4polar-STORM polarized super-resolution imaging of actin filament organization in cells

Abstract

Single-molecule localization microscopy provides insights into the nanometer-scale spatial organization of proteins in cells, however it does not provide information on their conformation and orientation, which are key functional signatures. Detecting single molecules' orientation in addition to their localization in cells is still a challenging task, in particular in dense cell samples. Here, we present a polarization-splitting scheme which combines Stochastic Optical Reconstruction Microscopy (STORM) with single molecule 2D orientation and wobbling measurements, without requiring a strong deformation of the imaged point spread function. This method called 4polar-STORM allows, thanks to a control of its detection numerical aperture, to determine both single molecules' localization and orientation in 2D and to infer their 3D orientation. 4polar-STORM is compatible with relatively high densities of diffraction-limited spots in an image, and is thus ideally placed for the investigation of dense protein assemblies in cells. We demonstrate the potential of this method in dense actin filament organizations driving cell adhesion and motility.

Fig. 1. 4polar-STORM allows single molecule orientational parameters retrieval in STORM imaging.

a Schematic representation of a wobbling Alexa Fluor 488 (AF488)-phalloidin conjugate labeling an actin filament (F-actin) (the fluorophore moiety is highlighted in red). (ρ,η): mean orientation of a single fluorophore in 3D; δ_3D: wobbling cone angle of the fluorophore in 3D; δ: projection in the sample plane; ξ: mean orientation of the fluorophore relative to the actin filament axis. b Schematic optical setup of 4polar-STORM imaging. BS, beam splitter; M, mirror; D, diaphragm; HWP, half-wave plate; PBS, polarizing beam splitter. c Monte Carlo simulations (see Methods) of the expected precision on ρ for various δ (η = 90°, ρ = 30°). Total intensities (ph, photons) and background levels (ph/pix, photons per pixel) are summed over all four polarized channels. d Corresponding simulations for the δ precision at different δ values (η = 90°) (the results do not depend on ρ). Same color code as in (c). e 4polar-STORM images of single AF488-phalloidin molecules labeling single actin filaments. Left panels depict single-molecule localization STORM images (blurred using a Gaussian filter width 0.3 pixel (=39 nm)). Middle and right panels depict single-molecule wobbling (δ) and orientation (ρ) measurements overlaid with the STORM image in grayscale. Each single molecule is represented as a stick whose orientation is ρ relative to the horizontal axis and whose color is the measured parameter (ρ or δ). Scale bars, 170 nm. Similar results were obtained on 5 different samples (about 5–10 filaments per sample). f Experimental δ histogram obtained on a straight region (see (e)) for thresholded intensity and localization precision (see Text). g Corresponding ρ histogram in both standard and polar-plot representations, relative to the average within the measured region of interest, Δρ = ρ − <ρ>. σ_Δρ is the standard deviation of Δρ. Source data are provided as a Source Data file.

Fig. 2. 4polar-STORM imaging of actin filament organization in fixed U2OS cells.

a Left: Spinning-disk fluorescence image of a U2OS cell (green, AF488-phalloidin-labeled F-actin; red, p-FAK). White arrows indicate focal adhesions (FAs) and the types of stress fibers (SFs) of interest. Right: z-stack mage of F-actin, z is color-coded as indicated. b Large field of view single-molecule (AF488) localization STORM image of the same cell. c Corresponding 4polar-STORM ρ stick image with color-coded orientation measurements. d Corresponding 4polar-STORM δ stick image with color-coded wobbling angle measurements. e STORM of zoomed regions of interest (ROI) (squares in (d)). ROI 1, ventral SF; ROI 2, FA; ROI 3, dorsal SF; ROI 4, meshwork. f δ stick images of zoomed regions (dashed squares in (e)). 4polar-STORM images were repeated independently on three different samples (3–5 cells per sample) with similar results. g Polar-plot histograms of ρ in the rectangle region indicated in (e) (whole region for ROI 4), with corresponding standard deviation σ_Δρ values of Δρ = ρ − <ρ>. h Histograms of δ values for ROIs 1–4, with corresponding values of <δ> over all measured molecules. For all images, intensities are thresholded above 1100 photons and localization precisions are thresholded below 0.15 pixels (σ_loc < 20 nm). Scale bars (a–d), 7 μm; (e), 800 nm; (f), 260 nm. Source data are provided as a Source Data file.

Fig. 3. Influence of the detection parameters on 4polar-STORM imaging.

a 2D histogram of (|Δρ|, δ) values obtained in single AF488 molecules (all detected intensities) present in ROI 1 (rectangle in Fig. 2e), with $△\rho =\rho -⟨\rho ⟩$. b 2D histogram of (PSF radius, δ) values of the same region. c 2D histogram of (intensity, |Δρ|) values of the same region. d Same 4polar-STORM images as in Fig. 2e, f, depicting δ sticks only for molecules for which r < 1.3 pixels (169 nm). The corresponding average <δ> over all measured molecules (present in the rectangle for ROIs 1–3 or the whole region for ROI 4 as in Fig. 2e) are shown. 4polar-STORM analyses were repeated independently on SF ROIs of three different samples (about 3–5 cells per sample) with similar results. e Same ROIs as in (d) showing ρ stick images only for molecules with δ < 110° (all PSF radii). f Corresponding polar-plot histograms of ρ for regions indicated in (e). The corresponding σ_Δρ values are shown. Scale bar (d, e), 800 nm. Source data are provided as a Source Data file.

Fig. 4. 4polar-STORM imaging of actin filament organization in different types of stress fibers.

a (left) σ_Δρ (δ < 110°) and (right) <δ> (all molecules, the percentage of the δ population with δ < 110° is indicated), measured in ROIs on different types of SFs. Eight cells were examined, giving a total number of measured ROIs of: n = 42 (ventral), n = 32 (peripheral), n = 30 (arc), n = 63 (FA), n = 28 (dorsal). The red bars represent mean values ± standard deviation. Statistical significance: ns (p > 0.05); ** (p < 0.01); *** (p < 0.001); **** (p < 0.0001) (two-sided unpaired two-sample T-test). p values from top bar to bottom, compared to the ventral SF population: (Left) p = 6.3e−12, 3.6e−8, 6.2e−4, 1.2e−5, 1055 molecules measured on average per ROI; (right) p = 2.6e−3, 8.2e−12, 0.3, 2.6e−7, 2994 molecules measured on average per ROI. b top: STORM images of control and blebbistatin-treated U2OS cells on micropatterns (see Methods). Scale bars, 6.5 μm; bottom: 4polar-STORM ρ stick images of zoomed regions (rectangles in the STORM images). Scale bars, 500 nm. Images were repeated on four different cells per condition, with similar results. c, d σ_Δρ and <δ> measured in four cells per condition (control, blebbistatin-treated), using in total n = 87 ROIs in control cells and n = 43 ROIs in treated cells. c σ_Δρ values (δ < 110°), p = 3.5e−8, 2333 molecules measured on average per ROI, (d) <δ> values (all δ selected), p = 4.1e−5, 6376 molecules measured on average per ROI. e Polar-plot histograms of ρ in the zooms shown in (b), for δ < 110° (left) and δ > 120° (right). The corresponding σ_Δρ values are shown. f Schematic representations of the SFs observed in control and blebbistatin-treated conditions, showing highly aligned actin filaments in 2D (top) and disorganized filaments in 3D (bottom). Source data are provided as a Source Data file.

Fig. 5. 4polar-STORM imaging of actin filament organization in lamellipodia.

a Single-molecule localization STORM image of a B16 cell labeled with AF488-phalloidin. b Corresponding 4polar-STORM ρ stick image with color-coded orientation measurements. c 4polar-STORM δ stick image with color-coded wobbling angle measurements. 4polar-STORM images were repeated independently on four different cells with similar results. d Examples of ρ stick images showing molecules with δ < 110° and corresponding STORM images in selected ROIs (squares in (b)). ROIs 1–9, regions in the lamellipodium; ROI 3,9, microspikes; ROI 10, SF. e Polar-plot histograms of ρ for the regions shown in (b). The condition δ < 110° is used, except for red-circled histograms for which δ >120° molecules are selected. The red lines in ROIs 1–6 are fits of the ρ histograms that resemble bimodal distributions. The blue lines in these same ROIs correspond to the normal direction of the cell membrane in the respective ROI. Angle between the two peaks around the membrane normal direction, obtained from the fits: ROI1 (60°), ROI2 (59°), ROI3 (50°), ROI4 (52°) ROI5 (70°), ROI6 (73°). Scale bars (a–c), 4 μm; (d), 260 nm. Source data are provided as a Source Data file.