Proteins of FAs organize as universal nanoclusters of similar size and containing a similar number of proteins.

(A) Representative dual-color STORM image of an HFF-1 cell fixed after 24 h on FN and labeled for α5β1 (magenta) and paxillin (green). Scale bar: 5 μm. (B) Representative images of all the adhesions identified for α5β1 with paxillin (left), talin (middle) and vinculin (right), with zoomed-in regions shown below each panel. Scale bar top row: 5 μm, bottom row: 1 μm. (C) Representative images of all the adhesions identified for αvβ3 with paxillin (left), talin (middle) and vinculin (right), with zoomed-in regions shown below each panel. Scale bar top row: 5 μm, bottom row: 1 μm. (D) Representative images all the adhesions identified for of active β1 (act β1) with paxillin (left), active β3 (act β3) with paxillin (right), with zoomed-in regions shown below each panel. Scale bar top row: 5 μm, bottom row: 1 μm. (E) Distribution for the number of localizations per secondary antibody on glass next to cell regions for both of the secondary antibodies used in this study, goat anti-mouse Alexa 405–Alexa 647 (Ch1) and goat anti-rabbit Cy3–Alexa 647 (Ch2). The median number of localizations per spot corresponds to: 5 for Alexa 405-Alexa 647-conjugated antibodies and 3 for Cy3–Alexa 647-conjugated antibodies (see Materials & Methods). (F, G) Box-and-whisker plots showing the number of localizations per nanocluster (F) and the nanocluster diameter (G) for the proteins in this study. The box represents the range from the 25th to the 75th percentiles and whiskers from 10th to 90th percentiles, with the median indicated by a horizontal line. Points outside these whiskers correspond to outliers. Individual points (points within the whiskers not shown) correspond to the median value over all nanoclusters for each cell (for N values, i.e., number of independent experiments, see SI Table S5). No statistical differences were found using the multi comparison one-way ANOVA test, where not significant (ns) corresponds to p>0.05.

Mesoscale organization of integrin nanoclusters and adaptor proteins inside FAs.

(A, B) Mean density of nanoclusters per unit area inside FAs (A) and on the cell membrane outside adhesions (B), with α5β1 and actβ1 in light and dark green, respectively; and αvβ3 and actβ3 in light and dark blue, respectively. The densities of integrins nanoclusters per unit area are shown as stacked graphs to allow comparison between the total number of integrin nanoclusters (α5β1 and αvβ3) and the adaptor proteins. The bars show the mean ± st. dev of the distribution. Statistical significance between data of all proteins was tested using a multi comparison one-way ANOVA, assuming Gaussian distribution of points. The comparison between α5β1 and αvβ3, (actβ1 and actβ3) was done with an unpaired t-test, where ns = not significant, p>0.05; *, p<0.05; **, p<0.01; ***, p<0.001. The insets show representative images of nanoclusters for α5β1 (magenta) and paxillin (green) in FAs (A) and outside adhesions (B). Scale bar: 1 μm. (C) Representative dual-color STORM image of the total populations of α5β1 (magenta) and αvβ3 (green) integrins inside FAs. The panel on the far-right side corresponds to a zoomed-in region of a single FA (white box) showing spatially segregated α5β1 and αvβ3 nanoclusters. Scale bars: 2 μm, and 500 nm for the zoomed-in region. (D) Histogram of the EED-NN values between α5β1 and the different adaptors (green) and, between αvβ3 and the adaptors (blue). The black line in each plot corresponds to the histogram for the simulated data sets. The dashed red line at zero corresponds to nanoclusters contiguous to each other. Percentage of overlapping nanoclusters: α5β1-paxillin: exp = 28.4%, sim = 24.4%; α5β1-talin: exp = 6.7%, sim = 8.1%; α5β1-vinculin: exp = 6.0%, sim = 7.7%; αvβ3 -paxillin: exp =31.7%, sim=29.2%; αvβ3 -talin: exp =15.7%, sim=16.5; αvβ3-vinculin: exp =14.9%, sim=17.4%. (E) Histogram of the NND values between nanoclusters of the same protein. The black line in each plot corresponds to the histogram for the simulated data sets. Bin width is 10 nm.

Integrin α5β1 nanoclusters are preferentially distributed on the edges of FAs.

(A) Distribution of distances from the center of mass (CoM) position of each individual α5β1 nanocluster to its closest FA edge obtained from DBSCAN-analyzed STORM images. The histogram represents the frequency distribution over all cells. Green lines correspond to the experimental data (Exp) and the simulated uniform distribution (Sim) is shown as a black histogram. Bin width is 10 nm. The inset shows the distribution for the first 250 nm. (B) Edge Proximity Factor (EPF) per cell for each of the proteins investigated. Data are shown as median values per individual cell (color dots). The mean ± st. dev. over all the cells are superimposed as black lines. Statistical significance between experimental and simulated data was tested using a paired t-test, where ns = not significant, p>0.05; *, p<0.05; **, p<0.01; ***, p<0.001. (C, D) Variation of the nanocluster distances from the FA edges with respect to random organization, as function of the distance to the FA edges for α5β1 and αvβ3 integrin nanoclusters (C) and for the adaptor proteins (D). (E) Percentage of nanoclusters proximal to FA edges (<100nm) for the total and active β1 nanocluster population (color dots) together with results from simulations of random organization (black dots). Each dot corresponds to the median value per cell, each cell containing multiple FAs. The mean ± st. dev. over all the cells is superimposed as red lines. Statistical significance between experimental and simulated data was tested using a paired t-test, and to test the significance between α5β1 and actβ1 the unpaired t-test was used, assuming a Gaussian distribution of data, where ns = not significant, p>0.05; *, p<0.05; **, p<0.01; ***, p<0.001.

Active β1 nanoclusters form multiprotein hubs at the FA edges.

(A) Population of active β1-talin nanocluster pairs as a function of their distance-to-the-FA edge. The data plotted in light green correspond to active β1 nanoclusters located close to talin (i.e., iNND < median), while the inset dark green corresponds to active β1 nanoclusters located far from talin (i.e., iNND > median). iNND distributions are shown in Fig. S4B. Black lines are for simulated data sets. Bin width is 20 nm. (B) Variation from random for active β1-talin nanocluster pairs as function of their distance-to-the-FA-edge. Filled bars correspond to active β1 nanoclusters located close to talin (iNND < median) and empty bars correspond to active β1 nanoclusters located far from talin (iNND > median) (C) Similar to (B) but for active β3 nanoclusters. (D-F) Variation from random for paxillin (D), talin (E), vinculin (F) active β1 nanocluster pairs, as function of their distance-to-the-FA-edge, with data segregated in two different distributions according to the pairs proximity. Bin width is 20 nm.

Spatial regulation and activation of integrin nanoclusters and protein adaptors inside focal adhesions.

α5β1, αvβ3, and main protein adaptors organize as small nanoclusters inside FAs physically segregated from each other. A considerable high fraction of α5β1 and αvβ3 integrins remain in an inactive state and do not intermix at the nanoscale. Active β1 nanoclusters (green) on the FA edge reside in close proximity to the adaptor protein nanoclusters and thus possibly engaged to actomyosin filaments (red and blue). Integrin nanoclusters in the FA center are more randomly distributed with respect to one another, most probably engaged with branched cortical actin. According to our results we propose that α5β1 nanoclusters located at the FA periphery could be primarily involved in cell substrate attachment, whereas α5β1 and αvβ3 nanoclusters at the center of FAs engage in mechanosensing and mechanotransduction. For simplification, we show a cross-section of a FA where active β1 integrins at the edge are engaged with talin. However, our data also show the presence of active β1 nanoclusters spatially separated from talin and thus possibly engaged to other binding partners linking β1 to the actin cytoskeleton (not shown in the cartoon for the sake of clarity).