Complex dynamics at the nanoscale in simple biomembranes

Nature is known to engineer complex compositional and dynamical platforms in biological membranes. Understanding this complex landscape requires techniques to simultaneously detect membrane re-organization and dynamics at the nanoscale. Using super-resolution stimulated emission depletion (STED) microscopy coupled with fluorescence correlation spectroscopy (FCS), we reveal direct experimental evidence of dynamic heterogeneity at the nanoscale in binary phospholipid-cholesterol bilayers. Domain formation on the length scale of ~200–600 nm due to local cholesterol compositional heterogeneity is found to be more prominent at high cholesterol content giving rise to distinct intra-domain lipid dynamics. STED-FCS reveals unique dynamical crossover phenomena at length scales of ~100–150 nm within each of these macroscopic regions. The extent of dynamic heterogeneity due to intra-domain hindered lipid diffusion as reflected from the crossover length scale, is driven by cholesterol packing and organization, uniquely influenced by phospholipid type. These results on simple binary model bilayer systems provide novel insights into pathways leading to the emergence of complex nanodomain substructures with implications for a wide variety of membrane mediated cellular events.


Pressure-area isotherms and layer-by-layer phospholipid monolayer transfer
for fabrication of SLBs: Surface pressure versus area-per-molecule isotherms were recorded using the KSV LB rectangular mini trough (area, 240 cm 2 ) equipped with a Wilhelmy balance. A platinum sensor of accuracy 0.1 mNm -1 was used to measure the interfacial surface pressure. A chloroform solution of either one component DOPC, POPC and DMPC lipid or binary mixture containing varied cholesterol content such as 3:1, 2:1 and 1:1 were spread on the air-water interface of a LB trough at 151 ºC using a precise Hamilton syringe to make a compact monolayer. After evaporation of chloroform, the isotherms were recorded at constant temperature and a barrier speed of 5 mm/min. Figure S1 shows the pressure-area isotherm of pristine DOPC, POPC and DMPC monolayer and with varied cholesterol composition starting from 3:1, 2:1 and 1:1. The mean molecular area of DOPC, POPC and DMPC are ~85, 73 and 55 Å 2 /molecule respectively. The decrease in mean molecular area is due to the decrease in hydration of the head group and higher van der Waals interaction among the alkyl chains. Upon addition of 25% cholesterol in the respective monolayer, the mean molecular area further decreases complying with the condensing effect of cholesterol. At higher percentages of cholesterol such as 33 and 50%, the mean molecular area increases and no demixing behavior was observed (note, the step-like feature observed for pure cholesterol (Fig. S1a, green) is absent in mixed lipid) even at the highest cholesterol concentration used. This provides partial evidence that cholesterol precipitation does not occur at these concentrations. The condensing effect of cholesterol is due to the intermolecular cooperative interaction between the cholesterol -OH group and lipid head group and degree of rigidity of the monolayer. Our results also further supported by the maximum compressibility = − 1 data as shown Figs. S1b, d and f.
3 Figure S1. (a-c) illustrate surface pressure-area isotherms of pristine DOPC, POPC and DMPC respectively and in each panel, lipid to cholesterol ratios are 3:1, 2:1 and 1:1. The isotherms and the LB transfer of binary phospholipid-cholesterol mixtures were carried out at 151 ºC.
Supported lipid bilayers were prepared at a highly condensed surface pressure of 35 mN/m as marked in the horizontal dotted line. Right panels show the compressibility plots of the respective monolayers derived from the π-A data.
Supported lipid bilayers of single-or two-component lipids were formed upon controlled transfer of lipid interfacial monolayers onto pre-treated glass substrates by employing the 4 Langmuir-Blodgett (LB) method. During the preparation of bilayer using the LB method, multiple compression-expansion cycles were followed before the collapse surface pressure and subsequently the bilayers were transferred at a highly condensed surface pressure of 35 mNm 1 to the hydrophilized glass slides by using layer-by-layer transfer. Prior to transfer, glass substrates (20mm20mm, Germany) were cleaned using "piranha solution" (a 30:70 mixture of 30% hydrogen peroxide and concentrated sulfuric acid at 80 ºC) for 30 min and washed multiple times with MilliQ DI water (resistivity ~ 18.2 M.cm). The first monolayer was transferred at an equilibrium pressure of 35 mN/m by vertical withdrawal of the substrate at a speed of 5 mm/min with a transfer ratio of 1.20.1. The second monolayer transferred at the same surface pressure by a vertical down stroke yielded a centro-symmetric bilayer (Y-type) on the support. To make the bilayer luminescent, dye tagged lipid (Atto488-PE, 510 -4 mol%) was mixed thoroughly with pristine phospholipids or phospholipid-cholesterol mixtures before spreading at the air-water interface. After transfer, the bilayers were transferred to a container under water and stored at 25ºC for further use. All measurements were done on the prepared bilayers within 4-5 h of the LB transfer at 242 ºC.

STED-FCS Nanoscope
For imaging and FCS, we applied STED-FCS nanoscopy using a commercial STED setup (SP5x, Leica Microsystems GmbH, Mannheim, Germany). Ar ion laser and a STED laser (592 nm) with a continuous-wave (CW) mode were aligned in such a way to accomplish a doughnutshaped focal intensity distribution featuring a central intensity zero and diffraction-unlimited spot. The master power of the Ar laser was set to 25-30% and subsequently excitation at 488 nm was used at 1-25% output power. The CW-STED 592 nm laser was operated at 0-100% output power (varying in the range 0-260 mW measured directly at the focal plane of a 10x air objective). Before each series of measurements, the auto-alignment procedure (super-imposing the excitation laser and the depletion lasers) was performed in 25nm×25nm pixel area. This procedure was repeated every 15 minutes. An oil immersion objective, 100x 1.4 NA envisage focusing of the superimposed excitation and STED laser beams as well as collection of the fluorescence emitted intensity. The emitted intensity was guided to the microscope objective back aperture through the confocal pinhole (set to 1 Airy unit) filtered by a 594 nm notch filter imaged onto a single-photon-counting avalanche photo-diode (APD; Micro Photon Devices, 5 PicoQuant, Berlin, Germany) in the external port of the microscope with a band-pass filter (BS 560) between 500-550 nm. All the images as shown in Fig. S2-4 are 512  512 pixels; line average was set at 2 and scan speed at 600 Hz.

Analyses of FCS:
The particle number fluctuations ( ) = 〈 〉 + ( ) of fluorescing molecules entering and leaving the focus of a confocal microscope during the excitation could be calculated by measuring the emitted intensity ( ) as ( ) = 〈 〉 + ( ); where 〈 〉 and ( ) represents a constant offset and the fluctuations respectively. In FCS, the autocorrelation function ( ) from the intensity signal ( ) measured in the microscope was calculated using where 〈… 〉 , denotes a time average over the time variable t. The correlation curves were fit by Origin 8.5.0 (OriginLab) using two-dimensional one-component diffusion model assuming a Gaussian-shaped fluorescence detection profile, where, denotes the average transit time through the focal spot diameter (or FWHM) , represents the diffusion coefficient and the anomaly coefficient. Note that in Eqn. S2, exponent is a fit parameter. For all the studied bilayer systems, was found to be 10.08 (see Fig. S5).
The values of in confocal mode used throughout our calculation for estimating D were found to be 200 nm (at 100x, oil immersion objective) and was obtained from PSF by scanning where is the transit times corresponding to each STED power. In the above equation, we assume that the lipid diffusion in pristine DOPC SLB undergoes two-dimensional Brownian free diffusion and the diffusion time scales proportionally with the diffusion area.

Spatio-temporal lipid dynamics upon β-cyclodextrin treatment:
We explore the role of cholesterol in regions S and F (see Fig.1 in main manuscript) and their differences between different lipid types using cholesterol extraction experiment with βcyclodextrin (CD). The bilayer was exposed to 2.5 mM solution of β-cyclodextrin in milli-Q water and the simultaneous fluorescence intensity as well as the FCS traces in confocal mode was monitored with respect to the time. Figure Figure   S10b). In the case of POPC, similar transit times from both regions S and F were observed over the duration of the CD experiment indicating that cholesterol was not sufficiently partitioned to create distinct domains (cf. Fig.S10c and d).

Lipid dynamics on polymer cushioned bilayer:
Poly(acrylic acid) (PAA) cushioned were prepared (see methods in main text) and the thickness of the film was characterized by X-ray reflectivity (XRR) and atomic force microscopy technique (AFM). AFM (NT-MDT, Russia) measurements were performed in tapping mode with a cantilever of force constant ∼5 N m −1 and radius of curvature ∼8 nm. XRR measurement was carried out using Rigaku's SmartLab® diffractometer at Cu K (1.54 Å) radiation. The alignment of sample is completely automated with Rigaku's proprietary Guidance data acquisition software. X-ray reflectivity profiles depict variation of measured reflectivity R with the perpendicular momentum transfer, = 4 sin , where is the wavelength of the X-rays and is the angle of incidence. The thickness (d) of the polymer was estimated by using the expression = 2 Δ ⁄ and found to be ~11.6 nm (see Fig. S13c) which is consistent with the height profile analysis from AFM measurement (see Fig. S13b). In the presence of PAA cushion, at low concentration of cholesterol content (25%), the DOPC-Chl bilayer is homogeneous (Fig. S14a) and FCS diffusion law reveals Brownian diffusion ( 0 = 0) with a diffusivity value which is slightly less than the uncushioned bilayer (cf. Fig. S14b).

18
At 50% cholesterol content, DOPC and DMPC bilayers, reveal a heterogeneous distribution of fluorescence intensity in the confocal images and a bimodal diffusivity (Fig. S15, SI). In cushioned platform, the differences in diffusivity values between the two domains are slightly different than that of the supported bilayers prepared on the glass substrates in the absence of cushion. From the FCS diffusion law data in STED mode, we also observed a crossover in diffusion behavior in these two morphologically distinct domains like the trends observed for bilayers on glass substrate (cf. Fig. S16, SI). The negative intercept values were found to be -2.5 and -1.9 for DOPC:Chl(1:1) and DMPC:Chl(1:1) bilayers respectively in slow phase. The estimated domain size () in the respective bilayers were found to be 912 and 943 nm.