Abstract

The existence of lipid rafts and their importance for immunoreceptor signaling is highly debated. By non-invasive single molecule imaging, we analyzed the dynamics of the T-cell antigen receptor (TCR), the lipid raft-associated glycosylphosphatidylinositol (GPI) proteins CD48 and CD59 and the major leukocyte phosphatase CD45 in living naive T lymphocytes. TCR triggering induced the immobilization of CD45 and CD48 at different positions within the T-cell interface. The second GPI protein, CD59, did not co-immobilize indicating lipid raft heterogeneity in living T lymphocytes. A novel biochemical approach confirmed that lipid raft components are not associated in the plasma membrane of resting cells, and variably associate with specific receptors to distinct lipid rafts upon activation.

Introduction

T-cell activation via the antigen-specific TCR–CD3 complex is the central process of the immune response initiation. Immediately after a T cell recognizes several cognate peptide–MHC complexes by its clonotypic TCR, numerous signal transduction pathways are initiated in the time window of seconds to minutes. During this initial period, the T cell reorganizes its plasma membrane receptors in a concerted fashion leading to the formation of a well-structured contact zone between the T cell and the antigen-presenting cell (APC) known as immunological synapse (1, 2). Triggered TCR complexes move from the peripheral ring into the so-called central supramolecular activation cluster. This process is driven by segregation of proteins according to their size (3), by actin cytoskeleton-mediated plasma membrane re-organization (4) and by vesicle recycling (5).

Simultaneously, membrane structures with higher lipid order accumulate within the developing synapse (6), accompanied by a change in the overall lipid composition (7). Indeed, several reports indicate that cholesterol- and sphingolipid-enriched plasma membrane microdomains—also termed lipid rafts—associate with the engaged TCR (8–14), which is supposed to represent a pivotal step for both antigen recognition and signaling. First, raft-resident co-receptors such as CD4/CD8 (15) have been found to substantially increase TCR sensitivity toward antigen, presumably via a stabilization of the rare TCR/agonistic peptide–MHC encounters. Second, major signaling enzymes are described as resident constituents of lipid rafts including Src family kinases (SFKs), heterotrimeric G proteins and numerous adaptor proteins such as LAT and PAG as well as glycosylphosphatidylinositol (GPI)-anchored proteins including CD48, CD55 and CD59 [for review see (16–18)]. In the resting state, membrane compartmentalization by lipid rafts facilitates the efficient separation of effector signaling molecules (e.g. the protein tyrosine kinase lck) from their substrates, i.e. transmembrane receptors. Upon activation this segregation is released, resulting in signal transduction and final naive T-cell activation. A variety of studies have recently described lipid raft heterogeneity (19–26), which might specifically influence the T-cell activation process in different stages of T-cell development.

However, all these hypotheses about existence, size and function of lipid rafts are still controversial mainly due to the lack of methods which allow for studying lipid rafts and the associated molecules directly in living cells without any experimental intervention, i.e. fixation, detergent solubilization or cholesterol extraction. Thus, despite the large number of studies performed, the structure and precise role of rafts for signaling via plasma membrane receptors is still highly debated, especially with regard to the initial activation period before their coalescence into larger clusters.

To elucidate the role of lipid rafts for T-cell activation, we studied interactions of putative raft-resident and -excluded molecules during the formation of the T-cell interface with a functionalized surface by a non-invasive single-molecule imaging approach. Clear differences in the association kinetics of the probe molecules with the TCR demonstrate for the first time in living cells, the heterogeneous nature of lipid rafts. A novel biochemical approach detecting the re-organization of molecular complexes associated with the TCR upon formation of the T-cell interface confirms these findings.

Methods

Antibodies

The mAbs of the MEM series used in this study were produced in the laboratory of Vaclav Horejsi (Institute of Molecular Genetics, Academy of Sciences of the Czech Republic, Prague, Czech Republic): MEM-57 (CD3e; IgG2a), MEM-31 (CD8; IgG2a), MEM-174 (CD11b; IgG2a), MEM-18 (CD14; IgG1), MEM-168 (CD16; IgM), MEM-181 (CD25; IgG1), MEM-28 (CD45; IgG1), MEM-102 (CD48; IgG1), MEM-188 (CD56; IgG2a), MEM-43/5 (CD59; IgG2b), MEM-147 (MHC class I; IgG1), MEM-36 (MHC class II; IgG1). The mAb B-D3 (CD19; IgG1) was a kind gift of John Wijdenes (Diaclone, Besancon, France), UCHL-1 (CD45RO; IgG2a) of Peter Beverley (Jenner Institute, Compton, UK), FN50 (CD69; IgG1) of Steinar Funderud (Norwegian Radium Hospital, Oslo, Norway). mAb LA45 to the free heavy chain of MHC class I was provided by Otto Majdic (Institute of Immunology, Medical University of Vienna). mAb OKT-3 to CD3 was obtained from Ortho Pharmaceuticals (Raritan, NJ, USA). The rabbit polyclonal antibody to the CD3e chain was from Santa Cruz Biotechnology (Santa Cruz, CA, USA), rabbit polyclonal antobodies to Lck and LAT and the anti-phosphotyrosine mAb 4G10 (horse radish peroxidase conjugated) were from Upstate Biotechnology (Lake Placid, NY, USA). Goat anti-rabbit IgG and rabbit anti-mouse IgG, both horse radish peroxidase conjugated, were from Bio-Rad (Richmond, CA, USA) and Sigma-Aldrich (St Louis, MO, USA), respectively.

Fab preparation

Fab fragments were produced by standard papain digestion of the parental mAb in the presence of 2-mercaptoethanol, followed by jodoacetamide for stopping the reaction. For labeling, the preparations were treated either with AlexaFlour555 or AlexaFlour647 (Molecular Probes, Eugene, OR, USA). Labeled Fabs were purified by gel filtration on a Superdex 200/30 HR column (Amersham Biosciences, Uppsala, Sweden). The Fabs eluted between 27 min and 31 min in comparison to low molecular digestion products (elution time >32 min) and the intact mAb (21–25 min). The purity of the isolated Fabs was further assessed by SDS–PAGE. To test whether the labeling procedure did not cause alteration in the specificity, the reactivities of the Fabs were compared with the untreated intact mAbs using a panel of positive and negative cell types. A particular attention was drawn to underlabel the Fab preparations in order to assure a <1:1 fluorophore/protein ratio, which is well suited for single-molecule experiments.

Cells

For single-molecule analysis, highly purified naive CD4+ T cells isolated from human peripheral blood buffy coats obtained from healthy volunteers (Red Cross, Linz/Vienna, Austria) were used throughout the study. In brief, after defibrination using glass beads, the mononuclear cells were collected using Ficoll-Paque centrifugation (Amersham Biosciences, Uppsala, Sweden). Afterwards, the cells were stained on ice with a cocktail of saturated PE-labeled mAbs for negative depletion (mAbs to CD8, CD11b, CD14, CD16, CD19, CD20, CD25, CD45RO, CD56, CD69 and MHC class II. All of these mAbs were conjugated to biotin and presaturated with streptavidin-PE as well as with 2 mg ml−1 Alexa555 or Alexa647-labeled Fab fragments to CD3, CD48, CD45 or CD59. In addition to negative sorting of PE-labeled cells, we used narrow gates at Alexa555 and Alexa647 channels to enrich for subsequent microscopic evaluation of CD4+ T cells expressing the particular Fab-labeled molecules at similar density. All sorting was performed on a two-laser (488, 635 nm) FACSAria (Becton Dickinson, Franklin Lakes, NJ, USA). The sorted cells were kept strictly on ice before placing them to the temperature-controlled slide holder POC Mini (Zeiss, Germany) for single-molecule microscopy. For imaging, the cells were immersed in HBSS supplemented with 10% autologous serum plus 2 mM CaCl2 and 1 mM MgCl2. RPMI medium with 10% autologous serum was used for the maintenance of mononuclear cells in a 5% CO2 atmosphere. Periodically, the purity of the sorted CD4+ T cells was confirmed to be higher than 95%.

For biochemical studies and confocal microscopy, we purified CD4+ T cell from mononuclear cell preparations by negative depletion using CD8, CD14, CD16, CD19, CD20, CD25 and CD56 mAbs and the MACS technique of Miltenyi Biotec (Bergisch Gladbach, Germany). Afterwards, the T cells were re-suspended in complete RPMI 1640 medium supplemented with 5% heat-inactivated FCS, 2 mM L-glutamine, 100 U ml−1 penicillin and 100 μg ml−1 streptomycin.

Coverslip functionalization

Epoxide-bearing glass slides were prepared as described (27). Briefly, glass slides of 150 μm nominal thickness were cleaned in CHCl3:methanol and deionized water, activated in Piranha solution (30:70, 30% aqueous H2O2:H2SO4) and washed with deionized water. The slides were then silanized by immersion in a 1% solution of 3-glycidoxypropyl trimethoxysilane in toluene for 8 h, followed by washing with acetone and drying in a stream of nitrogen. For quality purposes, the contact angles of water were determined for the epoxide surfaces and found to be between 65° and 75°. For the preparation of aldehyde-derivatized slides, the epoxide groups of the silanized surface were hydrolyzed using mild acidic conditions, and the resulting diols were subjected to oxidative cleavage by incubating the slides in an aqueous solution of NaIO4. The aldehyde slides were washed extensively with water and dried in a stream of nitrogen.

Immunoprecipitation, solid phase signalosome precipitation and western blotting

For immunoprecipitation, purified CD4+ T lymphocytes (2 × 107 ml−1) were lysed for 25 min at 4°C in ice-cold lysis buffer [20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA and protease inhibitors (5 μM aprotinin, 5 μM leupeptin, 5 μM pepstatin, 1 mM phenylmethylsulfonylfluoride (PMSF); all from Sigma-Aldrich) containing 1% NP-40 or 1% Brij58 (Pierce, Rockford, IL, USA)]. The insoluble material was removed by centrifugation at 13 000 rpm for 2 min at 4°C and the supernatant distributed to the wells of a 96-well plate coated with the respective mAbs (20 μg ml−1). After incubating the wells with the cell lysate (50 μl well−1) over night at 4°C, the wells were washed and 1× SDS–PAGE loading buffer was added (7–10 μl well−1) to collect and analyze the immunoprecipitates by SDS–PAGE and western blotting.

For solid phase signalosome precipitation (SPSP), purified CD4+ T lymphocytes (1 × 107 ml−1) were incubated in ice-cold RPMI 1640 with 5% heat-inactivated FCS, 2 mM L-glutamine, 100 U ml−1 penicillin, 100 μg ml−1 streptomycin and 20 mM HEPES for 1 h at 4°C. Afterwards, the cells were added to 6-well plates (1 ml well−1) coated with respective mAbs (10 μg ml−1) and blocked with RPMI 1640 supplemented with 10% heat-inactivated FCS. For synchronization of activation, the cells were shortly spun down using a plate rotor (3 sec at 450 rpm). Then the 6-well plates were incubated at 37°C for the indicated periods of time. After this stimulation period, the medium was carefully removed and the attached cells were lysed for 12 min at 4°C using ice-cold lysis buffer [20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA and protease inhibitors (5 μM aprotinin, 5 μM leupeptin, 5 μM pepstatin, 1 mM PMSF, 1 mM sodium orthovanadate, 50 mM NaF; all from Sigma-Aldrich] containing 1% NP-40 (Pierce). Subsequently, the lysate was collected and the insoluble material removed by centrifugation at 13 000 rpm for 2 min at 4°C. The 6-well plates with the mAb-induced and -precipitated cell surface protein complexes were washed three times with ice-cold washing buffer [20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 0.1 mM sodium orthovanadate] containing 0.1% Tweed 20 (Sigma-Aldrich). Immunoprecipitated protein complexes were harvested using 1× SDS–PAGE loading buffer, resolved together with the non-precipitated material by SDS–PAGE and analyzed by immunoblotting.

Confocal microscopy

Purified CD4+ T lymphocytes were incubated in RPMI 1640 medium for 2–3 h at 37°C, washed twice in 2% FCS/PBS and then re-suspended in HBSS with 1 mM Ca2+ and Mg2+ (2 × 106 cells ml−1). The cells were gently placed on CD3 mAb (MEM-57)-coated coverslips pre-treated with aldehyde and stimulated at 37°C for various time intervals. The incubation was stopped by fixing the cells with 3.75% PFA in PBS. After washing twice with 2% FCS/PBS, the cells were blocked with 2% beriglobin (Hoechst Pharmaceuticals, Frankfurt, Germany) in PBS for 20 min. Then, the cells were incubated for 30 min with directly labeled mAbs. Following staining, the cells were washed twice with 2% FCS/PBS and analyzed on a confocal microscope (LSM 510 Zeiss, Carl Zeiss AG, Jena, Germany) by making serial z stacks from bottom to top.

Single-molecule microscopy

For a comprehensive description of instrumentation required for single-molecule microscopy on living cells (28) and its methodological advantages (29–31), we refer to the recent literature. Experiments were performed on a modified epi-fluorescence microscope (Axiovert 200, Zeiss) equipped with a 100× NA = 1.4 Plan-Apochromat objective. Samples were illuminated via the epiport using Ar+ (Model 2020, Spectra Physics, USA) and Kr+ ion lasers (Innova, Coherent, USA) for selective fluorescence excitation of Alexa555 at 514 nm and Alexa647 at 647 nm, respectively. Proper alignment of the Köhler illumination allowed a homogenous excitation profile. Samples were illuminated with an intensity of 0.75–1.7 kW cm−2, which allowed for the reliable detection of individual fluorophores at illumination times of till = 1, 3 or 5 ms. An acousto-optic modulator was used to achieve exact timing of the laser illumination. The emitted light was split into two spectral channels (arbitrarily called here ‘green channel’ λ < 610 nm; ‘red channel’ λ > 630 nm) using a dichroic wedge (1° separation, Chroma, Rockingham, VT, USA) in the parallel beam path (32); the two images were recorded on a back-illuminated liquid nitrogen cooled CCD camera (Micro Max 1300-PB, Roper Scientific, Trenton, NJ, USA). Custom filter combinations (550DCRP and 514/633PC, Chroma; 700 nm short pass, Newport, USA) and wedge coatings were used to effectively block reflected and Rayleigh scattered light and to optimize the spectral separation of Alexa555 and Alexa647 emission.

Short delays between two images were enabled by operating the camera in kinetics mode: upon illumination, individual images within a time sequence were transferred into a masked region of the chip before readout of the full sequence. TTL pulse trains were used to synchronize camera line shift and laser illumination, which allowed arbitrary timing protocols. For two-color imaging, samples were illuminated consecutively with 514 nm and 647 nm light, separated by a delay of td = 1 ms. By this procedure, a quadruple of spectrally well-separated images was obtained, with three images containing red-shifted emission, which were used for further analysis: Alexa555 signal was detected in the green channel at 514 nm excitation, the Alexa647 signal in the red channel at 647 nm excitation. In these channels, cross-talk of the dyes was not observable. Due to its broad emission, cellular autofluorescence was detected in all three channels, in particular in the red channel at 514 nm excitation; we used this signal for correcting the Alexa555 and Alexa647 channels. For single dye tracing, six quadruples were recorded with time delays of tdelay = 30 ms–40 ms between two consecutive quadruples. All experiments were performed at 37°C in HBSS buffer supplemented with 2 mM Ca2+, 1 mM Mg2+ and 10% autologous serum.

The images were analyzed using in-house algorithms implemented in MATLAB (MathWorks, Natick, MA, USA). Individual diffraction limited signals were selected and analyzed by fitting with a Gaussian intensity profile, yielding the single-molecule position r(t) with an accuracy of typically 60 nm. Single-molecule trajectories were analyzed according to previous studies (33). The mean square displacement (MSD) forumla was calculated for all time lags tlag=2till+td+tdelay. Data were fitted according to MSD=4Dtlag+offset, with D, the two-dimensional diffusion constant; the offset includes contributions from diffusion during the illumination time and errors in localizing single molecules (33). For investigation of the time course of immobilization, an estimation of the diffusion constant of each observed molecule was calculated according to forumla A sliding average was calculated by convolution with a Gaussian function with a full width at half maximum of 10 data points.

Results

Biochemical analysis of the TCR complex composition

We first tested biochemically the association of the TCR–CD3 complex with lipid raft components (i.e. the GPI-anchored proteins CD48 and CD59, the adaptor molecule LAT and the SFK lck) using two non-ionic detergents, Brij58 and NP-40, at a concentration of 1%. Both are mild detergents of the polyoxyethylene type and commonly used for studies on molecules associated with lipid rafts (18). There was a substantial difference in the number of co-precipitated proteins between both detergents (Fig. 1). The CD3ε precipitate from resting T lymphocytes lysed in 1% Brij58 contained most of the molecules reported to be associated with lipid rafts (LAT, Lck, CD48, CD59). Also the protein tyrosine phosphatase CD45, described to temporarily associate with lipid rafts (34–36) was found in the precipitate. However, ZAP-70 known to be recruited to the TCR–CD3 complex upon triggering was not found. On the other hand, a more stringent lysis condition using 1% NP-40 resulted in no co-precipitation of any of these molecules. When analyzing the T-cell line Jurkat using 1% NP-40, also none of the tested molecules was pulled down with CD3ε. However, in contrast to naive T lymphocytes using 1% Brij58, CD3ε precipitation did not co-isolate CD48 and CD59 and pull down of lck, LAT and CD45 was weak.

Fig. 1

Analysis of the TCR–CD3 complex of CD4+CD25− naive T lymphocytes by standard immunoprecipitation. Resting CD4+ peripheral blood T lymphocytes and Jurkat T cells were lysed using 1% detergent Brij58 or NP-40 and precipitated by CD3 mAb OKT3. The immunoprecipitates were resolved by SDS–PAGE and immunoblotted with mAbs or antibodies to the indicated molecules (IP). As a control, also the crude lysate was analyzed by immunoblotting with the indicated mAbs (L).

Fig. 1

Analysis of the TCR–CD3 complex of CD4+CD25− naive T lymphocytes by standard immunoprecipitation. Resting CD4+ peripheral blood T lymphocytes and Jurkat T cells were lysed using 1% detergent Brij58 or NP-40 and precipitated by CD3 mAb OKT3. The immunoprecipitates were resolved by SDS–PAGE and immunoblotted with mAbs or antibodies to the indicated molecules (IP). As a control, also the crude lysate was analyzed by immunoblotting with the indicated mAbs (L).

This experiment shows the difficulty in the interpretation of results obtained from detergent-solubilized membranes. Dependent on the cell type and the detergent, different compositions of lipid raft components can be co-isolated with transmembrane receptors such as the TCR that raises the question whether these different complexes reflect heterogeneity of rafts dependent on the activation/differentiation state of the cell, fine architecture of rafts or are simply artifacts created during the solubilization procedure (37). Therefore, we established a technique, single-molecule imaging, to study the association of lipid raft components with the TCR in living T cells under non-invasive conditions and compare the results with the standard methods.

Single-molecule imaging of CD3 in unstimulated naive T cells

Evidence for the constitutive occurrence of lipid rafts in the membrane of T cells and their association with the TCR–CD3 complex upon activation is based primarily on studies with cell lines (1, 8,11–13, 38) or murine transgenic T cells (9, 10, 39) and only rarely in primary human naive T cells (14, 40). However, the response of cell lines such as Jurkat cells to CD3 stimulation is much faster in terms of Ca2+ mobilization and protein tyrosine phosphorylation in comparison to naive peripheral blood T lymphocytes (data not shown). Thus, one could speculate that cell lines are pre-activated containing altered signaling complexes. This assumption agrees with the different TCR–CD3 co-precipitations between Jurkat and naive T cells (Fig. 1). Therefore, we studied naive CD4+CD25− T lymphocytes isolated from human peripheral blood using single-molecule microscopy to obtain a picture of the constitutive situation in the membrane of resting T cells.

The movement of single molecules in living T cells was recorded with a system described in detail previously (33). Sorted naive human CD4+CD25− T lymphocytes were labeled with a mixture of an Alexa647-conjugated Fab fragment to CD3ε and an Alexa555-conjugated Fab fragment to CD45. As both membrane proteins are expressed at high surface density in the plasma membrane, we had to label only a minute fraction in order to resolve the signal of individual molecules; Fab concentrations were chosen such that single CD3ε could easily be followed in time (surface density <1 μm−2). We added the cells into a chamber preheated to 37°C and let them settle down onto non-coated epoxy-activated slides. Immediately after the first cells contacted the glass surface, we used the Alexa555 channel to focus onto the plasma membrane. Subsequently, data collection was started in the Alexa647 channel to record single-molecule movements of CD3ε; by this procedure, we eliminated any potential bias of the results due to photobleaching during focusing. For comparability, all data were recorded in the contact zone of the cell with the glass slide.

Figure 2A and B show a fluorescence and transmitted light image of a cell, respectively; its round shape is characteristic for an inactive T cell, which was found typically under the applied conditions. Individual fluorescent CD3ε molecules can be identified as diffraction-limited signals located in the cell membrane. From consecutive images, the single-molecule positions were determined with a precision of ∼60 nm by fitting to a Gaussian intensity profile, and the trajectories of single CD3 molecules were reconstructed (Fig. 2C).

Fig. 2

Single-molecule imaging on resting human naive CD4+CD25− T lymphocytes. Cells were specifically stained for CD3 using an Alexa647-conjugated Fab fragment of mAb MEM-57 and placed onto a glass coverslip. (A) and (B) show the fluorescence and corresponding transmitted light image of a single cell recorded on the bottom membrane adjacent to the glass slide, respectively. Single CD3 molecules can be easily identified as diffraction limited spots distributed randomly over the cell membrane. From consecutive images recorded with a delay of 45 ms, the trajectories of single CD3 molecules moving in the plasma membrane were reconstructed, exemplified for one molecule by a red line [see also magnification in (C)]. For illustration, the images in A, B and C were interpolated using bicubic interpolation. Analysis of 110 trajectories recorded from independent experiments on different cells yielded the MSD as a function of the time lag, tlag (D). The linear increase is in agreement with free Brownian motion of the molecules, yielding a diffusion constant forumla

Fig. 2

Single-molecule imaging on resting human naive CD4+CD25− T lymphocytes. Cells were specifically stained for CD3 using an Alexa647-conjugated Fab fragment of mAb MEM-57 and placed onto a glass coverslip. (A) and (B) show the fluorescence and corresponding transmitted light image of a single cell recorded on the bottom membrane adjacent to the glass slide, respectively. Single CD3 molecules can be easily identified as diffraction limited spots distributed randomly over the cell membrane. From consecutive images recorded with a delay of 45 ms, the trajectories of single CD3 molecules moving in the plasma membrane were reconstructed, exemplified for one molecule by a red line [see also magnification in (C)]. For illustration, the images in A, B and C were interpolated using bicubic interpolation. Analysis of 110 trajectories recorded from independent experiments on different cells yielded the MSD as a function of the time lag, tlag (D). The linear increase is in agreement with free Brownian motion of the molecules, yielding a diffusion constant forumla

The data were analyzed by plotting the MSDs as a function of the time lag (Fig. 2D). In this analysis, free Brownian motion is concomitant with a linear increase (MSD=4Dtlag), which can be discriminated from confined diffusion or active transport forumla respectively). In naive CD4+CD25− T cells, we found low but free mobility of CD3ε with a diffusion constant of forumla

Single-molecule imaging of the GPI-anchored proteins CD48 and CD59 in resting and activated naive T lymphocytes

To mimic the immunological synapse, we reliably induced the formation of T-cell interfaces with functionalized surfaces by placing naive CD4+CD25− T lymphocytes onto coverslips coated with CD3ε-specific whole mAb MEM-57. The clustering of CD3ε resulted in morphological changes characteristic for the formation of the T-cell interface on a timescale of 5–60 min. The distribution of CD3 in the formed contact zone was similar to an immunological synapse between APC and T cells (Fig. 4 and Fig. 1).

In unstimulated cells, both CD48 and CD59 were ∼4-fold more mobile forumla compared with the transmembrane protein CD3 (Fig. 3A). After stimulation, CD48 became strongly immobilized forumla while the mobility of CD59 was not affected by T-cell activation (forumla; Fig. 3A). We also analyzed the single-molecule mobility as a function of the activation time. For CD48, the time course of immobilization was directly observable; an exponential fit yielded a characteristic time constant of 8 min. For CD59, we recorded no systematic changes of the diffusion constant over time (Fig. 3B).

Fig. 3

Localization and mobility of CD48, CD59 and CD45 in naive CD4+CD25− T lymphocytes. (A) Analysis of single-molecule measurements on CD48, CD59 or CD45 recorded using Alexa647-labeled Fab fragments. All MSD data were found to increase linearly with the time lag, yielding corresponding diffusion constants D. Experiments were performed on non-stimulated cells (open circles; forumla) and stimulated cells (closed circles; forumla the number of analyzed trajectories, N, is indicated. (B) The time course of immobilization was further addressed by analyzing the single-molecule mobility as a function of the activation time. Here, time point zero was defined as the moment when the first cells were found to adhere to the activating surface. The diffusion constant obtained on resting cells is inserted as open circle. For CD48, the time course of immobilization was directly observable; an exponential fit yielded a characteristic time constant of 8 min.

Fig. 3

Localization and mobility of CD48, CD59 and CD45 in naive CD4+CD25− T lymphocytes. (A) Analysis of single-molecule measurements on CD48, CD59 or CD45 recorded using Alexa647-labeled Fab fragments. All MSD data were found to increase linearly with the time lag, yielding corresponding diffusion constants D. Experiments were performed on non-stimulated cells (open circles; forumla) and stimulated cells (closed circles; forumla the number of analyzed trajectories, N, is indicated. (B) The time course of immobilization was further addressed by analyzing the single-molecule mobility as a function of the activation time. Here, time point zero was defined as the moment when the first cells were found to adhere to the activating surface. The diffusion constant obtained on resting cells is inserted as open circle. For CD48, the time course of immobilization was directly observable; an exponential fit yielded a characteristic time constant of 8 min.

Single-molecule imaging of the leukocyte protein tyrosine phosphatase CD45 in resting and activated naive T lymphocytes

Besides the lipid raft components CD48 and CD59, we were also interested in the dynamic behavior of the major leukocyte phosphatase CD45, which regulates the activity of SFKs in the process of TCR signaling. In quiescent T lymphocytes, this rather large transmembrane molecule displayed a low basal mobility forumla that was only slightly higher than that of CD3. Upon stimulation, CD45 mobility became rapidly reduced by a factor of four (forumla; Fig 3A) at the periphery of the cell and did not change during the full observation period of 60 min (Fig. 3B).

Confocal microscopy of interface formation in naive T lymphocytes

To compare the single-molecule data with conventional confocal microscopy, we induced the formation of T-cell interfaces by placing naive T lymphocytes on identically treated glass slides. However, in contrast to the non-invasive single-molecule imaging, we fixed the cells after different time points of activation followed by staining with the specific mAbs. The cells spread immediately upon contact with the activating surface with a concomitant change in the CD3ε organization. Also by this method, we observed distinct behavior of the two GPI-anchored proteins CD48 and CD59. In accord with single-molecule imaging, CD48 was found to co-cluster with CD3ε more centrally. CD59 was localized predominantly at the periphery of the cell with respect to CD3ε (Fig. 4). Although the single-molecule approach revealed differences between CD48 and CD59, it did not indicate CD59 redistribution and clustering at the periphery, but in contrast free mobility over the whole interface region (Fig. 3). The observed difference between the two methods may be attributed to morphological changes induced by PFA fixation of the cells, which renders confocal images of fixed cells in general more susceptible to artifacts (41). With respect to CD45, both methods provided coincident results: confocal microscopy showed exclusion of CD45 from the center of the interface shortly after TCR engagement (Fig. 4), which is in agreement with the observation that single CD45 molecules slowed down preferentially at the periphery of the cell (Fig. 3).

Fig. 4

Confocal microscopy of the interface of CD4+CD25− T lymphocytes with functionalized surfaces. The cells were activated by CD3 mAb MEM-57 coated on aldehyde-treated coverslips for the indicated time points, fixed with 3.75% PFA and stained with Alexa647 (green)-conjugated mAb to CD3 (mAb MEM-57) and Alexa555 (red)-conjugated mAbs to CD48 (mAb MEM-102), CD59 (mAb MEM-43/5) or CD45 (mAb MEM-28).

Fig. 4

Confocal microscopy of the interface of CD4+CD25− T lymphocytes with functionalized surfaces. The cells were activated by CD3 mAb MEM-57 coated on aldehyde-treated coverslips for the indicated time points, fixed with 3.75% PFA and stained with Alexa647 (green)-conjugated mAb to CD3 (mAb MEM-57) and Alexa555 (red)-conjugated mAbs to CD48 (mAb MEM-102), CD59 (mAb MEM-43/5) or CD45 (mAb MEM-28).

Time-resolved biochemical analysis of the T-cell interface

Finally, we worked on establishing a biochemical method for characterizing the molecular re-organization of the TCR complex upon T-cell activation. We set up a precipitation assay that we termed here SPSP. It is based on the formation of the interface between the T cell and a stimulating surface. In contrast to standard immunoprecipitation, the immunosorbent is not only used for precipitation but also for activation, allowing kinetic analysis of alterations of the molecular microenvironment of resting and stimulated receptors. Peripheral blood T lymphocytes were shortly centrifuged onto the CD3ε mAb-coated plastic wells, which induced synchronized formation of T-cell interfaces. After different time periods of stimulation we lysed the cells with detergent 1% NP-40, which had not yielded any co-precipitation of the TCR with raft components in resting cells (Fig. 1). Then, both the material precipitated on the plate by the triggering CD3ε-specific mAb as well as the non-precipitated material were examined by SDS–PAGE and western blotting. Among the GPI-anchored proteins CD48 and CD59, we found that only CD48 became associated with CD3ε after TCR stimulation. We already saw a faint band at 2 min that gradually increased till 50 min after onset of triggering. CD59 could not be pulled down at any time point (Fig. 5). These data are in accord with the results from the single-molecules approach, which showed rapid co-immobilization of CD48 but not of CD59 with TCR–CD3.

Fig. 5

Analysis of the TCR–CD3 complex of CD4+CD25− naive T lymphocytes by SPSP. Freshly purified CD4+ peripheral blood T lymphocytes were subjected to SPSP for 2, 10, 20 and 50 min on plates coated with either the CD3 mAb OKT3 or the MHC class I mAb MEM-147 and then lysed with 1% NP-40. Both the precipitated (left) and the non-precipitated (right) materials were resolved by SDS–PAGE and immunoblotted (IB) with mAbs and antibodies to the indicated molecules.

Fig. 5

Analysis of the TCR–CD3 complex of CD4+CD25− naive T lymphocytes by SPSP. Freshly purified CD4+ peripheral blood T lymphocytes were subjected to SPSP for 2, 10, 20 and 50 min on plates coated with either the CD3 mAb OKT3 or the MHC class I mAb MEM-147 and then lysed with 1% NP-40. Both the precipitated (left) and the non-precipitated (right) materials were resolved by SDS–PAGE and immunoblotted (IB) with mAbs and antibodies to the indicated molecules.

We also analyzed co-precipitation of the other raft markers LAT and the SFK lck by this method, as well as CD45 and ZAP-70. The adaptor molecule LAT showed similar kinetics as CD48. The SFK lck, CD45 and ZAP-70 never showed up as co-precipitate of the TCR–CD3 complex. These data are consistent with the findings of Harder et al. (8), who showed that LAT but not lck accumulated at the contact zone of a T cell with a CD3-coated bead. However, lck was activated upon TCR–CD3 triggering as seen by the higher molecular weight lck band of 60 kDa that appeared in the non-precipitated lysate fraction. For control, we analyzed the co-precipitation profiles induced by triggering MHC class I, which was previously found to be partially associated with lipid rafts (42). At 10 min and 20 min of triggering, we co-precipitated weak bands of CD48 and LAT that decreased after 50 min (Fig. 5).

Discussion

In the present study, we developed a functional assay to measure the dynamical behavior of the TCR, the tyrosine phosphatase CD45 and two lipid raft constituents, the GPI-anchored proteins CD48 and CD59, during the initial phase of T-cell signaling under non-invasive conditions in living naive T lymphocytes. In resting cells, CD48 and CD59 were found to diffuse freely within the plasma membrane; their diffusion constants were similar but 4-fold higher than the value determined for the transmembrane TCR. In general, similar mobility ratios between GPI-anchored proteins and transmembrane proteins were reported in the literature by using fluorescence recovery after photobleaching (43). The clear difference in mobility demonstrates the absence of stable interactions between these molecules. Indeed, this result agrees with the biochemical finding that those molecules do not co-precipitate upon NP-40 solubilization. However, extraction with milder Brij58 detergent yielded a clear co-precipitation of CD48 and CD59 with CD3ε. Taken together, the results cannot reflect partitioning of these molecules into the same physical compartment in the native plasma membrane, but may well represent similar affinity to the same lipid microenvironment (10, 35).

Triggering the T lymphocyte via surface-bound CD3 antibody induced the formation of a T-cell interface concomitant with molecular association and segregation processes. For instance, tight regulation of effector molecule–substrate separation is important for controlling the activity of the tyrosine phosphatase CD45 during T-cell activation. CD45 initiates signaling by activation of lck via pY505 dephosphorylation. For sustained signaling, however, it is of critical importance that immune receptor tyrosine-based activation motifs on the TCR complex do not get immediately dephosphorylated by CD45, which is prevented by translocation of CD45 toward the periphery of the immunological synapse (Figs. 3 and 4). This process might be driven by the formation of tight cell–cell contact, which leads to size exclusion of CD45 due to its large ectodomain (36, 44). Indeed, it is known that the cytoplasmic domain of CD45 can interact via ankyrin with the actin cytoskeleton (45), which—similar to CD45—rearranges to the periphery during synapse formation (1). Thus, most likely, the observed immobilization and stabilization of CD45 in the periphery is due to its interaction with the actin ring formed along the pSMAC.

T-cell activation via surface-bound CD3 antibody further provides a methodological benefit, as the TCR becomes immobilized in the plasma membrane: interactions with other membrane proteins can now be addressed quantitatively by measuring the co-immobilization as a reduction in the mobility. Surprisingly, the two putative raft markers—CD48 and CD59—behaved differently: while CD48 became rapidly co-immobilized, the movement of CD59 was not affected. Again, our own biochemical experiments confirm these results: SPSP of CD3-stimulated T cells upon NP-40 lysis revealed clear co-precipitation of CD48 with CD3, but none for CD59. CD48 immobilization may be explained by the recently discovered immobile CD2 nanodomains, which are formed in similar interfaces between functionalized glass surfaces and Jurkat T cells (38): As CD2 represents a receptor for CD48 both in trans- (46) and cis-configuration (47), interaction with CD2 is expected to co-immobilize CD48.

More interesting is the finding that CD59 is not co-immobilized with CD48. If a homogenous population of lipid rafts existed with both CD48 and CD59 being constitutive members, immobilization of one partner should correlate with the immobilization of the other. The opposite finding obtained here could be interpreted as an indication for the non-existence of lipid rafts. Indeed, there is increasing evidence that protein interactions and not lipid-mediated interactions are required for the formation of signaling complexes (38, 48), which provoked a general skepticism on raft existence (48). On the contrary, the postulation of lipid-mediated protein associates has become well founded by different complementary methods (23, 26, 33,49–51). In the following, we will provide an argument that both findings—differential segregation of raft constituents and protein-interaction based formation of signaling complexes—are not contradicting but a consequence of the current picture of lipid rafts.

Obviously, the heterogeneous behavior of GPI proteins observed here at the single molecule level indicates heterogeneity in the composition of individual rafts. A short calculation shows that such raft heterogeneity would result naturally from the assumption of small platforms. For simplicity, let us assume lipid rafts would be circular platforms of radius r covering an area fraction η of the plasma membrane of a cell with radius R. If two molecular species—both expressed with a total number of n molecules per cell—are present exclusively in such rafts, the probability that any arbitrary raft contains at least one molecule of both species is given by forumla the average number of molecules of each species per raft. Assuming a typical size of naive T cells R = 2.5 μm, a medium expressor with n = 104 molecules per cell, a surface coverage η = 30% for rafts and a small raft size r = 2.5 nm—as measured in (49)—we calculate a negligible co-localization probability P = 0.007; this number increases for larger values of r, reaching P = 10% (P = 50%) for r = 17 nm (r = 30 nm). That means small rafts are heterogeneous in composition simply due to the fact that their loading is low. Further segregation of different GPI proteins into raft subspecies with a characteristic lipid content may contribute to the observed heterogeneity in raft composition (19–26).

What could be the function of such small entities? Due to loading fluctuations, no rafts would contain the same set of signaling molecules in resting cells. Still, the partition of the plasma membrane into raft and non-raft would be valid. We speculate that the major function of lipid rafts could be the separation of enzymes from their substrates, thereby inhibiting arbitrary phosphorylation and intracellular signaling in resting cells. Signaling would be initiated, when receptor engagement releases this separation and enables protein–protein interactions either via change in the raft affinity of enzyme or substrate (e.g. via palmitoylation/depalmitoylation) or via direct protein interactions exceeding the raft-mediated repulsion. In particular, ligation-induced association of the TCR with GPI-anchored CD48 is expected to create a more ordered lipid microenvironment (6), which releases the separation of the TCR from raft-associated effector molecules like the SFK lck.

Abbreviations

    Abbreviations
  • APC

    antigen-presenting cell

  • D

    diffusion constant

  • GPI

    glycosylphosphatidylinositol

  • MSD

    mean square displacement

  • PMSF

    phenylmethylsulfonylfluoride

  • SFK

    Src family kinase

  • SPSP

    solid phase signalosome precipitation

  • tlag

    time lag

This work was supported by the GEN-AU Program of the Austrian Federal Ministry of Education, Science and Culture, the Austrian Science Fund and the Higher Education Commission of Pakistan. We are grateful to Vaclav Horejsi for providing a huge number of invaluable mAbs, as well as to John Wijdenes, Peter Beverley, Steinar Funderud and Otto Majdic for their mAb gifts.

References

1
Bunnell
SC
Hong
DI
Kardon
JR
, et al.  . 
T cell receptor ligation induces the formation of dynamically regulated signaling assemblies
J. Cell Biol.
 , 
2002
, vol. 
158
 pg. 
1263
 
2
Grakoui
A
Bromley
SK
Sumen
C
, et al.  . 
The immunological synapse: a molecular machine controlling T cell activation
Science
 , 
1999
, vol. 
285
 pg. 
221
 
3
Choudhuri
K
Wiseman
D
Brown
MH
Gould
K
van der Merwe
PA
T-cell receptor triggering is critically dependent on the dimensions of its peptide-MHC ligand
Nature
 , 
2005
, vol. 
436
 pg. 
578
 
4
Barda-Saad
M
Braiman
A
Titerence
R
Bunnell
SC
Barr
VA
Samelson
LE
Dynamic molecular interactions linking the T cell antigen receptor to the actin cytoskeleton
Nat. Immun.
 , 
2005
, vol. 
6
 pg. 
80
 
5
Das
V
Nal
B
Dujeancourt
A
, et al.  . 
Activation-induced polarized recycling targets T cell antigen receptors to the immunological synapse; involvement of SNARE complexes
Immunity
 , 
2004
, vol. 
20
 pg. 
577
 
6
Gaus
K
Chklovskaia
E
Fazekas de St Groth
B
Jessup
W
Harder
T
Condensation of the plasma membrane at the site of T lymphocyte activation
J. Cell Biol.
 , 
2005
, vol. 
171
 pg. 
121
 
7
Burack
WR
Lee
KH
Holdorf
AD
Dustin
ML
Shaw
AS
Cutting edge: quantitative imaging of raft accumulation in the immunological synapse
J. Immunol.
 , 
2002
, vol. 
169
 pg. 
2837
 
8
Harder
T
Kuhn
M
Selective accumulation of raft-associated membrane protein LAT in T cell receptor signaling assemblies
J. Cell Biol.
 , 
2000
, vol. 
151
 pg. 
199
 
9
Montixi
C
Langlet
C
Bernard
AM
, et al.  . 
Engagement of T cell receptor triggers its recruitment to low-density detergent-insoluble membrane domains
EMBO J.
 , 
1998
, vol. 
17
 pg. 
5334
 
10
Drevot
P
Langlet
C
Guo
XJ
, et al.  . 
TCR signal initiation machinery is pre-assembled and activated in a subset of membrane rafts
EMBO J.
 , 
2002
, vol. 
21
 pg. 
1899
 
11
Moran
M
Miceli
MC
Engagement of GPI-linked CD48 contributes to TCR signals and cytoskeletal reorganization: a role for lipid rafts in T cell activation
Immunity
 , 
1998
, vol. 
9
 pg. 
787
 
12
Xavier
R
Brennan
T
Li
Q
McCormack
C
Seed
B
Membrane compartmentation is required for efficient T cell activation
Immunity
 , 
1998
, vol. 
8
 pg. 
723
 
13
Janes
PW
Ley
SC
Magee
AI
Aggregation of lipid rafts accompanies signaling via the T cell antigen receptor
J. Cell Biol.
 , 
1999
, vol. 
147
 pg. 
447
 
14
Viola
A
Schroeder
S
Sakakibara
Y
Lanzavecchia
A
T lymphocyte costimulation mediated by reorganization of membrane microdomains
Science
 , 
1999
, vol. 
283
 pg. 
680
 
15
Cinek
T
Hilgert
I
Horejsi
V
An alternative way of CD4 and CD8 association with protein kinases of the Src family
Immunogenetics
 , 
1995
, vol. 
41
 pg. 
110
 
16
Horejsi
V
Drbal
K
Cebecauer
M
, et al.  . 
GPI-microdomains: a role in signalling via immunoreceptors
Immunol. Today
 , 
1999
, vol. 
20
 pg. 
356
 
17
Dykstra
M
Cherukuri
A
Sohn
HW
Tzeng
SJ
Pierce
SK
Location is everything: lipid rafts and immune cell signaling
Annu. Rev. Immunol.
 , 
2003
, vol. 
21
 pg. 
457
 
18
Horejsi
V
The roles of membrane microdomains (rafts) in T cell activation
Immunol. Rev.
 , 
2003
, vol. 
191
 pg. 
148
 
19
Pike
LJ
Lipid rafts: heterogeneity on the high seas
Biochem. J.
 , 
2004
, vol. 
378
 pg. 
281
 
20
Gomez-Mouton
C
Abad
JL
Mira
E
, et al.  . 
Segregation of leading-edge and uropod components into specific lipid rafts during T cell polarization
Proc. Natl Acad. Sci. USA
 , 
2001
, vol. 
98
 pg. 
9642
 
21
Millan
J
Qaidi
M
Alonso
MA
Segregation of co-stimulatory components into specific T cell surface lipid rafts
Eur. J. Immunol.
 , 
2001
, vol. 
31
 pg. 
467
 
22
Schade
AE
Levine
AD
Lipid raft heterogeneity in human peripheral blood T lymphoblasts: a mechanism for regulating the initiation of TCR signal transduction
J. Immunol.
 , 
2002
, vol. 
168
 pg. 
2233
 
23
Wilson
BS
Steinberg
SL
Liederman
K
, et al.  . 
Markers for detergent-resistant lipid rafts occupy distinct and dynamic domains in native membranes
Mol. Biol. Cell
 , 
2004
, vol. 
15
 pg. 
2580
 
24
Bohuslav
J
Horejsi
V
Hansmann
C
, et al.  . 
Urokinase plasminogen activator receptor, beta 2-integrins, and Src-kinases within a single receptor complex of human monocytes
J. Exp. Med.
 , 
1995
, vol. 
181
 pg. 
1381
 
25
Kiyokawa
E
Baba
T
Otsuka
N
Makino
A
Ohno
S
Kobayashi
T
Spatial and functional heterogeneity of sphingolipid-rich membrane domains
J. Biol. Chem.
 , 
2005
, vol. 
280
 pg. 
24072
 
26
Brugger
B
Graham
C
Leibrecht
I
, et al.  . 
The membrane domains occupied by glycosylphosphatidylinositol-anchored prion protein and Thy-1 differ in lipid composition
J. Biol. Chem.
 , 
2004
, vol. 
279
 pg. 
7530
 
27
Pammer
P
Schlapak
R
Sonnleitner
M
, et al.  . 
Nanopatterning of biomolecules with microscale beads
Chemphyschem
 , 
2005
, vol. 
6
 pg. 
900
 
28
Moertelmaier
M
Axmann
M
Sonnleitner
A
, et al.  . 
Single molecule fluorescence microscopy on biomembranes
Recent Res. Dev. Biophys.
 , 
2003
, vol. 
2
 pg. 
95
 
29
Sako
Y
Yanagida
T
Single-molecule visualization in cell biology
Nat. Cell Biol.
 , 
2003
Suppl
pg. 
SS1
 
30
Mashanov
GI
Tacon
D
Knight
AE
Peckham
M
Molloy
JE
Visualizing single molecules inside living cells using total internal reflection fluorescence microscopy
Methods
 , 
2003
, vol. 
29
 pg. 
142
 
31
Sako
Y
Imaging single molecules in living cells for systems biology
Mol. Syst. Biol.
 , 
2003
, vol. 
2
 pg. 
56
 
32
Moertelmaier
MA
Kögler
EJ
Hesse
J
Sonnleitner
M
Huber
LA
Schütz
GJ
Single molecule microscopy in living cells: subtraction of autofluorescence based on two color recording
Single Mol.
 , 
2002
, vol. 
3
 pg. 
225
 
33
Schütz
GJ
Kada
G
Pastushenko
VP
Schindler
H
Properties of lipid microdomains in a muscle cell membrane visualized by single molecule microscopy
EMBO J.
 , 
2000
, vol. 
19
 pg. 
892
 
34
Cerny
J
Stockinger
H
Horejsi
V
Noncovalent associations of T lymphocyte surface proteins
Eur. J. Immunol.
 , 
1996
, vol. 
26
 pg. 
2335
 
35
Seveau
S
Eddy
RJ
Maxfield
FR
Pierini
LM
Cytoskeleton-dependent membrane domain segregation during neutrophil polarization
Mol. Biol. Cell
 , 
2001
, vol. 
12
 pg. 
3550
 
36
Zhang
M
Moran
M
Round
J
, et al.  . 
CD45 signals outside of lipid rafts to promote ERK activation, synaptic raft clustering, and IL-2 production
J. Immunol.
 , 
2005
, vol. 
174
 pg. 
1479
 
37
Heerklotz
H
Triton promotes domain formation in lipid raft mixtures
Biophys. J.
 , 
2002
, vol. 
83
 pg. 
2693
 
38
Douglass
AD
Vale
RD
Single-molecule microscopy reveals plasma membrane microdomains created by protein-protein networks that exclude or trap signaling molecules in T cells
Cell
 , 
2005
, vol. 
121
 pg. 
937
 
39
Lee
KH
Holdorf
AD
Dustin
ML
Chan
AC
Allen
PM
Shaw
AS
T cell receptor signaling precedes immunological synapse formation
Science
 , 
2002
, vol. 
295
 pg. 
1539
 
40
Tuosto
L
Parolini
I
Schroder
S
Sargiacomo
M
Lanzavecchia
A
Viola
A
Organization of plasma membrane functional rafts upon T cell activation
Eur. J. Immunol.
 , 
2001
, vol. 
31
 pg. 
345
 
41
Kakoi
H
Anniko
M
Erwall
C
Pettersson
CA
Jansson
B
Morphological changes in rat submandibular gland mucous cells during fixation with 10% formalin
Eur. Arch. Otorhinolaryngol.
 , 
1996
, vol. 
253
 pg. 
214
 
42
Vereb
G
Matko
J
Vamosi
G
, et al.  . 
Cholesterol-dependent clustering of IL-2Ralpha and its colocalization with HLA and CD48 on T lymphoma cells suggest their functional association with lipid rafts
Proc.Natl Acad. Sci. USA
 , 
2000
, vol. 
97
 pg. 
6013
 
43
Kenworthy
AK
Nichols
BJ
Remmert
CL
, et al.  . 
Dynamics of putative raft-associated proteins at the cell surface
J. Cell Biol.
 , 
2004
, vol. 
165
 pg. 
735
 
44
Irles
C
Symons
A
Michel
F
Bakker
TR
van der Merwe
PA
Acuto
O
CD45 ectodomain controls interaction with GEMs and Lck activity for optimal TCR signaling
Nat. Immunol.
 , 
2003
, vol. 
4
 pg. 
189
 
45
Pradhan
D
Morrow
J
The spectrin-ankyrin skeleton controls CD45 surface display and interleukin-2 production
Immunity
 , 
2002
, vol. 
17
 pg. 
303
 
46
Zhu
B
Davies
EA
van der Merwe
PA
Calvert
T
Leckband
DE
Direct measurements of heterotypic adhesion between the cell surface proteins CD2 and CD48
Biochemistry
 , 
2002
, vol. 
41
 pg. 
12163
 
47
Kaplan
AJ
Chavin
KD
Yagita
H
, et al.  . 
Production and characterization of soluble and transmembrane murine CD2. Demonstration that CD48 is a ligand for CD2 and that CD48 adhesion is regulated by CD2
J. Immunol.
 , 
1993
, vol. 
151
 pg. 
4022
 
48
Munro
S
Lipid rafts: elusive or illusive?
Cell
 , 
2003
, vol. 
115
 pg. 
377
 
49
Sharma
P
Varma
R
Sarasij
RC
, et al.  . 
Nanoscale organization of multiple GPI-anchored proteins in living cell membranes
Cell
 , 
2004
, vol. 
116
 pg. 
577
 
50
Pralle
A
Keller
P
Florin
EL
Simons
K
Horber
JK
Sphingolipid-cholesterol rafts diffuse as small entities in the plasma membrane of mammalian cells
J. Cell Biol.
 , 
2000
, vol. 
148
 pg. 
997
 
51
Lenne
PF
Wawrezinieck
L
Conchonaud
F
, et al.  . 
Dynamic molecular confinement in the plasma membrane by microdomains and the cytoskeleton meshwork
EMBO J.
 , 
2006
, vol. 
25
 pg. 
3245
 

Author notes

*
These authors contributed equally to this study.
Transmitting editor: L. Moretta