SEARCH

SEARCH BY CITATION

Supporting Material and Methods

Supporting Discussion

Figure S1: Confocal fluorescence recovery after photobleaching (FRAP) of Av-GPI in the plasma membrane of HeLa cells.

A) Sequence of images illustrating the fluorescence recovery after photobleaching in a 5μm radius circular ROI in the ventral membrane of a HeLa cell. Scale bar: 5μm. B) Averaged fluorescence recovery curves for several HeLa cells bleached in a 3μm radius ROI at 37°C (circle, n = 19) or at room temperature (RT, diamond, n = 17). Standard deviations for each data point were omitted for clarity but considered for the fit. Faster fluorescence recovery for smaller, 1.5μm radius bleached spots (result not shown) indicated that the recovery was dominated by diffusion. The apparent diffusion coeffcient (D) and the immobile fraction were thus derived by non-linear least square fit of the recovery curves to the lateral diffusion equations for uniform circular bleach spots. For both temperatures, the FRAP recovery curves were better described with a two-component lateral diffusion model (37°C, blue, F-test p < 0.0001; and RT, red, F-test p < 0.0001) than with a one-component model. At 37°C 79 ± 12% of Av-GPI appear to diffuse with while 16 ± 8% of Av-GPI diffuse with a diffusion coeffcient about 20 times smaller () and 5 ± 4% of GPI-test probes were found to be immobile. The apparent diffusion coeffcient of both fast and slow population were reduced about three-fold when FRAP measurements were performed at RT (∼ 27°C). At RT, 81 ± 12% of Av-GPI diffused with and 15 ± 7% diffused with . The fraction of immobile Av-GPI (4 ± 5%) remained unchanged. SD, standard deviation.

Figure S2: Analysis of the oligomeric state of Av-GPI in the membrane of HeLa cells.

A) Parallel Western blot analysis of native chicken avidin (right) and Av-GPI (left) extracted from the plasma membrane of HeLa cells and run on 15% SDS-PAGE. The blot shows that native avidin prepared in SDS buffer without boiling migrate as a mixture of high molecular weight species (lane 5). Upon boiling, the high molecular band disappeared and was replaced by a ∼17 kDa band corresponding to monomeric avidin (lane 7). As previously reported, the addition of biotin before boiling significantly enhanced the recovery of high molecular weight complexes (lane 8). A different behavior was observed for Av-GPI. In samples that were not boiled, the chimeric protein migrated as a homogenous high molecular weight complex with the predicted size for a ∼ 120 kDa tetramer (lanes 1 and 2). Upon boiling, as observed for avidin, the biotin-free complex was converted to a ∼ 30 kDa monomer (lane 3). This value is slightly above the theoretical molecular weight for the non-glycosylated and monomeric form of Av-GPI (∼ 23 kDa) and may indicate the presence of glycosylated moieties. The addition of biotin resulted in the predominant recovery of the tetramers (lane 4). B) Western blot analysis of Av-GPI extracted from the plasma membrane of HeLa cells after in situ acetylation with NHS-acetate and run on a 4–20% SDS-PAGE. Under non-denaturing conditions (no boiling), acetylated Av-GPI migrated essentially as ∼ 120 kDa band as expected for the tetrameric form of Av-GPI (lane 2). No obvious enrichment in multi-tetramers, trimers, dimers, or monomers of Av-GPI were observed, confirming that Av-GPI form stable tetramers in the membrane of HeLa cells. A boiled sample of acetylated Av-GPI (lane 3) was used as a control and compared to Av-GPI from membrane preparation not treated with NHS-acetate (lane 1). Both samples migrated between 20–30 kDa as expected for Av-GPI monomers (∼23 kDa). These results indicate that Av-GPI does not appear to form multi-oligomers compared to native avidin and most likely exists as a glycosylated homotetramer attached to the membrane.

Figure S3: Av-GPI labeled with quasi-monovalent or multivalent biotinylated qdots have similar diffusive behaviors.

A) Gel shift assay in 1% agarose to evaluate the binding properties of quasi-monovalent or multivalent qdots to Neutravidin at increasing concentrations of Neutravidin (10 000, 5000, 2500, 1250, 625, 312, 156, 78, 39, 19.5, 9.75, 5 nM). Qdot fluorescence was detected on a fluorescence gel scanner. indicates no Neutravidin. B) Distribution of diffusion coeffcients after single-qdot tracking of Av-GPI labeled with quasi-monovalent (top) or multivalent (bottom) biotinylated qdots in live HeLa cells. C) Examples of Av-GPI trajectories for quasi-monovalent (top) or multivalent qdots (bottom).

Figure S4: Single-dye tracking of Av-GPI with monovalent Alexa 488 biocytin at room temperature.

A) Example of Av-GPI trajectories after tracking with monovalent Alexa 488 biocytin. B) MSD plots for 81 Av-GPI labeled with Alexa 488 biocytin (blue curves) and ensemble MSD plot obtained by pooling together the square displacements of all trajectories (red curve). C) Example of Av-GPI trajectory with two diffusion coeffcients. The squared trajectory in (A) was suffciently long (4.6 s) to detect a change in diffusion when plotting the instantaneous diffusion coeffcient over time. D) Global PDSD analysis of the same 81 Av-GPI traces in (B). The PDSD curves were fitted with three fitting exponents for comparison with a similar global PDSD analysis of qdots labeled Av-GPI traces (data can be viewed at http://fpinaud.bol.ucla.edu/index_files/Traffc.htm). The three curves recovered were well fitted with a normal diffusion model < r2 >= 4Dt with D1 = 1.4 × 10−1 μm2/ s (18%), D2 = 4.0 × 10−2 μm2/ s (54%) and D3 = 5.7 × 10−3 μm2/ s (28%), for ) and , respectively.

Figure S5: Example study of Av-GPI colocalization with GM1-rich domains by MSD analysis of subtrajectories and PDSD analysis together with PDSD analysis of Monte Carlo simulated trajectories.

A) The trajectory of a tracked Av-GPI is overlaid with the green CTxB mean intensity projection image. Scale bar 500 nm. B) Colocalization of Av-GPI with a GM1-rich domain is confirmed by studying qdot (red) and CTxB signal (green) along the GPI-test probe trajectory. Periods during which signals are above background (gray) and overlap are selected (ROI). C) The sub-trajectory corresponding to the selected temporal ROI is automatically highlighted. D) MSD are then computed for the full trajectory (black), the selected ROI and colocalizing sub-trajectory (green) or the non colocalizing sub-trajectory (blue). MSD are then fitted over 10% with a simple Brownian diffusion model (red). The fit curve of the full trajectory MSD has been omitted for clarity. E) Diffusion coeffcients are determined from fits of 10% of the MSD in (D). For this particular Av-GPI molecule, the diffusion coeffcient within the GM1-rich domain is ∼ 30 times smaller than it is outside the domain. F) PDSD analysis on the same trajectory is then performed for consistency and to verify the diffusion values determined from sub-trajectory MSDs. The diffusion coeffcients fall within the fast and slow sub-populations of the diffusion histograms (Figure 3B). G) PDSD analysis of Monte Carlo simulated trajectories. (i) A simulated trajectory undergoing a change in diffusion mode from free diffusion (D1) to restricted diffusion (D2) into a circular domain of radius R = 100 nm. The diffusion constants chosen in this simulation correspond to the modal values of the fast and slow diffusion regimes reported in the text. Uncertainty in position determination was set to 30 nm, a typical value observed in our experiments. The particle was located in the slow domain for 60% of the trajectory duration (100 s). (ii) PDSD analysis of the trace in (i) over 3% of the time lags. (iii) Fit of PDSD in (ii) results in two r2i(t) curves (open circles). r21(t) could be fitted with a simple diffusion mode (black curve) resulting in a measured diffusion coeffcient D1,mes · r22(t) (red circles) was fitted with a restricted diffusion model (red curve) and yielded D2,mes and R in good agreement with the input values of the simulation. The reported uncertainty value is that associated with the restricted diffusion regime. The computed uncertainty associated with the simple diffusion regime is 58 nm. Notice that the graph is represented in logarithmic scale because of the large difference in D values.

Figure S6: Correlation of colocalization and diffusion coeffcients of Av-GPI and caveolae.

Three examples of simultaneous caveolae and Av-GPI tracking. Trajectories in (A) and (B) correspond to images of Figure 6E. The trajectory in (C) corresponds to one image of Figure 6C. Diffusion coeffcients were obtained by PDSD analysis as illustrated on the right. In (A) and (B), Av-GPI and caveola colocalize within the experimental uncertainty of the overlay of the green and red channel (∼1 pixel). The diffusion coeffcients are reasonably close and suggest trapping of Av-GPI in the caveolar pit. In (C) the GPI-test probe switches diffusion regimen in close proximity to a caveola.

Figure S7: Quantification of membrane cholesterol by filipin staining of Av-GPI expressing cells.

A) Confocal imaging of filipin in control cells, cells treated with 10μM lovastatin for 30 h or cells treated with 10 mM mβCD for 1 h. Acute cholesterol depletion with mβCD removes most of the membrane cholesterol and poor filipin staining is observed. Scale bar: 10μm. B) Quantification of membrane cholesterol from filipin fluorescence intensity. Lovastatin treatment did not result in a significant decrease in membrane cholesterol compared to control cells (86% ±22%, n = 47). After treatment with mβCD, however, the cholesterol content in cell membranes is strongly reduced (17 ± 7%, n = 34).

Figure S8: Effect of cholesterol depletion on the actin cytoskeleton and the distribution of GM1 in fixed HeLa cells expressing Av-GPI.

A) The organization of the actin cytoskeleton is impacted by cholesterol depleting drugs. Acute reduction in membrane cholesterol with 10 mM mβCD treatment for 1 h leads to a slight rounding up of adherent HeLa cells and reduced cell contacts. Although some stress fibers are present, they appear less defined than in untreated cells and actin-rich foci and microspikes are visible. Cholesterol depletion with 10μM lovastatin for 30 h led to less abundant and less defined stress fibers. The cortical actin network lining the inner surface of the plasma membrane is also irregular and cell contacts are reduced. B) The membrane distribution of GM1 is affected by cholesterol depleting drugs. While GM1 are distributed homogeneously in the plasma membrane of untreated cells, acute cholesterol depletion by mβCD redirects GM1 to perinuclear compartments (arrows) and significantly decreases GM1 membrane content. Milder cholesterol depletion with lovastatin does not strongly affect the distribution of GM1 in the membrane. Only limited perinuclear localization of GM1 is observed (arrows). Inserts: DIC images of micrographs. Scale bars: 20μm.

Figure S9: Cholesterol depletion with lovastatin has little effect on the diffusion of Av-GPI.

A) Confocal fluorescence image of Av-GPI in the plasma membrane of HeLa cells after 30 h cholesterol depletion with 10μM lovastatin. Insert: DIC image. Scale bar: 20μm. B) Distribution of Av-GPI diffusion coeffcients for lovastatin treated cells in the presence of the Alexa 488 labeled CTxB. The two populations of fast and slow Av-GPI have diffusion coeffcients similar to untreated cells with , 51%) and (SE: 0.8–1.4 × 10−4 μm2/ s, 36%). Only 7% of all Av-GPI repartitioned between the fast and slow populations during tracking. C) CTxB image of an ROI of the plasma membrane for a cell treated with lovastatin. Av-GPI trajectories are overlaid on the image. As observed for untreated cells stained with CTxB, about 65% of slow Av-GPI were colocalized with GM1-rich CTxB-labeled domains, while faster Av-GPI diffuse around these structures. Scale bar: 1μm. D) Modes of diffusion for Av-GPI after lovastatin treatment. The diffusion modes of fast and slow Av-GPI are not significantly influenced by lovastatin, and are similar to those of untreated cells.

Figure S10: Effects of actin cytoskeleton disruption on membrane organization and diffusion of Av-GPI and GM1 in HeLa cells.

Cells were incubated with 10 μM latrunculin-A for 45 min at the end of the serum-starving period and prepared for imaging as described in Materials and Methods. Latrunculin-A was kept in the cell medium at all steps including imaging. A) F-actin staining after treatment with latrunculin-A, fixation and permeabilization of HeLa cells. The actin cytoskeleton was completely disrupted. Membrane retraction and changes in cell shape are clearly visible (DIC image). Scale bar 20μm. B) Confocal fluorescence image of the distribution of Av-GPI and GM1 upon disruption of cortical actin. Both GPI-test probes and GM1 were located in bright, large and co-localizing membrane patches bound to the coverslip (arrows) or distributed on the cell membrane. C) Distribution of Av-GPI diffusion coeffcients for latrunculin-A treated cells in the presence of Alexa-488 CTxB. Only GPI-test probes that diffused clearly and did not change diffusion regime during the tracking were analyzed. The modal diffusion coeffcient was (SE: 1.4–2.1 × 10−1 μm2/ s). D) Examples of Av-GPI trajectories in latrunculin-A treated cells.

Table S1: Distributions and quantification of Av-GPI and transferrin receptor in flotation gradients.

Table S2: Diffusion models.

Table S3: Diffusion coeffcients of Av-GPI measured with various techniques and under different conditions.

Video 1: Imaging of qdot-labeled Av-GPI in the ventral membrane of HeLa cells (just before addition of CTxB) by TIRF microscopy. The typical on/off behavior of single qdots is observed. Some qdots freely diffusing in solution (between the coverslip and the membrane) can be seen binding to Av-GPI. These events are characterized by a sudden change from a very fast diffusion in three-dimension to a slower diffusion in two-dimension, in the plane of the plasma membrane. Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 2: Diffusion of Alexa 488 biocytin-labeled Av-GPI. Acquisition: 60 ms/frame; Display: 30 ms/frame.

Video 3: Dual-color TIRF imaging of qdot-labeled Av-GPI (red) and Alexa-488 CTxB labeled GM1 glycosphingolipids (green). Notice that the contrast of the point-spread-function of qdots was intentionally increased to facilitate visualization. Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 4: Example of qdot-labeled Av-GPI (red) diffusing either in stationary GM1-rich microdomains (green), outside these domains or partitioning in and out of the domains. The qdot channel was overlaid on the mean intensity projection image of the Alexa-488 CTxB-labeled GM1 channel. The contrast of the point-spread-function of qdots was intentionally increased to facilitate visualization. Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 5: Example of the entry and slowing down of a qdot-labeled Av-GPI (red) in a stationary GM1-rich domain (green). The qdot channel was overlaid on the mean intensity projection image of the Alexa-488 CTxB-labeled GM1 channel. Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 6: Example of exit and increased diffusion of a qdot-labeled Av-GPI (red) out of a stationary GM1-rich domain (green). The qdot channel was overlaid on the mean intensity projection image of the Alexa-488 CTxB-labeled GM1 channel. Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 7: Dual-color TIRF imaging of qdot-labeled Av-GPI (red) and Caveolin-1-EGFP labeled caveolae (green). Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 8: Dual-color TIRF imaging of qdot-labeled Av-GPI (red) and Alexa-488 CTxB-labeled GM1 (green) after treatment with lovastatin (10μM). Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 9: Dual-color TIRF imaging of qdot-labeled Av-GPI (red) and Alexa-488 CTxB-labeled GM1 (green) after treatment with mβCD (10 mM). Acquisition: 100 ms/frame; Display: 30 ms/frame.

Video 10: Dual-color TIRF imaging of qdot-labeled Av-GPI (red) and Alexa-488 CTxB cholera toxin B-labeled GM1 (green) after treatment with latrunculin-A (10μM). Acquisition: 100 ms/frame; Display: 30 ms/frame.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
TRA_902_sm_Movie1.mov4846KSupporting info item
TRA_902_sm_Movie2.mov4852KSupporting info item
TRA_902_sm_Movie3.mov4641KSupporting info item
TRA_902_sm_Movie4.mov2439KSupporting info item
TRA_902_sm_Movie5.mov1155KSupporting info item
TRA_902_sm_Movie6.mov1477KSupporting info item
TRA_902_sm_Movie7.mov4429KSupporting info item
TRA_902_sm_Movie8.mov4637KSupporting info item
TRA_902_sm_Movie9.mov4661KSupporting info item
TRA_902_sm_Movie10.mov4772KSupporting info item
TRA_902_sm_Movie1.mov4846KSupporting info item
TRA_902_sm_SupportingInformation.doc10448KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.