Skip to main content
. 2024 Nov 8;13:e101652. doi: 10.7554/eLife.101652

Figure 1. Epidermal growth factor receptor (EGFR), PI(4,5)P2, and phosphatidylserine (PS) are distributed nonrandomly in the plasma membrane.

(A) Images of EGFR–rsKame (cyan), PAmCherry–PI(4,5)P2 (magenta), and HMSiR–PS (yellow) before epidermal growth factor (EGF) stimulation. PAmCherry–PI(4,5)P2 and Halo–evt2–PH (Halo–PS) were transiently expressed in a cell line stably expressing EGFR–rsKame. Cells were incubated in serum-free medium in the presence of the HMSiR–Halo ligand overnight and treated with paraformaldehyde and glutaraldehyde. Left, a typical image of 15×15 µm. Right, enlarged image of the square region surrounded by a bold line in the left image. Enlarged images of other square regions surrounded by thin lines in the left image are shown in Figure 1—figure supplement 2C. (B) Distribution of the densities of EGFR (left) and PI(4,5)P2 cluster areas (right). Cells were incubated in serum-free medium overnight, stimulated with or without 20 nM EGF for 1 min, and treated with paraformaldehyde and glutaraldehyde. From the single-molecule localization microscopy (SMLM) images, the cluster areas and the number of clusters were measured. After the cluster number was normalized to the cell area (density), the cluster density was plotted as a function of the cluster area. Inset, enlarged graphs for the cluster areas of <0.03 µm2. Blue and red indicate before and after EGF stimulation, respectively. Data are means ± SEM of at least eight cells. (C) Univariate H(R) values of EGFR–rsKame (left), PAmCherry–PI(4,5)P2 (middle), and HMSiR–PS (right) in cells incubated in the absence (blue) or presence (red) of 20 nM EGF for 1 min. Ripley’s univariate H-function was calculated from the SMLM images. Data are means ± SEM of nine (EGFR–rsKame and PAmCherry–PI(4,5)P2) or 10 (HMSiR–PS) cells.

Figure 1—source data 1. Original image files displayed in Figure 1A.
Figure 1—source data 2. Raw data files displayed in Figure 1B and C.

Figure 1.

Figure 1—figure supplement 1. Multicolor single-molecule localization microscopy (SMLM) analysis workflow.

Figure 1—figure supplement 1.

(A) Typical images of 512×512 pixels (34.3×34.3 µm with a pixel size of 67 nm). Bright spots around the cells indicate 100 nm TetraSpeck Microspheres (beads), and those in the cells indicate fluorescence-labeled epidermal growth factor receptor (EGFR), PI(4,5)P2, or phosphatidylserine (PS) (samples). (B) Single-molecule tracking (SMT) analysis followed by alignment using affine transformation. An affine transformation matrix was calculated from the XY coordinates of the same beads in the nth frame (or color channel) and a target frame (or channel) using least-squares regression. Example trajectories of the same beads before (white) and after (yellow) drift correction during 1000 frames are shown in the left panel. A histogram of the X coordinates of the beads after affine transformation is shown under the trajectory. The difference in the XY coordinates of the same beads between two color channels, before (white) and after (yellow) affine transformation, is shown in the right panel. (C) Reconstituted images. After the alignment of localization with affine transformation, images were reconstituted by convolution with a Gaussian kernel around each localization of the sample. The 5×5 µm area is merged and expanded in the right panel. (D) Definition and calculation flow of Ripley’s K-function variants. Npi(r) is the number of points within distance r around the ith point. In the calculation of the univariate K-function, the number of points belonging to the same channel was counted. In the bivariate K-function calculation, the number of points belonging to the other channel was counted. Kpi(r) is a local K-function around the ith point, where λ is the mean density in the region of interest (ROI). K(r) is Ripley’s K-function, which is the ensemble average of all Kpi(r) for n points in the ROI. K(r) is πr2 if the spatial distribution of the n points in the ROI is random. Because intuitive comparison with πr2 is difficult, variants L(r) and H(r) are often used; L(r) and H(r) become r and 0 if random, respectively. G(r) is the difference function of K(r), where the number of points in the donut area between circles of radii rm+1 and rm is centered at each point. G0(r) is normalized G(r), where μ and σ are the mean and SD of G(r) in a random distribution. Unlike H(r), G(r) does not involve nonlinear computations, and therefore is suitable for the following spatial mapping analysis. (E) Automated multidistance spatial cluster analysis. After alignment with affine transformation, the image area was divided into ROIs with two rectangles (top panels). K(r) and its variants for the sample and for random simulation data were computed for the center points within the red region. In the calculation, the target points within both the red and blue areas were considered, to reduce the edge effect. After the calculations for all ROIs, ROIs containing cellular regions were detected with tiling images of the mean density of each ROI (lower left panel). The Otsu method was used to binarize the cellular and noncellular regions. The lower right panel shows an example of the mean H(r) for the cellular region. Data are shown as means ± SEM of N ROIs within the cellular regions. (F) G-function spatial map (G-SMAP) analysis. In the computation process in (E), the local K-function and its variants, including G0 pi(r), were calculated for all the sample and random simulation data points. G-SMAP images were generated by integrating the G0 pi(r) kernel around each point. Example G-SMAP images for sample and random simulation data in the same ROI are shown in pseudo rainbow colors (upper left panel). A pixel-intensity histogram of the G-SMAP images is shown in the upper right panel. Pixels above a threshold value (e.g. four SD of the random distribution) were defined as cluster regions, and closed cluster regions were numbered as individual clusters (pseudo rainbow colors in bottom panels).
Figure 1—figure supplement 2. Construction and evaluation of multicolor single-molecule localization microscopy (SMLM).

Figure 1—figure supplement 2.

(A) Amounts of phosphorylated-Tyr1068 epidermal growth factor receptor (EGFR) in the cell line. Top: Western blotting analysis of phosphorylated-Tyr1068 EGFR and total EGFR. Parental HeLa cells and EGFR-KO cells stably expressing EGFR–rsKame fusion protein were incubated in a serum-free medium overnight and stimulated with 20 nM epidermal growth factor (EGF) for 1 min. Phospho-EGFR and total EGFR were detected with anti-pY1068 EGFR and EGFR, respectively. Bottom: Ratio of phosphorylated-Tyr1068 EGFR/total EGFR. The ratio was normalized to the mean value of parental HeLa cells after EGF stimulation. Data are means ± SD of three experiments. (B) Amounts of phosphorylated extracellular signal-regulated kinase (ERK) in the cell line. Top: Western blotting analysis of phosphorylated ERK and total ERK. Cells were prepared as described in (A). Phospho-ERK and total ERK were detected with anti-pERK and ERK, respectively. Bottom: Ratio of phosphorylated-ERK/total ERK. The ratio was normalized to the mean value of parental HeLa cells after EGF stimulation. Data are means ± SD of three experiments. (C) Enlarged images of square regions surrounded by thin lines in Figure 1A. (D) Full-width half-maximum (FWHM) of the positional distributions for single immobile fluorescent beads. Data are means ± SEM of 10 beads. (E) Position accuracy of the dual-color immobile fluorescent beads. The accuracy of the superimposition of two images was calculated after affine transformation. Data are means ± SEM of 10 beads. NS (not significant) on Welch’s t-test.
Figure 1—figure supplement 2—source data 1. PDF file for western blotting analysis displayed in Figure 1—figure supplement 2A and B.
Figure 1—figure supplement 2—source data 2. Original files for western blotting analysis displayed in Figure 1—figure supplement 2A and B.