Geautomatiseerde kwantificering van hematopoietische cellen - stromale cel interacties in histologische afbeeldingen van ontkalkte Bone

The overall goal of this procedure is to quantify the associations between specific cell subsets in the bone marrow. This is accomplished by first collecting histological cryo sections from mouse femurs tumors. In the second step, the sections are stained for immunofluorescent analysis and confocal images are acquired.

The images are then analyzed digitally. Ultimately, the positions of the different cell types can be simulated to calculate the rates and frequency of contact between the specific immunological cells of interest and their microenvironment. We first had the idea for this method when we were analyzing microenvironmental niches for long lived plasma cells in the bone marrow, and we were a need for a method to unambiguously quantify the cellular associations within the bone marrow tissue.

RA Maya will demonstrate the procedure for you To cryo suction femoral bones from an adult mouse. First, set up the sample and blade temperature of a standard microtome equipped with a hard tissue blade to negative 24 degrees Celsius. After a 15 minute acclimation.

Fix a sample block to the metal sample holder with cryo embedding medium and adjust the orientation of the block as necessary. Trim the sample until the bone is fully opened and the marrow is visible. Then adjust a section thickness to seven microns.

Next, fix a piece of kawamoto adhesive tape with the sticky side on the top of a sample block using a dear leather covered wooden spatula Section the sample and using forceps. Turn the tape so that the section is positioned on the top side. Then transfer the tape to a glass microscope slide.

Fix the tape to the glass slide with scotch tape when the last section has been collected. Let the samples drive for at least 30 minutes and store the slides at negative 80 degrees celsius. Two women image the samples stain the thawed cryo sections according to common immunofluorescence protocols.

Making sure to include a nuclear stain for visualizing the tissue integrity. After the sections have been stained, add one drop of floor mount to each sample, followed by a number one glass cover slip on a laser scanning confocal microscope. Select the 20 x objective lens and set a field view of 708 point 15 by 708 point 15 microns.

Then image the samples one at a time at 2048 by 2048 pixels per image to keep the results comparable. Recording the images in one channel for visualizing the stromal structures, one channel for the nuclei and additional channels for the hematopoietic cells of interest as necessary. Record the images using line averaging.

Ideally capturing single adjacent images to cover the whole femoral section. Then save the images in a microscopy image file format. Then export one JPEG file per channel for each area imaged, and use an image analysis tool to perform the image segmentation quantification of colocalization and neighborhood analysis before executing the simulations of the random cell.

Positioning on the analyzed bone marrow images for each image to be simulated. Place the following files in a single folder, the single CSV files provided by the cell contact tool. The original DPI channel JPEG files and the original stromal channel JPEG files to perform batch simulations on the series of images from a single femoral sample to transfer all of the original JPEG files and their corresponding single CSV files into another folder at this time as well.

When all of the files are assembled, start the simulation tool. Then check the auto load image data box. With this option checked the CSV files corresponding to the analyzed images are opened automatically.

Next, enter the common tag for the CSV files and the number of image sets that should be used for batch mode simulation. Then load the J peg file generated from the S stroma channel with a file name ending in three to generate the mask for the DPI channel. Check the apply mask box and set the threshold for converting the DPI image into a binary mask to 10.

Check the eight bit and dilate boxes and set the grade of dilation to five pixels. Then check the width erosion box. Then enter the desired value for the vicinity radius used for the analysis of the simulated images, and check the use OSU box to use OSU algorithm for the automatic detection of the stromal structures for the hematopoietic cells.

Simulated as circular shapes. Measure the cell size distribution of the analyzed cell types with any image analysis software that includes the appropriate object segmentation functions, and determine the sigma for all of the hematopoietic types from these measurements. The cell size cutoff that is the diameter in pixels that describes the smallest object still recognized as the complete cell by the image analysis tool can be determined.

Then enter the cell size cutoff and the sigma in pixels for the simulation of the cell size distribution for the red, green, and blue cells. Next in the cells and mask tab, check the delete box to allow the program to choose a new position for a cell if it overlaps with at least one pixel within a no-go area of the mask. Under the cell exclusion tab, check the avoid cell overlap box as well.

Then enter the allowed minimum distance between the centers of two cells in pixels. If the overlap is defined by the minimum distance from the center of one cell to the center of another, set the maximum area of overlap to 100%Then set the simulation tool to 1000 repetitions. Then check the auto save sales box to save the coordinates of the simulated objects for all of the repetitions of each image of the batch as wrecked files in text format.

Enter a common tag for the saved files as well. Finally, click on run simulation and determine the average contact and vicinity frequencies of the sets of 1000 simulated images corresponding to each recorded image. For this divide, the average contact and vicinity counts by the recorded cell count per image, and compare the frequencies to the results of the automated colocalization analysis of the recorded image.

In this figure, a localization analysis of eosinophils, plasma cells and B cells with the stromal cell microenvironment with fluorescent chimeric bone marrow is presented. The CSV files generated by the cell contact and vicinity tools contain the cell count and total area for each cell population in pixels as well as the contact or vicinity counts. Before performing a simulation of the random cell positioning, the parameters that describe the cell size distribution as a Gaussian distribution have to be determined in this simulation, the relative values of the sigma measured for the three cell populations were maintained.

While the Sigma values defined in pixels were set to match the visual appearance of the recorded cells to define the areas where the cells can be placed, a mask was generated from the DAPI channel. A value of 10 was then determined to be the optimal density threshold for generating the binary images. The relevance of the contacts of the plasma cells, eosinophils, or B cells with the stromal cells was then tested as expected.

These analysis revealed significantly higher contact rates in the recorded images compared to the simulation for the plasma cells and B cells. In contrast, no clear difference observed for eosinophil stromal cell contact. This method can help to answer key questions in any field requiring the analysis of the complex tissue structure of the bone marrow, such as immunology, hematology, pathology, or stem cell biology.