Article Information

• PDF (649 kb)
• Full Text with Large Figures
• Supporting Material
• Cited by in Scopus (18)

PubMed

• Articles by Jérôme Wenger
• Articles by Pierre-François Lenne

Related Articles

• Fluorescence Correlation Spectroscopy Diffusion Laws to Prob...

  Fluorescence Correlation Spectroscopy Diffusion Laws to Probe the Submicron Cell Membrane Organization Biophysical Journal, Volume 89, Issue 6, 1 December 2005, Pages 4029-4042 Laure Wawrezinieck, Hervé Rigneault, Didier Marguet and Pierre-François Lenne Abstract To probe the complexity of the cell membrane organization and dynamics, it is important to obtain simple physical observables from experiments on live cells. Here we show that fluorescence correlation spectroscopy (FCS) measurements at different spatial scales enable distinguishing between different submicron confinement models. By plotting the diffusion time versus the transverse area of the confocal volume, we introduce the so-called FCS diffusion law, which is the key concept throughout this article. First, we report experimental FCS diffusion laws for two membrane constituents, which are respectively a putative raft marker and a cytoskeleton-hindered transmembrane protein. We find that these two constituents exhibit very distinct behaviors. To understand these results, we propose different models, which account for the diffusion of molecules either in a membrane comprising isolated microdomains or in a meshwork. By simulating FCS experiments for these two types of organization, we obtain FCS diffusion laws in agreement with our experimental observations. We also demonstrate that simple observables derived from these FCS diffusion laws are strongly related to confinement parameters such as the partition of molecules in microdomains and the average confinement time of molecules in a microdomain or a single mesh of a meshwork. Abstract | Full Text | PDF (516 kb)

• Statistical approach for subwavelength measurements with a c...

  Statistical approach for subwavelength measurements with a conventional light microscope Biophysical Journal, Volume 60, Issue 5, 1 November 1991, Pages 1147-1155 Daniel Palanker and Aaron Lewis Abstract A method is developed theoretically that will permit subwavelength measurements of objects that differ from the surroundings by any contrast enhancing parameter, such as fluorescence, second harmonic generation, reflection et cetera, using a statistical analysis of a picture obtained with a conventional light microscope through a set of subwavelength apertures or by repeated scanning of a laser beam over a defined area. It is demonstrated that with this methodology information can be obtained on microdomains that are thirty times less than the diameter of the aperture. For example, for apertures that are 0.3μm in diameter it is possible to measure the dimension of objects that are ∼10 nm. A technology is described by which it is possible to produce masks with the appropriate apertures. Instrumentation is described that would allow for the realization of these statistical methodologies with either apertures or scanning laser beams. Abstract | PDF (1085 kb)

• Pollen aperture evolution – a crucial factor for eudicot suc...

  Pollen aperture evolution – a crucial factor for eudicot success? Trends in Plant Science, Volume 9, Issue 3, 1 March 2004, Pages 154-158 Carol A. Furness and Paula J. Rudall Abstract Increased aperture number in angiosperm pollen grains offers a potential selective advantage because it increases the number of prospective germination sites, thus facilitating contact between at least one aperture and the stigmatic surface. Such an increase occurred at the base of the eudicot clade, coupled with an apparently fundamental shift in aperture position from polar to equatorial. This transition could represent a key innovation underlying eudicot success and subsequent radiations. There is a general trend in angiosperms to an increase in pollen aperture number, suggesting that pollen apertures are under strong selection pressure. This trend can be observed by comparing the aperture patterns in basal angiosperms (including monocots) with those of eudicots. Different developmental strategies operate for pollen apertures in basal angiosperms and eudicots; the eudicot strategy has been more successful in generating higher aperture numbers. Abstract | Full Text | PDF (819 kb)

• …more

Copyright © 2007 The Biophysical Society. All rights reserved.
Biophysical Journal, Volume 92, Issue 3, 913-919, 1 February 2007

doi:10.1529/biophysj.106.096586

Membranes

Diffusion Analysis within Single Nanometric Apertures Reveals the Ultrafine Cell Membrane Organization

Jérôme Wenger^*, †, Fabien Conchonaud^{‡, §, ¶}, José Dintinger^∥, **, Laure Wawrezinieck^{*, †, ‡, §, ¶}, Thomas W. Ebbesen^∥, **, Hervé Rigneault^*, †, Didier Marguet^{‡, §, ¶} and Pierre-François Lenne^*, †, ,   

^* Institut Fresnel, Mosaic Group, Université Paul Cézanne Aix-Marseille III, Domaine Universitaire de Saint Jérôme, Marseille Cedex, France
^† CNRS UMR 6133, Marseille, France
^‡ Centre d’Immunologie de Marseille Luminy, Université de la Méditerranée, Parc Scientifique de Luminy, Case 906, Marseille Cedex, France
^§ CNRS UMR 6102, Marseille, France
^¶ INSERM UMR 631, Marseille, France
^∥ ISIS, Université Louis Pasteur, Strasbourg, France
^** CNRS UMR 7006, Strasbourg, France

Address reprint requests to Pierre-François Lenne, Institut Fresnel, Domaine Universitaire de Saint Jérôme, 13397 Marseille Cedex 20, France. Tel.: 33-491-288-494; Fax: 33-491-288-067.

Understanding the cell membrane organization is a major biological issue, with implications in cell signaling, adhesion, and trafficking 1,2,3. Among the debated questions, most attention was paid to membrane heterogeneities, which are commonly investigated by monitoring the diffusion of lipids and membrane proteins. For years, most measurements of lateral diffusion have been made by fluorescence recovery after photobleaching (FRAP) 4,5, showing that membrane heterogeneities impede the diffusion of lipids and proteins 6. However, FRAP investigates an area at least ten times larger than the laser spot, so that it is therefore difficult to infer deviations from Brownian diffusion. Single particle tracking (SPT) offers a powerful alternative by tracking individual labeled molecules with a high spatial resolution of ∼10–30nm 7,8. SPT was successfully applied to reveal membrane corrals 9 and domains 10, but it appears presently limited by a low temporal resolution (often in the millisecond range) and a tedious data analysis. Fluorescence correlation spectroscopy (FCS) is a third method, used to monitor the mobility of individual fluorescent probes across a well-defined observation volume 11. Compared to SPT, FCS allows a better temporal resolution and also an easier data analysis, as statistical averaging is directly carried out in FCS. The main limitation of standard FCS is its spatial resolution of ∼300nm set by the diffraction of light. This drawback can be overcome by using nanometric apertures milled in a metallic film 12,13,14, but no quantitative application to the question of membrane organization has ever been demonstrated. Moreover, while performing FCS experiments with only a single nanoaperture size, it appears almost impossible to extract relevant information on membrane heterogeneities.

In this study, we describe the development of a methodology to probe the ultrafine organization of living cell membranes by combining FCS with single nanometric apertures of different sizes. The main innovation is that our strategy allows us to identify the mechanism controlling the diffusion of the different molecular components and also provides an estimate for the characteristic size of the nanometric membrane heterostructures. This yields a technique having both high spatial and temporal resolution together with a direct statistical analysis. Living cells with fluorescently labeled membrane components (lipid analogs and green fluorescent protein (GFP)-tagged proteins) are incubated over isolated nanoapertures milled in a metallic film with radii between 75 and 250nm. The single apertures act as pinholes directly located under the cell membrane and restrict the observation area below the diffraction limit (this is illustrated in Fig. 1). Performing FCS experiments with increasing aperture sizes, we clearly demonstrate that transient regimes are observed when the probed area is close to the size of the confining structures, revealing nanometric membrane heterogeneities. Based on numerical simulations and previous theoretical work 15, we identify the mechanism controlling the diffusion and give an estimate of the characteristic size of the heterogeneities. To validate the method, we compare the diffusion behaviors of different membrane components. For every molecule that we probed, the data obtained within larger nanoholes agrees well with the results using diffraction-limited spots and extensive drug treatments 16. This further rules out the possibility that the observed regimes are artifactual.

Display large version of this figure

Figure 1

(A) Electron micrograph image of a set of nanoapertures of various diameters under oblique illumination. This set was only intended for display, as isolated apertures were used in this study. (B) Experimental arrangement of the sample.

The GFP- glycosylphophatidylinositol (GPI) (gifted from A. LeBivic, IBDML, Marseille) and human transferrin receptor (TfR)-GFP constructs were previously described 16. All experiments were carried out on COS-7 cells (ATCC, CRL-1657). Cells were grown at 37°C, 7% CO₂, in DME supplemented with 10% fetal calf serum, glutamine, and sodium pyruvate. Transfections were performed with ExGen 500 as per the manufacturer's instructions (Euromedex, Souffelweyersheim, France) and stable expressing cells were cloned after selection for G418 resistance.

Optically opaque aluminum films (thickness 220nm) were deposited on standard microscope glass coverslips (thickness 150μm). Focused ion beam technique (FIB) was then used to mill circular nanometric apertures with radii ranging from 75 to 250nm. The samples were washed for 3min in an ultrasonic cleaner before rinsing with diluted ethanol (70°) and evaporation of the alcohol residue. We placed a drop of culture medium containing trypsinized COS-7 (ATCC, CRL-1657) cells 40h before the FCS experiments, to allow cell-substrate adhesion. Cells adhered to the metal surface and conformed to nanoapertures spontaneously.

Phosphatidylcholine (PC) and ganglioside (G_M1) lipids were labeled with the 4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene (BODIPY) fluorophore (Molecular Probes, Leiden, The Netherlands). Equimolar complexes of 0.05μM BODIPY phosphatidylcholine PC (FL-PC) or BODIPY ganglioside G_M1 (FL-G_M1) with defatted BSA were prepared in HBSS/HEPES buffer (Hanks’ buffered salt solution containing 10mM HEPES pH 7.4). Cell cultures were washed in HBSS/HEPES, incubated with the 0.05μM lipid/BSA complex in HBSS/HEPES for 30min at 20°C, rinsed and incubated subsequently in HBSS/HEPES at 37°C. FL-PC staining was done immediately previous to the FCS measurements and FL-G_M1 staining at least 12h before the measurements.

To modify the cholesterol composition of the plasma membrane, cells were treated with Streptomyces sp. cholesterol oxidase (COase, Calbiochem) in serum-free HBSS/HEPES buffer, incubated with COase at 1 U/mL for 60min at 37°C and finally rinsed in HBSS/HEPES buffer. All FCS measurements were completed within 60min after cellular treatments. In this condition, ∼20% of the total cell cholesterol was converted into cholestenone 16. We think that the effect of cholesterol oxidation on raft disorganization originates from a combined effect of cholesterol concentration decrease and production of cholestenone bearing antagonizing action to ordered domains 17.

FCS experiments were performed on a custom setup based on an Axiovert 200M microscope (Zeiss, Jena, Germany) with a Zeiss C-Apochromat 40×, numerical aperture 1.2 objective lens and a three-axis piezo-scanner (Physik Instrument, Karlsruhe, Germany). The laser power was lowered to 2.5μW to avoid cell damage and dye photobleaching. Fluorescence was collected by the same objective, filtered by a dichroic mirror and a confocal pinhole before detection by two avalanche photodiodes through a 50/50 splitter and 535±20nm bandpass filters. Cross correlations using a hardware correlator (ALV-GmBH, Langen, Germany) between the two photodiodes was used to reduce artifacts. For each aperture diameter, FCS measurements were performed on a minimum of 6 different cells. Each measurement was obtained from 20 runs of 5s duration. For experiments with diffraction-limited beams, the observation areas were calibrated using FCS measurements on rhodamine-6G in aqueous solution (the diffusion coefficient was taken as 280μm²/s). Due to technical concerns with the microscope apparatus at the beginning of this study, FCS measurements on living cells were carried out at 27°C for lipid analogs and at 37°C for GFP-tagged proteins.

For free Brownian two-dimensional diffusion in the case of a Gaussian molecular detection efficiency, the fluorescence autocorrelation function (ACF) is given by

(1)

For FCS measurements on lipids analogs, a single membrane component fit following Eq. (1) was implemented. Even for the smallest apertures used, this procedure efficiently fitted the experimental data, as shown on Fig. 2. In FCS experiments conducted on the chimeric proteins, two diffusive species were associated to the ACF, as already reported in 16. The faster component had diffusion times τ_fast of the order of 0.5–1ms (linked to species diffusing in the intracellular pool), whereas the longer component experienced clearly distinct diffusion times τ_d of ∼20ms (two-dimensional diffusion across the membrane). This is taken into account in the ACF numerical fits by introducing a supplementary free variable A to account for the relative weight of the slow (membrane-bound) species:

(2)

Display large version of this figure

Figure 2

(A) Normalized autocorrelation functions and numerical fits obtained for FL-PC using an open diffraction-limited beam (240-nm waist) or a 75-nm radius aperture. The upper graph shows the residuals of the fit for the 75-nm aperture case. (B) Same as A for the FL-G_M1 lipid analog. (C) FCS diffusion laws for FL-PC (squares) and FL-G_M1 (circles). The open markers correspond to open beam measurements, whereas the solid markers refer to the nanoaperture experiments (the dotted line is to guide the eyes). Solid lines are numerical fits to the open-beam experiments; these lines are extended below the diffraction limit by the dash-dotted lines. Errors bars represent the mean standard deviation.

In open-beam FCS experiments and for the wider apertures, A was found to be of the order of 65–70%. When the aperture radius was reduced, the depth of field was also restricted, which limited the influence of the cytosolic fluorescing species. Typically, for a 150-nm radius aperture, A was ∼75%. It increased to 85% for a 110-nm aperture, and to almost 100% for a 75-nm radius aperture. Thus for these smallest apertures, the diffusion time τ_d could be simply taken as the ACF lag time at half-maximum.

The method reveals different diffusion behaviors for lipid analogs FL-PC and FL-G_M1, indicating either free-like or hindered diffusion.

We first examine the diffusion of the fluorescent phosphatidylcholine FL-PC in a COS-7 cell membrane. Figure 2A shows two fluorescence ACFs obtained for FL-PC using a 75-nm radius nanohole and a diffraction-limited spot of 240-nm waist. The ACFs were numerically fitted to determine the diffusion time τ_d (see the Methods section for a description of the fitting procedure). The extracted diffusion times were then plotted against the aperture area and against the laser spot area (as in 15,16, this curve is referred to as FCS diffusion law, see Figure 2C). Aperture radii varied from 75 to 250nm in radius, whereas diffraction-limited spot waists were set from 240 to 350nm by changing the microscope objective's filling factor 15. Figure 2C shows that the diffusion time τ_d for FL-PC varies almost linearly with the observation area over the whole range being probed. A small slope change may be pointed out around r ∼ 125nm (area ∼5×10⁴nm²), but this comes close to the sensitivity of our apparatus, and appears a minor discrepancy compared to the transient regimes found for the other membrane components (see the discussion in the Supplementary Material online). Two conclusions can thus be drawn: i), FL-PC diffusion within the membrane is apparently unhindered by the nanoaperture; and ii), the observation volume is set by the aperture area.

Although cells, culture conditions and labeling methods were the same, ganglioside analogs FL-G_M1 have a completely different diffusion behavior, as reported in Figure 2C. At aperture radii below 100nm, τ_d increases almost linearly with the aperture area. FL-G_M1 diffusion then experiences a marked transition at a characteristic radius of ∼100nm, before taking an affine regime. From the slope at r<80nm, expressing τ_d as τ_d=area/(4πD), we estimate an effective coefficient D=0.36±0.10μm²/s. For larger observation areas (r>200nm), τ_d can be written as τ_d=area/(4πD)+t₀, with t₀=16.3±1.5ms and D=1.7±0.2μm²/s. The difference between these regimes enlightens the transition between the diffusion behaviors, even if D cannot be straightforwardly interpreted as the exact diffusion coefficients before and after the transition.

The relative change of diffusion behaviors measured for FL-PC and FL-G_M1 guarantees the relevance of the method. These lipid analogs exhibit very different diffusion laws, indicating that our observations depend on the probed component. For the range of nanoaperture diameters tested, this rules out the possibility that the observed regimes are artifactual. This point is reinforced by the fact that for every marker the diffusion times within larger nanoholes (r>200nm) agrees nicely with those obtained using diffraction-limited spots. It is probable that the membrane area within nanoholes does not equal the aperture area, but the correction could only account for slight changes of diffusion laws but cannot explain the different behaviors (see the discussion in the Supplementary Material online). Hereafter, we will extract information on the heterogeneities hindering the diffusion of FL-G_M1, but before we describe experimental data on the diffusion of GPI-anchored proteins.

We investigated the diffusion of the green fluorescent protein (GFP) anchored to a glycosylphophatidylinositol (GPI). Fig. 3 summarizes the results, showing a typical ACF function obtained for a 95-nm radius nanoaperture together with the corresponding curve of the average diffusion time versus the aperture area. Fig. S2 of the Supplementary Material shows nonnormalized ACFs in nanoapertures of various radii. The FCS diffusion law of GFP-GPI displays a marked transition at a characteristic radius of ∼120nm (Figure 3B). From the slope at r<100nm, we estimate an effective coefficient D=0.30±0.05μm²/s, whereas for r>140nm, we get D=0.9±0.2μm²/s, which is ∼3 times the value found for r<100nm (expressing τ_d=area/(4πD)+t₀, with t₀=8.7±1.3ms). Again, the results obtained for the largest apertures match well with those obtained using diffraction-limited beams, indicating that the probed area is defined by the aperture area and that the aperture does not strongly alter the diffusion process.

Display large version of this figure

Figure 3

(A) Normalized autocorrelation functions and numerical fits for GFP-GPI in a 95-nm radius aperture with (gray) or without (black) 1U/mL COase treatment. The upper graph displays the residuals of the fits. (B) FCS diffusion laws for untreated GFP-GPI (gray circles) and GFP-GPI with 1U/mL COase (black squares). Open markers correspond to open beam measurements, and solid markers refer to the nanoaperture experiments. Solid lines are numerical fits to the open-beam data. Errors bars represent the mean standard deviation.

We next modified the cholesterol composition of the plasma membrane, since this component is likely to play a significant role in the formation and stability of lipid domains. To achieve this, cholesterol was converted into cholestenone by a cholesterol oxidase (COase) treatment at 1 U/mL. This treatment had a dramatic effect on the diffusion of GFP-GPI, as a linear behavior of the diffusion time versus the probed area was recovered after COase treatment (see Figure 3B). This free-like two-dimensional diffusion held true for every spot size from 75 to 380nm, with an effective coefficient D=0.72±0.05μm²/s and an almost null intercept time at origin t₀=0.1±0.3ms consistent with unhindered diffusion 16. This effect confirms that the observed diffusion regimes are not induced by the nanoapertures. The diffusion of GFP-GPI is clearly shown to be cholesterol-dependent, which is consistent with previous studies 16,18.

We recently developed numerical models of diffusion into a membrane with confinement structures 15. These structures are areas which confine molecules transiently: their boundaries are sufficiently impenetrable so that the confinement time within the corral is longer than the diffusion time across it. We first consider the case of isolated domains, as indicated in Figure 4A. In a previous theoretical study, we have shown that diffusion in a such landscape leads to FCS diffusion laws with a marked transition from linear to affine with a positive shift 15. Briefly, isolated domains are assumed to be circular with radius a, periodically spaced and static. In this model, molecules can diffuse in and out of the domains with a prescribed probability. The simulations do not depend on the position of the observation area relative to the confining domains, as we averaged the simulated data for the different positions of the beam area. The simulated average diffusion time τ_d is displayed in Figure 4B versus the normalized area

Display large version of this figure

Figure 4

Simulated diffusion laws for isolated domains (A). The molecules diffuse inside and outside the domains, and are trapped transiently within them. Diffusion is measured by FCS within the probed area (gray disk), leading to the FCS diffusion law (B). Simulation parameters 15: domains density d=0.55, , D_in/D_out=1/5.

The features observed for FL-G_M1 and GFP-GPI (slope change and transient regime) are nicely reproduced by this microdomain model, the transitions being understood as the crossover from confined to normal diffusion. The characteristic transition in our simulations occurs at an observation area corresponding to . We can thus estimate the radii of the confining structures to be ∼40nm for GFP-GPI and 30nm for FL-G_M1. These results come fully within the framework of lipid rafts 19,20.

A cytoskeleton-mediated meshwork also plays a critical role in the membrane lateral organization 9,21. We thus investigated the diffusion of the human transferrin receptor TfR, a transmembrane protein was tagged with the fluorescent protein EGFP (TfR-GFP) whose diffusion is sensitive to the actin cytoskeleton. Fig. 5 displays a typical ACF function obtained for a 210-nm radius nanoaperture, together with the corresponding FCS diffusion law. The results for TfR-GFP clearly differ from the diffusion laws found for GFP-GPI and FL-G_M1. Figure 5B shows two transitions, respectively at r ∼ 150nm and r ∼ 230nm. As the probed area increases, the diffusion time grows first with an apparent coefficient D=0.23±0.02μm²/s, then levels off to D=0.7±0.1μm²/s, before increasing again with an effective diffusion rate D=0.27±0.03μm²/s and a negative time intercept at the origin t₀=-22±2ms.

Display large version of this figure

Figure 5

(A) Autocorrelation function and numerical fit for TfR-GFP in a 210-nm radius aperture, where the characteristic transition takes place. The upper graph displays the residuals of the fit. (B) FCS diffusion laws for TfR-GFP. The open markers correspond to open beam measurements, whereas the solid markers refer to the nanoaperture experiments (the dotted lines are to guide the eyes).

To describe the diffusion within the cytoskeleton-mediated meshwork, we developed a numerical model including a network of barriers, as displayed in Fig. 6. Molecules diffuse within corrals and hop between domains with a prescribed probability. As indicated in Figure 6B, the transition observed at large area for TfR-GFP comes well within the barrier network model, confirming that fences play a critical role in the compartmentalization of transmembrane proteins. For the meshwork model, the characteristic transition in the FCS law occurs at . From the measurements on TfR-GFP, the characteristic size of the network can be estimated to be ∼230nm, which appears consistent with the mesh size found for the actin meshwork in rat kidney fibroblast 21.

Display large version of this figure

Figure 6

Simulated diffusion laws for a network of barriers defining contiguous corrals (A). Molecules diffuse within corrals and move from one domain to a neighbored one with a prescribed probability. Diffusion is measured by FCS within the probed area (gray disk), leading to the diffusion law (B). We considered ; details of simulations can be found in Wawrezinieck et al. 15.

The shape of the FCS diffusion law in small radii nanoapertures suggests that confinement within discrete microdomains may also occur. This point is reinforced by previous studies using diffraction-limited laser beams 16, where it was shown that high cytochalasin D concentrations disrupted the actin meshwork and led to a positive y-axis intercept t₀. When both actin-based fences and cholesterol were simultaneously modified, the diffusion time was found to be linear with the probed area, thus recovering free-like diffusion behavior. It thus seems likely that the effects of fences and microdomain confinement add each other. From the results displayed in Figure 5B for probed areas below 10⁵nm² and the microdomain numerical model, we can thus estimate the radii of the microdomain confining structures to be ∼50nm for TfR-GFP.

We described what we believe is a novel optical approach to explore the ultrafine organization of the plasma membrane. Combining single nanometric apertures of different sizes with FCS, we observe different diffusion regimes, which reveal the kind and the size of the nanometric membrane heterostructures. Compared to conventional FCS, our method has a high spatial resolution, necessary to quantify the size of membrane heterogeneities at the submicron scale. Alternatively to single particle tracking, our approach takes advantage of a high temporal resolution at the microsecond range together with a simple data analysis. Thanks to the recent technical progress in nanotechnology and the development of numerous nanofabrication facilities, the well-established method of FCS can be readily improved by this method. There is thus a high immediate practical relevance for the quantitative study of membrane domains and molecular interactions in membranes.