Data From: Matrix tropism influences endometriotic cell attachment patterns

Due to the extended period for clinical diagnosis, the etiology of endometriotic lesion initiation is not well understood or characterized. Endometriotic lesions are most often found on pelvic tissues and organs, especially the ovaries. To investigate the role of tissue tropism on ovarian endometrioma initiation, we adapted a well-characterized polyacrylamide microarray system to investigate the role of tissue-specific extracellular matrix and adhesion motifs on endometriotic cell attachment, morphology, and size. We report the influence of cell origin (endometriotic vs. non-endometriotic), substrate stiffness mimicking aging and fibrosis, and the role of multicellular (epithelial-stromal) cohorts on cell attachment patterns. We identify multiple ovarian-specific attachment motifs that significantly increase endometriotic (vs. non-endometriotic) cell cohort attachment that could be implicated in early disease etiology. 

Methods
(As per described methods in the manuscript)  
Details on cell culture, Polyacrylamide (PA) microarray fabrication, and PA microarray experimentation were provided in the methods section of the associated paper. Slides were immunostained for actin and cell nuclei, cells prior to seeding were treated with Cell Tracker. Cellular microarrays were imaged at 10x and 20x magnification on the Zeiss AxioScan.Z1 Slide Scanner. Images of entire arrays were converted to 8-bit TIFF files per output channel using Fiji (ImageJ) and cropped into single island images utilizing dextran-rhodamine marker islands to identify the location of arrayed conditions within each image. CellProfiler (version 4.2.6) was used to identify and measure nuclei and fluorescent regions using the IdentifyPrimaryObjects, IdentifySecondaryObjects, and MeasureObjectSizeAndShape modules. Single-cell fluorescent intensity was quantified using the MeasureObjectIntensity module. The output from CellProfiler was loaded into RStudio for analysis. Cell location for patterning analysis was determined using coordinates output by CellProfiler. Cell location was assigned based on distance from the centroid. For co-culture analysis, the fluorescent intensity of the CellTracker dyes were rescaled to stretch each image to full intensity range using RescaleIntensity module and then binned for the expression of the pink and green channels by mean intensity using ClassifyObjects. Additionally, the comparison of mean intensity between the pink and green channels were classified as low-low, low-high, high-low, and high-high for mean intensity using the ClassifyObjects module. These classifications were used to create a decision tree that reliably assigned 95.41% of the total cells as either Stromal or Epithelial, with the remaining cells being categorized as Unassigned due to not fitting the criteria of the decision tree. For analysis, the number of cells attached to each island were counted and individual cell measurements were averaged together. Each representative island for a condition on a slide were averaged together to provide a per slide measurement.  Each island serves as a technical replicate of a condition on a slide. The multiple slides per experimental run serve as technical replicates of the culture conditions.