Fig 2: Identifying canonical EMT states in lung cancer cells through TGFβ time-course analysis.a Immunoblot of EMT markers in the presence or absence of TGFβ in three NSCLC cell lines (2 week treatment in HCC827 and H3255 cells, 1 week treatment in A549 cells). b Representative confocal images of HCC827 cells stained for E-Cadherin (red) and Vimentin (green) in the absence (a, b) or presence of TGFβ (c, d) for 10 days. Magnification ×40, scale bar 20 μm. White dotted arrows indicate cells that show loss of E-Cadherin and gain of Vimentin expression in untreated conditions. White solid arrows show enlarged cells on the periphery of epithelial islands that have acquired a pEMT phenotype (E-Cadherin+/Vimentin+); blue and red solid arrows indicate either enlarged and elongated or small and spindle-shaped cells that have acquired mesenchymal phenotypes (E-Cadherin−/Vimentin+). c Representative images of HCC827 cells treated with TGFβ (5 ng/mL) for 2, 6, and 10 days under continuous re-seeding conditions (see also Supplementary Figs. 1 and 2 and Methods). Magnification ×10, scale bar 200 μm. Below each image are shown respective E-Cadherin/Vimentin and CD44/CD24 flow cytometry plots. Arrows indicate changes in marker expression during EMT and numbers 1–4 designate the identified canonical EMT states, respectively: Epithelial (E), pEMT, pEMT/Mesenchymal (M), and M* (characterized by CD44hi/CD24lo, CSC-like cells). d Color-coded gated EMT canonical states depicted on E-Cadherin/Vimentin flow cytometry plots per experimental condition (see also Supplementary Fig. 1). e EMT state dynamics during TGFβ time course. Color-coded EMT states were calculated and depicted as % of total number of cells at each experimental time-point. f Heatmap summary of EMT marker fold (arsinh) changes relative to control condition when analyzed in all cells (left) and relative to the E state when analyzed in the four canonical EMT states individually at the 10-day TGFβ time-point (right). Source Data are provided as a Source Data file.

Fig 3: Mass cytometry time-series analysis offers deep single-cell resolution of canonical EMT states.a Workflow schematic of cell culture conditions for TGFß time-course (addition and withdrawal (W) time-points) and mass cytometry analysis. Arrows indicate times at which cells were collected for both analysis and re-seeding for the remaining time-points (see also Supplementary Table 1, Figs. 2 and 3, and Methods). b Representative images of HCC827 cells at each time-point (magnification ×10, scale bar 200 µm) with respective E-Cadherin/Vimentin and CD44/CD24 mass cytometry plots shown below. Arrows indicate the previously observed canonical EMT marker changes and states, respectively. c Histogram overlays illustrating EMT marker expression distributions in HCC827 cells undergoing EMT and MET. On the right, a heatmap summary of the canonical EMT marker fold (arsinh) changes relative to state E when analyzed in the four EMT states individually at the 10-day TGFß time-point. d Additional marker distributions in HCC827 cells undergoing EMT and MET (see Supplementary Fig. 3 for remaining markers and an independent biological replicate experiment).

Fig 4: Unsupervised analysis of mass cytometry data reveals computationally derived EMT and MET states.a Heatmaps depicting the expression of the six clustering markers in each cell (E-Cadherin, Vimentin, CD44, CD24, Twist, and MUC1) per computationally derived EMT and MET state (normalized global expression). Shown are the most prominent in number states that resulted from the CCAST algorithm performed on pooled mass cytometry HCC827 data. Dotted boxed regions illustrate subpopulations of Twist+ pEMT cells (see also Supplementary Fig. 6). b EMT and MET state dynamics. Each graph corresponds to the state heatmap above. States were calculated and depicted as % of total number of cells at each experimental time-point. c Graphs illustrating mean expression of the six clustering markers in EMT and MET states. Bars represent standard deviation (s.d.) of each marker within each downsampled state (E1 cells n = 2394, E2 cells n = 3284, E3 cells n = 1316, pEMT1 cells n = 3359, pEMT2 cells n = 2943, pEMT3 cells n = 2663, M cells n = 9609, MET cells n = 3498). d Time-resolved force-directed (FDL) layout of mass cytometry time-course data (left) and respective EMT and MET state annotations (right). Dotted area depicts subpopulation of M cells that exhibit CSC-like phenotypic characteristics (CD44hi/CD24lo, M*). e Time-resolved FDL layouts colored by protein expression of indicated markers (arsinh transformed data). See Supplementary Fig. 6 for remaining markers and analysis of an independent biological replicate experiment. Source Data are provided as a Source Data file.