Supplementary information text
Recommendations for optimizing workflow, handling of multinucleated cells, processing of tissue samples, and analysis of cells in suspension are discussed here. An explanation of CellProfiler codes and associated functions can also be found here.
CellProfiler Code nuclear quantification
CellProfiler code for the quantification of cells in mono-culture using a nuclear stain alone.
CellProfiler Code co-culture quantification
CellProfiler code for the quantification and characterization of two cell populations in co-culture using a nuclear, surface, and cytoplasmic stain.
ImageJ macro for automation
The following macro 1) removes the ring of fluorescence around the well edge, 2) enhances contrast of weakly stained cells, 3) crops whole-well images into sub-images, and 4) saves modified files.
Segmentation performance at well plate confluency
The ability to segment JC CRL 2116 cells at confluency in a 12-well plate was assessed using standard DAPI staining. A zoomed in image of the well and associated segmentation mask generated using CellProfiler reveal that nuclei are sufficiently spaced apart, even at confluency, to count individual cells.
Primary segmentation and quantification of surface and cytoplasmic stains
The same linear dilution series of J774.A1 cells that was used to assess the accuracy of secondary counts generated using nuclei as seeds in Figure 5 was instead primarily segmented. Cells were plated starting from 10,000 cells/well down to 1,000 cells/well on a 48-well plate and were stained with Vybrant CFDA SE (cytoplasmic stain), phycoerythrin (PE)-conjugated anti-CD11b antibodies (surface stain), and DAPI (nuclear stain). The system performs less than ideally when the fluorescent outline of the cell is used to identify cells instead of their nuclei and would likely have performed significantly worse had the J774.A1 cells not been relatively round. Error bars represent the standard deviation between triplicate conditions.
CellProfiler workflow for co-culture studies
(a) J774.A1 macrophages labeled with PE-conjugated anti-CD11b antibodies (red surface stain) were co-cultured with JC CRL 2116 tumor cells labeled with Vybrant (green cytoplasmic stain). Both cells were also stained with DAPI. (b) The representative image shown in (a) was then run through CellProfiler for processing. For demonstrative purposes, only the Vybrant stained JC CRL 2116 cells are shown in the sample workflow. First, illumination correction is performed on the i) original grayscale image to ii) correct for non-uniformities in illumination. iii) Cell classification of the cytoplasmic stain is then used to identify areas of fluorescence that correspond to the cell body. iv) Primary object identification is then used to fill in any holes generated during cell classification. The subsequently generated image serves as an inclusive mask that is applied to the v) original DAPI image in order to produce a new image vi) that contains only nuclei belonging to Vybrant stained cells. vii) Primary object identification is used once again to identify and quantify the remaining nuclei which then act as seeds for secondary object identification and cell body delineation.
Increasing the number of parameters/cells that can be assessed in a single experimental setup using a barcode approach
The same linear dilution series of J774.A1 cells that was used to assess the accuracy of secondary counts generated using nuclei as seeds in Figure 5 was instead used to assess the performance of a barcode approach to multiplex cell quantification. Cells were plated starting from 10,000 cells/well down to 1,000 cells/well on a 48-well plate and were stained with Vybrant CFDA SE (cytoplasmic stain), phycoerythrin (PE)-conjugated anti-CD11b antibodies (surface stain), and DAPI (nuclear stain). Plots of nuclei alone, nuclei delineated by an antibody surface mask, nuclei delineated by a cytoplasm mask, and nuclei delineated by both an antibody surface as well as cytoplasm mask were generated. There is only a marginal loss in performance when counting nuclei demarcated by two masks with accuracy primarily limited by the least accurate stain. Error bars represent the standard deviation between triplicate conditions.