preparation method, see the ESI†) were placed on top of themicroscope objective. The optimal z-position in the center ofthe SLB was chosen by maximizing the fluorescence intensitycounts in the plane of the bilayer. During the measurement,the adaptive focus control was enabled to avoid a drift in thez-direction between measurements. FCS data acquisition wasset for a total duration of 10 s for both confocal- and STED-FCSrecordings. For each optimized bilayer position, a fixed masterpower of the Ar laser (25%) and a full series of STED power(with an increment of 10% from 0 to 100%) were obtained. TheSTED power values in mW were measured using a power meterfrom the top of the 10� (air) objective. The detailed descriptionsof the analysis are given in the ESI.† At least 30 to 50 independent

measurements were made for each sample at different positionsand the FCS protocol was repeated for three independent samplesets to confirm reproducibility of the data.

3 Results3.1 Visualization of pore assemblies

Pores formed on model supported lipid membranes were initiallyvisualized by AFM at room temperature. Typical bilayers,composed of different lipids such as DOPC (Tm = �17 1C),POPC (Tm = �2 1C) and DMPC (Tm = 24 1C), with a fixedcholesterol (25%) were prepared using the Langmuir–Blodgett(LB) technique at highly liquid condensed (LC) surface pres-sures (see Fig. S1, ESI†). The topographic images of the pristinebilayers do not show any evidence of phase separation beforeLLO incubation (Fig. S2a–c and e, ESI†) indicative of the bilayersbeing, mostly, homogeneous in nature.42 After LLO incubation,the induced heterogeneities and pore-like features are observed forDOPC– and POPC–cholesterol bilayers as shown in AFM imagesin Fig. 1a and b respectively. In contrast, for DMPC:Chl bilayersno such features were observed as evidenced from Fig. 1c. It isfurther interesting to note that the pore–pore assembly, possiblydriven by pore coalescence, is more pronounced for the DOPC:Chlbilayer leading to a heterogeneous distribution of arcs, individualpores and coalesced pores (features highlighted in blue in Fig. 1a)with pore sizes ranging from B40 to 100 nm. The height profileanalyses (bottom panel, Fig. 1a) of the corresponding green lines(as shown in Fig. 1a–c) reveal that the protrusion of LLO domainsof B3.6� 0.2 nm are not perpendicular to the plane of the bilayersurface as expected from earlier observations,30 but slightlysplayed. This suggests that the pores are flexible and hence,

Fig. 1 Top panel shows the atomic force microscopy images of LLO incubated supported lipid bilayers of (a) DOPC:Chl, (b) POPC:Chl and (c) DMPC:Chl.In the DOPC:Chl (a) bilayers we observed arcs, fully formed pores and coalesced pores marked in blue as a, p and c respectively. Left and right panels in(b) illustrate isolated and colonies of assembled pores, respectively, in two different regions in the POPC:Chl bilayer. No evidence of pore formation wasobserved for DMPC:Chl bilayers (c). Bottom panel illustrates the height variation corresponding to the green line in the AFM topographical images (top).

PCCP Paper

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

Phys. Chem. Chem. Phys. This journal is© the Owner Societies 2016

could possibly, trigger the coalescence of oligomerized LLOpores in the DOPC:Chl host membrane (cf. Fig. S2d, ESI†).However, in the case of POPC:Chl individual pores (left panel inFig. 1b) and intact pore colonies (right panel in Fig. 1b) couldbe seen, similar to earlier observations.30 Further, the protru-sion above and the insertion below the membrane surface arelarger than that in DOPC membranes as evidenced from theheight profile shown in Fig. 1b (bottom panel). The ring shapedstructure of the pores along with the absence of lipids at theircenter strongly suggests that these could be fully inserted poresand are likely to be functional. In addition to the pore-likefeatures, we also observe arc-shaped aggregates along withthe ring-like structures for DOPC:Chl bilayers (cf. Fig. 1a andFig. S2d, ESI†), similar to the earlier observations for LLO.29

The wide variety of structures observed in our case are also in linewith the previous reports, where the membrane reorganizationupon protein/toxin incubation are directed by the lipidic host cellmembranes.43–45

In contrast to DOPC:Chl and POPC:Chl bilayer membranes,incubation of DMPC:Chl bilayers with LLO did not yield sucholigomerized pores or coalesced pore-like structures. The voidsthat are seen in Fig. 1c are possibly due to ejection of lipidsinduced by LLO binding without oligomerisation or pore for-mation. This is also clear from the corresponding typical heightprofile around such voids which indicate a depth of B4 nmwithout any significant protrusion or rim formation above themembrane as would be expected for pore-like structures formed bythe incorporation of LLO into the lipid bilayer membrane. Theabsence of such structures could be due to hydrophobic mismatchor due to the lower fluidity and greater ordering of the saturatedalkyl chains of DMPC lipid.8 We will discuss these aspects further.While the AFM images and height profile analyses of SLBs revealthe morphological perturbation induced in the bilayer membranesdue to LLO, no information about the corresponding changes inlipid dynamics is revealed. Fluorescence correlation spectroscopyis the ideal technique to reveal such microscopic information onmembrane lipid dynamics.46 However, before we proceed todiscuss lipid dynamics, we first demonstrate the morphologicalfeatures that are visible in confocal and STED super-resolutionmicroscopy, and correlate these features with the AFM images. Tovisualize the LLO induced perturbation, SLBs were stained initiallywith Atto 488 PE (0.001 mol%).

The confocal microscopy images of pristine bilayers ofDOPC:Chl, POPC:Chl and DMPC:Chl appear largely homo-geneous (cf. Fig. 2a–c). In contrast, confocal microscopy imagesof DOPC:Chl and POPC:Chl SLBs, after LLO incubation, showsignificant inhomogeneities due to the redistribution of lipidsas shown in Fig. 2d and e respectively. While, large scale(1–2 mm) structures are dominant in DOPC:Chl SLBs they aremuch less prominent in confocal images of LLO incubatedPOPC:Chl bilayers. Such large scale structures, which are alsoseen in our AFM images (Fig. 1), probably occur due to porecoalescence during LLO incubation.30 In STED mode images(at maximum power) for the same SLBs, we could clearlyidentify several nanoscale domains with B100 nm resolutionfor both DOPC:Chl (Fig. 2g) and POPC:Chl (Fig. 2h) which is

otherwise not clearly visible in their respective confocal images.The intensity line profiles from the region of interest (as markedin white lines) for both confocal and STED images for DOPC:Chland POPC:Chl are shown in Fig. 2j and k respectively. In contrastto DOPC and POPC lipids, we did not observe such nanoscaledomains and/or membrane perturbations for DMPC:Chl bilayers(Fig. 2f, i and l) after LLO incubation. The absence of suchstructures is consistent with our AFM images indicating ineffi-cient pore formation for DMPC:Chl bilayers after LLO incuba-tion. Our microscopy imaging results compare well with othertoxin–membrane interaction scenarios with, for instance,a-synuclein (AS) which when interacting with small unilamellarvesicles composed of DOPC and POPC showed moderate bindingaffinity but did not significantly bind to DMPC.47 Although theimages reveal the manifestation of LLO induced bilayer pertur-bation, the modulation of lipid dynamics around such proteo-lipid complexes can only be obtained from spatially resolvedFCS data.

3.2 Spatially resolved lipid mobility around pore complexesusing confocal FCS

To probe membrane lipid dynamics, we first performed spatiallyresolved FCS in confocal mode48–51 by carefully sampling dis-tinctly different regions of the membranes (for representativespatially resolved points, see Fig. S3, ESI†). Fig. 3a–c shows theintensity–intensity autocorrelation function (G(tc)) data for thecholesterol containing bilayers for DOPC, POPC and DMPC,respectively, before and after LLO incubation. The spatiotemporalcorrelation data (Fig. 3a–c) were fitted using eqn (1),33,52,53

G tcð Þ ¼1

N

� �1

1þ tc

tD

� �a (1)

where tD is the transit time, N is the particle number and a isthe anomaly coefficient. In all the confocal FCS correlationfunctions a B 1, indicating the presence of free Browniandiffusion at confocal spatial resolution. The diffusivity (D) insuch a case can be evaluated using eqn (2)33

tD ¼d2

D8ln2; (2)

where d is the diameter of the confocal spot (d = 200 nm).eqn (2) has a slightly different form from what is typically used inconfocal FCS measurements;33,38 however, we have used eqn (2)to be consistent with STED-FCS measurements to be dis-cussed later.

The diffusivity values for cholesterol containing SLBs madeup of DOPC, POPC and DMPC were found to follow a unimodaldistribution with mean values of 3.9 � 0.23, 2.8 � 0.2 and1.7 � 0.21 mm2 s�1, respectively, consistent with earlier resultson similar membranes.54 However, upon incubating DOPC:Chlbilayers with LLO, a bimodal distribution of diffusivities(Fig. 3d) with equal populations of slower and faster (but lowerin magnitude when compared with bilayers before LLO incuba-tion) diffusivities are obtained. The lowest diffusivity compo-nent (D = 1.1 � 0.21 mm2 s�1) is significantly smaller than the

Paper PCCP

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys.

diffusivities obtained for the corresponding LLO free bilayer.This lowered lipid diffusivity occurs mainly in the proximityof low fluorescence intensity regions (region I, Fig. 2d) whichwe attribute to the boundaries of the pore aggregates rangingfrom nanometer-to-micrometer in size. Lipids with diffusivity,D = 2.5 � 0.21 mm2 s�1, correspond to regions away from thepore aggregates (region II in Fig. 2d).

The mean D, in these regions, is also slightly smaller thanthat of the corresponding pristine DOPC:Chl SLBs and thisreduction could be due to the presence of isolated pores whichare not resolved in our diffraction-limited confocal microscopymeasurements. The equal probability of the smaller and largervalues of observed diffusivities could be correlated with thedifferent types and sizes of LLO pore structures includingcoalescence pores and pre-pore aggregates, which exist in ourAFM images, as discussed earlier. Interestingly, in the case of

POPC:Chl bilayers upon LLO exposure, although we alsoobserve a bimodal distribution of diffusivities (Fig. 3e), thedistribution is strongly skewed towards smaller diffusivities(D = 1.2 � 0.22 mm2 s�1). The contribution to the smallerdiffusivity component in DOPC:Chl SLBs could arise fromtightly bound lipids in the vicinity of fully inserted pores andhence the nature of the diffusivity distribution could be related toa larger population of fully inserted pores when compared with apreponderance of pre-pore and arc-like aggregates observed inPOPC:Chl SLBs.

In contrast to the more fluidic lipids such as DOPC andPOPC, diffusivities in the DMPC bilayers are smaller and did notalter significantly after LLO incubation. After LLO incubation, thediffusivity values were found to be 0.8 � 0.14 mm2 s�1 showing aunimodal distribution, overlapping with that of the LLO freebilayer. Although, we did not observe any pore-like features either

Fig. 2 Confocal microscopy image of pristine (a) DOPC:Chl, (b) POPC:Chl and (c) DMPC:Chl bilayers before LLO incubation. (d–f) Represent therespective confocal microscopy images after LLO incubation. (g–i) Are the corresponding STED microscopy images (STED power = B260 mW) after LLOincubation. The scale bar is 2 mm. (j)–(l) Represent the intensity line profile analyses for both confocal and STED images in the region of interest (ROI) asmarked in white lines shows the improvement of resolution as evidenced from decreasing FWHM as well as the nanosized domains can be discerned bySTED but are not separated by confocal microscopy. LLO induced membrane reorganization have yielded two distinct regions, I and II in (d–f) markedby ‘*’ and ‘+’ of different fluorescence intensity.

PCCP Paper

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

Phys. Chem. Chem. Phys. This journal is© the Owner Societies 2016

in AFM or confocal microscopic images, the decrease in diffu-sivities could be due to lipid–toxin interactions which led to theordering of alkyl chains as observed from our diffusing-wavespectroscopy (DWS) data shown in Fig. S4, ESI.† Our results areconsistent with recent observation by Sharma et al.55 where theaddition of melittin, an antimicrobial peptide, did not affect themembrane dynamics of a DMPC/cholesterol bilayer membrane.

3.3 Probing lipid dynamics around pore complexes usingSTED-FCS

Although it is well known that cholesterol is a necessaryingredient for binding of a CDC such as LLO,56 little is knownabout the molecular role played by different phospholipidstowards the propensity for CDCs to bind and initiate pore-formation.57,58 Overall, our confocal FCS data provide evidenceof large scale (Bmm) heterogeneous lipid dynamics induced byLLO interaction with DOPC:Chl and POPC:Chl membranes.Variations in the probability and width of the diffusivity distri-butions (Fig. 3d and e) after LLO binding onto POPC and DOPCmembranes suggest the presence of possible sub-micron scalelipid dynamics around pore assemblies which are not capturedby the optical diffraction limited FCS data. Therefore, probinglipid dynamics on the scale of single or coalesced pores couldprovide information about the rigidity or flexibility of the poresas well as their ability to re-organize the lipids around theseself-assembled structures. For this purpose, we performed FCS insuper-resolution STED mode33 by varying the STED excitationpower to reduce the observation volume to well below the diffrac-tion limit (Fig. S5, ESI†). Fig. 4a–c describes the dependence of thetransit time tD as a function of the various observation spots (d2)created by STED laser power, as extracted from the respective STEDpower dependent autocorrelation data before and after LLOincubation. In order to present a unified picture of the lipiddynamics in the absence and presence of LLO, and to highlightthe origin of crossover to the non-Brownian regime,59 thetD versus d2 variation can provide the underlying diffusionmechanism in the studied system. To unravel this, eqn (2)can be re-written as

tD ¼d2

8D ln2þ t0 (3)

where d is the diameter of the confocal or STED focal spot andt0 is an intercept.60–65 For free diffusion, t0 = 0 while it can takepositive or negative values according to various possible mecha-nisms of hindered diffusion involving the presence of nano-domains or a meshwork structure.53,60,66–69 The details of thefitting procedure used to obtain the values of tD and a are givenin Fig. S6 (ESI†) and Fig. 4d–f respectively.

The tD values, before LLO incubation, as shown in Fig. 4decrease in the order of DMPC:Chl 4 POPC:Chl 4 DOPC:Chland the magnitude of tD is always less when compared with theLLO incubated samples irrespective of any STED power used.The linear decrease in tD with the decrease in the focal spotarea with a zero intercept (t0 = 0) for pristine bilayers signifiesBrownian diffusion52,70–73 at all observed length scales (d) rangingfrom 80 to 200 nm (cf. open circle in Fig. 4a–c). In contrast,

for LLO incubated DOPC:Chl and POPC:Chl bilayers, a distinctdynamical crossover is observed at a length scale, x, of B125and B108 nm respectively (cf. vertical line in Fig. 4a and b).Further, the onset of this dynamical crossover is also associatedwith a decrease in the value of a for LLO incubated DOPC:Chland POPC:Chl bilayers, indicative of anomalous diffusion in thisregime. Such an anomalous diffusive regime is not observed forDMPC:Chl bilayers down to the smallest d2 values accessibleusing our STED microscope. Further, we fitted the data in therespective dynamical regimes using eqn (3) to extract the valuesof intercept, t0, as indicated in the respective panels of Fig. 4a–c.Interestingly, it is also clear that the region of the data belowx does not follow a Brownian diffusion law53,60,66,67 with clearpositive intercepts of t0 B 1.9 and B1 for DOPC:Chl andPOPC:Chl, after LLO incubation, respectively, suggestive of

Fig. 3 Representative auto-correlation functions in pristine bilayersbefore (red filled circle) and after LLO incubation (blue filled circle) of(a) DOPC:Chl, (b) POPC:Chl and (c) DMPC:Chl lipids measured by FCS. Theautocorrelation function (G(tc)) is plotted as a function of lag time (tc) andthe autocorrelation curve is fitted (solid lines) using the non-linear leastsquare fit method using eqn (1). (d–f) are the respective histograms ofdiffusion coefficient, D, for different lipids used in our measurements. Notethat, in the respective pristine bilayers the lipid diffusivity displays unimodaldistribution irrespective of the bilayer region sampled. The LLO incubatedbilayers of DOPC:Chl and POPC:Chl yield two distinct diffusivities denotedby * and + (corresponding regions in Fig. 2d and e) whereas, the DMPC:Chlyielded a unimodal distribution marked as * in (f). In all the cases of lipid–cholesterol bilayers, LLO incubation resulted in a lowering of the D valueswhen compared with the LLO free bilayers. The diffusivitiy data in (e and f)are collected from E30–50 measurements each corresponding to type Iand II regions.

Paper PCCP

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys.

the presence of nanodomains. No such crossover is observed forLLO incubated DMPC:Chl bilayers and the linear fit yielded a t0 ofB0.1 (Fig. 4c) indicative of a possible weak deviation from Brow-nian diffusion but it was not significant enough to warrant furtherevaluation. In addition, a distinct change of slope was not dis-cernible at the smaller d values. However, it is difficult to concludewhether a dynamical crossover is clearly absent in LLO incubatedDMPC:Chl bilayers or it lies below the spatial resolution of ourSTED microscope. Although the essential nature of membranelipid dynamics is captured by the data presented in Fig. 4, theuniversal properties of the lipid dynamics can be obtained fromthe diffusivity data. To ascertain the origin of crossover from theBrownian to the non-Brownian regime59 as seen in Fig. 4a and b,we continued to evaluate diffusion coefficient values using eqn (2).We refer to the diffusivity in the free Brownian region (open circlesin Fig. 5a–c) as D, and in the anomalous region (closed circles inFig. 5a and b) as Dapp which occurs at small length scales todistinguish it from the free Brownian diffusion71 as was doneearlier.33,35,36,39 Additionally we define D0 as the diffusivity ofthe pristine bilayer without LLO. Note that, for all our STED-FCSdata D0 = D for the case of LLO free bilayers.

Fig. 5 shows the extracted diffusion coefficient values as afunction of focal spot area for the lipids before LLO incubation (see,red, olive and magenta (open circles), respectively, for cholesterolcontaining DOPC, POPC and DMPC lipids). Consistent with theBrownian diffusion law (of a B 1, refer Fig. 4d–f) followed in thecorresponding data in Fig. 4, we find a length scale independentdiffusivity, D, for LLO free lipids down to the smallest values of d2

that could be obtained with our STED microscope. In contrast,

in the presence of LLO in the membranes the diffusivity no longerremains constant, as assumed in eqn (2),33,39 below x (cf. filledcircles in Fig. 5a and b), but a diffusivity which depends on theobservation time scale emerges, i.e., Dapp is actually D(tD) whena o 1. We recently observed such a dynamical crossover in DOPCbilayers with an increase in cholesterol concentration74 andascribed this to the emergence of nanoscale cholesterol enrichedlipid domains. The fact that very similar behavior is observed withLLO incubation, which, as mentioned already, has strong affinityfor cholesterol, strongly suggests possible nanoscale cholesterolenriched lipid domain formation around the LLO pores. In addi-tion, we have also extracted spatially averaged diffusivity from theslopes of the so-called diffusion law data in Fig. 4, using eqn (3) forthe non-Brownian regimes as well, and referred to this as aneffective diffusivity, Deff.

63,64 The extracted Deff values in thed2 o x regime are also shown for DOPC:Chl and POPC:Chl afterLLO incubation in Fig. 5a and b respectively. It is interesting tonote that the extent of change observed for DOPC:Chl bilayersupon LLO incubation is maximum when compared to that of thecorresponding POPC:Chl and DMPC:Chl bilayers. The decrease inthe magnitude of apparent diffusivities, Dapp, with LLO incubationas a function of different lipids is particularly revealing. The valueof D0/D decreases from 3.5 4 2.3 4 2 with decreasing fluidity oflipids DOPC 4 POPC 4 DMPC, respectively, as shown in the freeBrownian diffusion regime I (d 4 x) in Fig. 5d. For d o x (seenon-Brownian diffusion regime II, shaded region), the greatestperturbation of the lipid dynamics is observed in the case ofDOPC:Chl bilayers as compared to POPC:Chl and DMPC:Chlbilayers. When the phospholipid’s unsaturation decreases, i.e.,

Fig. 4 Dependence of transit time (tD) extracted from the correlation data of (a) DOPC:Chl, (b) POPC:Chl and (c) DMPC:Chl bilayers before and afterLLO incubation on the focal spot area, d2, using eqn (1). Open and closed circles represent data before and after LLO incubation, respectively, for eachbilayer. The dotted lines for data on pristine bilayers without LLO incubation are indicative of free diffusion with a zero intercept, t0. The vertical lines,in respective panels, indicate the crossover from a Brownian diffusive regime (t0 = 0 in eqn (3)) to an anomalous regime (t0 a 0) at smaller length scalesafter LLO incubation. Solid grey lines are the extrapolation of the fitted lines in the Brownian diffusion regime as observed at high d2. Dependence of a ond2 determined from STED-FCS data recorded at different STED power for pristine (open symbol) and LLO incubated (closed symbol) bilayers of(d) DOPC:Chl, (e) POPC:Chl and (f) DMPC:Chl respectively.

PCCP Paper

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

Phys. Chem. Chem. Phys. This journal is© the Owner Societies 2016

in the case of POPC:Chl bilayers, the extent of cholesterolsequestration also decreases. The observed decrease in lengthscale heterogeneities for POPC:Chl (see vertical line in Fig. 5b)than DOPC:Chl bilayers could be due to the differences in poreformation where pore coalescence is significantly perturbed forthe former bilayer. On the other hand, for fully saturated lipidbilayers (DMPC:Chl), we did not observe any pore-like featureseither in AFM or in confocal microscopic images.

4 Discussion

We now try to correlate the information we obtained about themorphology of LLO pores on the lipid membranes from AFM andconfocal/STED images with the results of the confocal- and STED-FCS data. It seems clear that LLO forms regular, more heteroge-neous and smaller compact pores in POPC:Chl SLBs as comparedto DOPC:Chl consistent with earlier observations.29 If we couplethis with the well established fact that LLO, in particular, andCDCs, in general, bind to cholesterol in membranes, it seemslogical to assume that the formation of oligomerized pre-poreassembly and inserted pores should result in cholesterol enrich-ment around these LLO structures. Our earlier work74 connects thischolesterol enrichment to the observed length scale dependent

emergence of anomalous dynamics in the DOPC membrane uponinteraction with LLO. The enrichment was found to be mostpronounced at 25% cholesterol which is the composition used inthis study.74 One can then view the role of LLO as a creator ofnanodomain regions where cholesterol concentration is enhancedaround the pore assemblies, although the native lipid–cholesterolcombination in two-component mixtures is not expected to showsuch domains or phase separation at the cholesterol concentrationsused in our study.75–77 Even this nanodomain formation is unde-tected in our conventional confocal FCS measurements except forthe occurrence of lipid regions with smaller diffusivity. However,STED-FCS not only reveals such domain formation but alsoprovides some quantitative estimation of the length scale, x, ofsuch domains below which the lipid diffusivities are significantlyperturbed. It is also possible to estimate the domain sizes from theFCS diffusion law (Fig. 4) using the value of the extracted intercept,t0, as has been alluded to earlier.53,67 However, we have refrainedfrom carrying out this analysis here. Further, we find that thisdynamical crossover length scale, for the same LLO and cholesterolconcentration in membranes, depends on the lipid mobility. Thissuggests that the ability to sequester cholesterol around themembrane by LLO protomers to facilitate their oligomerizationand subsequent pore formation depends crucially on the fluidity ofthe underlying membrane. Large lipid fluidity, as in the case of

Fig. 5 Dependence of the diffusion coefficients on d2 in different bilayers. D0 (=D) corresponds to the diffusivity in the LLO free bilayers. For LLOincubated bilayers, D (open circles) and Dapp (closed circles) correspond to the diffusivity in the Brownian (d2 4 x) and non-Brownian regimes (d2 o x),respectively for (a) DOPC:Chl, (b) POPC:Chl and (c) DMPC:Chl. Diffusion coefficients after LLO incubation are shown in blue circles. The horizontaldashed lines indicate the expected variation of diffusivity with d2 for Brownian diffusion. The vertical lines indicate the crossover length scale, x, for tworegimes; at low d2 filled blue circles represent the onset of anomalous diffusion and at high d2 open blue symbols represent Brownian diffusion. In panels(a) and (b) the asterisk symbol at the low d2 regime represents the estimated effective diffusion coefficient, Deff, values using eqn (3). (d) Illustrates thevariation in D0/D and D0/Dapp with d2 indicating free (Brownian) and anomalous (non-Brownian) diffusion, respectively.

Paper PCCP

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys.

DOPC, while being beneficial for cholesterol migration, and ease ofpore formation also imply highly mobile and less compact poreswhich can easily coalesce often resulting in larger pores with lowerstability as observed from our AFM images.29,30 Apart from theseaspects, it is also likely to lead to polydispersity in pore sizesand morphological heterogeneity and pore–pore migration asevidenced in our AFM and confocal images in Fig. 1 and 2 as wellas in other studies.45,78 On the other hand, POPC with intermediatefluidity allows sufficient mobility of cholesterol to form optimal,stable, compact pores but the reduced mobility of the pores(embedded in a lower mobility lipid membrane) significantlydecreases the opportunity for pore coalescence. What this alsoimplies is that the length scale, x, of crossover to anomalousdiffusion, is correlated with the region around LLO pores wherecholesterol and lipid density heterogeneity exist. The smaller valueof x for POPC membranes compared to DOPC signifies this subtleinterplay between lipid mobility, pore morphology and stability.Earlier it has also been proposed that the LLO could behave as apotent lipid domain aggregator79 but the mechanism of membraneresponse to LLO incubation was not clear therein due to the smalllength scale involved for both pore formation (10–50 nm) or lipidnanodomains (also of similar size) and the dynamic nature of thepotential interaction and response. With our STED-FCS measure-ments we are able to connect our length scale dependent data andour proposed mechanism of cholesterol sequestration, whichdepends solely on the nature and physical properties of the bilayerduring LLO interaction. Hence, our STED excitation power depen-dent FCS results clearly suggest that in the presence of LLO, theformation of nanoscale domains is driven by cholesterol sequestra-tion, when the fluidity of the bilayer is high. The lateral diffusivityand the length scale dependent dynamic heterogeneity at a sub-optimal cholesterol content are in line with the recent theoreticalpredictions by Hub et al.80 suggesting the variation of localcholesterol enrichment modulating the probability of pore forma-tion of LLO. It is interesting to note that an optimal membranelipid fluidity seems to be most suitable for stable, membraneinserted, LLO pore formation. Higher lipid mobility seems to leadto a larger length scale of cholesterol sequestration and a con-sequent formation of larger, possibly, partially inserted andunstable pores which are able to further fuse to create porecomplexes due to both higher lipid mobility and lipid inducedprotein–protein interactions.81,82 On the other hand, stiffer lipids orlipids with lower mobility such as DMPC are clearly not favorablesince they preclude cholesterol sequestration which is most criticalfor LLO (and other CDC) pore formation. Understanding the effectof cholesterol in stable model membrane systems and nanodomainformation due to lipid–protein interactions is critical and thoughtto play an important role in regulating signal transduction,inter- and intra-cellular transport, and lipid sorting.83,84

5 Conclusions

Using super-resolution STED-FCS we have elucidated the LLOinduced lipid reorganization pathways on supported bilayerplatforms. Our results reveal the intricate coupling between

protein oligomerization and emergence of nanoscale regions(B100 nm) around pore complexes. Using STED-FCS we haveshown that the induced lipid dynamical heterogeneity is mostprominent in bilayers with greater fluidity than the gel-likelipids upon incubation with LLO. Our microscopic imagesrevealed pore–pore coalescence which is more prominent forlipids with greater fluidity and is completely arrested when thefluidity of the bilayer is decreased. Thus there seems to be aconnection which emerges between membrane fluidity, onsetof cholesterol and LLO mediated lipid dynamical heterogeneityand the stability and nature of the resultant pores formed. Ourstudy highlights the subtle interplay of lipid membrane mediatedprotein assembly and lipid fluidity in determining proteo-lipidiccomplexes formed in biomembranes and the significant insightthat STED microscopy provides in unraveling the critical aspectsof nanoscale membrane biophysics.

Acknowledgements

The authors thank the Department of Science and Technology,New Delhi, India, for the financial support through a specialproject (DST-IRHPA) and a JC Bose Fellowship to SSV. N. K. S.acknowledges the Dr D. S. Kothari Postdoctoral fellowship fromUGC, Govt. of India. N. K. S. acknowledges Mr Pradeep S andI. I. P for assistance with LLO purification processes. We wouldlike to acknowledge constructive and valuable feedback by ananonymous reviewer which has helped in considerably improvingthe quality of the manuscript.

References

1 R. J. Gilbert, Biochim. Biophys. Acta, 2016, 1858, 487–499.2 C.-Y. Hsia, M. J. Richards and S. Daniel, Anal. Methods, 2015,

7, 7076–7094.3 J. Rossy, Y. Ma and K. Gaus, Curr. Opin. Chem. Biol., 2014,

20, 54–59.4 P. Lague, M. J. Zuckermann and B. Roux, Biophys. J., 2001,

81, 276–284.5 G. B. Irvine, O. M. El-Agnaf, G. M. Shankar and D. M. Walsh,

Mol. Med., 2008, 14, 451–464.6 K. C. Rice and K. W. Bayles, Microbiol. Mol. Biol. Rev., 2008,

72, 85–109.7 M. D. Peraro and F. G. van der Goot, Nat. Rev. Microbiol.,

2016, 14, 77–92.8 N. Rojko and G. Anderluh, Acc. Chem. Res., 2015, 48,

3073–3079.9 A. F.-P. Sonnen and P. Henneke, Clin. Dev. Immunol., 2013,

2013, 608456.10 V. Solovyeva, L. Johannes and A. C. Simonsen, Soft Matter,

2015, 11, 186–192.11 M. R. Gonzalez, M. Bischofberger, L. Pernot, F. G. van der Goot

and B. Freche, Cell. Mol. Life Sci., 2008, 65, 493–507.12 A. G. Lee, Biochim. Biophys. Acta, 2004, 1666, 62–87.13 K. A. Burke, E. A. Yates and J. Legleiter, Front. Neurol., 2013,

4, 1–17.

PCCP Paper

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

Phys. Chem. Chem. Phys. This journal is© the Owner Societies 2016

14 W. Niedermeyer, Toxicon, 1985, 23, 425–439.15 R. J. Gilbert, Structure, 2005, 13, 1097–1106.16 M. A. Dunstone and R. K. Tweten, Curr. Opin. Struct. Biol.,

2012, 22, 342–349.17 R. K. Tweten, Infect Immunol., 2005, 73, 6199–6209.18 E. Sackmann, Science, 1996, 271, 43–48.19 M. Tanaka and E. Sackmann, Nature, 2005, 437, 656–663.20 M. Marchioretto, M. Podobnik, M. Dalla Serra and G. Anderluh,

Biophys. Chem., 2013, 182, 64–70.21 M. Schulz, A. Olubummo and W. H. Binder, Soft Matter,

2012, 8, 4849–4864.22 B. Wang, L. Zhang, S. C. Bae and S. Granick, Proc. Natl. Acad.

Sci. U. S. A., 2008, 105, 18171–18175.23 N. Biswas, R. Bhattacharya, A. Saha, N. R. Jana and J. K. Basu,

Phys. Chem. Chem. Phys., 2015, 17, 24238–24247.24 R. Bhattacharya, S. Kanchi, C. Roobala, A. Lakshminarayanan,

O. H. Seeck, P. K. Maiti, K. G. Ayappa, N. Jayaraman andJ. K. Basu, Soft Matter, 2014, 10, 7577–7587.

25 R. Bhattacharya, C. Indukuri, N. Begam, O. H. Seeck andJ. K. Basu, Adv. Funct. Mater., 2015, 25, 7233–7242.

26 E. T. Castellana and P. S. Cremer, Surf. Sci. Rep., 2006, 61,429–444.

27 M. Deleu, J. M. Crowet, M. N. Nasir and L. Lins, Biochim.Biophys. Acta, 2014, 1838, 3171–3190.

28 N. K. Sarangi and A. Patnaik, ChemPhysChem, 2012, 13,4258–4270.

29 E. Mulvihill, K. van Pee, S. A. Mari, D. J. Muller andO. Yildiz, Nano Lett., 2015, 15, 6965–6973.

30 M. Podobnik, M. Marchioretto, M. Zanetti, A. Bavdek,M. Kisovec, M. Mojca Cajnko, L. Lunelli, M. Dalla Serraand G. Anderluh, Sci. Rep., 2015, 5, 9623.

31 S. W. Hell and J. Wichmann, Opt. Lett., 1994, 19, 780–782.32 S. W. Hell, Nat. Methods, 2009, 6, 24–32.33 C. Eggeling, C. Ringemann, R. Medda, G. Schwarzmann,

K. Sandhoff, S. Polyakova, V. N. Belov, B. Hein, C. vonMiddendorff, A. Schonle and S. W. Hell, Nature, 2009, 457,1159–1162.

34 L. Kastrup, H. Blom, C. Eggeling and S. W. Hell, Phys. Rev.Lett., 2005, 94, 178104.

35 A. Honigmann, V. Mueller, S. W. Hell and C. Eggeling,Faraday Discuss., 2013, 161, 77–89.

36 A. Honigmann, V. Mueller, H. Ta, A. Schoenle, E. Sezgin,S. W. Hell and C. Eggeling, Nat. Commun., 2014, 161, 5412.

37 D. M. Andrade, M. P. Clausen, J. Keller, V. Mueller, C. Wu,J. E. Bear, S. W. Hell, B. C. Lagerholm and C. Eggeling, Sci.Rep., 2015, 5, 11454.

38 S. K. Saka, A. Honigmann, C. Eggeling, S. W. Hell, T. Langand S. O. Rizzoli, Nat. Commun., 2014, 5, 4509.

39 V. Mueller, C. Ringemann, A. Honigmann, G. Schwarzmann,R. Medda, M. Leutenegger, S. Polyakova, V. N. Belov, S. W. Helland C. Eggeling, Biophys. J., 2011, 101, 1651–1660.

40 S. G. Boxer, Curr. Opin. Chem. Biol., 2000, 4, 407–709.41 A. Bavdek, R. Kostanjsek, V. Antonini, J. H. Lakey,

M. Dalla Serra, R. J. Gilbert and G. Anderluh, FEBS J.,2012, 279, 126–141.

42 J. Zhong, Integr. Biol., 2011, 3, 632–644.

43 P. Miles and D. Frankel, Soft Matter, 2014, 10, 9562–9567.44 D. Nordin, O. Yarkoni, N. Savinykh, L. Donlon and D. Frankel,

Soft Matter, 2011, 7, 10666–10675.45 N. Yilmaz and T. Kobayashi, ACS Nano, 2015, 9, 7960–7967.46 M. Eigen and R. Rigler, Proc. Natl. Acad. Sci. U. S. A., 1994,

91, 5740–5747.47 V. V. Shvadchak, L. J. Falomir-Lockhart, D. A. Yushchenko

and T. M. Jovin, J. Biol. Chem., 2011, 286, 13023–13032.48 L. Zhang and S. Granick, Proc. Natl. Acad. Sci. U. S. A., 2005,

102, 9118–9121.49 S. A. Kim, K. G. Heinze and P. Schwille, Nat. Methods, 2007,

4, 963–973.50 V. Betaneli and P. Schwille, Methods Mol. Biol., 2013, 974,

253–278.51 J. Ries and P. Schwille, BioEssays, 2012, 34, 361–368.52 P. Schwille, J. Korlach and W. W. Webb, Cytometry, 1999, 36,

176–182.53 L. Wawrezinieck, H. Rigneault, D. Marguet and P. F. Lenne,

Biophys. J., 2005, 89, 4029–4042.54 G. Lindblom and G. Oradd, Biochim. Biophys. Acta, 2009,

1788, 234–244.55 V. K. Sharma, E. Mamontov, D. B. Anunciado, H. O’Neill and

V. S. Urban, Soft Matter, 2015, 11, 6755–6767.56 K. S. Giddings, A. E. Johnson and R. K. Tweten, Proc. Natl.

Acad. Sci. U. S. A., 2003, 20, 11315–11320.57 Y. Ohno-Iwashita, M. Iwamoto, S. Ando and S. Iwashita,

Biochim. Biophys. Acta, 1992, 1109, 81–90.58 A. Zitzer, E. J. Westover, D. F. Covey and M. Palmer, FEBS Lett.,

2003, 553, 229–231.59 M. J. Saxton, Biophys. J., 1996, 70, 1250–1262.60 R. Machan and M. Hof, Biochim. Biophys. Acta, 2010, 1798,

1377–1391.61 H. Basit, S. G. Lopez and T. E. Keyes, Methods, 2010, 68,

286–299.62 S. Ganguly and A. Chattopadhyay, Biophys. J., 2010, 99,

1397–1407.63 N. Bag, X. W. Ng, J. Sankaran and T. Wohland, Methods

Appl. Fluoresc., 2016, 4, 034003.64 N. Bag, D. H. X. Yap and T. Wohland, Biochim. Biophys. Acta,

2014, 1838, 802–813.65 H. Huang, M. F. Simsek, W. Jin and A. Pralle, PLoS One,

2015, 10, e0121777.66 H. T. He and D. Marguet, Annu. Rev. Phys. Chem., 2011, 62,

417–436.67 P.-F. Lenne, L. Wawrezinieck, F. Conchonaud, O. Wurtz,

A. Boned, X.-J. Guo, H. Rigneault, H.-T. He and D. Marguet,EMBO J., 2006, 25, 3245–3256.

68 H. Soula, B. Care, G. Beslon and H. Berry, Biophys. J., 2013,105, 2064–2073.

69 R. Metzler, J.-H. Jeon and A. G. Cherstvy, Biochim. Biophys.Acta, 2016, 1858, 2451–2467.

70 C. Favard, J. Wenger, P.-F. Lenne and H. Rigneault, Biophys.J., 2011, 100, 1242–1251.

71 D. S. Banks and C. Fradin, Biophys. J., 2005, 89, 2960–2971.72 M. Wachsmuth, W. Waldeck and J. Langowski, J. Mol. Biol.,

2000, 298, 677–689.

Paper PCCP

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online

This journal is© the Owner Societies 2016 Phys. Chem. Chem. Phys.

73 V. Mueller, A. Honigmann, C. Ringemann, R. Medda,G. Schwarzmann and C. Eggeling, Methods Enzymol., 2013,519, 1–38.

74 N. K. Sarangi, I. P. Ilanila, K. G. Ayappa, S. S. Visweswariahand J. K. Basu, Langmuir, 2016, 32, 9649–9657.

75 L. Toppozini, S. Meinhardt, C. L. Armstrong, Z. Yamani,N. Kucerka, F. Schmid and M. C. Rheinstadter, Phys. Rev.Lett., 2014, 113, 228101.

76 D. Marsh, Biochim. Biophys. Acta, Biomembr., 2010, 1798,688–699.

77 H. Heerklotz and A. Tsamaloukas, Biophys. J., 2006, 91,600–607.

78 P. D. Rakowska, H. Jiang, S. Ray, A. Pyne, B. Lamarre,M. Carr, P. J. Judge, J. Ravi, U. I. M. Gerling, B. Koksch,

G. J. Martyna, B. W. Hoogenboom, A. Watts, J. Crain,C. R. M. Grovenor and M. G. Ryadnova, Proc. Natl. Acad.Sci. U. S. A., 2013, 110, 8918–8923.

79 N. O. Gekara, T. Jacobs, T. Chakraborty and S. Weiss,Cell. Microbiol., 2005, 7, 1345–1356.

80 C. Dıaz-Tejada, I. Ariz-Extreme, N. Awasthi and J. S. Hub,J. Phys. Chem. Lett., 2015, 6, 4799–4803.

81 I. Vorobyov, L. Li and T. W. Allen, J. Phys. Chem. B, 2008,112, 9588–9602.

82 M. Murate and T. Kobayashi, Chem. Phys. Lipids, 2016, 194,58–71.

83 K. Simons and E. Ikonen, Nature, 1997, 387, 569–572.84 E. L. Elson, E. Fried, J. E. Dolbow and G. M. Genin,

Annu. Rev. Biophys., 2010, 39, 207–226.

PCCP Paper

Publ

ishe

d on

05

Oct

ober

201

6. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Scie

nce

on 0

2/11

/201

6 10

:54:

59.

View Article Online