Fig 2: Vismodegib induces initial tumor response followed by rapid recurrence.a, b Medulloblastoma in sagittal hindbrain sections stained for pRB, from representative a P15 M-Smo mice, treated with three daily doses of vehicle or vismodegib, or b P25 M-Smo mice treated for 2 weeks with vismodegib, and quantification of pRB+ fraction in the indicated groups. c Kaplan–Meier curve comparing survival of M-Smo mice treated with vismodegib or vehicle. d Schematic showing the timing of luciferase imaging and vismodegib administration. e Luciferase signal driven by Gli-luc at indicated times. e' Luminescence fold change over the indicated intervals in replicate vismodegib-treated and control mice. Each dot represents the value for a specific replicate animal. Horizontal lines indicate the means, and error bars indicate SEM. p Values determined by two-sided Student’s t-test. Scale bars = 2 mm, except in insets, where scale bars = 50 µm.

Fig 3: Integration of Dye Drop assays with CyCIF and time series.a Schematic and representative image from technical triplicates of multiplexing Deep Dye Drop assays with cyclic immunofluorescence: the Hoechst (gray-scale), EdU (red), pH3 (green), beta-catenin (cyan), phospho-pRb (blue), and p21 (yellow) signals are displayed, and contrast was adjusted for visualization purposes only. Scale bar is 100 µm. b UMAP representation of MCF7 and MCF 10 A cells treated with BMS-265246 (1 µM, 10 µM), ribociclib (10 µM) or DMSO stained with Deep Dye Drop and cyclic immunofluorescence. c The number of MCF7 cells in S-phase following treatment with increasing concentrations of ribociclib, d palbociclib, and e abemaciclib after 6, 24, 48, and 72 h. Data are presented as mean +/- standard deviation of four technical replicates. EdU versus DNA content scatter plots show the single cell cell-cycle distributions from one well at 1 µM doses of each drug at the time points indicated, the percentage of cells in S-phase (average from four technical replicates) is indicated on each plot. The scatter plots show all cells in a single, representative well of a 384-well plate for each condition. f Illustration of possible patterns of the emergence of resistant cells. A, clonal, genetic resistance or B, non-clonal, non-genetic adaptation followed by representative images from biological duplicates performed in technical quadruplicate of MCF7 cells treated with 1 µM palbociclib for 24 h or 72 h. Nuclei are visualized with Hoechst in white and EdU positive cells are shown in red, scale bars are 100 µm. Source data are provided as a Source Data file and on Synapse.

Fig 4: HES1 and MYOD1 mark vismodegib-responsive and resistant subsets of cells.a t-SNE projections as in Fig. 6a, with Gli1+ cells indicated. b MALDESI images of cholesterol and vismodegib distribution in sagittal brain sections including both tumor and normal brain. c t-SNE projections as in a with Hes1+ cells indicated. d IHC for HES1 and pRB in sagittal hindbrain sections from representative M-Smo mice treated as indicated. e t-SNE projections as in a, with Myod1+ cells indicated. f IHC for MYOD1 and pRB in sagittal hindbrain sections from representative M-Smo mice treated as indicated. g Quantification of data from c–f. Dots represent individual replicate mice. Horizontal lines represent means and error bars indicate SEM. h Analysis of MYOD1, pRB and cell cycle phase by flow cytometry. Graphs formatted as in g. i, j Expression microarray data on HES1 and MYOD1 expression in human medulloblastoma samples, as presented in ref. 44, with boxes indicating the mean, 25th and 75th percentile and whiskers indicating the 10th and 90th percentiles. Scale bars = 2 mm, except in insets where scale bars = 100 µm. p Values determined by two-sided Student’s t-test.