Computational protocol: Iterative sorting of apical and basolateral cargo in Madin–Darby canine kidney cells

Similar protocols

Protocol publication

[…] Fluorescence peaks detected by the CCD were analyzed based on their intensity and the shape of their pixel distributions. Intensity mapping transformation using the Redist module for the ImageMagick Suite was applied to redistribute the intensity histogram of each image to fit a Gaussian curve (in gray scale) to lower noise from low-intensity pixel values that affect correlative measurements. The Gaussian distribution was based on the cumulative histogram of each image’s intensities, and the same Gaussian fitting parameters were applied to each image. The ImageMagick Suite is available free of charge under a GPLv3 license from ImageMagick Studios, LLC (www.imagemagick.org/). The Redist module is a copyright of Fred Weinhaus and is available free of charge for noncommercial use and is also subject to the ImageMagick Studios, LLC license (www.fmwconcepts.com/imagemagick/redist/). Fluorescence peaks from the redistributed images were then processed with TGNPeakMaster module. TGNPeakMaster fit a Gaussian along x- and y-axes over each localized fluorescence cluster of minimum size of 2 pixels. The circularity of each tentative peak was assessed as the difference in x and y Gaussian widths at half-maxima, and the Gaussian fit was further assessed by fitting to the minimum and maximum threshold values (Supplemental Figure S4). Thus peaks that were either noncircular or resulted in poor Gaussian fit due to irregular, low, or saturating intensities were discarded from the analysis (Supplemental Figure S5). Intensity values for all peaks that passed the Gaussian fitting step were recorded, and peaks with overlap of >15% were reported as colocalized. A “raw” Pearson correlation coefficient was obtained by plotting the intensities of the colocalized, calnexin-negative peaks, and a “raw” relative colocalized and single fraction abundance was quantified for each cargo. Differences in centroid coordinates between peaks in different channels were assessed to make sure that pixel shift between channels was not contributing to colocalization readout. [...] Correlative data obtained from membrane fractions in the TGNPeakMaster analysis were corrected for the differences in expression observed in cells sampled just before fractionation. Coverslips placed along the edge of the plate of cells used in the fractionation experiments were incubated on the plate from the time of plating and removed and fixed before homogenization. The coverslips were processed for immunofluorescence of cargo and costained with the Golgi marker giantin (except for mRFP-ST–expressing cells, which had a stably expressed Golgi marker). Using ImageJ 1.49 (National Institutes of Health, Bethesda, MD), confocal stacks of 177 μm × 177 μm fields of view of the cells were summed, and a mask of the Golgi was made by thresholding the giantin or mRFP-ST channel. The mask was applied over the summed stacks of the two cargo channels after they were background corrected with a linear median value subtraction, and the resulting images were processed by JACOP plug-in () to obtain Pearson and Manders coefficients for the two cargo proteins. An average of several Pearson measurements was used to backcorrect the Pearson correlation measured for the membrane fractions processed by TGNPeakMaster, and an average of several Manders coefficients was used to backcorrect the relative number of colocalized versus single peaks. […]

Pipeline specifications

Software tools ImageMagick, ImageJ, JACoP
Applications Miscellaneous, Microscopic phenotype analysis
Organisms Canis lupus familiaris