Cortical actin networks induce spatio-temporal confinement of phospholipids in the plasma membrane--a minimally invasive investigation by STED-FCS

Abstract

Important discoveries in the last decades have changed our view of the plasma membrane organisation. Specifically, the cortical cytoskeleton has emerged as a key modulator of the lateral diffusion of membrane proteins. Cytoskeleton-dependent compartmentalised lipid diffusion has been proposed, but this concept remains controversial because this phenomenon has thus far only been observed with artefact-prone probes in combination with a single technique: single particle tracking. In this paper, we report the first direct observation of compartmentalised phospholipid diffusion in the plasma membrane of living cells using a minimally invasive, fluorescent dye labelled lipid analogue. These observations were made using optical STED nanoscopy in combination with fluorescence correlation spectroscopy (STED-FCS), a technique which allows the study of membrane dynamics on a sub-millisecond time-scale and with a spatial resolution of down to 40 nm. Specifically, we find that compartmentalised phospholipid diffusion depends on the cortical actin cytoskeleton, and that this constrained diffusion is directly dependent on the F-actin branching nucleator Arp2/3. These findings provide solid evidence that the Arp2/3-dependent cortical actin cytoskeleton plays a pivotal role in the dynamic organisation of the plasma membrane, potentially regulating fundamental cellular processes.

Figure 1. Detecting compartmentalised diffusion with a small lipid probe.

(a) Schematic showing branched actin networks (magenta) and associated membrane achors (orange), which partially confine two-dimensional diffusion of molecules. As exemplarily shown by single-molecule diffusion tracks (blue), molecules are assumed to diffuse freely within compartments, and in the event of hitting the compartment boundaries, transposition to the adjacent compartment occurs with a certain hopping probability P_hop. (b) Schematic of lipid probes used in SPT and in STED-FCS, put in perspective (from left to right): gold particle (yellow, ~40 nm in diameter) linked to a lipid (orange: chains, light red: head group) by Fab antibody fragment (green), and QD (red, ~20 nm in diameter) linked to two lipids via streptavidin (blue), as often used in SPT; and a fluorescent lipid analogue (dark red: organic dye, ~1 nm in diameter), as used in STED-FCS. Possible oligomerisation induced by SPT probes is illustrated for the QD. The membrane bilayer is shown in orange and the actin cytoskeleton in grey.

Figure 2. STED-FCS simulation of compartmentalised diffusion.

(a) In STED-FCS, the apparent diffusion coefficient D_app is determined for different sizes of the observation spot (given by the diameter d), as formed by varying the STED laser power (red: STED light, green: effective observation or fluorescence area). (b) In silico STED-FCS experiments: Simulations show characteristic dependencies of D_app on the diameter d of the observation area, assuming a model for compartmentalised diffusion as depicted in Fig. 1a with D_free = 0.8µm²/s. As d is increased, D_app decreases. Characteristic compartment size of length L, free diffusion coefficient D_free, and hopping probability P_hop define the diffusion model. These simulations (using D_free = 0.8µm²/s and P_hop and L as given) show that only strong confinement (small P_hop) renders clear patterns of compartmentalised diffusion whereas weaker confinement (for example, P_hop = 0.5) closely resembles free diffusion.

Figure 3. Experimental observation of lipid compartmentalised diffusion by STED-FCS.

D_app(d) dependencies (blue) for DPPE-Atto647N diffusion in NRK (a) and IA32 cells (b). Clear compartmentalised diffusion patterns are observed. **P < 0.01 (unpaired t test). Error bars are s.e.m. In a, n = 32 cells; in b, n = 33 cells. In a and b, r = 10 (n stands for the number of cells, from r samples). Fitting of the experimental data using Monte-Carlo simulations (orange dotted lines) resulted for NRK cells, D_free = 0.8 ( ±0.03) μm²/s, P_hop = 0.1 ( ±0.01) and L = 80 ( ±8) nm, and for IA32 cells, D_free = 0.8 ( ±0.02) μm²/s, P_hop = 0.1 ( ±0.01) and L = 150 ( ±12) nm. Insets: Representative Voronoi lattices (red) relative to the correspondent compartment sizes as well as simulated diffusion trajectories (blue) correspondent to the fitted parameters. Scale bars: 250 nm.

Figure 4. Investigation of molecular mechanisms underlying compartmentalised lipid diffusion via STED-FCS: Pharmacological treatments.

D_app(d) dependencies: (a,b) Cytoskeleton depletion in NRK (a) and IA32 (b) cells, respectively: treatment with Latrunculin B and CK-666 as labelled, (c,d) Cholesterol depletion and myosin II inhibition in NRK (c) and IA32 (d) cells, respectively: treatment with Cholesterol Oxidase and Blebbistatin as labelled. (e,f) Summary of NRK and IA32 data, respectively, showing values of D_app for confocal (d = 240 nm, filled columns) and STED recordings (d ~ 40 nm, open columns) – the increase in D_app from 240 to 40 nm indicates the extent of compartmentalised diffusion. Error bars are s.e.m. Symbols on top of the columns represent results of the statistical test (**P < 0.01, *P < 0.05, NS not significant; two-tailed unpaired t test): for d ~ 40 nm comparison with the value representing d = 240 nm in the same experiment, and for d = 240 nm comparison with the respective value for d = 240 nm in the control (untreated) experiment. n stands for the number of cells, from r samples.

Figure 5. STED-FCS of lipid diffusion in different parts of the cell and in PtK2 cells.

(a) D_app(d) dependencies in NRK cells: diffusion near the cell edge (lamellipodia) and under the cell body (as labelled). (b) Values of D_app for confocal (d = 240 nm, filled columns) and STED recordings (d ~ 40 nm, open columns) for diffusion probed near the cell edge (control, blue) and under the cell body (purple) – the increase in D_app from 240 to 40 nm indicates the extent of compartmentalised diffusion. Compartmentalised diffusion is more pronounced near the cell edge. (c) Results from measurements on PtK2 cells: Experimental diffusion coefficient of the fluorescent DPPE analogue in the plasma membrane of untreated and CK-666-treated PtK2 cells (d = 240 nm). We have previously shown that diffusion of this DPPE analogue in PtK2 cells is apparently free, therefore the measurements corresponding to d = 240 nm are expected to effectively represent the range (d = 40 nm to d = 240 nm). Error bars are s.e.m. Symbols on top of the columns represent results of the statistical test (**P < 0.01, *P < 0.05, NS not significant; two-tailed unpaired t test): for d ~ 40 nm comparison with the value representing d = 240 nm in the same experiment, and for d = 240 nm comparison with the respective value for d = 240 nm in the control experiment. Here, n stands for the number of cells, from r samples.

Figure 6. Comparing compartmentalisation in Arp2/3 knock-down cells.

(a) D_app(d) dependencies in IA32 MEFs and IA32 2xKD MEFs (Arp 2/3 knock-down), respectively. (b) Values of D_app for confocal (d = 240 nm, filled columns) and STED recordings (d ~ 40 nm, open columns) in IA32 MEFs (control, dark blue) and IA32 2xKD MEFs (IA32 2xKD, light blue) – the increase in D_app from 240 to 40 nm indicates the extent of compartmentalised diffusion. Error bars are s.e.m. Symbols on top of the columns represent results of the statistical test (**P < 0.01, *P < 0.05, NS not significant; two-tailed unpaired t test): for d ~ 40 nm comparison with the value representing d = 240 nm in the same experiment, and for d = 240 nm comparison with the respective value for d = 240 nm in the control experiment. n stands for the number of cells, from r samples.