Step-by-step quantitative analysis of focal adhesions

1MaterialsFluorescence imagesImageJ software

2Fluorescence images

3ImageJ software

4Critical pointsSample should be properly immunostained so that good quality fluorescence images can be captured.The images should be of high quality; unfocused images and inadequate resolution hinder the quality of the analysis.Fluorescence images should ideally be captured at 100× magnification.If color images are captured, single channel images should first be created with the ImageJ split channel command.Use raw image data which are greyscale images without any special preprocessing.

5Sample should be properly immunostained so that good quality fluorescence images can be captured.

6The images should be of high quality; unfocused images and inadequate resolution hinder the quality of the analysis.

7Fluorescence images should ideally be captured at 100× magnification.

8If color images are captured, single channel images should first be created with the ImageJ split channel command.

9Use raw image data which are greyscale images without any special preprocessing.

10Here, we have analyzed focal adhesions of MDA-MB-231 cells. However, our method can be applied to not only any cell type such as fibroblasts, endothelial cells or leucocytes but also to any structure in the cell with a localized signal such as focal adhesions, endosomes or invadopodia. Focal adhesions were immunostained for vinculin using a vinculin specific primary antibody (Sigma Cat no: V9131), followed by a fluorophore conjugated secondary antibody. Other markers of focal adhesions such as paxillin or focal adhesion kinase could also be easily used for analysis. 8-bit images were captured using an Olympus epi-fluorescence microscope with an Olympus infinity corrected 100× oil UPlanSApo objective with a numerical aperture of 1.4 and an Olympus SLR E-330 camera such that each image was 3136 pixels by 2352 pixels and 37 pixels corresponded to 2 μm. All steps of image processing were carried out using ImageJ which is freely available and in the public domain. ImageJ provides a wide range of processing and analysis approaches possible via not only built-in functions but also numerous plugins. ImageJ has been in use for some time by researchers. However, choosing the right combination and order of processing steps is crucial for obtaining meaningful quantitative results.

11The image processing optimized here aims to identify FAs immunostained for vinculin. The raw fluorescent images are processed in steps as follows:Step 1: Apply subtract background. Selection of sliding paraboloid option changes the rolling ball to a parabolic with the same ball radius in pixels but with sharper curvature. A parabola slides in different directions over the image and calculates and subtracts local background from the original image. We chose the sliding paraboloid option with the rolling ball radius set to 50 pixels.Step 2: Enhance the local contrast of the image by running clahe (Contrast Limited Adaptive Histogram Equalization). The plugin clahe offers three critical parameters which are block size, histogram bins and maximum slope to clarify desired objects. We used the following values: block size = 19, histogram bins = 256, maximum slope = 6, no mask and fast.Step 3: Apply mathematical exponential (exp) to further minimize the background.Step 4: Adjust brightness & contrast automatically. The brightness & contrast tool updates the lookup table of the image based on an analysis of the histogram of the image.Step 5: Run Log3D (Laplacian of Gaussian or Mexican Hat) filter. The log3D plugin filters the image based on user predefined parameters which are standard deviations in X, Y and Z directions. For 2D images Z direction is not relevant and slice per slice processing is used. We defined the size of log3D filter as sigma X = 5 and sigma Y = 5.Step 6: Run the threshold command. The threshold command converts a grayscale image to a binary image with two pixel values, 255 (white) and 0 (black). For the threshold command, there are options such as Huang, triangle, mean or default. In addition, the user can adjust minimum and maximum threshold values or set the threshold levels according to an analysis of the histogram of the current image automatically. Here we used the default method and threshold was adjusted automatically.Step 7: Execute analyze particles command. The analyze particles command scans the thresholded (binary) image and finds the edges of objects or particles. Essentially, particle analyzer counts and measures particles according to user pre-defined parameters which are size and circularity. We set these parameters as follows: size = 50-infinity and circularity = 0.00–0.99.

12Step 1: Apply subtract background. Selection of sliding paraboloid option changes the rolling ball to a parabolic with the same ball radius in pixels but with sharper curvature. A parabola slides in different directions over the image and calculates and subtracts local background from the original image. We chose the sliding paraboloid option with the rolling ball radius set to 50 pixels.

13Step 2: Enhance the local contrast of the image by running clahe (Contrast Limited Adaptive Histogram Equalization). The plugin clahe offers three critical parameters which are block size, histogram bins and maximum slope to clarify desired objects. We used the following values: block size = 19, histogram bins = 256, maximum slope = 6, no mask and fast.

14Step 3: Apply mathematical exponential (exp) to further minimize the background.

15Step 4: Adjust brightness & contrast automatically. The brightness & contrast tool updates the lookup table of the image based on an analysis of the histogram of the image.

16Step 5: Run Log3D (Laplacian of Gaussian or Mexican Hat) filter. The log3D plugin filters the image based on user predefined parameters which are standard deviations in X, Y and Z directions. For 2D images Z direction is not relevant and slice per slice processing is used. We defined the size of log3D filter as sigma X = 5 and sigma Y = 5.

17Step 6: Run the threshold command. The threshold command converts a grayscale image to a binary image with two pixel values, 255 (white) and 0 (black). For the threshold command, there are options such as Huang, triangle, mean or default. In addition, the user can adjust minimum and maximum threshold values or set the threshold levels according to an analysis of the histogram of the current image automatically. Here we used the default method and threshold was adjusted automatically.

18Step 7: Execute analyze particles command. The analyze particles command scans the thresholded (binary) image and finds the edges of objects or particles. Essentially, particle analyzer counts and measures particles according to user pre-defined parameters which are size and circularity. We set these parameters as follows: size = 50-infinity and circularity = 0.00–0.99.

19As a result of the analyze particles process, outlines of the particles found can be saved and overlaid on the original image to visually confirm faithful detection of FAs.

20Our analysis on FAs of MDA-MB-231 cells cultured on fibronectin, an extracellular matrix protein that promotes cell adhesion and on K-casein, a blocking protein that prevents cell adhesion, determined the following:

21As expected, cells form more and larger FAs on a fibronectin coated surface compared with a K-casein coated surface (p < 0.05).

22Focal adhesions (FAs) are multifunctional structures comprising over a hundred proteins. FAs provide not only physical contact between a cell and its extracellular environment but also mediate inside-out and outside-in signaling. Therefore, quantitative analysis of FAs is crucial for understanding important cellular phenomena such as cell morphology, motility, tissue organization, growth and differentiation. Since cells can form many and micrometer scale FAs, their quantitative analysis demands well-optimized image analysis approaches. The dimensions of focal adhesions range from 0.25 to 10 μm depending on the maturation stage of the FA as well as the cell type. Mature focal adhesions are also associated with the cytoskeleton. There can be more than a hundred proteins associated with a focal adhesion and the identification of focal adhesion is possible by labeling protein of interest within this complex. Labeling of a mechanical stabilizer protein, vinculin, is the best suitable method technically and commercially.

23ImageJ software: http://rsbweb.nih.gov/ij/download.htmlLog3D plugin: http://bigwww.epfl.ch/sage/soft/LoG3D/CLAHE plugin: http://rsbweb.nih.gov/ij/plugins/clahe/index.htmlSimilar image processing package: http://fiji.sc/Fiji

24ImageJ software: http://rsbweb.nih.gov/ij/download.html

25Log3D plugin: http://bigwww.epfl.ch/sage/soft/LoG3D/

26CLAHE plugin: http://rsbweb.nih.gov/ij/plugins/clahe/index.html

27Similar image processing package: http://fiji.sc/Fiji