Fig 2: Gene expression of markers identified for stemness. (A) Feature plots of Epcam, Itga6, and Ly6a, which represents for the proteins Epcam, Cd49f, and Sca-1 (B) Feature plot expression of the key gene markers after filtering for cells expressing a minimal threshold of 0.3 for all markers. (C) Pie plot visualizing population broken down by cluster. Between the entire Comma-1D population and P8, there is increased occupancy of luminal 1 and myoepithelial 4 with a significant decrease of myoepithelial two and loss of the fibroblast cell population.

Fig 3: Changes in MaSCs with Loss of AP-2?(A) Cell cycle shown as proportion of cells in G1, S, and G2/M by mammary cell cluster for CO- and 4OHT-treated mice.(B) Expression pattern for the Itga6, Itgb1, Cd24a, and Lgr5 genes from CO- and 4OHT-treated mice. Middle panel: magnified view of boxed area focused on basal clusters 4 and 10; right panel: heatmap showing relative expression of Itga6, Itgb1, Cd24a, and Lgr5 in basal MMEC clusters in CO- versus 4OHT-treated mice.

Fig 4: Phenotypic assay overview. (A) A subsection of a beacon chip across five time points of a culture period. The bar on the bottom indicates heterogeneity in observed growth response. Red indicates dying cells, white is no significant observed response, and green is for cells with detected proliferation (B) The top graph is plotting the growth ratio across identified subpopulations. The bottom graph indicates growth differences depending on initial cell concentration. 5% error bars are visualized for each plot. (C) Sample pen images of various cells across imaging channels. The colored outline is to highlight expression (green) or lack of (red). The top row is a known P8 population imaged at three channels with expression detected for all three markers. UA and UB indicate sample images of unknown cell types with marker expression pattern resolving subtypes P8 and P5, respectively (D) Overview of MATLAB script pipeline to identify cell populations by protein expression, automating observations described in (C). (Di) Illustrates regions of interest (ROIs) representing each nanopen. (Dii) Shows possible cell detection locations based on circle detection. (Diii) Visualizes filtered cell detections leveraging OEP fluorescence thresholds. Fluorescence intensity above 65,535 is set as the threshold for true positive cell detection. (Div) Visualizes normalized expression scores for two cell detection locations in the TRED channel. Normalized scoring accounts for variability in background fluorescence between FOVs and between experimental chips. Readings are imported into a matrix for each detected cell for each marker and channel. (Dv)A sample readout of cells subtype based on the measured expression. Numbers “5” and “7” indicate populations P5 and P7, respectively. (Dvi)A scatter plot of all the detected cells plotted based on their expression of Cd49f and Epcam and the subtype they are predicted to belong to. 88.3% of FACS sorted cells pseudo-bin to same population on-chip. Each color represents the pseudo-binned subtype for that cell, as indicated in the legend. Nanopens with multiple cells at t0 are sources of noise and background fluorescence that misidentify some cells. (E) ROI images of randomly selected P8 identified cells from each imaging channel. The heatmap on the bottom compares expression across channels and cells. This functionality provides verification to the pseudo binning results, allows for clustering and threshold adjustment, and provides easy visualization for every cell subpopulation of interest.