Fig 1.
Actin-pl-clusters visualized by “p”halloidin and “l”ifeact in fixed NRK cells.
(A) A representative confocal-based SRM (Olympus FV-OSR) image of cellular actin structures, including stress fibers, thinner filamentous mesh-like structures, and small actin clusters, visualized by Alexa647-phalloidin in NRK cells. The region within the yellow square is magnified on the right, with superimposed arrowheads showing elongated dot-like actin clusters. (B) A representative dual-color FV-OSR observation of NRK cells transfected with Lifeact-mGFP (green in the rightmost image) and stained with Alexa647-phalloidin (red). The corresponding region indicated by the yellow box in the top images is magnified in the bottom images.
Fig 2.
Live-cell super-resolution images of cortical actin structures, showing the existence of many actin-pl-clusters as well as the cortical fine actin meshwork and stress fibers.
A representative snapshot from image sequences of NRK cells transfected with Lifeact-mGFP and observed by (A) a spinning-disk confocal-based SRM (SDSRM; Olympus SD-OSR system) operated at 2 Hz (a time resolution of 0.5 s with a signal integration time of 0.5 s) for a period of 50 s, and by (B) structured illumination microscopy (3D-SIM mode of Nikon N-SIM system) operated at 0.44 Hz (a time resolution of 2.3 s with a signal integration time of 0.1 s) for a period of 60 s. The regions within the yellow boxes in the images on the left are magnified on the right, with arrowheads showing actin-pl-clusters. For the original image sequences, see S1 Movie (SDSRM) and S2 Movie (3D-SIM).
Fig 3.
3D-SIM tomographic images of actin-pl-clusters in fixed NRK cells, showing that actin structures visualized by Lifeact-mGFP are localized within ~400 nm from the PM cytoplasmic surface.
3D-SIM observations of fixed NRK cells transfected with Lifeact-mGFP using a Nikon N-SIM system every 120 nm from the glass surface (Z: 0 nm) up to 720 nm, with z-resolution of ± 200 nm.
Fig 4.
Actin-pl-clusters exhibited continuous dynamic morphological changes and movements on and along the cortical actin meshwork, linking actin filaments (acting as nodes in the meshwork).
(A) Images of actin-pl-clusters visualized by Lifeact-mGFP were obtained using a Nikon N-SIM system operated at a time resolution of 2.3 s (clipped from S2 Movie). In the first image sequence, the cluster split into two separate clusters. In the second sequence, the cluster indicated by the arrowheads translocated along the mesh and merged into the larger cluster. (B) Images of actin-pl-clusters visualized by Lifeact-mGFP were obtained using an Olympus SD-OSR system operated at a time resolution of 0.5 s, and were sampled every 5 s (sampled and clipped from S1 Movie). In the first image sequence, the cluster in the center elongated and the cluster in the left-bottom corner underwent merging, elongation, and shrinkage. In the second image sequence, the cluster indicated by the arrowhead in the image at 0 s elongated and spread to form a fork morphology, and then split into two fragments. One of these fragments underwent shrinkage, as indicated by the second arrowhead in the image at 25 s. (C) Individual actin-pl-clusters that fit into a square region of 0.6 × 0.6 μm, as observed using Olympus SD-OSR system at a time resolution of 0.5 s, were traced using the single fluorescent-molecule tracking software we previously developed, and their trajectories were superimposed on the SRM image (top-right region in S1 Movie). (D) The trajectories in the yellow regions in C were magnified by a factor of 3.3 and were color-coded into different colors every 5 s (in the order of purple, blue, green, orange, red, and then back to purple). The black and white arrowheads indicate the start and end positions, respectively.
Fig 5.
Comparison of actin-pl-cluster images observed by TIRFM with those by SRM.
(Top) Observations of NRK cells expressing Lifeact-mGFP after fixation was performed in the same field of view, first (Top-left) with a Zeiss Airyscan SR microscope (38 nm/pixel) and then (Top-right) with our custom-built TIRF microscope based on a Nikon Ti-E inverted microscope (68 nm/pixel). Arrowheads indicate the actin-pl-clusters observed in each image. (Bottom) The regions within the yellow boxes in the top images are magnified in the bottom images. The magenta arrowheads indicate the actin-pl-clusters observed in both the SRM and TIRFM images and the cyan arrowheads indicate an actin-pl-cluster that could be observed clearly in the SRM image, but was blurred in the TIRFM image.
Fig 6.
The actin-pl-clusters observed using SRM and TIRFM exhibited very similar dynamics, which was greatly suppressed by the myosin inhibitor blebbistatin.
The trajectories of individual actin-pl-clusters that fit into a square region of 0.6 × 0.6 μm were obtained using SRM (Olympus SD-OSR) and TIRFM at a frame rate of 2 Hz (the actual integration times were 0.5 s and 0.167 s, respectively). The ensemble-averaged MSD-Δt plots (with error bars indicating standard errors) were obtained for the actin-pl-clusters observed by SRM (green triangles; 124 trajectories in five cells), TIRFM (purple squares; 120 trajectories in five cells), and TIRFM after the treatment with 25 μM blebbistatin, an inhibitor of myosin II (blue circles; 112 trajectories in five cells). The plots for SRM and TIRFM (before blebbistatin addition) were fitted by the equation MSD(Δt) = 4DΔt + (vΔt)2 representing the model of Brownian diffusion + directed motion, where D is the diffusion coefficient and v is the drift velocity. The fitting of the SRM data yielded the dark green curve with v = 0.011 ± 0.00016 μm/s and D = 0.0013 ± 0.000014 μm2/s and the fitting of the TIRFM data yielded the dark magenta curve with v = 0.013 ± 0.00019 μm/s and D = 0.0010 ± 0.000018 μm2/s (the error bars indicate the 68.3% confidence limit of the fitting).
Fig 7.
A newly developed probe, Lifeact-TM, detected actin-pl-clusters closely apposed to the ventral membrane.
(A) Schematic diagram of a newly developed probe for the detection of actin-concentrated structures (including mesh-like structures, stress fibers, and actin-pl-clusters) in close proximity to the PM cytoplasmic surface, named Lifeact-TM, with an ACP tag for fluorescence labeling as the extracellular domain. (B) Snapshot images from two-color TIRFM observations of (Left) Lifeact-mGFP to visualize actin-pl-clusters and (Right) single Lifeact-TM molecules labeled with SeTau647. (C) (Top) The region within the yellow square of the Lifeact-mGFP image shown in B Left. The image was binarized to optimally extract (Middle) actin-pl-clusters and (Bottom) stress fibers. (D) The area size distribution of binarized actin-pl-clusters (85 clusters in five cells). The arrowhead indicates the mean value. (E) (Left) The single-molecule trajectories of Lifeact-TM obtained at 60 Hz (16.7-ms time resolution) are superimposed on the Lifeact-mGFP image shown in C Top. Temporary immobilizations, or TALL events, were detected within each trajectory (see the TALL Detection section in the Materials and Methods), and were color-coded based on their mobility and location: blue indicating the mobile period, and magenta, yellow, and green indicating the TALL periods that occurred in the regions of actin-pl-clusters, stress fibers, and elsewhere, respectively. For the original image sequence, see S3 Movie. (Right) Another representative TALL analysis result in a different cell. (F) The distribution of the diffusion coefficients on the time scale of 16.7–50 ms (D33ms) of Lifeact-TM during the mobile period (239 trajectories in five cells). The arrowhead indicates the median value.
Table 1.
The percentage of TALL events that Lifeact-TM molecules exhibited on actin-pl-clusters, stress fibers, and the cortical actin meshwork (the PM regions on which these structures are projected), and the percentage of at least one TALL event once Lifeact-TM entered the PM region on the three actin structures (the projected PM regions).
Fig 8.
Myosin II and filamin A, the components of actin nodes, did not colocalize with actin-pl-clusters.
Two-color TIRFM observations of (Top row) Lifeact-mGFP (green) and Halo-filamin A labeled with TMR (red), and (Second row) Lifeact-Halo labeled with TMR (green) and EGFP-MRLC (red) were performed at 60 Hz (16.7-ms time resolution), so the monomers of Halo-filamin A and EGFP-MRLC and their clusters of up to five molecules could be visualized within the dynamic range of the camera. In the rightmost column, the top values indicate the percentages of actin-pl-clusters that colocalized with actin-node-related cytoplasmic proteins (Halo-filamin A and EGFP-MRLC). The values in the brackets were obtained as controls to evaluate incidental colocalization, indicating the colocalization percentages when the images of Lifeact-mGFP were rotated 180 degrees. The bottom values are the P values of a Student t-test for the colocalization percentages between the correct superimpositions and the rotated superimpositions and the number of cells examined. No statistically significant colocalizations were found for these actin-node-related cytoplasmic proteins.
Fig 9.
Actin-pl-clusters were found in intact HeLa cells as well as in the NRK cells extensively used in this study: After the addition of 200 nM latrunculin A, actin-pl-clusters disappeared in 1~2 min in both HeLa and NRK cells, but in HeLa cells, actin clusters much larger than the actin-pl-clusters started appearing between 600 and 750 s after latrunculin addition.
The figure shows representative snapshots from image sequences of HeLa cells (Left) and NRK cells (Right) transfected with Lifeact-mGFP and observed by SRM (an Olympus SD-OSR system). After the snapshots at time = 0 (0 s), the cells were treated with 200 nM latrunculin A and time-lapse observations were performed for about 1,000 s.
Fig 10.
Podosome-related proteins N-WASP, cortactin, Tks4, and Tks5 colocalized with actin-pl-clusters.
Two-color TIRFM observations of (Top row) Lifeact-mGFP (green) and Halo-N-WASP labeled with TMR (red) (see S4 Movie), (Second row) Lifeact-Halo labeled with TMR (green) and EGFP-cortactin (red), (Third row) Lifeact-mGFP (green) and Halo-Tks4 labeled with TMR (red) (see S5 Movie), (Fourth row) Lifeact-mGFP (green) and Halo-Tks5 labeled with TMR (red) (see S6 Movie) and (Bottom row) Lifeact-mGFP (green) and Halo-paxillin labeled with TMR. The observation conditions and the analyses were the same as in Fig 8. Statistically significant colocalizations with actin-pl-clusters were found for N-WASP, cortactin, Tks4, and Tks5.
Fig 11.
Residency times of N-WASP, Tks4, and Tks5 at the actin-pl-clusters and at the PM outside the actin-pl-clusters.
(Top) The histograms show the distributions of residency times of individual N-WASP, Tks4, and Tks5 molecules at the actin-pl-clusters. Each distribution could be fitted well by the sum of two exponential functions, yielding the two exponential decay constants with the fraction (%) of each population. The decay constants were then corrected for the photobleaching lifetime of TMR bound to the Halo protein (6.2 s), and the shorter residency time of τ1 and the longer residency time of τ2 were obtained. For N-WASP, τ1 = 0.020 ± 0.0040 (60%) and τ2 = 0.17 ± 0.017 (40%), for Tks4, τ1 = 0.064 ± 0.0018 (81%) and τ2 = 0.36 ± 0.059 (19%) and for Tks5, τ1 = 0.084 ± 0.026 (36%) and τ2 = 0.20 ± 0.038 (64%). The error bars indicate the 68.3% confidence limit of the fitting, and the numbers of recruitment events (n = 395 for N-WASP, n = 530 for Tks4 and n = 295 for Tks5) were obtained from five cells for each molecule. Since trajectories as short as 1–3 frames (shorter than 50 ms at 60 fps) often include those produced by transient background noise, these short trajectories were excluded from the analysis to avoid overestimation of the number of recruitment events with short residency lifetimes (thus the x-axes of the graphs start from 67 ms). (Bottom) The histograms show the distributions of residency times of individual N-WASP, Tks4, and Tks5 molecules at the PM outside the actin-pl-clusters. For N-WASP, τ1 = 0.064 ± 0.0098 (68%) and τ2 = 0.56 ± 0.22 (32%), for Tks4, τ1 = 0.066 ± 0.0037 (69%) and τ2 = 0.35 ± 0.044 (31%) and for Tks5, τ1 = 0.092 ± 0.0047 (68%) and τ2 = 0.71 ± 0.15 (32%). The recruitment events (n = 959 for N-WASP, n = 530 for Tks4 and n = 262 for Tks5) were obtained from five cells for each molecule.
Fig 12.
The effect of CK-666, an inhibitor of Arp2/3, on actin-pl-clusters, showing that the number of actin-pl-clusters quickly decreased, within 20 s, to the levels of ~20% and ~7.2% of those found before the CK-666 addition.
(A) Representative snapshots from image sequences of NRK cells transfected with Lifeact-mGFP and observed by TIRFM before (Left) and 20 s after the addition of CK-666 (Right). Control, treated only with DMSO. (B) The number of actin-pl-clusters observed per cell at 20 s after the CK-666 treatment vs. that before the treatment. The error bars indicate the standard errors (n = 6 cells for each condition).
Fig 13.
Actin-pl-clusters exhibit dynamic morphological changes and podosome-related proteins are recruited to actin-pl-clusters.
Actin-pl-clusters underwent dynamic motion and morphological changes on or near the PM. These include (a) elongation, (b) splitting, (c) directed movement along actin meshes, (d) merging, (e) shrinking, and (f) spreading (often forming a fork-like morphology). N-WASP, Tks4, and Tks5 molecules were either directly recruited to the actin-pl-cluster from the cytoplasm (60–70% of molecules) or first arrived at the PM outside the cluster and were then recruited to the cluster by lateral diffusion on the PM cytoplasmic surface. They left the actin-pl-clusters either by direct dissociation into the cytoplasm (approximately 70% of molecules) or by lateral diffusion on the PM cytoplasmic surface.